Ucalgary 2019 Ramezanpour Mohsen PDF
Ucalgary 2019 Ramezanpour Mohsen PDF
Ucalgary 2019 Ramezanpour Mohsen PDF
2019-03-12
Ramezanpour, Mohsen
University of Calgary graduate students retain copyright ownership and moral rights for their
thesis. You may use this material in any way that is permitted by the Copyright Act or through
licensing that has been assigned to the document. For uses that are not allowable under
copyright legislation or licensing, you are required to seek permission.
Downloaded from PRISM: https://prism.ucalgary.ca
UNIVERSITY OF CALGARY
by
Mohsen Ramezanpour
A THESIS
CALGARY, ALBERTA
MARCH, 2019
Biomolecules, including lipids and enzymes, are of special interest in biotechnology and nano-medicine
applications. A knowledge of structure, dynamics, and function of these biomolecular systems is required
for development and improvement of drugs and drug delivery systems. Computer simulation, as a
details on the molecular systems. In this thesis, molecular dynamics (MD) simulation technique was used
composed of zwitterionic and ionizable cationic lipids, 2) inverted hexagonal (HII) phases composed of
First, lipid bilayers composed of POPC and DLin-KC2-DMA (also known as KC2) ionizable cationic
lipids were studied as a function of pH, temperature, and mixing ratio. Simulations suggest that neutral
KC2 are segregated in the presence of POPC. This segregation was proposed to affect both the internal
Next, DOPE and POPE HII phases were constructed, simulated, and their structural properties as a
function of hydration level and temperature were studied using MD simulations. HII systems are usually
challenging to study experimentally, especially at low hydration regimes. Our findings suggest that MD
simulation could successfully reproduce structural properties of HII phase in a good agreement with
experimental data. Furthermore, a computational protocol for construction of HII molecular systems
Finally, the auto-inhibition and activation mechanism of CCT enzyme upon attachment or detachment of
two autoinhibitory helices was studied. Attachment of autoinhibitory helices were shown to affect the
dynamics and consequently the catalytic function of the CCT enzyme. Accordingly, a novel two-part
ii
Preface
Chapter 1 provides the reader with a general introduction required for understanding the results
presented in the following chapters. The questions of interest in this study and their importance are also
briefly discussed.
In Chapter 2, molecular dynamics (MD) simulation technique and its underlying algorithms are
discussed. Next, the concept of biomolecular force field, the protocol used to develop force field
parameters for KC2 and its protonated state (KC2H), as well as the protocols used to construct the
Chapter 3 is reproduced with permission from a review article published in the Biochimica et Biophysica
In Chapter 4, lipid bilayers composed of POPC and KC2 were studied using MD simulations. KC2 is an
ionizable cationic lipid of interest in lipid nanoparticle formulations developed for drug and gene delivery
applications.
In Chapter 5, the structural properties of DOPE and POPE HII phases were studied as a function of
hydration level and temperature. HII systems are of interest in nano-medicine field of study.
Chapter 6 is reproduced with permission from a research article published in the Journal of Biological
Finally, in Chapter 7 the general conclusions are made, and directions for future studies are discussed.
iii
Chapters 3 and 6 are based on the following publications reproduced with permission from the
publishing journals.
Chapter 3:
Ramezanpour, M., Leung, S. S. W., Delgado-Magnero, K. H., Bashe, B. Y. M., Thewalt, J., &
Chapter 6:
Ramezanpour, M.*, Lee, J.*, Taneva, S. G., Tieleman, D. P., & Cornell, R. B. (2018). An auto-
inhibitory helix in CTP:phosphocholine cytidylyltransferase hi-jacks the catalytic residue and constrains a
pliable, domain-bridging helix pair. Journal of Biological Chemistry, 293(18), 7070 –7084. (*
contributed to work equally). Copyright © the American Society for Biochemistry and Molecular
Biology.
contributions on these chapters, as well as the names of co-authors/collaborators are briefly described at
the beginning of each corresponding chapter. Briefly, I designed, ran, analyzed, and interpreted all the
simulations, and developed the computational protocols and models in this study, although
iv
Acknowledgements
Special thanks to my supervisor, Dr. Peter Tieleman for his support and mentoring. Peter, it was an
honour to be a Ph.D. student in your research lab. I have learned a great deal from you in all scientific,
management and mentoring aspects. Thanks for providing me with all the resources, collaborations, and
travels needed for conducting my research. None of these achievements would have been possible
Thanks to my supervisory committee, Dr. Justin MacCallum and Dr. Vanina Zaremberg, for their
I would like to thank the external examiners, Dr. Peter Kusalik and Dr. Michel Lafleur, for agreeing to be
The last four years was an honour for me to work on several projects in collaboration with Dr. Pieter
Cullis’s research group at the University of British Columbia, Dr. Jenifer Thewalt’s and Dr. Rosemary
Cornell’s research groups at the Simon Fraser University, and Dr. Paul Harper’s research group at Calvin
College. I enjoyed every minute of our collaborations and learned a lot through these collaborations.
Thank-you all.
Thank-you Dr. Elmar Prenner, Dr. Gregory Moorhead, and Dr. Hamid Habibi for your career advices and
support.
v
Thanks to all the former and current lab members, as well as other members of the Centre for Molecular
Simulation (CMS) for creating an active, scientific and friendly environment suitable for learning.
Thanks to the support staff at the department of biological sciences for providing a friendly environment
for graduate students and assisting them in any possible way they could. I would also like to thank the
Faculty of Graduate Studies (FGS) at the University of Calgary for providing all the resources a graduate
I would like to thank both the funding agencies, including NSERC, and Compute Canada for their
Last but not least, a great thanks to my parents who dedicated their whole lives to help me succeed.
Without any doubt, I would not be able to achieve these without your support. I would also like to thank
vi
Dedication
vii
Table of Contents
Abstract........................................................................................................................................................ ii
Preface......................................................................................................................................................... iii
Acknowledgements ..................................................................................................................................... v
viii
1.2.4.2 DOPE and POPE. ........................................................................................................... 15
2.5 Construction of Binary and Ternary Mixtures in Lamellar and HII Phase ...................................... 51
ix
Chapter Three: Computational and Experimental Approaches for Investigating Nanoparticle-Based
Chapter Four: Ionizable Amino Lipid Interactions with POPC: Implications for Lipid Nanoparticle
x
4.3.2.2 Molecular dynamics simulations. ................................................................................. 162
Chapter Five: Structural Properties of Inverted Hexagonal Phase: A Hybrid Computational and
5.3.2.3 Electron density reconstruction and water core radius estimation. ............................... 203
xi
5.3.2.4 Deuterium NMR (2H NMR). ........................................................................................ 204
5.4.2 Structural differences between DOPE and POPE HII systems. ................................... 209
5.4.6 Temperature effect on DOPE HII lattice distance and maximum hydration. ............... 220
5.5.1 MD predicts the HII structural parameters as a function of hydration. ........................ 223
5.5.2 HII water channels are not straight and cylindrical at low hydrations.......................... 225
5.5.3 MD reproduces the differences between DOPE and POPE in HII. .............................. 226
5.5.4 Both dehydration and raising temperature reduces SCD in HII. .................................... 226
5.5.5 Temperature affects the maximum hydration and lattice distances in HII. .................. 227
5.5.6 Protocol for construction of equivalent HII systems to experiments. ........................... 228
Catalytic Residue and Constrains a Pliable, Domain-Bridging Helix Pair ........................................ 240
6.3.1 Optimal catalysis requires a flexible Loop L2 at Lys-122, enabled by Gly-123. ......... 246
xii
6.3.2 The AI helices selectively repress the dynamics of helix αE........................................ 249
6.3.3 The AI helix supplies H-bonding partners for Lys-122 that compete with CTP. ......... 253
6.3.4 Removal of the constraining AI helices allows bending of the E helices. ................. 256
6.3.5 Membrane binding reorganizes the E helices, a pre-requisite for activation. ............ 260
6.4.2 Novel backbone competition for the catalytic lysine. ................................................... 266
xiii
7.2 Future Directions ............................................................................................................................ 289
Appendix A: Parameters for the Parameterized Segment in the KC2 and KC2H ................................... 376
B.1 KC2 and KC2H Parameterization in the CHARMM36 Force Field .............................................. 390
B.2 KC2(H) Effects on POPC as a Function of Temperature and Mixing Ratio .................................. 393
B.4 KC2H has Little to no Effect on the POPC Bilayer Thickness ...................................................... 398
B.5 KC2H Repels Na Ions from and Attracts Cl Ions to the Lipid-Water Interface ............................. 400
xiv
List of Tables
Table 3-1: Questions relevant to drug delivery system design that can be answered by computational
simulations. ........................................................................................................................................ 67
Table 3-2: Experimental methods used to study drug delivery systems. .................................................... 69
Table 5-2: Maximum hydrations for DOPE and POPE HII systems at several temperatures. ................. 220
Table A-1: Parameters for the ring segment in KC2 and KC2H. ............................................................. 378
Table A-2: Parameters for the amino segments in KC2 and KC2H. ........................................................ 378
Table A-3: Parameters for the hydrogen atoms in KC2 and KC2H. ........................................................ 379
Table A-4: Parameters for several bonds defined in KC2 and KC2H. ..................................................... 379
Table A-5: Parameters for several angles defined in KC2 and KC2H...................................................... 380
Table A-6: Parameters for several dihedrals defined in KC2 and KC2H. ................................................ 381
xv
List of Figures
Figure 1.3: The two-dimensional structure of KC2 and its protonated state, KC2H. ................................. 12
Figure 1.5: Two-dimensional structure of the lipids studied in this work. ................................................. 14
Figure 2.2: KC2(H) breakdown into two segments for parameterization. .................................................. 45
Figure 3.3: Categories of simulation methods and their respective spatio-temporal domains of
applicability. ....................................................................................................................................... 79
Figure 3.4: PAMAM dendrimer-DNA interaction and deformation as a function of generation number. . 85
Figure 3.5: Proposed mechanism of endosomal escape for pH-responsive dendrimers. ............................ 88
Figure 3.6: Time evolution of nanoparticle-polymer complex endocytosis as a function of pH. ............... 93
xvi
Figure 3.10: Effect of PEGylation method, and PEG chain size and grafting density on the conformation
Figure 4.1: KC2 segregation from POPC as a function of pH. ................................................................. 171
Figure 4.2: Systematic increase in POPC inter-leaflet distance upon rising the KC2 concentration at basic
Figure 5.2: Definitions for several structural parameters calculated from MD simulations. .................... 198
Figure 5.3: Structural parameters for DOPE HII system as a function of hydration. ............................... 207
Figure 5.4: Radius of water core from reconstructed electron density profiles and MD simulation. ....... 209
Figure 5.5: Structural parameters of DOPE and POPE HII systems as a function of hydration. ............. 210
Figure 5.6: Cylinder length in DOPE and POPE HII systems as a function of hydration. ....................... 212
Figure 5.7: Effect of hydration level on lipid tubules shape in DOPE HII systems. ................................ 214
Figure 5.8: Deuterium order parameter (SCD) for DOPE and POPE in HII phase. ................................... 217
Figure 5.9: Estimation of maximum hydration in HII systems at several temperatures. .......................... 219
Figure 5.10: Lattice distance as a function of temperature for DOPE HII systems. ................................. 222
Figure 5.11: A proposed protocol for molecule parameterization in HII phase. ...................................... 230
Figure 6.1: Structure of mammalian CCT and its active site. ................................................................... 245
Figure 6.2: CCT relies on a single lysine followed by glycine in loop L2 for catalysis. .......................... 248
Figure 6.3: Effect of the AI helix and CTP on the dynamics of active-site loops and side chains. .......... 251
Figure 6.5: αE conformational change triggered by removal of the AI helices. ....................................... 259
xvii
Figure 6.6: Lipid vesicles reduce interchain cross-linking efficiency between the two αEc helices. ....... 261
Figure 6.7: A disulfide bond linking the αE helices at Cys-217 inhibits CCT activity. ........................... 263
Figure A.1: Atom index numbers for the parameterized segment of KC2(H). ......................................... 377
Figure B.1: Validation of developed models for KC2 and KC2H. ........................................................... 392
Figure B.2: SCD parameters for the palmitoyl chain (sn-1) of POPC at 288 K and 293 K. ...................... 394
Figure B.3: SCD parameters for the palmitoyl chain (sn-1) of POPC at 303 K and 313 K. ...................... 396
Figure B.4: SAXS curves for POPC/KC2 multilamellar vesicles at several basic pH values. ................. 398
Figure B.5: Effect of adding KC2H on POPC inter-leaflet distance. ....................................................... 399
Figure B.7: Effect of KC2 and KC2H on POPC P→N vector angle distribution. .................................... 403
Figure C.1: Lattice distance as a function of hydration for DOPE HII systems (semi-isotropic versus
Figure D.1: Dynamics of active site loops and side chains based on the 20 simulations with positional
Figure D.3: Correlations between K122 backbone dihedrals and H-bonding partners............................. 416
Figure D.4: Stability of 4-helix bundle during 1000 ns simulation time. ................................................. 417
Figure D.5: Alpha-E hinge dynamics enables contacts with the active site. ............................................ 418
Figure D.7: Lipid vesicles reduce inter-chain cross-linking between αEc................................................ 420
xviii
Figure D.8: CCT(A217C) binding to lipid vesicles via domain M is not impaired by oxidative disulfide
xix
Chapter One: General Introduction and Thesis Outline
1.1 Abstract
Biological molecules, including lipids, proteins and nucleic acids, are of special interest for their
therapeutic applications. They can be used as targets for drugs, as therapeutics, or as delivery systems for
Rational design of new therapeutics and development of effective drug/gene delivery systems
could benefit from deeper insights on the role, structure, dynamics, function, and regulatory mechanisms
of these biomolecular systems. The current knowledge in this field of research has been gained mainly
through extensive experimental research. Over the last few decades, however, there have been many
studies demonstrating how computer simulations can lead to a deeper understanding of molecular
systems, such as drug delivery systems [5]. Computational techniques can provide insights into the nano-
system in an atomic resolution and a femtosecond time scales, a spatiotemporal resolution which is not
easily achievable otherwise. Among all the computational techniques, molecular dynamics (MD)
simulation is the most commonly used technique for investigating the structure and dynamics of
In this thesis, MD simulations were used to study the structure of lipid nano-aggregates in both
lamellar and inverted/reverse hexagonal (HII) phases, and to investigate the structure and dynamics of a
This chapter provides the reader with the background information on the problems studied and
the results presented in Chapters 3-6. In the last section of this chapter, i.e. thesis outline, both the goals
and their importance will be briefly explained for each chapter. This chapter is mainly focused on the
1
parts which either were not discussed or briefly discussed in the following chapters. For more information
on each topic, the reader is referred to the articles cited in this and corresponding chapters.
1.2 Introduction
Cellular membranes are among the most common targets for many drugs [1, 4]. Most drugs either interact
with cell membranes as their target or require passing across them to reach their site of action, i.e.
Biomembranes are mainly composed of lipids and proteins [7]. There are many different lipid
types found in cellular membranes, and the roles of these lipid polymorphisms still remain to be
(PSs), and sphingomyelins are the common lipids found in the plasma membrane of eukaryotic cells.
These lipids span a wide variety of lengths and unsaturation degrees in their acyl chains. Cell membranes
also include sterols, with cholesterol the major sterol in mammalian cell membranes. Membrane proteins
are the proteins which interact with the lipid membrane. These proteins could be part of the membrane,
e.g. integral proteins, or interact with the membrane and its integral proteins temporarily. Lipid content
and organization, lipid-protein and protein-protein interactions, as well as the proper function of
membrane proteins are known to be critical for a correct cell function and its survival. These make
Drugs, which their biomolecular target is inside the cell, need to reach and be internalized to the
target cells. Therapeutics with a high molecular weight and/or being highly charged, e.g. short interfering
RNA (siRNA), are not stable in the circulation and cannot readily get into the cell. As a solution, drug
delivery systems (DDSs) can be used to protect such drugs in the bloodstream and pass them across the
2
cell membranes [3]. When inside the cell, drugs could target a variety of biomolecules, including
Targeting and degrading mRNAs, and consequently preventing the expression of disease-causing genes is
a promising therapeutic approach called RNA-interference [2]. Short interfering RNAs (siRNAs) are post
transcriptional silencing agents experimentally designed to prevent the expression of problematic genes
by degrading the corresponding mRNAs. siRNAs have a great potential in treatment of a wide range of
diseases, including cancer [10]. This class of therapeutics are highly selective, and capable of affecting
Despite their promising applications, siRNAs have difficulty to reach the cytoplasm of target cells
[11]. Given their large size (~22 nucleotides), high molecular weight (~14 kDa in mass), and negatively
charged structure, naked siRNAs must overcome several barriers on their way to the cytoplasm. They are
rapidly degraded by RNases and cleared from the bloodstream. They activate the immune responses in the
body and cannot be internalized into the cell via passive diffusion.
There are many techniques developed to assist in siRNA delivery, such as chemical modifications
of siRNA strands, conjugating them with receptor-binding ligands, conjugation with peptides, and loading
siRNAs into/on the nanoparticles [12]. Lipid nanoparticles (LNPs) are among the clinically successful
DDSs for both drugs and gene materials, e.g. siRNA (see Section 1.2.3).
Enzymes are one of the most commonly targeted biomolecules, which affecting their catalytic activities
could have therapeutic effects [1]. As an example of interest in this study, inhibiting lipid biosynthesis
3
pathways via inhibiting CCT enzyme, a critical enzyme which catalyzes the rate limiting step in PC
biosynthesis, is known to be correlated with apoptosis [13]. Furthermore, mutations in this enzyme are
known to cause several disorders related to bone growth, vision, and lipid metabolism [14-17]. Finally,
inhibiting the PC biosynthesis pathway by blocking the CCT enzyme in Plasmodium falciparum is
considered a potential target for drug development against malaria [18, 19]. Therefore, to develop
effective drugs and inhibitors, or to reveal the underlying reasons by which these mutations cause genetic
disorders, a deeper knowledge of the CCT activation and autoinhibition is required. For instance, when
the catalytic activity of CCT enzyme is a valuable target, the underlying structural factors playing roles in
its catalytic activity, activation and inhibition mechanisms are important factors to be studied and
determined.
Lipids are amphiphilic molecules, i.e. molecules with both hydrophobic and hydrophilic segments, which
are well known for their presence and vital roles in living cells [8]. Hydrated lipids are capable of self-
assembly into a variety of nano-structures [20, 21]. The phase diagram, i.e. the structure of a molecular
system at different temperatures, concentrations, and mixing ratios, for a variety of lipid-water systems
has been constructed experimentally [22, 23]. These lipid phases are usually classified according to the
lattice type and symmetry, and conformation of lipids hydrocarbon chains in the systems. Lamellar (L),
hexagonal (H), and cubic phases (Q) are among the commonly studied structural phases, and of a high
interest in nanotechnology, including nano-medicine [24-26]. Except the lamellar phase, other phases
could exist in either normal (type I or oil-in-water) or inverted (type II or water-in-oil) types. For instance,
there are two types of hexagonal phases: HI (normal) and HII (inverted) phases. In the normal phase,
lipids hydrocarbon chains form hydrophobic rods which are separated form the outside solvent by the
lipid-water interface. For the inverted hexagonal phase [Figure 1.1-right], water is trapped inside the
4
lipid cylinders while the lipid hydrocarbons fill the space between the hexagonally packed cylinders [22,
23].
The amphiphilic nature of the lipids results in their specific packing in solvent so that the free
energy of the system is minimized [27]. For instance, in a lipid bilayer, the polar headgroups face the
aqueous media, while the hydrophobic hydrocarbon chains pack together in the bilayer interior to
There are a wide variety of factors which could favor a specific structural phase or promote the
transitions from one to another [28, 29]. Lipid composition, lipid type, hydration level, ionic strength of
the media, and temperature are among the main determining factors.
Lipid type is of great importance on its own because it includes other factors such as headgroup
size, charge, hydrocarbon chain length and tail unsaturation degree. All the above-mentioned factors can
affect the “lipid shape” [28, 30], a concept upon which the observed polymorphism preference of the
lipids could be explained. The effective shape of a lipid molecule can be quantified by the “critical
where the 𝑣 is the hydrocarbon volume, 𝑙𝑐 is the critical chain length (i.e. the length of the fully extended
acyl chain), and 𝑎0 is the optimal area per lipid molecule (i.e. the area per headgroup which the
interaction energy between the lipids is at its minimum). The molecules with a packing parameter of S ~ 1
have an effective cylindrical shape and prefer the lamellar phase. The lipids with a S > 1 and S < 1 are
referred to as “cone” and “inverse cone” shaped lipids, respectively [28]. The cone-shaped lipids form
inverted phases [Figure 1.1-right], whereas the inverse cone-shaped lipids prefer normal phases.
5
Figure 1.1: Lipid phase as a function of lipid shape.
Lipid shapes and corresponding S parameters are shown (Middle). Lipids with S ~ 1 have a cylindrical
shape and support a lamellar phase (Left), whereas the lipids with S > 1 are cone-shaped and support the
HII phase (Right). Spheres represent the lipid headgroups. Blue areas represent the solvents while the
grey areas are representative of hydrophobic parts formed by lipid acyl chains. In the HII phase, unit cell
spacing (or d-spacing) and radius of water core are shown as parameters 𝑎 and 𝑅𝑤 , respectively. Inkscape
A change in the effective molecular shape could result in a phase preference for the lipidic system
and subsequent phase transition to the more favorable phase [28]. Taking the lamellar-to-HII phase
transition as an example, several factors are known to influence this transition [33]. An increase in
temperature, unsaturation degree and length of the lipids’ acyl chains will push the system towards the
6
HII phase, whereas an increase in hydration, headgroup size and charge favors the lamellar phase.
Charged headgroups have a larger effective 𝑎0 compared to their neutral states due to the electrostatic
repulsions between the charged headgroups. For instance, phosphatidylserines (PSs) with a net charge of
-e have an effective cylindrical shape which favors lamellar phase. However, lowering the pH to less than
3 results in the carboxyl group protonation which reduces the electrostatic repulsions, and consequently
reduces the optimal area per lipid. This change in the shape makes the HII phase as the phase of
preference for PS lipids in this pH regime [34]. Adding cations to other anionic lipids, e.g. cardiolipins,
has been also shown to promote the HII phase due to screening the headgroup charges and reducing the
Lamellar phase is referred to a system with several lipid bilayers stacked on the top of each other where
solvent separates each of two adjacent bilayers [23, 35]. Lipid bilayers, themselves, are membranes
composed of two layers of lipids tightly bound to each other, in which the lipid headgroups are facing the
solvent while the acyl chains form a hydrophobic layer in the middle [Figure 1.1]. This hydrophobic
layer acts like a barrier and blocks the water-soluble molecules to pass through it. Assuming the bilayer is
in the XY-plane, the lipids are in average parallel to the normal vector to the membrane, i.e. the Z axis.
Lipid bilayers have been extensively studied, both experimentally and computationally, and much is
known about both their structure and dynamics [36-39]. Given an average length of ca. 2 𝑛𝑚 for a
7
Figure 1.2: Phosphate-to-phosphate (P-P) thickness for POPC bilayer.
Right) A snapshot taken from POPC bilayer simulation. All lipids are colored in orange, and the
phosphorous atom of each lipid is shown as a red sphere. The blue box represents the simulation box.
Water and ions on both sides of the bilayer are not shown for clarity. Left) The corresponding electron
density of the phosphorous atoms along the bilayer normal, with respect to the bilayer center, is shown.
Inverted/reverse hexagonal phase is simply made of several lipid cylinders packed in a two-dimensional
hexagonal geometry [Figure 1.1]. Each cylinder has a hydrophobic shell and an aqueous core. Lipids are
oriented perpendicular to the cylinder surface with their headgroups point inward to the solvent core,
while the lipid tails point outward. The structure of HII and other non-lamellar phases, as well as the
dynamics of the comprising lipids have been extensively studied experimentally, and to a much lesser
extent, computationally [21, 23, 40-42]. The lipid components of HII phase assume an effective cone
shape, and the radius for the aqueous core in HII phase is more than 1 𝑛𝑚 [28]. One of the structural
8
parameters of interest in the HII phase is 𝑎, the d-spacing or the unit cell spacing, defined as the distance
between the centers of two adjacent water cores [43]. This parameter can both be measured using small
angle X-ray scattering (SAXS) experiments and be directly calculated from the simulations by ensemble
averaging of the simulation box size. Given the hexagonal arrangement of the cylinders, the 𝑎 parameter
can be converted to the HII lattice distance (𝑑ℎ𝑒𝑥 ) according to the Equation 1.2:
𝜋 Equation 1.2
𝑑ℎ𝑒𝑥 = 𝑎(𝑐𝑜𝑠 ( ))
6
Assuming a cylindrical geometry for the water cores in HII phase, the d-spacing parameter, the
radius of water core (𝑅𝑤 ), and the water (core water) volume fraction (𝜑𝑤 ) are related through the
Equation 1.3
𝑅𝑤 = 𝑎 √𝜑𝑤 ( √3 / 2 𝜋)
Therefore, the approximate (𝑅𝑤 ) could be calculated if both (𝑎) and (𝜑𝑤 ) are provided. Although (𝑎) can
be obtained from SAXS experiments, it is not easy to estimate (𝜑𝑤 ) for a given HII phase when
experiments are done in the presence of excess water. In simulations, however, this quantity can be
To measure 𝜑𝑤 in simulations, the volume of simulation box (𝑉𝑡𝑜𝑡𝑎𝑙 ) should be measured first.
The volume taken by the water molecules inside the lipid cylinder, i.e. 𝑉𝑐𝑜𝑟𝑒−𝑤𝑎𝑡𝑒𝑟 , can also be
9
𝑉𝑐𝑜𝑟𝑒−𝑤𝑎𝑡𝑒𝑟 = 𝑁𝑤𝑎𝑡𝑒𝑟 𝑉𝑤 Equation 1.4
with 𝑁𝑤𝑎𝑡𝑒𝑟 and 𝑉𝑤 be the number of water molecules inside the lipid cylinder and volume of one single
water molecule at the temperature at which simulations were conducted, respectively. Thus, 𝜑𝑤 can be
calculated as:
Since the systems are composed of water and lipids, the lipid volume fraction (𝜑𝑙 ) in the sample can be
𝜑𝑙 = 1 − 𝜑𝑤 Equation 1.6
Also, the volume taken by a single lipid in the system is equal to:
Finally, given that all these parameters (i.e. 𝑅𝑤 , 𝜑𝑙 , 𝑑ℎ𝑒𝑥 , 𝑉𝑙 ) are known, the area per lipid (APL)
on the cylinder surface of radius 𝑅𝑤 can be calculated using the Equation 1.8 [45].
10
𝐴PL = (√3 𝜋 𝑅𝑤 𝑉𝑙 )⁄(𝑑ℎ𝑒𝑥 2 𝜑𝑙 ) Equation 1.8
Nanoparticles based on lipids have a wide range of applications in nanotechnology, including DDSs [46,
47]. Lipid nanoparticles (LNPs) can be used for delivery of different types of therapeutics, to the target
cells, e.g. cancer cells. Indeed, LNPs containing optimized ionizable cationic lipids (ICLs) are currently
the most potent systems and the lead carriers for siRNAs [48-50].
LNPs containing optimized ICLs are currently the most effective in liver-related diseases, e.g. cancer,
lower are one of the main components of LNPs with key roles in siRNA encapsulation, siRNA-LNP
structure and stability, and endosomal escape [51]. This pKa range is known to be critical so that it
ensures that ICLs are neutral in the physiological pH levels inside the bloodstream, while are protonated
and positively charged inside the endosomes where the pH is more acidic. This way, these ionizable lipids
assist in LNP stability and prolong its circulation time in bloodstream, while assisting in the endosomal
membrane destabilization and siRNA release into the cytoplasm, respectively [50, 52].
with an apparent pKa of ca. 6.7 is an optimized and highly efficient ICL in silencing hepatic genes [51,
53]. Despite the importance, its role in LNP size, internal structure, stability, and siRNA release is not
known. The role of KC2 in each of the above-mentioned aspects could be determined by studying its
11
Figure 1.3: The two-dimensional structure of KC2 and its protonated state, KC2H.
Top) KC2, Bottom) KC2H. Hydrogen atoms, except the protonation one, are not shown for clarity.
Avogadro [54] and Chimera [55] were used to generate the figure.
A schematic view of the internal structure for a typical siRNA-LNP containing KC2 is shown in
Figure 1.4. These siRNA-LNPs have a diameter of about 100 nm, mainly composed of KC2,
phosphatidylcholines (PCs), cholesterol, polyethylene glycol (PEG)-lipids, targeting ligands, and siRNAs
[56]. Although the exact distribution of lipids and siRNAs inside the LNPs is not known, these siRNA-
LNPs are shown to have an electron dense core [57-61]. An initial model for these systems was proposed
based on the results from simulations and observed electron dense cores in the experiments [Figure 1.4-
Left] [56, 59]. According to that model, this electron dense core was due to the siRNAs entrapped inside
inverted micelles comprised of protonated KC2 (KC2H). However, the recent experiments [57] suggest
otherwise. According to this study, the neutral KC2 molecules are segregated from other components in
12
the siRNA-LNPs and form an oil droplet in the core of LNPs, whereas the siRNAs are packed between
lipid bilayers surrounding this oily core. Based on these results, a new model for the siRNA-LNPs has
Left) Previous model. Figure adapted from (Tam, Yuen, Sam Chen, and Pieter Cullis. Advances in lipid
nanoparticles for siRNA delivery. Pharmaceutics 2013, 5(3): 498-507) with permission from the Authors
owing the copyrights. Right) Updated model. Reprinted (adapted) with permission from (Kulkarni,
Jayesh A., Maria M. Darjuan, Joanne E. Mercer, Sam Chen, Roy van der Meel, Jenifer L. Thewalt, Yuen
Yi C. Tam, and Pieter R. Cullis. On the Formation and Morphology of Lipid Nanoparticles Containing
Ionizable Cationic Lipids and siRNA. ACS nano. 2018, 12(5): 4787-4795). Copyright (2018) American
Chemical Society. Inkscape was used to generate or modify the models [32].
13
1.2.4 Lipids of interest in this study.
Chapters 4 and 5 are based upon the results from simulation of bilayers and HII structures, respectively.
These lipid aggregates are composed of one or more of the lipids shown in Figure 1.5. This group of
14
From left to right: DOPE, POPE, POPC, KC2H, and KC2. Except the KC2H which are positively
charged, the other lipids have a net charge of zero. Oxygen, nitrogen, phosphorus, carbon, and hydrogen
atoms are represented in red, blue, orange, light yellow, and white, respectively. Most of the hydrogen
atoms are not shown for clarity. A few of the hydrogen atoms in the lipid tails are shown to locate the
C=C bonds in them. Avogadro [54] and Chimera [55] were used to generate the figure.
1.2.4.1 POPC.
Phosphatidylcholines (PCs) are the major component of the mammalian plasma membranes [8].
Moreover, given to their bilayer forming preference, they are one of the main components in the siRNA-
LNP formulations [56]. Therefore, PCs are both of a biological importance and of interest for practical
From a computational point of view, the parameters for these lipids are optimized and well-tested
in CHARMM36 force field (C36 FF) (see Section 2.3) and many other force fields [62]. POPC bilayers
are also one of the most frequently studied systems in both simulations and experiments, providing a
considerable amount of data in the literature to compare our simulations with. Taken together, POPC is a
good candidate for KC2 and KC2H parameterization and model validation, and for looking at interactions
Phosphatidylethanolamine (PE) bilayers are commonly used to model the inner leaflet of the plasma
membrane because PEs are the major lipids in the cell membrane inner (cytoplasmic) leaflet [63]. DOPE
and POPE are both capable of forming the HII phase at high temperatures due to their cone-shaped
15
structures [20, 64]. These lipids are also known to promote the lamellar-to-HII transition as helper lipids
in mixed lipid systems [52]. These neutral lipids are well studied in HII phases, especially using
experimental techniques [44, 65-67], and the parameters for them are well optimized and validated
through many computational studies [62]. These factors make them good candidates to study the HII
structures and validate the system setup and protocols developed in this study.
KC2 is another lipid of interest in this thesis. This ICL is one of the critical components of the siRNA-
LNPs [51]. At the physiological pH level, the amino group in the KC2 headgroup is deprotonated and
KC2 has a net charge of zero. Upon a drop in the pH, corresponding to the acidic condition inside the
endosomes, this amino group is protonated and KC2H with a net charge of +e forms. Where KC2 is
localized in the nano-formulation and how it interacts with PC lipids as a function of pH and mixing ratio
is one of the aims in this thesis. Force field parameters for these lipids, however, were not available and
Following are brief descriptions of what will be discussed in each chapter and the importance of each
study.
Chapter 2: Methods
MD simulation using atomistic force fields (FFs) was used for all the simulations conducted in this thesis.
In this chapter, the theory of molecular dynamics simulation, as well as the underlying concepts and
algorithms implemented within will be briefly introduced first. Next, common biomolecular FFs, e.g.
16
CHARMM36 FF [62] and AMBER99SB-ILDN FF [68], and the reason for choosing each FF will be
briefly discussed. The protocol which was developed and used for parameterization of KC2 and KC2H in
CHARMM36 FF, will be presented next. In the last section of this chapter, construction of molecular
systems, e.g. bilayers and HII phases, composed of two or three molecule types will be explained.
delivery systems
Nanoparticles are of special interest in several fields of studies, including nanomedicine [69]. To achieve
the maximum potential of these systems as drug/gene delivery systems, molecular structures and
This chapter is a comprehensive review of the previous studies which successfully implemented
variety of nano-particles commonly used for drug/gene delivery purposes. Reading through this review
article, the reader will find out the type of questions in the nano-medicine field of research which could be
investigated using computational techniques. We have further discussed the experimental techniques
which are commonly used in nanomedicine field of research, and how computer simulations could
potentially assist experimentalists in the rational design of nano-particulate drug delivery systems [5].
Chapter 4: Ionizable amino lipid interactions with POPC: Implications for lipid nanoparticle
function
Lipid nanoparticles (LNPs) containing ICLs are one of the molecular systems of interest in this thesis.
These systems are clinically validated non-viral delivery systems [51, 57], with ONPATTROTM
(Patisiran) as the first siRNA-based drug recently approved by the Food and Drug Administration (FDA).
17
KC2 is one of the most efficacious ICLs which is known to play critical roles in determining the internal
structure and size of this class of LNPs, their stability in the bloodstream, and drug release from
endosomes. However, the KC2 distribution in LNPs and its effect on the structure, stability and drug
The main goal of this study was to look at the interactions between KC2 and its protonated state
(KC2H) with other lipids (e.g. PCs) in the LNPs. Investigation of these molecular interactions could
provide deeper insight into the KC2 distribution in the LNPs, and the molecular nature of the electron-
dense core observed in the structure of this class of LNPs. These interactions could be studied using MD
simulation technique. However, computational study of these systems requires the parameters for the
amino lipids. Thus, the FF parameters for these synthetic lipids were developed and validated in this
study. Given these parameters, computer simulation of POPC/KC2 binary mixtures as a function of pH
experimental approach
Inverted/reverse hexagonal phase is a non-lamellar phase of interest in the field of nanomedicine [24].
The current knowledge in this phase is mainly due to extensive experimental research, and a limited
number of computational studies. Experiments on HII phase are usually conducted in presence of excess
water. In these situations, the number of water molecules taken up by the HII lattice is difficult to be
determined. Therefore, determination of internal dimensions of HII phase in these conditions are
challenging. Conducting experiments in low hydration regimes are also challenging and could be time
The main goal of this study was to develop computational protocols for investigation of HII
phases composed of amphiphilic molecules e.g. lipids, and to examine the effect of hydration level and
18
temperature on the structural properties of HII systems. Inverted hexagonal phases composed of DOPE
and POPE were constructed and simulated for several hydration levels and temperatures. Structural
properties, including the internal dimensions, of HII systems were calculated from simulations and
CCT enzyme is a key enzyme in phosphatidylcholine (PC) synthesis. Catalytic activity of CCT enzyme is
critical for the PC homeostasis and cell proliferation [70]. Inhibition of PC biosynthesis by antitumor
phospholipid analogues have been shown to cause apoptosis, with some of these inhibitors directly
targeting the CCT enzyme [13]. Moreover, mutations in CCT enzyme, e.g. Arg223Ser, have been
previously reported to result in genetic disorders [14-17], although the underlying reason is not known.
CCT is known to be activated only when it is bond to a PC-deficient membrane. This activation is
known to be due to disassociation of two auto-inhibitory (AI) helices in this enzyme. The underlying
mechanism by which these AI helices can suppress the catalytic activity of CCT enzyme is not known
either. The main goal of this study was to investigate the effect of AI helices on the structure and
This chapter will summarize the main findings of this study. The results reported, the constructed
molecular systems, as well as the developed methods and force field parameters can be used for future
studies. The last section of this chapter provides directions for future studies.
19
1.4 Bibliography
[1]. Santos, R., Ursu, O., Gaulton, A., Bento, A.P., Donadi, R.S., Bologa, C.G., Karlsson, A., Al-
Lazikani, B., Hersey, A., Oprea, T.I. & Overington, J. P. (2017). A comprehensive map of
[3]. Allen, T. M., & Cullis, P. R. (2004). Drug delivery systems: Entering the mainstream. Science,
303(5665), 1818-1822.
[4]. Sudhahar, C. G., Haney, R. M., Xue, Y., & Stahelin, R. V. (2008). Cellular membranes and lipid-
binding domains as attractive targets for drug development. Current Drug Targets, 9(8), 603-613.
[5]. Ramezanpour, M., Leung, S. S. W., Delgado-Magnero, K. H., Bashe, B. Y. M., Thewalt, J., &
1858(7), 1688-1709.
[6]. Schlick, T., Collepardo-Guevara, R., Halvorsen, L. A., Jung, S., & Xiao, X. (2011). Biomolecular
modeling and simulation: A field coming of age. Quarterly Reviews of Biophysics, 44(2), 191-
228.
[7]. Engelman, D. M. (2005). Membranes are more mosaic than fluid. Nature, 438(7068), 578-580.
[8]. Van Meer, G., Voelker, D. R., & Feigenson, G. W. (2008). Membrane lipids: Where they are and
how they behave. Nature Reviews Molecular Cell Biology, 9(2), 112-124.
[9]. Van Deenen, L. L. M., & De Gier, J. (1974). Lipids of the red cell membrane. The Red Blood
Cell, 1, 147-211.
[10]. Izquierdo, M. (2005). Short interfering RNAs as a tool for cancer gene therapy. Cancer Gene
20
[11]. Dowdy, S. F. (2017). Overcoming cellular barriers for RNA therapeutics. Nature Biotechnology,
35(3), 222-229
[12]. Yin, H., Kanasty, R. L., Eltoukhy, A. A., Vegas, A. J., Dorkin, J. R., & Anderson, D. G. (2014).
Non-viral vectors for gene-based therapy. Nature Reviews Genetics, 15(8), 541-555.
[13]. Ramos, B., El Mouedden, M., Claro, E., & Jackowski, S. (2002). Inhibition of
[14]. Hoover-Fong, J., Sobreira, N., Jurgens, J., Modaff, P., Blout, C., Moser, A., Kim, O.H., Cho, T.J.,
Cho, S.Y., Kim, S.J. & Jin, D. K. (2014). Mutations in PCYT1A, encoding a key regulator of
[15]. Testa, F., Filippelli, M., Brunetti-Pierri, R., Di Fruscio, G., Di Iorio, V., Pizzo, M., Torella, A.,
Barillari, M.R., Nigro, V., Brunetti-Pierri, N. & Simonelli, F. (2017). Mutations in the PCYT1A
gene are responsible for isolated forms of retinal dystrophy. European Journal of Human
[16]. Yamamoto, G.L., Baratela, W.A., Almeida, T.F., Lazar, M., Afonso, C.L., Oyamada, M.K.,
Suzuki, L., Oliveira, L.A., Ramos, E.S., Kim, C.A. & Passos-Bueno, M. R. (2014). Mutations in
PCYT1A cause spondylometaphyseal dysplasia with cone-rod dystrophy. The American Journal
[17]. Payne, F., Lim, K., Girousse, A., Brown, R.J., Kory, N., Robbins, A., Xue, Y., Sleigh, A.,
Cochran, E., Adams, C. & Borman, A. D. (2014). Mutations disrupting the Kennedy
phosphatidylcholine pathway in humans with congenital lipodystrophy and fatty liver disease.
21
[18]. Contet, A., Pihan, E., Lavigne, M., Wengelnik, K., Maheshwari, S., Vial, H., Douguet, D. &
two functional catalytic domains and is inhibited by a CDP-choline analog selected from a virtual
[19]. González-Bulnes, P., Bobenchik, A. M., Augagneur, Y., Cerdan, R., Vial, H., Llebaria, A., &
Mamoun, C. B. (2011). PG12, a phospholipid analog with potent antimalarial activity inhibits
[20]. Cullis, P. R., & Kruijff, B. D. (1979). Lipid polymorphism and the functional roles of lipids in
399-420.
[21]. Seddon, J. M. (1990). Structure of the inverted hexagonal (HII) phase, and non-lamellar phase
1-69.
[22]. Luzzati, V., & Tardieu, A. (1974). Lipid phases: Structure and structural transitions. Annual
[23]. Luzzati, V., & Husson, F. (1962). The structure of the liquid-crystalline phases of lipid-water
[24]. Hirlekar, R., Jain, S., Patel, M., Garse, H., & Kadam, V. (2010). Hexosomes: A novel drug
[25]. Chen, Y., Ma, P., & Gui, S. (2014). Cubic and hexagonal liquid crystals as drug delivery systems.
[26]. Barauskas, J., Johnsson, M., & Tiberg, F. (2005). Self-assembled lipid superstructures: Beyond
22
[27]. Greaves, T. L., & Drummond, C. J. (2008). Ionic liquids as amphiphile self-assembly media.
[28]. Cullis, P. R., Hope, M. J., & Tilcock, C. P. (1986). Lipid polymorphism and the roles of lipids in
[29]. Kaasgaard, T., & Drummond, C. J. (2006). Ordered 2-D and 3-D nanostructured amphiphile self-
assembly materials stable in excess solvent. Physical Chemistry Chemical Physics, 8(43), 4957-
4975.
[30]. Israelachvili, J. N., Marčelja, S., & Horn, R. G. (1980). Physical principles of membrane
[31]. Israelachvili, J. N., Mitchell, D. J., & Ninham, B. W. (1976). Theory of self-assembly of
hydrocarbon amphiphiles into micelles and bilayers. Journal of the Chemical Society, Faraday
[32]. Team, I. (2004). Inkscape: A vector drawing tool. URL http://www. inkscape. org.
[33]. Gruner, S. M., Cullis, P. R., Hope, M. J., & Tilcock, C. P. S. (1985). Lipid polymorphism: The
molecular basis of nonbilayer phases. Annual Review of Biophysics and Biophysical Chemistry,
14(1), 211-238.
[34]. Cullis, P. R., Verkleij, A. J., & Ververgaert, P. H. (1978). Polymorphic phase behavior of
cardiolipin as detected by 31P NMR and freeze-fracture techniques. Effects of calcium, dibucaine
[35]. Tresset, G. (2009). The multiple faces of self-assembled lipidic systems. PMC Biophysics, 2(1),
3.
[36]. Nagle, J. F., & Tristram-Nagle, S. (2000). Structure of lipid bilayers. Biochimica et Biophysica
23
[37]. Venable, R. M., Brown, F. L., & Pastor, R. W. (2015). Mechanical properties of lipid bilayers
from molecular dynamics simulation. Chemistry and Physics of Lipids, 192, 60-74.
[38]. Tieleman, D. P., Marrink, S. J., & Berendsen, H. J. (1997). A computer perspective of
membranes: Molecular dynamics studies of lipid bilayer systems. Biochimica et Biophysica Acta
[39]. Pastor, R. W., Venable, R. M., & Feller, S. E. (2002). Lipid bilayers, NMR relaxation, and
[40]. Marrink, S. J., & Mark, A. E. (2004). Molecular view of hexagonal phase formation in
[41]. Perutková, Š., Daniel, M., Dolinar, G., Rappolt, M., Kralj‐Iglič, V., & Iglič, A. (2009). Stability
of the inverted hexagonal phase. Advances in Planar Lipid Bilayers and Liposomes, 9, 237-278.
[42]. Sodt, A. J., & Pastor, R. W. (2013). Bending free energy from simulation: Correspondence of
planar and inverse hexagonal lipid phases. Biophysical Journal, 104(10), 2202-2211.
[43]. Turner, D. C., & Gruner, S. M. (1992). X-ray diffraction reconstruction of the inverted hexagonal
[44]. Tate, M. W., & Gruner, S. M. (1989). Temperature dependence of the structural dimensions of
[45]. Rand, R. P., & Fuller, N. L. (1994). Structural dimensions and their changes in a reentrant
[46]. Pattni, B. S., Chupin, V. V., & Torchilin, V. P. (2015). New developments in liposomal drug
[47]. Allen, T. M., & Cullis, P. R. (2013). Liposomal drug delivery systems: From concept to clinical
24
[48]. Zimmermann, T.S., Lee, A.C., Akinc, A., Bramlage, B., Bumcrot, D., Fedoruk, M.N., Harborth,
J., Heyes, J.A., Jeffs, L.B., John, M., Judge, A. D. et al. (2006). RNAi-mediated gene silencing in
[49]. Arteta, M.Y., Kjellman, T., Bartesaghi, S., Wallin, S., Wu, X., Kvist, A.J., Dabkowska, A.,
Székely, N., Radulescu, A., Bergenholtz, J. & Lindfors, L. (2018). Successful reprogramming of
[50]. Wan, C., Allen, T. M., & Cullis, P. R. (2014). Lipid nanoparticle delivery systems for siRNA-
[51]. Semple, S.C., Akinc, A., Chen, J., Sandhu, A.P., Mui, B.L., Cho, C.K., Sah, D.W., Stebbing, D.,
Crosley, E.J., Yaworski, E., Hafez, I. M., et al. (2010). Rational design of cationic lipids for
[52]. Hafez, I. M., Maurer, N., & Cullis, P. R. (2001). On the mechanism whereby cationic lipids
[53]. Jayaraman, M., Ansell, S.M., Mui, B.L., Tam, Y.K., Chen, J., Du, X., Butler, D., Eltepu, L.,
Matsuda, S., Narayanannair, J.K. & Rajeev, K. G. (2012). Maximizing the potency of siRNA
lipid nanoparticles for hepatic gene silencing in vivo. Angewandte Chemie International Edition,
51(34), 8529-8533.
[54]. Hanwell, M. D., Curtis, D. E., Lonie, D. C., Vandermeersch, T., Zurek, E., & Hutchison, G. R.
(2012). Avogadro: An advanced semantic chemical editor, visualization, and analysis platform.
[55]. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., &
Ferrin, T. E. (2004). UCSF Chimera—a visualization system for exploratory research and
25
[56]. Tam, Y., Chen, S., & Cullis, P. R. (2013). Advances in lipid nanoparticles for siRNA delivery.
[57]. Kulkarni, J.A., Darjuan, M.M., Mercer, J.E., Chen, S., van der Meel, R., Thewalt, J.L., Tam,
Y.Y.C. & Cullis, P. R. (2018). On the formation and morphology of lipid nanoparticles
containing ionizable cationic lipids and siRNA. ACS nano, 12(5), 4787-4795.
[58]. Chen, S., Tam, Y. Y. C., Lin, P. J., Sung, M. M., Tam, Y. K., & Cullis, P. R. (2016). Influence of
particle size on the in vivo potency of lipid nanoparticle formulations of siRNA. Journal of
[59]. Leung, A.K., Hafez, I.M., Baoukina, S., Belliveau, N.M., Zhigaltsev, I.V., Afshinmanesh, E.,
Tieleman, D.P., Hansen, C.L., Hope, M.J. & Cullis, P. R. (2012). Lipid nanoparticles containing
[60]. Leung, A. K., Tam, Y. Y. C., Chen, S., Hafez, I. M., & Cullis, P. R. (2015). Microfluidic mixing:
A general method for encapsulating macromolecules in lipid nanoparticle systems. The Journal of
[61]. Belliveau, N.M., Huft, J., Lin, P.J., Chen, S., Leung, A.K., Leaver, T.J., Wild, A.W., Lee, J.B.,
Taylor, R.J., Tam, Y.K. & Hansen, C. L. (2012). Microfluidic synthesis of highly potent limit-
size lipid nanoparticles for in vivo delivery of siRNA. Molecular Therapy-Nucleic Acids, 1, e37.
[62]. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez,
C., Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom
additive force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B,
114(23), 7830-7843.
26
[64]. McIntosh, T. J. (1996). Hydration properties of lamellar and non-lamellar phases of
131.
[65]. Gawrisch, K., Parsegian, V. A., Hajduk, D. A., Tate, M. W., Gruner, S. M., Fuller, N. L., & Rand,
[66]. Lafleur, M., Bloom, M., Eikenberry, E. F., Gruner, S. M., Han, Y., & Cullis, P. R. (1996).
Correlation between lipid plane curvature and lipid chain order. Biophysical Journal, 70(6),
2747-2757.
[67]. Lafleur, M., Cullis, P. R., Fine, B., & Bloom, M. (1990). Comparison of the orientational order of
lipid chains in the L.alpha. and HII phases. Biochemistry, 29(36), 8325-8333.
[68]. Lindorff‐Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., & Shaw, D.
E. (2010). Improved side‐chain torsion potentials for the Amber ff99SB protein force field.
[69]. Blanco, E., Shen, H., & Ferrari, M. (2015). Principles of nanoparticle design for overcoming
regulation, and structure of an amphitropic enzyme required for membrane biogenesis. Progress
27
Chapter Two: Methods
2.1 Abstract
Molecular dynamics (MD) simulation is a powerful technique for study of molecular systems. This
technique was used to study both the structure and dynamics of biomolecules of interest: lipid aggregates
in Chapters 4 and 5, and CCT enzyme in Chapter 6. In Section 2.2, the MD simulation methodology
and some of its underlying algorithms will be explained. The material presented in that section are mainly
based on the GROMACS user manual [1] and a book written by Andrew R. Leach on molecular
Classical MD simulations are based on classical descriptions of atoms in the system via so-called
classical force fields (FF). Thus, in Section 2.3, the main concepts in the biomolecular force fields will be
discussed, and their functional forms as well as different functional terms used in the common force fields
will be explained. In this thesis, CHARMM36 FF was used for simulations in Chapters 4 and 5, and
AMBER99SB-ILDN for simulations in Chapter 6. A short description of each of these two FFs, and the
reason why these FFs were chosen for each study will be provided next.
MD simulations require force field parameters for all the molecules in the system to be well
defined. Since these parameters for KC2 and KC2H did not exist in any force field, the parameters for
these two key lipids were developed and validated. The protocol developed and used for parameterization
of these cationic lipids, as well as the discussions on parameter validation will be explained in detail in
Section 2.4.
Finally, in Section 2.5, construction of molecular systems simulated in this study will be
explained.
28
2.2 Molecular Dynamics (MD) Simulation
Molecular dynamics (MD) simulation is a well-known computational method for studying molecular
systems [3]. This technique can provide insights into structural, dynamic, and thermodynamic properties
of biomolecular systems.
𝑚𝑖 (𝑑 2 𝒓𝒊 / 𝑑𝑡 2 ) = 𝑭𝒊 , i = 1 . . . N Equation 2.1
Where, 𝑁 is the number of particles in the system, and 𝑭𝑖 is the force applied to an atom 𝑖 in the system
by other particles in the system. The forces are given by the negative derivative of the potential energy
function (𝑈(𝑹)) - which (𝑹) refers to ( 𝒓1 , 𝒓2 , … , 𝒓𝑁 ) - with respect to the atomic coordinates [Equation
2.2].
The total simulation time (𝑡) - ranging from a few nanoseconds to tens of microseconds or even
milliseconds, depending on the question asked – is broken into a number of time steps (𝑛) with a step size
of 𝛥𝑡. At a given position in space, each atom experiences a net force (𝑭𝑖 ) because of its interactions with
other surrounding atoms in the system, although the net force on the whole system will be zero. A change
in the system configuration after a time step results in a change in the net force on each atom. Therefore,
29
at each time step, the net force on each atom (𝑖) is calculated and subsequently used to update both their
description of the interaction energy of the system and is a function of atomic positions (𝒓𝑖 ). These
classical potential functions are relying on the Born-Oppenheimer approximation which allows the
separation of motions of electrons from nuclei in a molecule. This approximation is valid due to the mass
difference between nuclei and electrons because it causes the electrons to move much faster than atomic
nuclei. Both the function and the parameters required to calculate the potential energy of a molecular
To start a MD simulation, an initial structure (preferably close to the equilibrium state of interest)
is needed. This initial structure could be either taken from the ones determined by experiments (e.g. X-ray
or NMR), or be constructed using molecular modeling packages, e.g. GROMACS [4]. In addition, initial
velocities corresponding to the desired temperature should be assigned to each particle in the system.
Starting from this state (𝒓𝑖 and 𝒗𝑖 , for 𝑖 = 1,2,3, . . . , 𝑁 with 𝑁 as the number of atoms in the system),
atoms interact with each other and their positions and velocities are updated over time producing a so-
called trajectory. This trajectory can be further analysed by statistical mechanics methods to extract the
macroscopic properties of interest from these atomic motions (microscopic data). This task is possible
based on the ergodic hypothesis. According to this hypothesis, when a system is simulated for a long
time, it will span all the accessible microstates in the phase state in an equiprobable way. Thus, the
trajectory will be equivalent to an ensemble of microstates for the system of interest. This indicates that
the time average over trajectory will be equivalent to the ensemble average. This equivalency, however,
is valid when the system is simulated for enough time so that the system can sample all the corresponding
microstates to that specific total energy [2]. The lack of sampling is one of the major limitations and could
30
result in inaccurately averaged thermodynamic properties calculated from computer simulation
There are several molecular simulation packages such as GROMACS [4], CHARMM [6], LAMMPS [7],
AMBER [8], and NAMD [9], which support MD simulations. GROMACS is a fast, flexible, efficient,
scalable, and free package for molecular simulations and trajectory analysis. It is one of the most popular
packages for simulations of biomolecules, and supports the use of several major FFs. The reported studies
in this thesis were conducted using several versions of the GROMACS MD packages - the version
All the simulations in this study were conducted under constant number of particle (𝑁), temperature (𝑇),
and pressure (𝑃) conditions. These conditions were corresponding to the experimental conditions in the
laboratory where the temperature is the room temperature and the pressure is equal to atmospheric
pressure. The thermodynamic ensemble, i.e. the set of all the microstates accessible to and sampled by the
𝑁𝑃𝑇 ensemble in computer simulations. In some cases, such as the first equilibration phase of the
simulations or when the experiments/simulations in constant volume (𝑉) is required, the canonical (𝑁𝑉𝑇)
ensemble is used. In principle, in the thermodynamic or macroscopic limit, i.e. where the systems are
composed of a large/infinite number of particles, all the ensembles are expected to result in the same
thermodynamic properties for the system. However, this is not the case in the simulations because only a
There are several algorithms in the MD simulation which can be used to control the temperature
and pressure of the molecular system. In the following section, these algorithms will be briefly explained.
31
For the pressure coupling, different types of coupling such as isotropic, semi-isotropic, and anisotropic,
Temperature of the system (consist of 𝑁 particles) is calculated from the ensemble average (shown as
angle brackets) of the total kinetic energy < 𝐸𝑘𝑖𝑛 >, and the total number of degrees of freedom 𝑁𝑑𝑓 in
the system [Equation 2.3]. In this equation 𝑘𝐵 represents the Boltzmann’s constant, 𝑚𝑖 the atomic mass
for atom 𝑖, and 𝑣𝑖 stands for the velocity of atom 𝑖. The left part of this equation is relying on the
equipartition theorem which states that each degree of freedom in the system contributes as 1/2 𝑘𝐵 and
1/2 𝑘𝐵 𝑇 to the system’s heat capacity and total kinetic energy, respectively. For a system consisting of 𝑁
atoms, 3𝑁 degrees of freedom exist. However, 𝑁𝑑𝑓 is usually less than 3𝑁 due to the applied constraints
(𝑁𝑐 ), e.g. applied constraints on bonds and angles, and center-of-mass (COM) motion removal (𝑁𝐶𝑂𝑀 = 3)
𝑁 Equation 2.3
< 𝐸𝑘𝑖𝑛 >𝑁𝑉𝑇 = (1/2) 𝑁𝑑𝑓 𝑘𝐵 𝑇 = (1/2) ∑ 𝑚𝑖 𝑣𝑖2
𝑖=1
There are several ways to maintain a constant temperature during the simulation course (required
for 𝑁𝑉𝑇 and 𝑁𝑃𝑇 simulations), including Berendsen [10], velocity-rescaling [11], and extended Nose-
Hoover [12, 13] schemes. In these methods, the system is coupled to a heat bath with a given temperature
32
of 𝑇𝑏𝑎𝑡ℎ which is the reference temperature desired for the system. This way, systems can exchange heat
Berendsen coupling is a weak coupling method based on rescaling the atomic velocities at each
time step. The scaling factor (𝜆 𝑇 ) is calculated from the difference between the temperature of the system
and the reference temperature at each time step [Equation 2.5], where 𝛥𝑡 and 𝜏 are the time step and the
Although the Berendsen algorithm is a good choice for the equilibration phase to adjust the
system temperature to the reference temperature, it is known to fail to generate a correct canonical
ensemble. Therefore, for the production runs, other algorithms with the ability to sample a correct
canonical ensemble are recommended. Velocity-rescaling and Nose-Hoover algorithms are two
temperature coupling algorithms used for the production runs in the simulations conducted in this thesis.
Velocity rescaling algorithm is based on stochastic terms added to the kinetic energy calculations. In
Nose-Hoover algorithm, the bath is considered as part of the system (extended systems) and is included in
the system Hamiltonian and consequently into the particles’ equations of motion.
For a molecular system, the pressure is defined as the average perpendicular force applied on the
container walls caused by random particle collisions with those walls. In simulations, the macroscopic
pressure tensor 𝑃 is the ensemble/time average of the microscopic (or instantaneous) pressure tensor Ƥ
33
𝑃 = < Ƥ >𝑒𝑛𝑠𝑒𝑚𝑏𝑙𝑒 Equation 2.6
In this equation, the ⊗ stands for direct product of two vectors, whereas 𝒓𝑖𝑗 and 𝑭𝑖𝑗 are the interatomic
Like temperature coupling, there are several types of barostats which can maintain the pressure of
the system close to a reference value of Ƥ0 . The pressure can be controlled by rescaling the simulation
box volume, and consequently the atomic coordinates. Berendsen [10] and Parrinello-Rahman [14, 15]
algorithms are the most frequently used methods for controlling the system pressure in the MD
simulations. In both algorithms, the system is coupled to a pressure bath. Berendsen algorithm is a weak
coupling algorithm, where the box volume and atomic coordinates are scaled by a scaling matrix which is
where the pressure bath is considered as a piston acting on the system and is part of the system.
Therefore, the particles’ equations of motion are modified as in the case of the Nose-Hoover algorithm.
Herein again, the Berendsen weak coupling can be used for the equilibration phase, while for the
production run or when the box vector fluctuations are important, Parrinello-Rahman coupling is
recommended. This is because the Berendsen coupling is not capable of generating the accurate ensemble
and therefore, will affect the ensemble averaged thermodynamic properties extracted from the
simulations.
The pressure of the molecular system can be controlled in three different ways; isotropic, semi-
34
In the isotropic scheme, all three box vectors are scaled with the same scaling factor, and the
system pressure is an average over all three directions X, Y, and Z. The isotropic pressure coupling is
usually used for the simulation of proteins or other solutes in solution [5]. In Chapter 6 we used isotropic
In the semi-isotropic scheme, box size in two directions (e.g. X and Y) are scaled together but
treated independently from the third direction (i.e. Z). When lipid membranes or other interfaces are of
interest, the symmetry of the system suggests that the semi-isotropic pressure coupling is the best choice
for pressure control. This way, both the height and the surface area of the simulation box will be better
controlled by the coupling algorithm. Thus, as recommended in the literature [5], this scheme was used
for all the simulations conducted on lipid bilayers in this thesis (e.g. Chapter 4).
In addition to the lipid bilayer, the HII systems should also be simulated using the semi-isotropic
scheme. This is mainly because HII is not homogenous in all three directions (thus, the isotropic scheme
is not appropriate for this case) and there is a hexagonal symmetry in the way lipid cylinders are arranged
in space. Assuming lipid cylinders are parallel to the Z axis, the pressure in XY-plane could be treated
separately from the pressure in Z direction [Figure 1.1]. This allows the lipid cylinder height and
diameter be freely adjusted to optimize the system free energy. In Chapter 5, several PE HII systems
The third scheme is the anisotropic pressure coupling in which pressure in each direction is
treated independently. The anisotropic coupling is not usually recommended because like semi-isotropic
coupling it does not create the physical ensemble [5]. Furthermore, it allows large deformations of the box
size in each direction, and the box shape. However, this suggests that anisotropic pressure coupling might
be a choice where phase transitions or system deformations from their equilibrium states are expected to
happen or are desired (see [16-18] as examples). For instance, a lamellar-to-HII phase transition in a lipid
system requires special conditions such as water/lipid ratio, box size and shape, to be fulfilled so that a
35
HII phase can form. Given the lipid density and temperature in the system, the box should be able to
change in all three directions to form the unit cell suitable for HII phase formation. In Chapter 5 of this
thesis, an anisotropic pressure coupling was used to study DOPE and POPE HII structures. The results
were also compared with simulations conducted using a semi-isotropic coupling method.
Both semi-isotropic and anisotropic pressure couplings do not correspond to any experimental
ensemble [19]. This could result in an incorrect ensemble and affect the comparisons between
system. This will, indeed, minimize the artifacts caused by a finite-size molecular system (the
To apply the PBC, a space-filling box enclosing all the molecules in the system is constructed
first. Next, this unit cell (or the original box) is replicated in space (in the desired directions) for an
infinite number of times (these copies are also called images) to mimic the large system. According to the
minimum-image convention, each particle in the unit cell will only interact with the nearest image of other
particles (could be in either the original box or one of the adjacent cells) when short-range non-bonded
interactions (e.g. van der Waals interactions) are calculated. For the long-range electrostatic interactions,
however, the effect of particles in other image cells might be critical to consider. As a solution to this,
there are a few algorithms, e.g. Particle Mesh Ewald (PME), which could take advantage of this
periodicity and consider these long-range non-bonded pair interactions in an efficient way (see Section
2.2.5).
36
There are several box types, e.g. triclinic, cubic, and dodecahedron, supported in GROMACS
MD package. Depending on the biomolecule to be simulated and the geometry/symmetry of the system (if
Simulation of solute proteins are often conducted in rhombic dodecahedron box shape. This is
mainly because this box is the smallest unit cell capable of fully covering the protein and its solvation
shell, while still saving a lot of computational time by minimizing the required number of water
molecules to fill the box. Thus, in Chapter 6 of this thesis, where the simulation of CCT enzyme was of
interest, the enzyme-substrate complex was enclosed and simulated in this box type.
For the simulation of lipid bilayers (e.g. simulations represented in Chapter 4), a rectangular box
type is the most appropriate and the recommended box type. Using this simulation box, the bilayer is
usually oriented on XY-plane and the vector normal to the bilayer surface is parallel to the Z axis.
For the simulation of lipid aggregates in HII phase (e.g. simulations in Chapter 5) the situation is
a bit complicated. As discussed earlier, in HII phase the lipid cylinders are arranged in a hexagonal
geometry. So, when the original box is copied in the space by PBCs, the infinite system must have this
hexagonal symmetry. Unfortunately, a hexagonal box shape is not supported in GROMACS yet.
However, simulation of HII phase is still possible using either rectangular or triclinic box types.
Both the electrostatic (∝ 1/𝑟𝑖𝑗 ) and the dispersion part of the van der Waals (∝ 1/𝑟𝑖𝑗6 ) interactions are
long-range interactions. In principle, these nonbonded interactions between each pair of atoms (except the
ones separated by 1, 2, or 3 bonds) in the system (i.e. particles in both the original and image boxes)
should be considered in the potential energy and atomic force calculations. However, this is too time-
consuming in practice. Assuming the system is constituted of 𝑁 particles, the number of pair interactions
37
to be calculated is an order of 𝑁 2 . Therefore, to save computational time, the interactions between the
pairs are cut-off at a certain distance (usually between 1 to 2 𝑛𝑚 ). This sudden truncation of the
interactions results in a discontinuity in the potential energy, force, and its derivative, and consequently
cause significant artifacts. For the Lennard-Jones potentials, modifying the potential energy or the force
functions using either “switch” or “shift” functions could greatly reduce these artifacts. These methods,
however, are not recommended for treating the coulomb potential. For electrostatic part, the interactions
are cut-off at the given distance and the effect of other particles beyond this cut-off is considered using
other algorithms such as reaction field [20] or Particle Mesh Ewald (PME) [21, 22]. The PME method is
the recommended algorithm for treating the long-range electrostatic interactions, especially for the highly
charged systems such as DNA, and nonhomogeneous systems such as lipid bilayers [5, 23]. Therefore,
the PME method was used in all the simulations presented in this thesis. For more details on reaction field
and PME methods, the reader is referred to [24] and other references within.
MD simulation works by the numerical integration of Newton’s equations of motion for each particle in
the system at each time step. There are several well-established integrators, including leap-frog and
velocity Verlet algorithms, used in MD simulations which are available and supported in GROMACS
package. The leap-frog algorithm is the default integrator in GROMACS and is the algorithm used in all
the simulations presented in this thesis. In this method, the positions of atoms at time 𝑡 + 𝛥𝑡 are
calculated from the atomic positions at time 𝑡 and and their velocities at 𝑡 + (𝛥𝑡/2):
38
Therefore, the velocities at 𝑡 + (𝛥𝑡/2) need to be calculated first. The 𝒗𝑖 (𝑡 + (𝛥𝑡/2)) are calculated
Now, given both 𝒗𝑖 (𝑡 + (𝛥𝑡/2)) and 𝒗𝑖 (𝑡 − (𝛥𝑡/2)), the velocities at time 𝑡 are calculated as:
This process is done for each time step. The accelerations at time 𝑡 are calculated as 𝒂𝑖 (𝑡) =
𝑭𝑖 (𝑡)/𝑚𝑖 , i.e. the force applied on particle 𝑖 at that time step divided by particle mass. The 𝑭𝑖 (𝑡) can also
be calculated from the atomic positions at time 𝑡 and the potential functions 𝑈(𝑹) provided in force fields
Force fields (FF) are referred to a set of functions and associated parameters required to model
all the comprising molecules, and the way each atom in the system interacts with other atoms in
the system [2, 23, 25]. Molecular mechanic techniques, including MD simulation, make use of
the classical FFs; the FFs which are based on classical description of the potential energy of the
system. There are several FFs such as CHARMM [26], AMBER [8, 27], OPLS [28] and
GROMOS [29] which are used for simulation of biological molecules. These FFs include all the
parameters required for the simulation of main classes of biological molecules, including
proteins and lipids. There are other FFs, e.g. Berger lipids [30], Slipids [31, 32], and Lipid14
39
[33], which are usually modified/combined versions of the FFs mentioned above and are
developed for simulation of specific class of these biomolecules, e.g. lipids. Irrespective of slight
differences between these FFs, they are similar in their functional forms [5, 6] [Equation 2.10].
∑𝑏𝑜𝑛𝑑𝑠 𝑘𝑏 (𝑏 − 𝑏0 )2 + ∑𝑎𝑛𝑔𝑙𝑒𝑠 𝑘𝜃 (𝜃 − 𝜃0 )2 +
This function is a sum over six terms, each describing either a bonded (terms 1-4) or non-bonded (terms
The first and second terms are the energy contributions for bond and angle deformations from
their reference values (𝑑0 , 𝜃0), respectively. These are harmonic functions with force constants of (𝑘𝑏 )
The third and fourth terms are the energy contributions for the proper and improper dihedrals in
the system. The proper dihedrals (term 3) are sinusoidal expansions - usually are expanded up to six terms
for each individual dihedral angle (torsion) about a certain bond - where (𝑘𝜑 ), (𝑛), and (𝛿) are the force
constant, multiplicity, and phase shift, respectively. The improper dihedral term (term 4) is a harmonic
function with a force constant of (𝑘𝜔 ) and a reference value of (𝜔0), which is a required term in the
energy function to maintain the planarity of the rings, and to keep the chirality of chiral atoms the same
Finally, the last two terms in Equation 2.10 are the contributions for the van der Waals (term 5)
and electrostatic (term 6) non-bonded interactions, using Lennard-Jones (L-J) and columbic potential
40
functions, respectively. 𝑖 and 𝑗 subscriptions are the atomic indices for the interacting atoms. In L-J term,
𝑚𝑖𝑛 𝑚𝑖𝑛
the 𝜀𝑖𝑗 is the well depth, 𝑅𝑖𝑗 is the distance between two interacting atoms in which the L-J energy is
minimum, and 𝑟𝑖𝑗 is the interatomic distance [Figure 2.1]. In the coulombic term, 𝑞𝑖 and 𝑞𝑗 are the point
charges assigned to atoms 𝑖 and 𝑗, whereas 𝜀0 and 𝜀 are the permittivity of vacuum and relative dielectric
constants, respectively.
Lennard-Jones potential energy (𝑉 (𝑟𝑖𝑗 )) as a function of interatomic distances 𝑟𝑖𝑗 . Inkscape was used to
41
Most FFs consist of these six terms, although there are some which have additional terms,
Urey-Bradley is also a harmonic function of distance between 1-3 atoms in a three-bond atom group
forming an angle. This term is additional to the angle term (term 2 in Equation 2.10) and is a rarely used
term except that vibrational spectra be of interest. Similar to the bond and angle terms, the 𝑆0 and 𝑘𝑈𝐵 are
the reference 1-3 distance and the force constant, respectively. The CMAP term is a correction term
applied to the protein mainchain torsion energies resulting in a significantly improved reproduction of
Each FF, regardless of its functional form and parameter set, has been developed and validated
based on a certain number of quantities, and using a limited group of model small molecules. Extensive
computational studies and comparison with corresponding experimental data available in the literature
suggests that some FFs are more successful in reproduction of certain structural, dynamic, or
thermodynamic properties of molecular systems [35]. As mentioned earlier, there are also FFs, e.g.
Slipids [31, 32], which have been optimized for a specific class of biomolecules which makes them
capable of reproducing experimental data in a good to excellent level. Moreover, each FF was originally
developed in a MD package, e.g. GROMACS or NAMD, with a specific model for the water molecules,
e.g. SPC/E or TIP3P, and specific choices for other parameters such as cut-offs and algorithms for
treatment of interactions [36]. Therefore, choosing the most appropriate FF in each simulation is highly
dependent to the molecular system under study and the questions of interest in that study. Depending on
the type of biomolecules present in the system, as well as the physical and thermodynamic properties of
interest to be extracted from simulations, different FFs might be the best candidate. Finally, depending on
42
the spatio-temporal scale and the level of resolution necessary to get the properties of interest from
The behavior of a lipid aggregate is highly sensitive to the assigned parameters of the comprising lipids
[38, 39]. There are several atomistic force fields (FF) which could be used in computer simulations of
biomembranes [5]. There are studies which suggest C36 FF is among the best atomistic FFs at
reproducing the structural data for lipids [40]. For instance, C36 is known to yield improved lipid volume
and bilayer thickness for PC lipids [35]. C36 was also shown to reproduce the lipid diffusion with very
For simulations conducted in Chapters 4 and 5, the systems are merely composed of either one
or more lipid molecules [Figure 1.5] dispersed in water, making the C36 FF as one the best FF choices
for this study. The parameters for PC and PE are currently available and highly optimised in the C36 FF
[41]. These parameters can successfully reproduce several corresponding experimental data sets,
including deuterium order parameters (𝑆𝐶𝐷 ), in reasonable agreement. However, there are no parameters
Protein parameters are available in several FFs, including CHARMM [26] and AMBER FFs [8, 27]. The
performance of both FFs have been slightly improved by incorporating so-called CMAP backbone
corrections - corrections to the protein “mainchain” dihedral energy [42, 43]. Further improvements were
implemented into AMBER FF in 2010, resulting in a new version of AMBER FF called AMBER99SB-
ILDN [44]. In this modified version, the torsion energy of four amino acids (isoleucine, leucine, aspartate,
43
and asparagine) “side chains” were optimized. Due to these corrections and improvements compared to
other FFs, the use of AMBER99SB-ILDN has been recommended for the simulation of proteins in the
long (micro-second and beyond) simulations [44], where the side chain rotations are possible.
In Chapter 6 of this thesis, the structure and dynamics of CCT enzyme, as well as its interactions
with its substrate cytidine triphosphate (CTP) was studied using MD simulations. Since the stability and
dynamics of the enzyme, as well as the dynamics of the loops and its side chains were critical, and the
required time-scale was estimated to be 1 microsecond, AMBER99SB-ILDN was used for the
CHARMM36 FF is the chosen FF to study the lipid aggregates in this thesis (Chapters 4 and 5).
Although the parameters for most of the lipids of interest in this thesis [Figure 1.5] are available,
optimized and well-tested in C36 FF [41], the parameters for KC2 and KC2H are not present and need to
be developed. This section briefly describes the process by which these lipids were parameterized, as well
as provides the reader with the parameters for future simulations using these models.
lipids which their bulk properties are highly sensitive to the assigned parameters. Such a parameterization
for novel molecules is usually done by analogy to other parameterized molecules in the chosen FF.
Unfortunately, there are no functional groups analogous to the KC2 and KC2H headgroups, and the linker
segment connecting the headgroup and the acyl chains, which are already parameterized in C36 FF.
There are several computational tools, e.g. CHARMM-GUI [45], GAAMP [46], and ParamChem
[47, 48], which could assist in parameterization of novel molecules in C36 FF. Although these tools are
not designed and meant to parameterize the lipid molecules, they can facilitate the parameterization
process to a high degree. It is the user experience and chemical intuition to make efficient use of these
44
tools, adjust the parameters if required, and combine them properly to achieve a final set of parameters.
These parameters should generate data with an acceptable level of agreement with corresponding
experimental data. The protocol described in the next section could be used as a guide for
parameterization of new lipids in C36 FF in a time efficient way using freely available tools.
The KC2H molecule was first divided into two main segments: lipid tails and lipid headgroup; the
headgroup includes the first two CH2 atoms in each acyl chain to consider the effect of headgroup on the
The parameters for the lipid tails were simply taken from DUPC lipid downloaded from
45
Figure 2.3: KC2(H) acyl chain parameters.
The parameters for the tails are assigned based on available parameters for DUPC lipid on CHARMM-
GUI [45].
Next, the headgroup segment was parameterized using GAAMP server [46] as follows [Figure
2.4]. The headgroup itself was divided into two smaller segments: the ring segment and the amino
segment. The amino segment was parameterized twice, once for neutral amino group and once for the
protonated case. Parameters for each segment were calculated separately. To do so, a few constraints were
applied to, I) identify the equivalent atoms in the segment and II) assign a partial charge of 0.090 to each
non-polar hydrogen atom in that segment. The 0.090 value is the standard value for the partial charge of
aliphatic hydrogen atoms in C36 FF, and was required to be consistent with the C36 FF charge
assignment rules [48]. These constraints were included in the input file submitted to the GAAMP server.
46
GAAMP uses the previously parameterized molecules to automatically assign the initial parameters to
each atom in a small molecule, while applying all the given constraints [46]. Starting from this initial
parameter set, it then optimizes the parameters using the ab initio quantum mechanical calculations as the
target data. We used these optimized parameters given by GAAMP as the initial parameters for each
segment. To obtain the optimized model for each segment, these parameters were further modified and
These fragments had some overlapping CH2 and CH3 groups (shown as orange and pink spheres,
47
The parameters developed for the segments were subsequently merged to give the parameter set
for the whole headgroups of KC2 and KC2H. To facilitate the merging process of segments in the next
step, extra CH2 and CH3 groups were added to the segments. The partial charges for these extra CH2 and
CH3 groups were taken to be the same as the aliphatic CH2 and CH3 groups of lipids available in C36
FF.
So far, all the atom types for the acyl chains are taken from the parent FF, i.e. C36FF [41],
whereas the ones for the headgroup are taken from CHARMM General force field (CGenFF) [47-49]. The
last step, therefore, is to assign missing bonded parameters for the parts where atom types from CGenFF
and C36FF are combined through bonded interactions, e.g. the dihedral of HGA2-CG321-CTL2-HAL2.
This step could be done by analogy between the atom type definitions in CGenFF and C36FF and taking
the parameters from either C36FF or CGenFF, while the preference was with C36FF. Based on these
definitions, CG321, HGA2, CG331, and HGA3 in CGenFF are equivalent to the CTL2, HAL2, CTL3,
HAL2 dihedral, C36FF was searched for HAL2-CTL2-CTL2-HAL2 first. Since there was no such
combination in C36FF, HGA2-CG321-CG321-HGA2 was searched instead, found, and was assigned to
The final force field parameters (atom types, partial charges, and the bonded parameters) for the
parameterized segment [Figure A.1] in KC2 and KC2H are provided in Appendix A as reference.
There are several physicochemical properties of lipids and lipid bilayers which could be targeted for
parameterization purposes. Deuterium order parameter (SCD), area per lipid, and bilayer thickness are
48
examples of such experimental data which could be potentially used for lipid model validation [39]. The
amount of target data to validate parameters against are limited for the case of KC2 and KC2H.
In this study, the effect of adding KC2(H) on POPC palmitoyl chain order parameters in
The SCD profiles for the POPC palmitoyl chain extracted from 2H NMR spectra 1 and
calculated from simulations of POPC/KC2(H) binary mixtures are shown in Figure 2.5. In
simulations, systems with KC2 and KC2H correspond to the 2H NMR experiments conducted at
The SCD parameters from simulations are in good agreement with the experimental data.
The overall trend in SCD profiles is similar between simulations and experiments. The acyl chain
order gradually decreases along the acyl chain towards the terminal methyl group. Also, in both
simulations and experiments, the pH seems to have no or little effect on acyl chain order
parameter at 313 K. The only difference was observed between pure POPC bilayers at two pH
values. We note that the experimental POPC palmitoyl chain order parameters are smaller at pH
4.4 than at pH 8.1. This is likely due to the higher interfacial surface tension observed in POPC
membranes at pH 4.4 [50]. In simulations, however, the SCD parameters for palmitoyl chain in
pure POPC systems are the same between two pH levels. That is mainly because the effect of
buffer and the higher surface tension was not considered in our simulations.
Finally, the SCD parameters calculated from simulations are higher than the corresponding
experimental values. This is also true for POPC which the parameters are well optimized in C36
1
Data provided by Dr. Jenifer Thewalt research group at the Simon Fraser University
49
FF. To conclude, the models developed here are valid and can be used for the purpose of simulations in
Chapter 4.
Smoothed deuterium order parameter (SCD) for the palmitoyl chain of POPC in POPC/KC2(H) binary
mixtures were measured using 2H NMR experiments (Left column). The same parameters were
calculated from simulations and smoothed (Right column). This comparison was done for two pH levels
50
2.5 Construction of Binary and Ternary Mixtures in Lamellar and HII Phase
Construction of a molecular system is the first, and usually a time-consuming step, in computer
simulations. This process usually takes more time for complicated structures; i.e. the systems composed
of arbitrary molecules, with a desired mixing ratio, and with complex architectures such as HII and cubic
phases. Given that the force field parameters for all the comprising molecules in the system are available
and validated, reaching thermodynamically stable and equilibrated systems in the desired state is the next
challenge to overcome [51]. In this study, both lamellar (bilayers) and non-lamellar (HII) phases are of
interest. Systems with several mixing ratios, at several water per lipid regimes and ion concentrations
There are several computational tools [52-55] designed for construction of molecular systems
required for doing molecular simulations. However, there are shortcomings in each one of these tools
when complicated structures are of interest. Therefore, GROMACS modules [4] (see below) were used to
generate the molecular systems according to the protocol described in the next two sections.
2.5.1 Bilayers.
Construction of the bilayers was relatively straightforward. The only challenge was to make bilayers
composed of KC2(H) with other lipids such as POPC, at several mixing ratios. Consider the POPC/KC2H
system for example. To do so, two monolayers each composed of only one lipid type are built first
[Figure 2.6]. These two monolayers both must be composed of the same number of lipids, e.g. 100,
which are distributed on a two-dimensional grid (10 by 10) on a square-shaped surface with the same total
area. Two GROMACS modules, gmx editconf and gmx genconf, were used to translate and rotate the
lipids on each grid point, to build these pure monolayers. Each grid point on these surfaces will be
51
assigned a number in the range of 1 to 100, which is equal to the residue number of each lipid. The next
step is to generate a set of random numbers from 1 to 100 so that they satisfy the desired mixing ratio. For
numbers will be used to both keep the KC2H in these grid points, and to delete the POPC in the same grid
points. Finally, the coordinate files from two modified monolayers are merged together resulting in one
mixed monolayer with a desired mixing ratio; which was taken as the upper leaflet [Figure 2.6].
Assuming that this monolayer is in the XY-plane, its rotation (using gmx editconf module) around either
Adjusting the center of both leaflets to (0,0,0) and subsequently merging the coordinate files for
these two leaflets results in the POPC/KC2H (80/20) bilayer. Waters and ions were subsequently added
on the top and bottom of the system. The same approach can be applied to make bilayers composed of
52
2.5.2 Inverted hexagonal (HII) phase.
Considering the DSPS/KC2H/cholesterol (35/35/30) system as an example, first, a single lipid tube for
each lipid type (DSPS, KC2H, and cholesterol) was built and oriented in the Z direction [56]. Each lipid
tube was composed of 360 lipid molecules distributed on a cylinder with a radius of ca. 2.5 𝑛𝑚 and a
height of ca. 17 𝑛𝑚. Lipids were distributed on 15 rings, with each ring composed of 24 lipids molecules.
Two GROMCS modules, gmx editconf and gmx genconf, were used to translate and rotate the lipids. The
angular distance between lipids on a ring was 15 degrees, equals to 360 degrees divided by the number of
lipid molecules per ring (i.e. 24). The distance between the adjacent rings were 1.2 nm, to prevent bad
Given these pure cylinders for each lipid type, the lipid tube with the desired mixing ratio was
constructed as described for bilayers in Section 2.5.1. Next, a water cylinder of the same dimensions and
orientation was built using CHARMM-GUI Solvator module [45]. These two systems, i.e. the mixed lipid
tube and the water cylinder, were subsequently merged to give a water cylinder covered by lipids
[Figure 2.7]. Water molecules were gradually deleted so that the systems had 30 or fewer water per
lipids (nw) when desired. For the systems with hydration levels less than 20 and 10, the relaxed HII
systems with 20 and 10 water per lipid were used to remove the water from, respectively. A triclinic box
was employed for the simulations and periodic boundary conditions (PBC) were applied in all three
This system was energy minimized using the steepest descent algorithm and used for
equilibration and production runs using either semi-isotropic or anisotropic pressure couplings. Using
these pressure coupling methods, each system could further adjust its height and diameter, as
energetically favourable. When systems with more than one lipid cylinder was of interest, the lipid
53
Figure 2.7: Construction of the DSPS/KC2H/cholesterol (35/35/30) HII system.
Left) a single lipid tube composed of DSPS, KC2H and cholesterols, filled with water molecules. Red,
dark blue, and green spheres represent the nitrogen atom of DSPS, nitrogen atom of KC2H, and oxygen
atom in cholesterol molecules, respectively. Orange and light blue lines represent the DSPS and KC2H
tails, whereas the cholesterol body are colored in tan. Water represented as the cyan surface. Hydrogens
are not shown for clarity. Right) a HII system composed of such single water channel. Lipid cylinders are
arranged in a hexagonal geometry. Blue box represents the triclinic simulation box. System is periodic in
all X, Y, and Z directions. There are empty spaces (or voids) in the corners in the initial structures. During
the equilibration step, both the box size and lipid tails are adjusted in a way to fill these gaps. VMD [57]
54
Given the background information and methodology provided in Chapters 1 and 2, the next
chapter presents a comprehensive review on computational nanomedicine field of study (Chapter 3).
Next, the results for three research studies are presented in Chapters 4, 5, and 6. Finally, the conclusions
2.6 Bibliography
[1]. Spoel, D.V.D., Lindahl, E., Hess, B., Buuren, A.R.V., Apol, E., Meulenhoff, P.J., Tieleman, D.P.,
Sijbers, A.L.T.M., Feenstra, K.A., Drunen, R.V. & Berendsen, H. J. C. (2014). GROMACS User
[2]. Leach, A. R. (2001). Molecular modelling: Principles and applications. Pearson Education.
[3]. Schlick, T., Collepardo-Guevara, R., Halvorsen, L. A., Jung, S., & Xiao, X. (2011). Biomolecular
modeling and simulation: A field coming of age. Quarterly Reviews of Biophysics, 44(2), 191-
228.
[4]. Hess, B., Kutzner, C., Van Der Spoel, D., & Lindahl, E. (2008). GROMACS 4: Algorithms for
highly efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory
[5]. Tieleman, D. P. (2010). Methods and parameters for membrane simulations. In Molecular
[6]. Brooks, B.R., Brooks, C.L., Mackerell Jr, A.D., Nilsson, L., Petrella, R.J., Roux, B., Won, Y.,
Archontis, G., Bartels, C., Boresch, S. & Caflisch, A. (2009). CHARMM: The biomolecular
[7]. Plimpton, S. (1995). Fast parallel algorithms for short-range molecular dynamics. Journal of
55
[8]. Case, D.A., Cheatham, T.E., Darden, T., Gohlke, H., Luo, R., Merz, K.M., Onufriev, A.,
Simmerling, C., Wang, B. & Woods, R. J. (2005). The Amber biomolecular simulation programs.
[9]. Phillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel,
R.D., Kale, L. & Schulten, K. (2005). Scalable molecular dynamics with NAMD. Journal of
[10]. Berendsen, H. J., Postma, J. V., van Gunsteren, W. F., DiNola, A. R. H. J., & Haak, J. R. (1984).
Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics, 81(8),
3684-3690.
[11]. Bussi, G., Donadio, D., & Parrinello, M. (2007). Canonical sampling through velocity rescaling.
[12]. Nosé, S. (1984). A molecular dynamics method for simulations in the canonical ensemble.
[14]. Parrinello, M., & Rahman, A. (1981). Polymorphic transitions in single crystals: A new
[15]. Nosé, S., & Klein, M. L. (1983). Constant pressure molecular dynamics for molecular systems.
[16]. Marrink, S. J., & Mark, A. E. (2004). Molecular view of hexagonal phase formation in
[17]. Marrink, S. J., De Vries, A. H., & Mark, A. E. (2004). Coarse grained model for semiquantitative
56
[18]. Knecht, V., Mark, A. E., & Marrink, S. J. (2006). Phase behavior of a phospholipid/fatty
acid/water mixture studied in atomic detail. Journal of the American Chemical Society, 128(6),
2030-2034.
[19]. Zhang, Y., Feller, S. E., Brooks, B. R., & Pastor, R. W. (1995). Computer simulation of
[20]. Tironi, I. G., Sperb, R., Smith, P. E., & van Gunsteren, W. F. (1995). A generalized reaction field
method for molecular dynamics simulations. The Journal of Chemical Physics, 102(13), 5451-
5459.
[21]. Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N⋅log (N) method for
Ewald sums in large systems. The Journal of Chemical Physics, 98(12), 10089-10092.
[22]. Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., & Pedersen, L. G. (1995). A
smooth particle mesh Ewald method. The Journal of Chemical Physics, 103(19), 8577-8593.
[23]. Monticelli, L., & Tieleman, D. P. (2013). Force fields for classical molecular dynamics. In
[24]. Koehl, P. (2006). Electrostatics calculations: Latest methodological advances. Current Opinion in
[25]. MacKerell Jr, A. D. (2004). Empirical force fields for biological macromolecules: Overview and
[26]. MacKerell Jr, A.D., Bashford, D., Bellott, M.L.D.R., Dunbrack Jr, R.L., Evanseck, J.D., Field,
M.J., Fischer, S., Gao, J., Guo, H., Ha, S. & Joseph-McCarthy, D. (1998). All-atom empirical
potential for molecular modeling and dynamics studies of proteins. The Journal of Physical
57
[27]. Ponder, J. W., & Case, D. A. (2003). Force fields for protein simulations. In Advances in Protein
[28]. Jorgensen, W. L., Maxwell, D. S., & Tirado-Rives, J. (1996). Development and testing of the
OPLS all-atom force field on conformational energetics and properties of organic liquids. Journal
[29]. Scott, W.R., Hünenberger, P.H., Tironi, I.G., Mark, A.E., Billeter, S.R., Fennen, J., Torda, A.E.,
Huber, T., Krüger, P. & van Gunsteren, W. F. (1999). The GROMOS biomolecular simulation
[30]. Berger, O., Edholm, O., & Jähnig, F. (1997). Molecular dynamics simulations of a fluid bilayer of
[31]. Jämbeck, J. P., & Lyubartsev, A. P. (2012). Derivation and systematic validation of a refined all-
atom force field for phosphatidylcholine lipids. The Journal of Physical Chemistry B, 116(10),
3164-3179.
[32]. Jämbeck, J. P., & Lyubartsev, A. P. (2012). An extension and further validation of an all-
atomistic force field for biological membranes. Journal of Chemical Theory and Computation,
8(8), 2938-2948.
[33]. Dickson, C. J., Madej, B. D., Skjevik, Å. A., Betz, R. M., Teigen, K., Gould, I. R., & Walker, R.
C. (2014). Lipid14: The Amber lipid force field. Journal of Chemical Theory and Computation,
10(2), 865-879.
[34]. Team, I. (2004). Inkscape: A vector drawing tool. URL http://www. inkscape. org.
[35]. Pluhackova, K., Kirsch, S. A., Han, J., Sun, L., Jiang, Z., Unruh, T., & Böckmann, R. A. (2016).
A critical comparison of biomembrane force fields: Structure and dynamics of model DMPC,
POPC, and POPE bilayers. The Journal of Physical Chemistry B, 120(16), 3888-3903.
58
[36]. Lee, J., Cheng, X., Swails, J.M., Yeom, M.S., Eastman, P.K., Lemkul, J.A., Wei, S., Buckner, J.,
Jeong, J.C., Qi, Y. & Jo, S. (2015). CHARMM-GUI input generator for NAMD, GROMACS,
[37]. Ingólfsson, H. I., Arnarez, C., Periole, X., & Marrink, S. J. (2016). Computational ‘microscopy’of
[38]. Lyubartsev, A. P., & Rabinovich, A. L. (2016). Force field development for lipid membrane
[39]. Poger, D., Caron, B., & Mark, A. E. (2016). Validating lipid force fields against experimental
[40]. Zhuang, X., Dávila-Contreras, E. M., Beaven, A. H., Im, W., & Klauda, J. B. (2016). An
extensive simulation study of lipid bilayer properties with different head groups, acyl chain
3093-3104.
[41]. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez,
C., Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom
additive force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B,
114(23), 7830-7843.
[42]. Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., & Simmerling, C. (2006).
Comparison of multiple Amber force fields and development of improved protein backbone
[43]. Mackerell Jr, A. D., Feig, M., & Brooks III, C. L. (2004). Extending the treatment of backbone
59
protein conformational distributions in molecular dynamics simulations. Journal of
[44]. Lindorff‐Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., & Shaw, D.
E. (2010). Improved side‐chain torsion potentials for the Amber ff99SB protein force field.
[45]. Jo, S., Kim, T., Iyer, V. G., & Im, W. (2008). CHARMM‐GUI: A web‐based graphical user
[46]. Huang, L., & Roux, B. (2013). Automated force field parameterization for nonpolarizable and
polarizable atomic models based on ab initio target data. Journal of Chemical Theory and
[47]. Vanommeslaeghe, K., & MacKerell Jr, A. D. (2012). Automation of the CHARMM General
Force Field (CGenFF) I: Bond perception and atom typing. Journal of Chemical Information and
[48]. Vanommeslaeghe, K., Raman, E. P., & MacKerell Jr, A. D. (2012). Automation of the
CHARMM General Force Field (CGenFF) II: Assignment of bonded parameters and partial
[49]. Vanommeslaeghe, K., Hatcher, E., Acharya, C., Kundu, S., Zhong, S., Shim, J., Darian, E.,
Guvench, O., Lopes, P., Vorobyov, I. & Mackerell Jr, A. D. (2010). CHARMM general force
field: A force field for drug‐like molecules compatible with the CHARMM all‐atom additive
[50]. Petelska, A. D., & Figaszewski, Z. A. (2000). Effect of pH on the interfacial tension of lipid
[51]. Marrink, S. J., & Tieleman, D. P. (2001). Molecular dynamics simulation of a lipid diamond
60
[52]. Boyd, K. J., & May, E. R. (2018). BUMPy: A model-independent tool for constructing lipid
bilayers of varying curvature and composition. Journal of Chemical Theory and Computation,
14(12), 6642-6652.
[53]. Durrant, J. D., & Amaro, R. E. (2014). LipidWrapper: An algorithm for generating large-scale
[54]. Martínez, L., Andrade, R., Birgin, E. G., & Martínez, J. M. (2009). PACKMOL: A package for
[55]. Jo, S., Cheng, X., Lee, J., Kim, S., Park, S.J., Patel, D.S., Beaven, A.H., Lee, K.I., Rui, H., Park,
S. & Lee, H. S. (2017). CHARMM‐GUI 10 years for biomolecular modeling and simulation.
[56]. Kolev, V., Ivanova, A., Madjarova, G., Aserin, A., & Garti, N. (2012). Molecular dynamics
approach to water structure of HII mesophase of monoolein. The Journal of Chemical Physics,
136(7), 074509.
[57]. Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: Visual molecular dynamics. Journal of
61
Chapter Three: Computational and Experimental Approaches for Investigating
Copyrights:
This research was originally published in the Journal of Biochimica et Biophysica Acta (BBA)-
Thewalt, J., & Tieleman, D. P. (2016). Computational and experimental approaches for investigating
Contributions:
I had a major contribution in writing this review article: most of the computational sections, as well as the
abstract and part of the introduction. I created all the figures in this manuscript, except the ones taken
(with permission) from cited publications. The section on experimental techniques and studies, as well as
part of the introduction section were provided by Dr. Jenifer L. Thewalt, Dr. Sherry S.W. Leung, and
Bashe Y.M. Bashe at the Simon Fraser University. The section on lipid-based delivery systems was
Abbreviations:
5-FU, 5-fluorouracil; AuNPs, gold nanoparticles; BSA, bovine serum albumin; CD, cyclodextrin; CG,
phosphatidylcholine; CPNT, cyclic peptide based nanotube; CPP, cell penetrating peptide; DDS, drug
delivery system; DLS, dynamic light scattering; DOX, doxorubicin; DPD, Dissipative Particle Dynamics;
62
DPPC, dipalmitoylphosphatidylcholine; DSC, differential scanning calorimetry; ESR, electron spin
glutamic acid); HSA, human serum albumin; IDP, intrinsically disordered proteins; IFN, interferon alpha-
1b; IR, infrared; ITC, isothermal titration calorimetry; LD, laser diffraction; LNP, lipid nanoparticle; MD,
Molecular Dynamics; NMR, nuclear magnetic resonance spectroscopy; NP, nanoparticle; p-THPP,
PPI, poly(propylene imine); QD, quantum dot; SAXS, small angle X-ray scattering; SCPs, star
copolymers; SEM, scanning electron microscopy; SSMs, sterically stabilized micelles; TAT, trans-
activator of transcription; TEM, transmission electron microscopy; VIP, vasoactive intestinal peptide;
3.1 Abstract
Most therapeutic agents suffer from poor solubility, rapid clearance from the bloodstream, a lack of
targeting, and often poor translocation ability across cell membranes. Drug/gene delivery systems (DDSs)
are capable of overcoming some of these barriers to enhance delivery of drugs to their right place of
computational studies have enhanced our understanding of the mechanism of action of nanocarriers and
their underlying interactions with drugs, biomembranes and other biological molecules. We review key
biophysical aspects of DDSs and discuss how computer modeling can assist in rational design of DDSs
with improved and optimized properties. We summarize commonly used experimental techniques for the
study of DDSs. Then we review computational studies for several major categories of nanocarriers,
including dendrimers and dendrons, polymer-, peptide-, nucleic acid-, lipid-, and carbon-based DDSs, and
gold nanoparticles.
63
3.2 Introduction
In medicine, the delivery of a drug can be as important as the drug itself. Physiology poses key challenges
to effective drug delivery; an administered drug must penetrate obstacles such as endo- or epithelial
membranes and also survive the host's defenses in order to be effective. Addressing such challenges
requires some form of drug encapsulation, and the entity forming the capsule, which has a defined
molecular architecture, is known as a "drug delivery system (DDS)." From a functional point of view, an
ideal DDS is easy to administer, non-toxic, carries the drug to its desired destination, and then releases it
[Figure 3.1] [1, 2]. Practically, an ideal DDS is cheap and straightforward to make, as well as stable
prior to its administration. Intense research into DDS design is providing increasingly effective disease
treatment.
A central aim for specialized DDS development is the optimization of tiny drug encapsulation
vehicles known as nanoparticles (NPs). NPs typically have diameters in the range of 10 to 100 nm [3].
These small DDS can circulate freely even in capillaries [4], and are intrinsically better at traversing
biological barriers than larger DDS. It is worth mentioning that nanomedicines, however, have a more
complicated and time-consuming FDA approval process than parent unimolecular therapeutics [5]. In
addition to efficacy and side effect evaluations common between both small molecules and
nanomedicines, there are other concerns about nanocarriers which need further evaluation. Nanocarrier
aggregation in vivo and their drug release, as well as evaluation of each component of the formulation are
some of those concerns. FDA regulations for nanomedicines, the complexity of nanocarriers, and the
prospect of only a small increase in performance upon reformulating small drugs in nanocarriers can
discourage pharmaceutical companies from investing in these systems. In fact, it seems that investigations
into DDS are much more successful in generating papers than in developing new treatment methods [5].
64
However, considering the number of approved drugs and current ones in clinical trials [6, 7], as well as
clearer guidelines for nanomedicine approval by the FDA, there is growing momentum in transferring
After the DDSs enter the bloodstream (1), drug (green circle) leakage might occur if the drug-DDS
complex is unstable (2). Depending on the surface chemistry, DDSs might interact with biomolecules (3)
and ions (4) in circulation. Eventually, they will extravasate to the extracellular space (5). After reaching
the target cell, they interact with the cell membrane (6) and can be internalized either via direct
penetration (7) or endocytosis (8). In the latter case, the nanocarrier will be entrapped in endosomes and
must be released (9) before endosomal maturation. Drug-DDS dissociation (10) is necessary for drugs to
65
be effective in the cytoplasm or nucleus (11). This dissociation can be mediated by interactions with
Many different types of NPs exist, including liposomes [7, 8], dendrimers [9-11], carbon
nanotubes (CNTs) [12-14], inorganic [15-20], and polymer-based [21, 22] NPs, with each having its own
unique properties [Figure 3.2]. Physicochemical properties including size, shape, deformability, surface
charge and chemical composition affect how well they evade phagocytosis and how they interact with
vasculature, traverse cell membranes and escape from endosomes prior to drug degradation [1]. Thus,
From top-left to bottom-right, monolayer-protected gold nanoparticle, polymeric micelle, DNA cage,
carbon nanotube, solid lipid nanoparticle, polymer, dendrimer and dendron, liposome, and protein-based
66
DDS. For dendrimer: orange, green, red, and blue represents G0, G1, G2, and G3, respectively. Dendron
Despite the increasing sophistication of experimental efforts to measure, design and optimize the
structure and dynamics of NPs, such research inevitably faces intrinsic and practical limitations.
Resolution of structural details can be difficult or impossible, and systematic variation of properties such
as composition, size and surface charge can be prohibitively time consuming and expensive. As a result,
general biophysical principles connecting the effectiveness of a DDS with its composition are difficult to
elucidate. Predicting optimum DDS design solely on the basis of experimental research is therefore
unlikely. Another challenge in DDS development is that many DDSs show promise in vitro but fail in
vivo [6]. This is mainly because of the lack of mechanistic insight obtainable by experiments which are
based on trial and error. Theoretical methods, both analytical and computational, can allow primary
screening of variables in order to predict suitable conditions for further experiments [23]. Computer
modeling techniques provide detailed information about molecular interactions and other
physicochemical properties of drugs and carriers [Table 3-1] and have found broad applications in
Table 3-1: Questions relevant to drug delivery system design that can be answered by
computational simulations.
67
Structure and dynamics of drug-DDS complex
Drug loading
Drug distribution
Drug-DDS interactions
Functionalization
Drug leakage
Aggregation
Stability
Cellular internalization
Nucleus translocation
design of new formulations with improved efficacies. In this chapter, we describe the main experimental
approaches to characterize NP-based DDSs and focus in more detail on current applications of computer
68
In the next sections, we first give a brief overview of commonly used experimental techniques,
followed by a section that describes several common computational approaches. Section 3.5 highlights
recent computational studies of what currently are the major classes of NP-based DDSs. We conclude
with a brief outlook on the role of computer simulations in drug delivery technology.
Here we provide a brief summary of techniques regularly used to characterize NP size, zeta potential,
surface morphology, and the existence of colloidal structures [Table 3-2]. We first describe the
importance of each of these properties and the experimental techniques used to measure them. Next, we
describe some sophisticated physical chemistry tools that are used in DDS development to provide data
that can parameterize and validate computational simulations. Experiments used to visualize cellular
uptake, NP-cell interactions, as well as those used to determine cell viability will be omitted, even though
they are vital to pharmaceutical development and are also commonly performed, since details on these
Technique
microscopy
69
Scanning electron SEM Particle size, dispersity, shape, surface morphology
microscopy
Atomic force microscopy AFM Particle size, dispersity, shape, surface morphology
spectroscopy
light spectroscopy
High performance liquid HPLC Separate, identify and quantify components in mixture
chromatography
calorimetry
70
Particle size is an important property because the size of the carrier can affect circulation time,
encapsulation efficiency, and cellular uptake [1, 26, 27]. For example, in the case of cancer treatment,
nano-drug carriers of the appropriate sizes can extravasate from the bloodstream to tumor tissues (380 –
780 nm) [1, 28], but not from the tighter capillaries (50 – 200 nm) in healthy tissues [29]. This contributes
to the well-known enhanced permeability and retention effect [30-33]. Light scattering techniques such as
laser diffraction (LD) and dynamic light scattering (DLS) can be used to measure NP size [34]. In both of
these experiments, a laser beam is directed at a dilute sample (e.g. a suspension or solution of NPs). In
LD, the diffraction angle is used to determine the particle size [35]. In DLS, the time dependence of the
scattered light intensity is detected at a known scattering angle. The Brownian motion of suspended
particles causes the scattered light intensity to fluctuate. These fluctuations are correlated with the
particle’s velocity and can be used to determine particle size via an appropriate theoretical model (e.g.
Stokes-Einstein equation). Most models assume the particles are spherical and monodisperse; artifacts
may arise if they are not. DLS can be used to measure the size of particles in the 2 nm – 3 μm range, and
LD can be used in the 20 nm – 2 mm range [34, 36]. Resolution of small differences in particle size can
be improved by combining light scattering techniques with a separation technique, such as field flow
microscopy (SEM), and atomic force microscopy can be used to measure NP size, as well as size
distribution (dispersity), shape, and surface morphology. Dispersity – also known as polydispersity in
older literature – in a formulation can result in varying body-residence time and immunogenicity [39, 40].
Particle shape can affect cellular uptake mechanisms and kinetics [41-45] as well as margination
dynamics – the lateral drift of NPs to endothelial walls [1]. The particle surface is the first point of contact
between the NP and its environment, and surface morphology is known to affect drug release kinetics
[46]. In electron microscopy, an electron beam is directed at the sample, and differences in electron
71
density in the sample gives contrast to structures. Since electron microscopy can give high-resolution
images approaching molecular resolution, all motions on the supramolecular scale have to be halted. For
SEM, the NPs are deposited onto a carbon conductive tape, dried, and covered with a thin layer of metal
[47]. In cryo-TEM, samples are applied onto an electron microscopy grid, plunge-frozen rapidly in ethane
to prevent the formation of damaging cubic ice and then imaged under cryogenic conditions to prevent
radiation damage to the sample during imaging [48]. Magnifications on the order of 50,000 ×, with a
resolution of 0.3 nm, are possible [49]. Atomic force microscopy probes surface terrain using a nanoscale
A NP’s shape can influence the physiological fate of its cargo [1]. Particle shape can be an
indication of undesirable aggregate structures. Aggregate structures can also cause inaccurate particle size
determination by LD or DLS [34]. Light microscopy can be used for determining if microparticles are
aggregates of smaller particles [36, 50], even though it cannot be used for directly observing NP structure
due to limitations in optical resolution (~ 200 nm). Isothermal titration calorimetry (ITC), which will be
described in greater detail below, has been used to study aggregate formation in polycationic nucleic acid
carriers [51]. Aggregation of NPs can be prevented via electrostatic repulsion [31, 52]. Surface charge
Surface charge can affect cellular uptake, cytotoxicity, and circulation times [1, 31, 53, 54]. Zeta
potential is used to characterize the surface charge of NPs. When a particle moves in a liquid, a thin layer
of liquid moves with it. The zeta potential is defined as the electrical potential at the boundary of this
moving liquid, defined as the shear plane. It is a function of surface charge density, shear plane location,
and surface structure [55]. Zeta potential is determined by measuring light scattering caused by particle
motion in an applied electric field. Charged particles with larger zeta potential (|zeta| > 25 mV) are less
likely to form aggregates but neutral particles generally have longer circulation times [56]. Surface charge
72
effects are very complex; positive and negative zeta potentials correlate with higher cellular uptake in
non-phagocytic and phagocytic cells, respectively [54]. Future DDS designs may require NPs to switch
zeta potential at the target site to maximize circulation time while targeting a particular type of cell [1]. It
should be noted that protein adsorption in cellular media can change the surface charge [52]. Conclusions
on surface-charge effects, therefore, are only valid when comparing functionalized or non-functionalized
Encapsulation efficiency – the ratio of encapsulated drug to the amount of drug used during
formulation preparation – is an important parameter [46]. Low encapsulation efficiency implies that more
carrier material is needed, heightening the risk of carrier material toxicity [2, 57]. Encapsulation
efficiency can be determined in a number of ways, depending on the particular type of NP. Drug loading
can be determined by weight [16]. Gel electrophoresis has been used to determine encapsulation
(LNPs). Physical chemistry techniques can also be used to identify and quantify the degree to which
components have been encapsulated. For example, fluorescence spectroscopy can be used if the drug is
fluorescent [16]. Ultraviolet-visible (UV-Vis) light spectroscopy can be used to determine the amount of
cargo (e.g. siRNA, anticancer drug doxorubicin (DOX)) present [58, 59]. High performance liquid
chromatography can be used to separate, identify, and quantify the components of a mixture [46, 60].
Novel single molecule measurements have been applied to measure encapsulation efficiency in individual
An array of physical chemistry techniques has been instrumental in evaluating the chemical and
physical properties of NP components. Dendrimer structures have been simulated based on experimental
data from nuclear magnetic resonance spectroscopy (NMR) and X-ray [62]. LNP matrix state,
polymorphism, and phase behavior have been characterized by differential scanning calorimetry (DSC),
X-ray diffraction (XRD), and neutron scattering [63]. Lipid crystallinity of solid LNPs has been studied
73
using X-ray scattering, DSC, NMR, and electron spin resonance spectroscopy (ESR) [64]. CNTs have
been characterized using XRD, UV-Vis spectroscopy, Fourier transform-infrared spectroscopy (FT-IR),
ESR, SEM, and energy dispersive X-ray diffraction [65]. Iron oxide NPs, which are magnetic, can be
studied using magnetometry, NMR relaxation dispersion profiles [65], and magnetic field flow
fractionation [66]. Some more common physical chemistry tools will be briefly described here.
characteristic of the molecular structures present. UV-Vis light is absorbed if it supplies the right amount
of energy to excite valence electrons, while infrared (IR) light is absorbed if it excites bond vibrations.
FT-IR is a more efficient variant of IR spectroscopy where all frequencies are sampled simultaneously. It
has been used to monitor cholesterol-induced changes in liposomal membrane structure [67] and to verify
conjugation of block copolymers onto iron oxide NPs [68]. UV-Vis spectroscopy has been used
extensively in NP characterization. For example it was used to confirm the presence of cleavable disulfide
bond in CNT-nucleic acid conjugates [69]; monitor the metal to ligand charge transfer band (5d(Pt11) to
π*(diimine)) in dendrimer synthesis [70]; measure dye and drug encapsulation in poly(lactic-co-glycolic
acid) polymeric carriers [71, 72] and study drug release kinetics of liposomal formulations [67].
NMR and ESR both rely on spin manipulations in the presence of an applied external magnetic
field. NMR monitors the nuclear spin of nuclei with non-zero nuclear spins, and ESR monitors the
electron spin of excited unpaired electrons [73]. NMR can provide a wealth of information on molecular
arrangements. NMR is sensitive to small changes in the local environment: it can be used to differentiate
between various layers up to the fourth generation in dendrimers [74], and between neighboring carbon
atoms in lipids [75]. 2D NMR techniques can be used to ascertain chemical connectivity (e.g. COSY,
HMQC, HSQC) and distances (e.g. NOESY and ROESY) between specific atoms [74, 76, 77]. Dynamic
information is also accessible to NMR. Pulsed field-gradient spin echo 1H NMR and diffusion-ordered
spectroscopy have been used to determine diffusion coefficients in aliphatic polyester and poly(propylene
74
31
imine) (PPI) dendrimers [74, 78]. P NMR has been used to determine the mobility of siRNA
encapsulated in LNPs [79]. NMR relaxation was used to monitor molecular tumbling activities [80] and
this information, in turn, was used to determine the location of a spin-labeled anticancer drug in micelles
formed by PEG modified with long alkyl groups and of paclitaxel in cyclodextrin (CD) vesicles [77, 80].
Most DDSs are invisible to ESR, requiring the addition of paramagnetic spin probes [81]. Lipophilic ESR
probes have been used as model lipophilic drugs to help determine where these drugs localized in solid
LNPs [35]. In gadolinium poly(amido amine) (PAMAM) dendrimer containing spin probes, ESR was
used to determine the location and concentration of the magnetic resonance imaging contrast agent
gadolinium [78]. ESR has also been used to prove that the terminal interbranch H-bonded groups preclude
backfolding in PAMAM dendrimers [78]. Since ESR can probe membrane fluidity, ESR has been used to
study interactions between polymeric micelles and lipid membranes, which are models for cell
membranes [68]. ESR can also give microviscosity and micropolarity information [81]. Viscoelastic
properties of dispersions can affect DDS interactions with the body. For example, capillary blockage can
occur with solid LNPs, which are less deformable than nano-emulsions [64].
Scattering techniques can be used to study the structure of colloidal systems. In scattering studies,
a monochromatic beam of light, neutrons, or X-ray is focused on the sample, and the intensity of the
scattered beam is measured. Light scattering was covered earlier in this section. X-ray scattering results
from variations in electron density and neutron scattering results from variations in the spatial distribution
of atomic nuclei [82]. The interference pattern formed by the scattered beam can be used to determine
characteristic lengths using Bragg’s Law. Larger angle measurements contain information for shorter
length scales: small and wide-angle X-ray scattering (SAXS and WAXS) are used to look at liquids and
solids with structures on the length scale of 1-200 nm and sub-nm, respectively. For example, SAXS is
used to monitor monolayer and bilayer repeat spacing, and hydrophobic thickness in lipid membranes,
while WAXS is used to determine chain packing and extent of lipid motion [83, 84]. SAXS has been used
75
to study NPs with cubosome structures [85, 86], to provide micelle shape, size and density of nonsteroidal
anti-inflammatory drug celecoxib-loaded protein micelles [50], and to observe in situ shell growth of
polymeric nanocapsules [87]. WAXS has been used to study lipid polymorphic transformations in solid
LNPs and crystallinity of encapsulated drugs [88], and crystallinity of polyelectrolyte complexes made
with polysaccharides [89]. SANS, useful for studying structures of 1-100 nm [90], has been used to
determine molecular weight and end group locations of dendrimers [74]. Neutron scattering can also be
used to study drug diffusion and internal molecular motions in LNPs [63].
excitation of fluorescent species. An extrinsic fluorescent probe is added when no fluorescent moieties are
present in the material of interest. Absorption of a photon excites the fluorophore and fluorescence occurs
when a photon is later emitted. Between excitation and emission, the fluorophore’s motions and local
environment (e.g. viscosity, polarity, temperature, pH, ion concentration) influence the fluorophore’s
spectrum [91]. Polarity sensitive fluorescent probes are commonly used to determine the critical
surfactant micelles [92]. Fluorescence quenching assays have been done to evaluate nucleic acid-
polycation binding in polycationic nucleic acid carriers [51] and dye leakage assays have been used to
Thermal analysis is a versatile tool for DDS characterization [94, 95]. DSC measures enthalpy
changes – the heat required for a sample to undergo a physical phase transition (e.g. glass transition,
melting, decomposition, isomerization or heats of solution, water sorption-desorption) [95, 96]. DSC was
used to monitor the physical state of the antineoplastic drug paclitaxel inside polymeric microspheres, for
example [46]. More specific to DDS design, DSC has been used to measure the gel-to-liquid crystalline
transition temperature of lipids in liposomal formulations, which can be used to predict release rates of
encapsulated drugs [95, 97]. In a closely related technique, ITC, the evolved heat is measured as
76
concentrated aliquots of one substance are added to a solution of a second substance. The released heat is
directly proportional to the amount of substance added, and can be used to measure enthalpy changes,
stoichiometry, and binding constants. ITC has been used to measure cyclodextran-guest molecule
NP DDSs are complex systems that can evolve over time. Dynamic processes cannot be captured
in steady-state measurements and can pose as a challenge to characterization [35]. For example,
PEGylated NPs are known to lose their PEG coating before reaching target cells [99-101]. Time-resolved
DLS combined with TEM was used to investigate morphology transition kinetics of multiblock
copolymer micelles [102]. Time-resolved fluorescence lifetime measurements have been used to follow
the release of fluorescent compounds from polymeric carriers [71], and fluorescence lifetime imaging
microscopy could extend drug release studies to cellular systems [103, 104].
Computer simulations in general describe the physics of materials at a suitable level of detail for a
particular application. A simulation is described by the level of detail in the physics used to model the
system of interest, and the algorithms used to generate enough simulation data to draw statistically valid
conclusions about the behavior of the system of interest. NPs can be described at different levels of detail
[Figure 3.3], but for most of the applications in this review the appropriate levels of detail are limited to
At the highest level of detail, a system is described by quantum mechanics in one of its
approximations. Quantum mechanical calculations have been widely applied to CNTs and play a role in
optimizing less detailed simulations, but in practice they are only useful for small systems with limited
degrees of freedom and generally not in solution. They are essential to correctly model electronic
77
properties and can be used to calculate electrical, optical, magnetic and mechanical properties of CNTs,
quantum dots (QDs), magnetic NPs and similar systems. For soft-condensed matter systems such as
rather than taking them into account explicitly, we significantly simplify calculations. At this level of
detail, atoms are described as point charges that interact through simplified empirical potential terms
including Coulomb interactions, Van der Waals interactions, and simplified terms describing bonds,
angles, and dihedrals as harmonic terms and cosine expansions [105]. The resulting technique is called
Molecular Dynamics (MD) and is widely used for biological and soft-condensed matter systems. The vast
majority of the papers reviewed below use this approach, which accounts for a substantial fraction of
super computer time on the worlds’ major academic computer centers. MD is a mature technique that is
implemented in a number of widely-used software packages, including GROMACS [106], NAMD [107],
LAMPPS [108], CHARMM [109], and AMBER [110]. MD simulations essentially generate a movie of a
set of molecules over time, incorporating every degree or nearly every degree of freedom, from which
through statistical mechanics thermodynamic properties can be calculated. Its main limitations include the
computational cost and the corresponding limited time scale (realistically, microseconds in most cases)
and length scale (~10 nm per dimension) and the limitations implicit in the physics that describes the
interactions between atoms: point charges mean electronic effects like polarization, pi-electron clouds
possibly important for CNTs, and charge transfer are typically ignored. In addition, the simplified
78
Figure 3.3: Categories of simulation methods and their respective spatio-temporal domains of
applicability.
The question of interest dictates the level of resolution required. Briefly, if electronic motion play an
important role in the property we are studying, quantum mechanics (QM) level is appropriate. However,
if there is no bond formation and cleavage, all atom (AA) level works well; we can safely ignore
electronic motion by assigning point charges to each atom. If electrostatic and hydrophobic interactions
are the dominant contributors, coarse grain models (CG) can be used: individual atoms can be ignored
and a group of atoms (e.g. 4 heavy atoms in MARTINI) can be treated as one interaction point/bead. CG
models allow the exploration of a larger area in phase space, at the expense of losing atomistic details.
Coarser levels of simulations (e.g. DPD and continuum models) are appropriate for systems and processes
79
For many properties of interest, especially in mesoscopic-scale systems like NPs with a size of
20-200 nm, atomistic detail may not always be necessary. Many papers below use coarse-grained (CG)
simulations, often based on the MARTINI model [111]. Here CG means that several atoms have been
grouped into interaction sites, which are no longer recognizable as atoms but instead are ‘beads’ that
represent molecular fragments. Depending on the level of coarse-graining, these beads can still maintain a
significant amount of chemical specificity, or they can represent very generic properties such as those that
represent a lipid by one head group bead and two tail beads. The same physics applies as in atomistic
algorithm to accurately calculate thermodynamic and structural properties. This can be done by MD,
Monte Carlo, or various other algorithms including Langevin dynamics or Dissipative Particle Dynamics
(DPD) [105].
At the CG level, individual molecules and usually fragments of molecules are still clearly
recognizable. In material science, there exists a whole hierarchy of additional models. Beyond molecules,
DPD solves hydrodynamics equations for materials by representing materials as discrete particles that can
be much larger than individual molecules. DPD blurs the line with CG MD a little by its choice of particle
volume: if the volume is chosen to coincide with molecules or molecular fragments its resolution is
similar to CG MD, although it typically treats problems that are more generic. Beyond DPD there are
many other possibilities, including field-based treatments where a system is described by density fields.
Simulations coarser than DPD are beyond the scope of this review.
80
3.5 Computational Studies on Drug and Gene Delivery Systems
Computer modeling, as complementary tools to experiments, is excellent in shedding light into the
structural and dynamical properties of systems of interest at atomistic or molecular levels of detail [112,
113]. Applied to DDSs, computer simulations have been used to address a broad range of questions
[Table 3-1]. Computer simulations can be used to study self-assembly [114-116], the structural and
dynamical characteristics of the resulting aggregates [117, 118], drug loading capacity, mechanism and
rate [119-121], drug distribution/localization in DDS [121], complex stability [118, 122], drug retention,
release mechanism and release rate [123-125], dominant drug-DDS interactions [119, 126, 127], and to
design or optimize DDSs targeting capabilities [128, 129]. Environmental conditions, e.g. pH,
temperature, salt type and concentration, counterions [126, 130], and external stimulus such as external
magnetic fields [131-133], as well as interactions with other biomolecules (e.g. serum proteins, miRNA,
heparin), all might affect the aforementioned aspects of DDSs [126, 134-137] and can be studied
computationally.
The way DDSs interact with cell membranes is one of the most commonly studied steps in drug
delivery by computational studies [138, 139]. Computer modeling can investigate the driving forces for
NP-membrane interactions, as well as how factors such as design parameters and environmental
conditions affect these. These parameters include size, shape, surface chemistry, and concentration of
NPs, as well as mechanical and elastic properties of both NPs and membranes [138-143]. Surface
chemistry refers to NPs hydrophobicity/hydrophilicity, charge density and distribution, as well as coating
ligand’s length, grafting density and distribution, rigidity, and their affinity to receptors [144-147]. Since,
in real systems the membrane often interacts with more than one NP simultaneously, NP aggregation and
interaction of multiple NPs with the membrane also is a relevant factor [138, 139]. Membrane properties
such as lipid phase, membrane composition, surface tension, charge density, receptor types and density
81
also play important roles [148-150]. In addition to all of these, external macromolecules, e.g. proteins in
the bloodstream, and different environmental conditions in different cell types [134], as well as
differences between external and internal cellular environments, are also of great importance and worth
further investigation. Of course, many of these factors are coupled and have to be taken into account
NPs are usually functionalized by coating with polymers, lipids, or ligands, to increase their
stability, targeting capability, cellular uptake efficiency, and also to reduce their toxicity [131]. Both the
type of coating molecule and the coating pattern and density strongly affect the physicochemical
properties of these NPs, and as a result their interactions with their cargo and target lipid membranes
[147, 151-155]. Simulations can be used to study these coatings, how they affect NP properties including
their interactions with other molecules [152, 156], and can be used to optimize these coatings towards
DDS with improved delivery capabilities [151, 154]. Functionalization with PEG polymers, a process
called PEGylation, is a good example of functionalization [157, 158]. Drugs/proteins/nucleic acids and
NPs’ surfaces are usually PEGylated to improve their water solubility, stability, and circulation lifetime,
and to reduce their aggregation, toxicity, and immune response [157, 159].
Considering the importance of PEG and its wide usage in drug delivery applications, PEG is
worth further discussion. PEG is the most commonly used non-ionic polymer with stealth behavior for
drug delivery purposes [160]. These biocompatible hydrophilic polymers can solubilize hydrophobic
molecules and reduce renal clearance and toxicity of drugs and nanoparticles. They also inhibit the
aggregation of PEGylated agents by steric stabilization, as well as mask therapeutic agents from the host
immune system and consequently suppress immunogenicity and antigenicity. This protective polymer
layer, which is known as stealth sheath, can reduce the fast recognition by the immune system and reduce
the non-specific interactions with blood components, resulting in longer blood circulation times. These
properties have made it the gold standard polymer for drug delivery and other biomedical applications
82
and have been the topic of several computational modeling studies reviewed in more detail below.
Despite the widespread use of PEG, there are several drawback to PEG including immunogenicity and
antigenicity, e.g. hypersensitivity reactions and the development of antibodies against PEGs [160], which
adversely affects pharmacokinetics [161], and the search for alternatives, e.g. poly(amino acids),
In addition to different aspects of DDS mechanism, different types of DDS have been simulated.
There are many types of nanoparticulate DDS. DDSs based on polymers, peptides, nucleic acids, lipids,
carbon, dendrimers and dendrons, and gold have been investigated via computer modeling and are
described in more detail below. Combined, these applications show that computer simulation has become
a powerful technique for rational design and optimization of DDS and demonstrates the growing
For the rest of this chapter, we will present examples of computational studies on each aspect of
interest for each nanoparticulate DDS type in separate sections, and then comment on challenges in this
field. We will focus on atomistic, CG MD simulations, and DPD simulations of nanoparticulate DDSs;
these fall mainly in categories defined as dendrimers/dendrons, polymer-, protein/peptide-, nucleic acid-,
lipid-, carbon-based DDS, as well as gold nanoparticles (AuNPs). We emphasize studies published since
2010. There are computational studies on other DDSs that are outside the scope of this review, e.g.
graphene oxides and reduced graphene oxides [19, 162], silicon NPs [19, 163], nanodiamond [164, 165],
layered drug delivery carriers [166, 167], zeolites [168-170], metal-organic frameworks [171, 172], and
other porous materials. These are excluded because they are not NPs, there are limited computational
studies available, or because their focus is different from the studies described below.
83
3.5.1 Dendrimers and dendrons.
Dendrimers are a class of branched globular materials with broad application in nanotechnology and
nanomedicine, including for drug/gene delivery [9-11]. Their monodispersity, modifiable surface, shell
and core characteristics, and their high loading capacity make them suitable carriers for therapeutic
agents, with promising applications in gene delivery through complex formation with nucleic acids, so-
overcome solubility problems and to improve targeting and cellular uptake. Dendrimers have been studied
in many computational papers to characterize their structure in solution, their interactions with drugs and
biomolecules [62, 175], and their response to external stimuli [178]. All parts of dendrimers can be
modified chemically; core, shell and surface. The surface can be optimized for different biological
activities by functionalizing the terminal groups. Thus, dendrimers are a flexible class of materials with
Different dendrimers have been modeled including PPI, triazines and PAMAM. Simulations can
be used to investigate their interactions with other molecules, the effect of solvents and counter-ions, pH,
salt concentration, different generation number, the nature of terminal groups and chemical modifications.
Computer simulations have also been used to study deformation and conformational changes of
dendrimers upon interaction with nanopores [179], lipid membranes [180], and nucleic acids [181] as a
function of generation number, pH, functionalization, and ionic strength. For example, Nandy et al. [181]
simulated a 38 base pair double stranded DNA and three different generations of PAMAM dendrimer
(G3, G4, G5). They found that DNA caused deformations in lower-generation dendrimers (G3) while
higher generation PAMAM (G5) bent DNA [Figure 3.4]. Generation number, pH, and architecture have
been shown to affect dendron’s flexibility and rigidity, and as a result influence their binding efficiency to
84
nucleic acids [182, 183]. Several studies have investigated the effect of chemical modification on self-
number.
a) For the G3 dendrimer, very little deformation in DNA was observed, while the dendrimer was bent
considerably by the DNA. b) For G4, both DNA and dendrimer were deformed. c) For G5, DNA was
deformed, but the dendrimer behaved almost as a rigid body. Adapted with permission from B. Nandy,
P.K. Maiti, A. Bunker, Force Biased Molecular Dynamics Simulation Study of Effect of Dendrimer
Generation on Interaction with DNA, (2013). Copyright 2013 American Chemical Society.
Self-assembly of Janus dendrimers, comprised of hydrophobic and hydrophilic parts, has been
studied both experimentally and computationally. They form onion-like dendrimers and other complex
85
Computer simulation has been used to study how chemical modification and surface
functionalization (e.g. PEGylation, acetylation, folic acid groups, peptides, and targeting ligands of the
surface) influences the structure [189], and the interactions with cargos [190, 191]. Computer modeling
can also be used to determine the optimal grafting densities and patterns based on structural and drug-
loading properties, and interactions with target membranes [154, 192-195]. Effect of PEGylation size and
grafting density on the structure of dendrimers, and also the structure of the PEG layer itself have been
widely studied [154, 196-198]. PEGylation expands the core of dendrimers, effectively reduces their
surface charge density and increases the overall size of dendrimers [196]. The effect of PEGylation on
structure and drug loading capacity of PAMAM-G4 dendrimers was investigated by an all-atom MD
simulation [154]. Different PEG densities (25%, 50%, 75% and 100%), were tested for complexation and
loading capacity for 5-fluorouracil (5-FU), as a model anticancer drug. The simulations suggested that
25% PEG is the most suitable PEGylation density for drug delivery purposes. This grafting density can
retain the internal drug complexation capability, and also assist in retaining drugs. For higher PEGylation
degrees than 25%, 5-FU molecules to a high extent interact with external PEG chains than dendrimer
branches. Pearson et al. [193] used MD simulation to understand why PEGylated dendron-based
copolymers functionalized with different terminal groups, -NH2, -COOH, and –Ac, do not exhibit a
charge-dependent cellular interaction. Simulations suggest that interactions between positively charged
terminal groups with PEG molecules sequester the charges and cause this low level of cellular interaction
of generation number (size), protonation state, functionalization, and dendrimer concentration, as well as
the effect of bilayer asymmetry, bilayer tension, lipid tail length, and lipid phase (temperature) all have
been studied by simulations [124, 180, 195, 199-204]. Tian and Ma [124] studied interactions of G4-
PAMAM dendrimers with both symmetric and asymmetric negatively charged bilayers, with and without
86
tension, at both neutral and low pH. It was shown that both membrane tension and electrostatic
interactions between charged dendrimers and tensed membrane play important role in dendrimer
penetration through membrane. Based on their simulation results authors proposed a mechanism of
endosomal escape for pH-responsive gene delivery vectors [Figure 3.5]. Xie et al. [204] found that low
generations leave gel-phase membranes intact while higher generations fluidize the membrane and cause
a gel-fluid phase transition around the dendrimer binding site. Higher generation cationic dendrimers are
capable of membrane disruption in both low and high temperatures [204]. Functionalization can influence
87
Figure 3.5: Proposed mechanism of endosomal escape for pH-responsive dendrimers.
In endosomes (pH ~ 5), dendrimers are protonated, leading to increasing osmotic pressure and dendrimer
swelling, putting the membrane under tension. Meanwhile, electrostatic interactions between charged
dendrimers and asymmetric negatively charged endosomal membrane cause a drop in the critical
membrane tension required for membrane disruption, allowing NP escape from endosomes. Bottom row:
membrane (with lipid tails represented with cyan beads, and lipid headgroups represented with green and
blue beads) at different pH levels. Reproduced from [124] with permission from The Royal Society of
Chemistry.
88
Dendrimers can condense DNA and siRNA, and stabilize small molecules in their core, shell and
interface. Several studies have considered the loading, release, interactions and distribution of small
molecules in dendrimers [174, 205-213]. For example, Jain et al. [206] used all atom MD simulations and
molecular mechanics Poisson-Boltzmann surface area analysis to study the solubility and drug release
profile of the hydrophobic drugs famotidine and indomethacin in G5 ethylenediamine cored PPI
dendrimers under different pH conditions. They found fast drug release in low pH, intermediate values in
neutral pH and a slow release in high pH. These results and further simulations on the effect of dendrimer
chemistry and topology on drug loading and release suggest that both the pKa of the drug and pH-
dependent changes in the dendrimer influence these dendrimer-drug interactions and complex stability.
Loading and release of the antibiotic rifampicin, from G4-PAMAM dendrimer has been studied by Bellini
et al. [205] in a combined experimental and simulation approach. Experimentally, approximately 20 drug
molecules were found as the maximum potential loading capacity of this dendrimer at neutral pH. In
simulation, the model of the dendrimer with 20 drug molecules in neutral pH was more stable than in low
pH, where drug molecules got expelled rapidly and simultaneously to the solvent. This suggests a pH
dependent drug release desirable for drug delivery applications. Simulations can also be used to study the
effect of PEG lipids on loading capacity of dendrimers, and on surface ligand’s targeting capability [154,
211, 214].
optimal as vector for gene delivery, based on PAMAM interactions with bilayers and siRNA
complexation [203, 215]. For gene delivery applications, generation number (size), flexibility, hydrogen
bonding capability, pH, ion concentration and salt type have been considered in simulations [183, 216-
219]. Electrostatic interactions and the entropy gain from counterion release upon binding have been
89
proposed as important driving forces determining the interactions between dendrimers and nucleic acids
As one final example, we would like to design drug specific nanocarriers by including optimal
drug-binding molecules in the structure. Shi et al. [220] used de novo design of dendrimers via
optimizing building blocks and including optimal drug-binding molecules with a final goal the design of a
drug-specific nanocarrier. They combined virtual screening with molecular docking and MD simulations
on DOX-dendrimer complexes. Using Rhein, a molecule with strong DOX binding, they achieved Rhein-
Polymeric NPs are of interest in drug delivery applications and more generally because of their stability
and easily modifiable surface [21, 31]. Block-copolymers [92] composed of hydrophobic and hydrophilic
parts can form a variety of self-assembled structures [221] of interest in drug delivery applications. These
structures, e.g. polymeric micelles [222], are capable of accumulating drug molecules in their core,
interface and shell [121]. There are several reviews of computational studies on different aspects or
specific type of polymers [23, 221, 223-225]. Here, we will discuss some examples.
Amphiphilic polymers form a variety of structures in solvents, e.g. core-shell, Janus particles,
micelles, and rod-like micelles [102, 226, 227]. Several factors affecting aggregate structures, stability
and phase transitions between different structures have been studied: for example, solvent polarity [130],
polymer architecture, concentration and physicochemical properties [226, 228, 229], mixture components
and mixing ratios of comprising polymers [230, 231], pH, temperature [226, 232], and the addition of
components, e.g. lipids, cholesterol and ligands [102, 233, 234]. Huynh et al. [122] performed all atom
MD simulations on star copolymers (SCPs) to understand why SCP micelles are unstable and form
multimolecular micelles in solution, and to assist in the rational design of stable SCP unimolecular
90
micelles in solution. Each SCP has a central connecting part and six attached arms, with each arm
comprised of a hydrophilic part (PEG) and a hydrophobic one (PCL). These SCPs were different in their
PCL and PEG length and the resulting molecular weight. They form core-shell structures, in which PCL
blocks form a dense hydrophobic core while PEG blocks form a shell around the PCL core. The
simulations suggested that partial water exposure of the PCL core drives aggregation into multimolecular
micelles. Increasing PEG length protects the PCL core from exposure to the water but it also increases the
size of micelles. Therefore, since smaller micelles are preferred in DDS applications, predicting the
minimum number of PEG units for a specific number of PCL blocks required to fully protect the PCL
core is important. The authors found a quantitative relation between hydration of PCL core and both
The internal structure of the thermoresponsive polymer blend NPs, and the phase transition
between core-shell and Janus structures were investigated by DPD simulations by Guo et al. [226]. Chen
and Ruckenstein [230] looked at the structure and mechanism of formation and the degradation process of
multicomponent multicore micelles via DPD simulations. Taresco et al. [231] used a combined
formed by two different types of monomers. Wang et al. [233] studied the structural characterization of
and DPD simulations showed that introducing a second hydrophilic phosphatidylcholine group into the
polymer chains causes a phase transition from sphere to rod-like in multiblock polyurethane micelles.
Hybrid systems composed of polymers and other NPs, e.g. AuNPs and QDs, also show
interesting structural and phase transition behaviour [227, 235]. Using both experiments and DPD
simulations, Cai et al. [227] studied the effect of adding AuNPs on block copolymer self-assembly.
Adding AuNPs caused a transformation from long cylindrical micelles to short cylindrical micelles and
91
Interactions with and translocation through cell membranes for polymer-based DDSs have been
studied both by experiment and simulations [236-238]. Li et al. [236] studied the interactions between a
novel amphiphilic polymer, PMAL, and siRNAs, combining experiment and CG MD simulations.
Potential of mean force calculations showed that siRNA by itself experiences a large free energy barrier
for translocation while the siRNA-PMAL complex entered the membrane spontaneously, consistent with
experiment. The PMAL polymers induced pore formation to facilitate siRNA translocation. Srinivas et al.
[237] used CG MD simulation and showed that these patchy polymeric micelles are capable of
accommodating and transporting hydrophilic contents across a lipid membrane into a lipid vesicle.
Ding and Ma [238], using DPD simulations, investigated receptor-mediated endocytosis for a
new type of pH-responsive DDS composed of NP (radius of 4 nm) and some pH-sensitive polymers (of
twelve beads length). Beads on the NP surface were treated as ligands and assigned +e charges. The
polymer had N randomly distributed negatively charged (-e) beads, where N depends on the pKa of the
polymer and the system’s pH. Half of the model lipids in the membrane were treated as receptors by
replacing charged beads in the lipid’s head group with neutral beads. They introduced a modified
three different pH conditions combined with potential of mean force calculations showed a triple-pH-
response. The NP can only be engulfed by cell membrane when the pH is higher or lower than the
polymer’s pKa, whereas endocytosis is blocked when the pH equals the polymer’s pKa. It was also found
that the properties of NPs, pH-sensitive polymer and cell membranes, and the external environment, all
affect NP engulfment by the membrane, at least in this simplified DPD model [Figure 3.6].
92
Figure 3.6: Time evolution of nanoparticle-polymer complex endocytosis as a function of pH.
absorbed on its surface) depends on the pH. For pH values lower (a) and higher (c) than the polymer’s
pKa, the nanoparticle can be fully engulfed by the membrane. However, for pH = pKa (b), endocytosis is
blocked. Membrane lipid head groups are shown in green, purple, and blue correspond to +e, -e, and
neutral beads, respectively. Lipid tail are in orange, receptor heads are in red, and nanoparticle beads are
in yellow. Polymer beads are in cyan (-e) and pink. Reprinted by permission from Macmillan Publishers
Ltd: Scientific reports H. Ding, Y. Ma, Controlling Cellular Uptake of Nanoparticles with pH-Sensitive
The effect of functionalization on interactions with the membrane, uptake mechanism, as well as
effect on the structure and conformation of drugs/polymer complexes have been studied by simulation
[118, 239-243]. Liao et al. [240] used both experiment and all atom MD simulations to study the uptake
93
of chitosan/DNA complexes coated with anionic poly(gamma-glutamic acid) (gamma-PGA). This coating
enhanced the cellular uptake of chitosan/DNA complexes via a specific protein-mediated endocytosis.
Both the structure of this ternary complex (CS/DNA/PGA), as well as the interaction between gamma-
PGA and glutamyl transpeptidase proteins were studied by MD simulations. Sun et al. [239, 244] studied
the effect of polyethylenimine polymer (PEI) modification by lipids on nucleic acid compaction and
aggregation. They found that lipid association as an additional mechanism of aggregation resulted in more
Several studies have focused on interactions between polymeric DDS and small molecule cargos,
investigating driving forces for partitioning [60, 245-249], complex stability [250], drug loading capacity
and rate, and drug distribution and release [121, 247, 251-255] as a function of temperature [256], pH
[228, 232, 257, 258], concentration of drugs [232], physicochemical and structural properties of polymers
and cargos [259, 260]. CD is a well-known family of cyclic oligosaccharides with great promise as
solubilizing agent for poorly soluble drugs [246, 261]. He et al. [246] performed MD simulations to
investigate the binding mode and affinities of the antifungal drug amphotericin B drug with γ- and β-CD.
Consistent with experiments, simulations showed a significantly higher binding affinity for γ-CD than β-
CD. The binding mechanism of the drugs bexarotene and human vasoactive intestinal peptide (VIP) to
highly PEGylated sterically stabilized micelles (SSMs) has been investigated by Vukovic et al. [121] both
experimentally and computationally. Using free energy profiles, they predicted the drug distribution in
SSMs. Single bexarotene, as a poorly water-soluble drug, resides in the ionic interface with its polar end
exposed to solvent. With multiple bexarotene molecules, the preferred distribution is as clusters in the
alkane core of SSM [Figure 3.7]. Dominated by electrostatic interactions, VIP molecules, with two
regions with positively charged residues, interact with negatively charged -PO4 groups in the ionic
interface. These simulations suggest the importance of the balance between hydrophobic and Coulomb
94
Figure 3.7: Solubilization of bexarotene in sterically stabilized micelles (SSM).
Left) Free energy profiles for systems composed of 1, 3, and 5 bexarotenes in SSMs. SSMs are either
composed of 10 (small) or 90 (large) monomers, in water and 0.16 M NaCl, respectively. Single drugs
prefer the ionic interface, represented with vertical arrows, in both SSMs. For multiple drugs accumulated
in the SSM core (shown as a gold surface), a deeper minimum develops in the core. This is because
several bexarotene (e.g. 5) cluster together (Right) via a hydrogen bond network between their –COOH
groups. Reprinted (adapted) with permission from L. Vukovic, A. Madriaga, A. Kuzmis, A. Banerjee, A.
Tang, K. Tao, et al., Solubilization of Therapeutic Agents in Micellar Nanomedicines, (2013). Copyright
The interaction and complexation of nucleic acids, including DNA and siRNA with cationic
polymers has been studied extensively [223]. PEI, as an example of cationic polymers, is one of the most
promising polymers which has been widely utilized in studies for gene delivery applications. Both
experimental and computational studies have provided a better understanding of PEIs’ binding modes
95
with DNA/siRNA [223]. Molecular modeling has had a significant impact on our understanding of
critical steps in gene delivery, including complexation of carriers with nucleic acids, as well as nucleic
acid condensation and aggregation by gene carriers [177]. Simulations can be used to investigate the
factors determining interactions between cationic polymers and nucleic acids, including complexation and
decomplexation. Such factors, similar to those in other DDSs, include chemical surface composition as
amine-to-phosphate ratio [51, 262], charge density, protonation state, polymer molecular weight, polymer
chemical structure and charge distribution [262-267], the effect of endogenous molecules [136], and
multivalent ions, e.g. Fe(III) [137, 268]. miRNA as endogenous molecule might cause decomplexation.
To test this hypothesis, Meneksedag-Erol et al. [136] used both experiment and all atom MD to study the
interactions between miRNA and pre-formed PEI-siRNA complexes. PEI was found to bind more
strongly to miRNA than siRNA. However, miRNA could not disrupt the integrity of PEI-siRNA but
Interactions between polymeric DDS and nucleic acids for gene delivery purposes can be studied
indirectly using polymers. Polycations and polyanions can be considered as models for DDS and nucleic
acids, respectively [269-271]. Zhao et al. [271] utilized polycations and polyanions and studied the effect
of charge distribution along the chain on complexation behavior and the structure of the resulting
complexes. They found that charge distribution has a significant influence on aggregation and the
complex structure.
There are several peptide-based DDSs, e.g. protein- [272] and cyclic peptide-based [273, 274], and cell
penetrating peptides (CPPs) [275]. There are also DDSs based on nucleic acids. RNA and DNA have
been used as either building blocks or surface ligands to design a variety of three-dimensional
nanostructures, including rings, tubes, cubes, cages, and spheres [276-279]. As with other DDSs, the
96
structural properties of these systems, and their interactions with endogenous biological molecules, as a
function of environmental conditions are of interest [280-282]. Badu et al. [281] performed atomistic MD
simulations on RNA nanotubes and investigated their structural properties. Juul et al. [282] studied DNA
cages loaded with active enzymes both experimentally and by all-atom MD simulation and found a
97
Figure 3.8: DNA nanocage with temperature-controlled conformational transition capability.
The cage is closed at 4 C (a), and in a more open conformation at 37 C (b). Heating the nanocage to
37 C in presence of horseradish peroxidase (HRP), followed by cooling to 4 C, allows HRP enzymes to
be encapsulated in nanocages (c). Reheating to 37 C causes enzyme (orange) release (not shown).
Reprinted with permission from S. Juul, F. Iacovelli, M. Falconi, S.L. Kragh, B. Christensen, R. Frøhlich,
et al., Encapsulation and Release of an Active Enzyme in the Cavity of a Self-Assembled DNA
biocompatibility, biodegradability, low cytotoxicity, and non-antigenicity [283, 284]. Some proteins, e.g.
albumin, can be used for targeted delivery to tumor cells and inflamed tissues since they are preferentially
taken up by these types of cells [284]. Luo et al. [272] using both theory and experimentations
investigated the binding mode, affinity and interactions between bovine serum albumin (BSA) and
98
interferon alpha-1b (IFN) as delivery system and protein drug, respectively. Simulations predicted domain
III of BSA as the most probable binding sites, as well as hydrogen bonds and salt bridges as the main
contributors in binding between BSA and IFN. Cyclic peptide-based nanotubes (CPNTs) can be used as a
transporter for ions and small molecules [273, 274]. Vijayaraj et al. [273] investigated the transport
mechanism of 5-FU through a variety of CPNTs and determined the driving forces and free energy
barriers affecting this transport. They found that transport was driven by direct or water-mediated
CPPs are of interest because of their intrinsic ability to enter cells, their low cytotoxicity, and
their versatility. They can facilitate the transport of small molecules (e.g. drugs, imaging agents),
macromolecules (e.g. DNA, siRNA, proteins, peptides) and NPs/DDS. Therefore, CPPs can be used as
drug/gene delivery systems, as well as carriers for imaging agents, proteins and peptides [275]. Recently,
it has been of great interest to combine the benefits of nanomaterials and CPPs. CPPs can be
functionalized with other delivery systems, like QDs, liposomes, and dendrimers, potentially improving
CPPs intracellular drug release ability and decreasing their toxicity. DDSs such as polymeric micelles,
dendrimers, liposomes, QDs, and inorganic nanocarriers (e.g. gold-, silver-, and iron-based NPs) can be
functionalized with CPPs to enable their cellular uptake [275]. CPPs' unique physicochemical properties,
their uptake mechanisms, classifications, and applications in medicine and biotechnology have recently
Key elements involved in CPPs' roles in drug delivery have been studied computationally [285-
288]. Atomistic MD simulations revealed the importance of arginine, lysine and tryptophan in the
interaction between two CPPs, penetratin [285] and transportan [286], and
conjugation can help translocate hydrophobic and hydrophilic drugs across lipid membranes [289, 290].
Teixeira et al. [290], via both MD simulations and experiments, found that lysine-based surfactants
99
enhanced the transdermal permeation of two topically administered hydrophilic drugs, tetracaine and
ropivacaine hydrochloride. As mentioned earlier, NPs can be functionalized with CPPs. The
concentration and distribution of CPPs on the NP's surface influence the conformation of the peptide
layer, which in turn affects the CPP-functionalized NPs’ activity and the efficiency of cellular
internalization. So, designing a DDS of desired activity requires knowledge of the structure and dynamics
of CPPs on NPs in solution prior to interacting with membranes. Todorova et al. [291] using both
experimental and atomistic MD simulations investigated the effect of the grafting density and surface
They found a correlation between the surface properties, e.g. positive charge distribution, of these TAT
New DDS design strategies combine the benefits of several systems such as block copolymers
and CPPs [292], or mimic natural systems such as intrinsically disordered proteins (IDPs) [293]. Sanchez-
Sanchez et al. [293], using both experiments and MD simulations, designed and characterized single
chain NPs called artificial IDP mimetics. These IDP mimetics are capable of simultaneous release of two
dermal bioactive molecules, folic acid and hinokitiol, into aqueous solution in a pH-dependent manner.
Computer simulations can provide insight into drug loading, distribution/complex structure and
release from peptide self-assemblies [294, 295], as well as the influence of pH and temperature [282, 294,
296] on these processes. Guo et al., using both experiments and DPD simulations investigated the effect
Arginine10) peptides conjugated with cholesterol, either loaded or not with DOX. Micelles had more
compact structure at pH > 6 [Figure 3.9]. For pH < 6, channels in swollen micelles might cause drugs to
diffuse out of micelles. This was verified experimentally as the DOX release rate was influenced by pH
100
Figure 3.9: Effect of pH on microstructure of self-assembled micelles.
Starting from a homogeneous state (a), cholesterol conjugated peptides form compact core/shell micelles
at pH > 6.0 (b). Decreasing the pH loosens these micelles (c) and the swelling facilitates DOX release.
Arginine, histidine, and cholesterol, are shown in green, brawn, and black, respectively. Water has not
been shown for clarity. Reprinted with permission from X.D. Guo, L.J. Zhang, Z.M. Wu, Y. Qian,
Dissipative Particle Dynamics Studies on Microstructure of pH-Sensitive Micelles for Sustained Drug
Finally, physicochemical properties of peptides and solvent conditions can govern self-assembly
and structural properties of peptide-based carriers. Computer simulation can be used to design peptides
which can self-assemble, and to study the mechanism of self-assembly into nanostructures.
Computational approaches can also be used to study how factors such as solvent and temperature
101
3.5.4 Carbon-based delivery systems.
Several different types of carbon-based nanoparticles (CNPs) have the potential to act as DDS [12, 302].
Here we focus on DDS based on CNTs and fullerenes but DDS using graphene, graphene oxide, and
Computational research has played a major role in exploring drug loading of [164, 303] and
release [132, 133, 304, 305] from CNPs, as well as interactions of cargos with carbon nanocarriers [306]
as a function of internal and external stimulus, e.g. pH, temperature, torsion, and external magnetic field.
CNTs are capable of loading therapeutic agents on their surface or inside their cylindrical hollow [19,
307]. Saikia et al. [305] performed MD simulation to study the C60 fullerene-mediated release of
multiple pyrazinamide molecules from the interior of a single walled CNT as a function of temperature.
Chaban et al. [304] investigated the effect of temperature and concentration on drug release from CNTs
using MD simulations. The effect of external magnetic fields has been studied on hybrid systems
Understanding the interactions of CNTs with biological macromolecules is also of great interest
[308-313]. Wu et al. [313] studied DNA ejection from CNTs as a function of temperature, torsion loading
and CNT size by MD simulation. Santosh et al. [309] studied the binding of siRNA and DNA to the
surface of single walled CNTs via MD simulations. Chen et al. [311] studied the dynamic mechanism of
encapsulation of HIV replication inhibitor peptide into CNTs using MD simulation. It is also important to
understand the interactions of carbon-based DDS with human serum albumin (HSA), the most abundant
protein in blood. Interaction of DDS with HSA is crucial in DDS absorption, distribution and metabolism
[314].
The membrane associations of carbon-based DDS have been studied by computer simulation
[315-317]. Several factors affect these associations: DDS structural properties [318-321]; clustering [320-
322]; functionalization [319, 323, 324]; concentration [319] and lipid membrane properties [321, 325].
102
The effects of size and clustering of fullerene NPs, and fullerene concentration on a DPPC monolayer as a
model for pulmonary surfactant was investigated by CG MD simulation by Chiu et al. [321]. Free energy
calculations suggest that all fullerene systems in this study (C60, C180, C540, and cluster of five C60
fullerenes) spontaneously diffuse into the hydrophobic region of both monolayer and bilayer membranes.
Large fullerene molecules were found to prefer partitioning into bilayers rather than monolayers,
however, which can influence the monolayer-to-bilayer transition in the respiratory cycle. This may
The effects of using polymers, peptides, and lipids as surface functionalization agents in carbon-
based DDS were studied by computational methods. Chehel Amirani and Tang [308] comprehensively
reviewed studies, mainly at the quantum mechanical level, of graphene and CNT functionalization, and
their binding energies with nucleobases. Lai and Barnard [131] have reviewed recent studies on
functionalized nanodiamonds and their importance for biomedical applications. Functionalization can
prevent NPs from aggregating [323, 326], increase their stability [327], and influence their interactions
with lipid membranes [323, 326, 328]. Surface functionalization can also promote drug delivery through
improving translocation across lipid bilayers [328]. Sridhar et al. [328] used CG MD simulations to look
at effects on fullerene (C60) translocation through the membrane due to polar and nonpolar
functionalization, temperature, and fullerene concentration. Although none of their models showed
complete translocation, they found that Janus particles having half the surface modified by polar groups
were the most promising form of functionalized fullerenes in terms of bilayer translocation.
A NP's behavior depends on the conformation of the functionalization layer on its surface. The
coating mechanism used to functionalize the NP, e.g. covalent vs. non-covalent attachment, as well as the
grafting density are two factors which can affect this conformation [329, 330]. For example, Lee [330]
found, using CG MD simulation, that the PEGylation method can influence PEG distribution on CNTs.
They also found that PEG size and grafting density affected the PEG layer's conformation [Figure 3.10].
103
Designing functionalized NPs with desired properties will require a detailed understanding of these
factors.
Figure 3.10: Effect of PEGylation method, and PEG chain size and grafting density on the
Non-covalently modified single walled CNT (SWNT) (Left) is more exposed to water than covalently
modified SWNT (Right). PEG (red) size and grafting density also influence the conformation of PEG
layer on SWNT (grey cylinder). RF is the mushroom radius and L is the thickness of brush state. SWNT is
shown as grey cylinders. Reprinted with permission from H. Lee, Molecular Dynamics Studies of
PEGylated Single-Walled Carbon Nanotubes: The Effect of PEG Size and Grafting Density, (2013).
104
3.5.5 Lipid-based delivery systems.
Recently, lipid-based DDS such as liposomes, micelles and LNPs have attracted much attention [7, 8,
331, 332] due to their ability to encapsulate and transport drugs as well as biomolecules, the versatility of
their structures and compositions, and their inherent selectivity to tumor cells and inflammation sites.
There are currently more than ten approved lipid-based drug formulations and many more in clinical trials
[7, 8]. Specifically, sterically stabilized liposomes, i.e. PEGylated liposomes, are widely used as DDS.
PEGylation achieves long circulation times for liposomes in vivo due to the formation of a protective PEG
layer over the liposomal surface. This layer sterically prevents the coating of liposomes by opsonins, thus
reducing drug uptake by cells of the immune system [333]. However, the mechanism through which
PEGylated lipids interact with different components in the liposomes, its function in drug distribution,
localization and retention as well as the influence of external environmental factors such as salt
Since atomistic simulations of full liposomes are computationally expensive, a common approach
is to infer results on liposomes from lipid bilayer simulations. Dzieciuch et al., [334] for instance,
performed atomistic MD simulation to study the effect of liposome PEGylation on their drug-loading
efficiency. They compared the effect of zwitterionic and PEGylated membranes on the location and
THPP). p-THPP enters both types of lipid bilayers, which agreed with experimental results, but in
PEGylated liposomes p-THPP also localizes to the outer PEG corona where porphyrins are wrapped by
PEG chains. Thus, in PEGylated liposomes p-THPP showed a greater exposure to the water than in
zwitterionic liposomes. These results highlight that PEGylation enhances the drug-loading efficiency of
membranes and support fluorescence experiments carried out in this study [334]. Simulations also
predicted strong interaction between PEG lipids and porphyrin [334]. This interaction is particularly
important in drug delivery studies as it may be an extra barrier to drug release. This finding supports
105
previous computational studies that showed that PEG interacts with hydrophobic molecules [127, 335,
336]. Similarly, MD simulations combined with free energy calculations were used to elucidate binding
mechanisms and divulge the exact location and organization of two small drugs, bexarotene and human
PEGs strongly interact with salts. This property has been utilized in lithium ion batteries with
PEGs as polymer electrolytes [337, 338]. PEGs are soluble in a wide variety of both polar and nonpolar
solvents [339]. Although both of these general properties have been shown previously by computational
studies [337, 338, 340] and experiments [341, 342], for DDS applications it is important to understand
how the PEG layer on DDS, e.g. PEGylated liposomes, interacts with salts in physiological conditions
and how these interactions affect the structure of the PEG layer [118, 343, 344]. Stepniewski et al. [344]
modeled the effect of NaCl on the surface structure of PEGylated liposomes. They varied the salt
concentration in Langmuir monolayer film experiments and added ions at the physiological level in
atomistic MD simulation studies. The results showed that the PEG surface layer should not be treated as
generic hydrophilic molecules completely outside the bilayer, nor should it be considered as totally
neutral as was previously accepted. PEG molecules are able to penetrate into the liquid-crystalline lipid
bilayer, which may affect the permeability and structure of the membrane. It was also noted that Na + ions
bind to PEG chains, changing the surface charge of the liposome and enhance opsonization, whereas Ca 2+
did not interact with PEG [343]. These results are not surprising, however. In fact, when PEGs are
attached on DDS surface, e.g. liposomes, they seem to show similar properties as previously reported for
other applications [337, 338, 340-342]. As another example, Pannuzo et al. [345] performed CG
simulations to study the effect of Ca2+ and PEG molecules on membrane fusion. They showed that for a
rapid fusion between negatively charged apposed membranes both Ca2+ and PEG were required.
An important property governing drug release from liposomal DDS is bilayer permeability [346,
347]. Magarkar et al. [346] performed atomistic simulations using the OPLS-AA force field to investigate
106
the high permeability observed for "inverse-phosphatidylcholine" (CPe) liposomes in experimental
studies [348]. CPe is a synthetic analog of phosphatidylcholine (PC) with a reversed zwitterionic
headgroup and has been proposed for use in liposomal DDS formulations. The larger surface area
observed for the CPe bilayer compared with the PC bilayer may cause CPe's higher permeability. Adding
cholesterol to liposomal formulations reduces leakage from liposomes due to its well-known role in lipid
packing and stability, but the mechanism for this is not understood. Magarkar et al. [336], by performing
MD simulation using OPLS force field, proposed a model to explain the interaction between PEG
molecules and cholesterol. PEG molecules tend to interact with the ß side of cholesterol, disrupting the
membrane structure. Contrary to the expectation that cholesterol should make the membranes more stable
and compact, adding cholesterol in the presence of PEG caused membrane destabilization. This result
may explain possible effects of cholesterol on the permeability and compressibility of PEGylated
liposomes.
understanding provides a framework for developing new nanoscale materials with desirable properties
[116]. CG and DPD simulations have been performed for the investigation of lipid-based aggregates
relevant for drug delivery [115, 349-352]. For example, Lee and Pastor [349] used the Martini CG force
field to study mixtures of lipids and PEGylated lipids in water at different sizes and concentrations of
PEGylated lipids. They found that the mixtures self-assembled to liposomes, bicelles and micelles. The
analyses and simulations indicated that the average aggregate sizes decreased when the concentration of
PEGylated lipid increased, in agreement with experimental results. Janke et al. [352] used CG MD
simulations to study the phase behavior of oleic acid aggregation at various concentration and protonation
states. They observed a range of structures including micelles, vesicles and oil phases depending on the
107
Simulations can be used to calculate drug release rates from lipid-based DDS [353, 354]. For
instance, the release of encapsulated materials from systems including emulsions (100% liquid lipid
phase) and nanostructured lipid carriers, which have a combination of solid and liquid lipid domains, was
investigated by Dan using Monte Carlo simulations [353]. The results obtained suggest that the size of
lipid domains does not significantly affect the rate of release, but the location and distribution of the solid
domains have a notable impact. When solid domains are concentrated near the solution interface,
transport is inhibited due to a reduction in the accessible surface area. However, when the solid domain is
located in the center of the LNP the release rate is increased and may even be higher than in the
corresponding emulsion particle. In another study by Dan [354], Monte Carlo simulations were performed
There are also other interesting studies, three of which are mentioned briefly here. Computer
simulations were combined with experiments to investigate the effect of penetration and membrane
behavior of three terpenes, which are effective chemical permeation enhancers in nanostructured lipid
carriers [355]. A detailed model of the distribution of drug molecules inside a liposome was obtained
using the Martini force field [356]. Jämbeck et al. [356] performed large scale simulations of hypericin, a
photosensitizing drug, in a DPPC liposome [Figure 3.11] and calculated the distribution and orientation
of the drugs in the lipid bilayer. Hoof et al. [357] studied the encapsulation of proteins in spontaneously
forming vesicles using MD simulation with a CG force field. They showed that the interactions of
proteins with membranes govern the encapsulation efficiency, but the size of the encapsulated proteins
and the speed of the vesicle formation did not seem to have a significant effect.
108
Figure 3.11: Distribution and orientation of hypericin within liposome.
a) Snapshots from MD simulations taken after 10 μs. Phosphorus groups (orange), choline groups (blue)
and hydrophobic tails (green) of the lipids are represented, with different numbers of drug molecules
(yellow). b) Liposomes with some lipids removed to show the binding of hypericin to DPPC; the
hydrophilic parts of hypericin are shown in red and the hydrophobic parts in yellow. Reprinted with
permission from J.P.M. Jämbeck, E.S.E. Eriksson, A. Laaksonen, A.P. Lyubartsev, L.A. Eriksson,
Validation and Simulations with a Coarse-Grained Model, (2013). Copyright 2013 American Chemical
Society.
109
A promising gene therapy approach is based on siRNA. Sophisticated delivery systems are
required to protect and deliver siRNA effectively to the target tissues [358, 359]. CG molecular
simulations together with a variety of experimental techniques were able to give a more detailed view of
the structure of a LNP containing siRNA [Figure 3.12] [79, 360]. Simulations of small systems
containing 8 duplex 12 base pair DNA complexes, DLinKC2-DMA (an ionizable cationic lipid), DSPC,
cholesterol, a PEG-lipid, water and ion molecules were performed. These molecules self-assembled and
organized in an ordered structure where, for instance, the PEG-lipids were distributed in the outer layer of
the LNP. This was the first study to show that LNP siRNA systems likely have a nanostructured core,
with the encapsulated siRNA located in internalized inverted micelles complexed to cationic lipids. The
in silico finding agreed with 31P NMR, FRET, cryo-TEM, and RNase digestion data and it was possible to
understand the mechanism whereby LNP siRNA systems are formed. This insight is being applied to the
design and construction of new LNP systems. In addition, both CG and atomistic computer simulation
have been used to study the DNA adsorption onto anionic lipid membranes induced by multivalent
cations [361].
110
Figure 3.12: Lipid nanoparticle.
External and a cross-sectional view of the LNP (top). Components are represented by their molecular
densities: protective PEG-lipid (magenta), therapeutic siRNA (red), cholesterol (green), DSPC (yellow),
ionizable cationic lipid (DLin-KC2-DMA) (blue) and water (cyan). Adapted from D. Rozmanov, S.
Baoukina, D.P. Tieleman, Density Based Visualization for Molecular Simulation., Faraday Discuss. 169
(2014) 225–43. doi:10.1039/c3fd00124e. under a Creative Commons Attribution 3.0 Unported License.
3.5.6 Gold-nanoparticles.
AuNPs are of particular importance in drug delivery research due to their unique properties including
their stability, ease of synthesis, and the ability to manufacture a range of sizes [15, 362]. They can be
loaded with various therapeutics including small molecules, peptides, proteins, and nucleic acids [19],
either conjugated to AuNPs covalently or noncovalently. There are several computational studies on the
properties of AuNPs and their interactions with other molecules [135, 227, 363-370].
111
The effect of AuNP conjugation on peptide conformational flexibility and structure was studied
by Lee and Ytreberg [366]. They examined the structure and dynamics of six peptides that were either
free or conjugated to AuNPs in water. Conjugation affected both structure and dynamics in an amino acid
sequence dependent way. Peptides with little or no secondary structure in solution were adsorbed on the
AuNP surface. This causes peptides to lose their specific interactions with cell components. For drug
delivery purposes, this suggests that peptides with significant secondary structures in solution are suitable
candidates for peptide-NP conjugation. Interactions between ubiquitin and AuNPs have also been studied
Van Lehn and co-workers performed a series of computational studies on AuNPs coated by
lipids. Among other properties, they investigated the structure of the monolayer coating the AuNP under
different conditions [369], lipid composition and AuNP size [368], the interactions of a monolayer-coated
AuNP with model membranes relevant for uptake [367], suggesting similarities in uptake process with
We have reviewed common biophysical and computational approaches to studying DDSs, focusing in
particular on computational studies of several major classes of NPs used in drug delivery research. The
strength of computer simulations in general is their ability to give very detailed insight into the structure
of NPs and their interactions with drugs and model membranes under a range of conditions. Clearly, the
efficiency of DDSs depends on a large number of other parameters, including stability in the bloodstream,
targeting to the desired tissues, uptake by endocytosis or other mechanisms, and final release of the drugs.
Although experimental biophysical and computational studies can characterize important aspects of
DDSs, a major challenge for the near future is to go beyond individual drug carriers and to investigate at a
mechanistic level the larger-scale processes that determine the success of DDSs.
112
At the computational level, currently we are limited by some of the standard limitations of
biomolecular simulation. In general, these involve limited sampling time, limited length scale, and
expensive or difficult access to important thermodynamic variables. Simulations can access time scales of
the order of microseconds or for coarser models tens or hundreds of microseconds, at length scales of ca.
10 × 10 nm for atomistic and up to 150 × 150 nm for CG simulations. Thus, it is now possible to simulate
entire nanocarriers, which will open up new areas of study for simulations. However, the process of NP
formation by e.g. microfluidics remains outside reach of direct detailed simulations. Questions regarding
drug loading, the stability of complexes, and the interactions of NPs with membranes essentially involve
free energies of binding, partitioning, and membrane pore formation. These can all be studied by free
energy methods, but the current state of the literature remains somewhat qualitative. This is an obvious
area for future work, which is already feasible with current methods and computational resources. Some
of the most interesting simulations currently available use CG MD or DPD. This is an exciting
development, but also comes with its own limitations; in particular, CG models need to be carefully tested
for their ability to reproduce the effect of important factors in DDSs such as salt concentration, the
distribution of chemical functional groups over the surface of a DDS, and the effect of pH.
DDSs are an interesting example of multi-scale computational problems, with a growing amount
of data from biophysical characterization. They have an obvious biomedical and biotechnological
importance, and bridge chemistry, nanoscience, materials, basic biology and medical applications in a
unique way. Although there already is a large body of literature as reviewed in this paper, in our
assessment the impact of computational work in this important area is likely to grow significantly in the
near future as model systems become more realistic and more tightly coupled to experiment.
113
3.7 Bibliography
[1]. Blanco, E., Shen, H., & Ferrari, M. (2015). Principles of nanoparticle design for overcoming
[2]. Allen, T. M., & Cullis, P. R. (2004). Drug delivery systems: Entering the mainstream. Science,
303(5665), 1818-1822.
[3]. Parveen, S., Misra, R., & Sahoo, S. K. (2012). Nanoparticles: A boon to drug delivery,
8(2), 147–166.
[4]. Kohane, D. S. (2007). Microparticles and nanoparticles for drug delivery. Biotechnology and
[5]. Venditto, V. J., & Szoka Jr, F. C. (2013). Cancer nanomedicines: So many papers and so few
[6]. Zhang, Y., Chan, H. F., & Leong, K. W. (2013). Advanced Materials and processing for drug
delivery: The past and the future. Advanced Drug Delivery Reviews, 65(1), 104-120.
[7]. Pattni, B. S., Chupin, V. V., & Torchilin, V. P. (2015). New developments in liposomal drug
[8]. Allen, T. M., & Cullis, P. R. (2013). Liposomal drug delivery systems: From concept to clinical
[9]. Gillies, E. R., & Frechet, J. M. (2005). Dendrimers and dendritic polymers in drug delivery. Drug
[10]. Lee, C. C., MacKay, J. A., Fréchet, J. M., & Szoka, F. C. (2005). Designing dendrimers for
[11]. Svenson, S., & Tomalia, D. A. (2005). Dendrimers in biomedical applications-reflections on the
114
[12]. Bianco, A., Kostarelos, K., & Prato, M. (2005). Applications of carbon nanotubes in drug
[13]. Lacerda, L., Bianco, A., Prato, M., & Kostarelos, K. (2006). Carbon nanotubes as nanomedicines:
[14]. Bianco, A., Kostarelos, K., Partidos, C. D., & Prato, M. (2005). Biomedical applications of
[15]. Ghosh, P., Han, G., De, M., Kim, C. K., & Rotello, V. M. (2008). Gold nanoparticles in delivery
[16]. Jain, T. K., Morales, M. A., Sahoo, S. K., Leslie-Pelecky, D. L., & Labhasetwar, V. (2005). Iron
oxide nanoparticles for sustained delivery of anticancer agents. Molecular Pharmaceutics, 2(3),
194-205.
[17]. Bao, G., Mitragotri, S., & Tong, S. (2013). Multifunctional nanoparticles for drug delivery and
[18]. Sun, C., Lee, J. S., & Zhang, M. (2008). Magnetic nanoparticles in MR imaging and drug
[19]. Sun, X., Feng, Z., Hou, T., & Li, Y. (2015). Computational simulation of inorganic nanoparticle
Molecular Modelling in Drug Delivery (pp. 149-168). John Wiley & Sons.
[20]. Kim, C. S., Tonga, G. Y., Solfiell, D., & Rotello, V. M. (2013). Inorganic nanosystems for
therapeutic delivery: Status and prospects. Advanced Drug Delivery Reviews, 65(1), 93-99.
[21]. Duncan, R. (2003). The dawning era of polymer therapeutics. Nature Reviews Drug Discovery,
2(5), 347–360.
115
[22]. Soppimath, K. S., Aminabhavi, T. M., Kulkarni, A. R., & Rudzinski, W. E. (2001).
70(1-2), 1-20.
[23]. Huynh, L., Neale, C., Pomès, R., & Allen, C. (2012). Computational approaches to the rational
design of nanoemulsions, polymeric micelles, and dendrimers for drug delivery. Nanomedicine:
[24]. Van Gunsteren, W.F., Bakowies, D., Baron, R., Chandrasekhar, I., Christen, M., Daura, X., Gee,
P., Geerke, D.P., Glättli, A., Hünenberger, P.H. & Kastenholz, M. A. (2006). Biomolecular
4064-4092.
[25]. Kann, B., Offerhaus, H. L., Windbergs, M., & Otto, C. (2015). Raman microscopy for cellular
investigations—From single cell imaging to drug carrier uptake visualization. Advanced Drug
[26]. Sharma, A., & Sharma, U. S. (1997). Liposomes in drug delivery: Progress and limitations.
[27]. Niven, R. W., Speer, M., & Schreier, H. (1991). Nebulization of liposomes. II. The effects of size
[28]. Hobbs, S. K., Monsky, W. L., Yuan, F., Roberts, W. G., Griffith, L., Torchilin, V. P., & Jain, R.
K. (1998). Regulation of transport pathways in tumor vessels: Role of tumor type and
[29]. Dass, C. R. (2008). Drug delivery in cancer using liposomes. In Drug Delivery Systems (pp. 177-
[30]. Lu, P. Y., & Woodle, M. C. (2008). Delivering small interfering RNA for novel therapeutics. In
116
[31]. Singh, R., & Lillard Jr, J. W. (2009). Nanoparticle-based targeted drug delivery. Experimental
[32]. Jang, S. H., Wientjes, M. G., Lu, D., & Au, J. L. S. (2003). Drug delivery and transport to solid
[33]. Kirtane, A. R., Siegel, R. A., & Panyam, J. (2015). A pharmacokinetic model for quantifying the
effect of vascular permeability on the choice of drug carrier: A framework for personalized
[34]. Brar, S. K., & Verma, M. (2011). Measurement of nanoparticles by light-scattering techniques.
[35]. Mehnert, W., & Mäder, K. (2012). Solid lipid nanoparticles: Production, characterization and
[36]. Kübart, S. A., & Keck, C. M. (2013). Laser diffractometry of nanoparticles: Frequent pitfalls &
overlooked opportunities. Journal of Pharmaceutical Technology and Drug Research, 2(1), 17.
[37]. Moquin, A., & Winnik, F. M. (2012). The use of field-flow fractionation for the analysis of drug
and gene delivery systems. In Field-Flow Fractionation in Biopolymer Analysis (pp. 187-206).
Springer, Vienna.
[38]. Zattoni, A., Roda, B., Borghi, F., Marassi, V., & Reschiglian, P. (2014). Flow field-flow
fractionation for the analysis of nanoparticles used in drug delivery. Journal of Pharmaceutical
[39]. Veronese, F. M., & Pasut, G. (2005). PEGylation, successful approach to drug delivery. Drug
[40]. Berchane, N. S., Carson, K. H., Rice-Ficht, A. C., & Andrews, M. J. (2007). Effect of mean
diameter and polydispersity of PLG microspheres on drug release: Experiment and theory.
117
[41]. Beddoes, C. M., Case, C. P., & Briscoe, W. H. (2015). Understanding nanoparticle cellular entry:
[42]. Müllner, M., Dodds, S. J., Nguyen, T. H., Senyschyn, D., Porter, C. J., Boyd, B. J., & Caruso, F.
(2015). Size and rigidity of cylindrical polymer brushes dictate long circulating properties in
[43]. Gratton, S. E., Ropp, P. A., Pohlhaus, P. D., Luft, J. C., Madden, V. J., Napier, M. E., &
[44]. Champion, J. A., & Mitragotri, S. (2006). Role of target geometry in phagocytosis. Proceedings
[45]. Fusco, S., Huang, H.W., Peyer, K.E., Peters, C., Haberli, M., Ulbers, A., Spyrogianni, A.,
Pellicer, E., Sort, J., Pratsinis, S.E. & Nelson, B. J. (2015). Shape-switching microrobots for
medical applications: The influence of shape in drug delivery and locomotion. ACS Applied
[46]. Ruan, G., & Feng, S. S. (2003). Preparation and characterization of poly (lactic acid)–poly
[47]. Fattal, E., Hillaireau, H., Mura, S., Nicolas, J., & Tsapis, N. (2012). Targeted delivery using
[48]. Milne, J.L., Borgnia, M.J., Bartesaghi, A., Tran, E.E., Earl, L.A., Schauder, D.M., Lengyel, J.,
Pierson, J., Patwardhan, A. & Subramaniam, S. (2013). Cryo‐electron microscopy–a primer for
[49]. Bartesaghi, A., Merk, A., Banerjee, S., Matthies, D., Wu, X., Milne, J. L., & Subramaniam, S.
118
(2015). 2.2 Å resolution cryo-EM structure of β-galactosidase in complex with a cell-permeant
[50]. Turovsky, T., Khalfin, R., Kababya, S., Schmidt, A., Barenholz, Y., & Danino, D. (2015).
[51]. Zheng, M., Pavan, G.M., Neeb, M., Schaper, A.K., Danani, A., Klebe, G., Merkel, O.M. &
Kissel, T. (2012). Targeting the blind spot of polycationic nanocarrier-based siRNA delivery.
[52]. Alkilany, A. M., & Murphy, C. J. (2010). Toxicity and cellular uptake of gold nanoparticles:
[53]. Murugan, K., Choonara, Y. E., Kumar, P., Bijukumar, D., du Toit, L. C., & Pillay, V. (2015).
[54]. Fröhlich, E. (2012). The role of surface charge in cellular uptake and cytotoxicity of medical
[55]. Di Marco, M., Sadun, C., Port, M., Guilbert, I., Couvreur, P., & Dubernet, C. (2007).
for biomedical application as MRI contrast agents. International Journal of Nanomedicine, 2(4),
609–622.
[56]. Xiao, K., Li, Y., Luo, J., Lee, J.S., Xiao, W., Gonik, A.M., Agarwal, R.G. & Lam, K. S. (2011).
The effect of surface charge on in vivo biodistribution of PEG-oligocholic acid based micellar
[57]. Mendes, L. P., Delgado, J. M. F., Costa, A. D. A., Vieira, M. S., Benfica, P. L., Lima, E. M., &
Valadares, M. C. (2015). Biodegradable nanoparticles designed for drug delivery: The number of
119
nanoparticles impacts on cytotoxicity. Toxicology in Vitro, 29(6), 1268-1274.
[58]. Law, M., Jafari, M., & Chen, P. (2008). Physicochemical characterization of siRNA‐peptide
[59]. Chen, A. M., Zhang, M., Wei, D., Stueber, D., Taratula, O., Minko, T., & He, H. (2009). Co‐
delivery of doxorubicin and Bcl‐2 siRNA by mesoporous silica nanoparticles enhances the
[60]. Wang, X. Y., Zhang, L., Wei, X. H., & Wang, Q. (2013). Molecular dynamics of paclitaxel
1843-1851.
[61]. Sun, B., & Chiu, D. T. (2005). Determination of the encapsulation efficiency of individual
[62]. Martinho, N., Florindo, H., Silva, L., Brocchini, S., Zloh, M., & Barata, T. (2014). Molecular
[63]. Bunjes, H., & Unruh, T. (2007). Characterization of lipid nanoparticles by differential scanning
calorimetry, X-ray and neutron scattering. Advanced Drug Delivery Reviews, 59(6), 379-402.
[64]. Mehnert, W., & Mäder, K. (2001). Solid lipid nanoparticles: Production, characterization and
[65]. Deborah, M., Jawahar, A., Mathavan, T., Dhas, M. K., & Benial, A. M. F. (2015). Spectroscopic
valine. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 139, 138-144.
[66]. Williams, P. S., Carpino, F., & Zborowski, M. (2009). Magnetic nanoparticle drug carriers and
1290-1306.
120
[67]. Briuglia, M. L., Rotella, C., McFarlane, A., & Lamprou, D. A. (2015). Influence of cholesterol on
liposome stability and on in vitro drug release. Drug Delivery and Translational Research, 5(3),
231-242.
[68]. Wang, J., Wang, Y., & Liang, W. (2012). Delivery of drugs to cell membranes by encapsulation
[69]. Kam, N. W. S., Liu, Z., & Dai, H. (2005). Functionalization of carbon nanotubes via cleavable
disulfide bonds for efficient intracellular delivery of siRNA and potent gene silencing. Journal of
[70]. Achar, S., & Puddephatt, R. J. (1994). Organoplatinum dendrimers formed by oxidative addition.
[71]. Viger, M. L., Sheng, W., McFearin, C. L., Berezin, M. Y., & Almutairi, A. (2013). Application of
time-resolved fluorescence for direct and continuous probing of release from polymeric delivery
[72]. Metwally, A. A., & Hathout, R. M. (2015). Computer-assisted drug formulation design: Novel
[73]. Sankaram, M. B., & Thompson, T. E. (1990). Interaction of cholesterol with various
[74]. Caminade, A. M., Laurent, R., & Majoral, J. P. (2005). Characterization of dendrimers. Advanced
[75]. Hsueh, Y. W., Zuckermann, M., & Thewalt, J. (2005). Phase diagram determination for
121
[76]. Io, T., Fukami, T., Yamamoto, K., Suzuki, T., Xu, J., Tomono, K., & Ramamoorthy, A. (2009).
[77]. Sun, T., Guo, Q., Zhang, C., Hao, J., Xing, P., Su, J., Li, S., Hao, A. & Liu, G. (2012). Self-
assembled vesicles prepared from amphiphilic cyclodextrins as drug carriers. Langmuir, 28(23),
8625-8636.
[78]. Astruc, D., Boisselier, E., & Ornelas, C. (2010). Dendrimers designed for functions: From
[79]. Leung, A.K., Hafez, I.M., Baoukina, S., Belliveau, N.M., Zhigaltsev, I.V., Afshinmanesh, E.,
Tieleman, D.P., Hansen, C.L., Hope, M.J. & Cullis, P. R. (2012). Lipid nanoparticles containing
[80]. Mathias, E. V., Liu, X., Franco, O., Khan, I., Ba, Y., & Kornfield, J. A. (2008). Model of drug-
19
loaded fluorocarbon-based micelles studied by electron-spin induced F relaxation NMR and
[81]. Martini, G., & Ciani, L. (2009). Electron spin resonance spectroscopy in drug delivery. Physical
[82]. Evans, D. F., & Wennerström, H. (1999). The colloidal domain: Where physics, chemistry,
[83]. Nicolini, C., Kraineva, J., Khurana, M., Periasamy, N., Funari, S. S., & Winter, R. (2006).
POPC/SM/cholesterol model raft mixtures—a FT-IR, SAXS, DSC, PPC and Laurdan
122
248-258.
[84]. Mills, T.T., Tristram-Nagle, S., Heberle, F.A., Morales, N.F., Zhao, J., Wu, J., Toombes, G.E.,
Nagle, J.F. & Feigenson, G. W. (2008). Liquid-liquid domains in bilayers detected by wide angle
[85]. Meli, V., Caltagirone, C., Falchi, A.M., Hyde, S.T., Lippolis, V., Monduzzi, M., Obiols-Rabasa,
M., Rosa, A., Schmidt, J., Talmon, Y. & Murgia, S. (2015). Docetaxel-loaded fluorescent liquid-
[86]. Angelova, A., Angelov, B., Mutafchieva, R., Lesieur, S., & Couvreur, P. (2010). Self-assembled
multicompartment liquid crystalline lipid carriers for protein, peptide, and nucleic acid drug
[87]. Utama, R. H., Dulle, M., Förster, S., Stenzel, M. H., & Zetterlund, P. B. (2015). SAXS analysis
of shell formation during nanocapsule synthesis via inverse miniemulsion periphery RAFT
[88]. Müller, R. H., Runge, S. A., Ravelli, V., Thünemann, A. F., Mehnert, W., & Souto, E. B. (2008).
[89]. Martins, A. F., Pereira, A. G., Fajardo, A. R., Rubira, A. F., & Muniz, E. C. (2011).
[90]. Pan, J., Heberle, F. A., Petruzielo, R. S., & Katsaras, J. (2013). Using small-angle neutron
scattering to detect nanoscopic lipid domains. Chemistry and Physics of Lipids, 170, 19-32.
[91]. Lakowicz, J. R. (2006). Principles of fluorescence spectroscopy. Springer Science & Business
Media.
123
[92]. Kataoka, K., Harada, A., & Nagasaki, Y. (2012). Block copolymer micelles for drug delivery:
Design, characterization and biological significance. Advanced Drug Delivery Reviews, 64, 37-
48.
[93]. Pygall, S. R., Whetstone, J., Timmins, P., & Melia, C. D. (2007). Pharmaceutical applications of
[95]. Giron, D. (2002). Applications of thermal analysis and coupled techniques in pharmaceutical
[96]. Reinl, H., Brumm, T., & Bayerl, T. M. (1992). Changes of the physical properties of the liquid-
ordered phase with temperature in binary mixtures of DPPC with cholesterol: A 2H NMR, FT-IR,
[97]. Labiris, N. R., & Dolovich, M. B. (2003). Pulmonary drug delivery. Part I: Physiological factors
[98]. Bouchemal, K. (2008). New challenges for pharmaceutical formulations and drug delivery
systems characterization using isothermal titration calorimetry. Drug Discovery Today, 13(21-
22), 960-972.
[99]. Gref, R., Domb, A., Quellec, P., Blunk, T., Müller, R. H., Verbavatz, J. M., & Langer, R. (1995).
[100]. Avgoustakis, K., Beletsi, A., Panagi, Z., Klepetsanis, P., Karydas, A. G., & Ithakissios, D. S.
release and in vivo drug residence in blood properties. Journal of Controlled Release, 79(1-3),
124
123-135.
[101]. Mui, B.L., Tam, Y.K., Jayaraman, M., Ansell, S.M., Du, X., Tam, Y.Y.C., Lin, P.J., Chen, S.,
Narayanannair, J.K., Rajeev, K.G. & Manoharan, M. (2013). Influence of polyethylene glycol
[102]. Tan, H., Wang, Z., Li, J., Pan, Z., Ding, M., & Fu, Q. (2013). An approach for the sphere-to-rod
[103]. Almutairi, A., Akers, W. J., Berezin, M. Y., Achilefu, S., & Fréchet, J. M. (2008). Monitoring the
[104]. Dai, X., Yue, Z., Eccleston, M. E., Swartling, J., Slater, N. K., & Kaminski, C. F. (2008).
[105]. Frenkel, D., & Smit, B. (2001). Understanding Molecular Simulation: From Algorithms to
[106]. Hess, B., Kutzner, C., Van Der Spoel, D., & Lindahl, E. (2008). GROMACS 4: Algorithms for
highly efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory
[107]. Phillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel,
R.D., Kale, L. & Schulten, K. (2005). Scalable molecular dynamics with NAMD. Journal of
[108]. Plimpton, S. (1995). Fast parallel algorithms for short-range molecular dynamics. Journal of
[109]. Brooks, B.R., Brooks, C.L., Mackerell Jr, A.D., Nilsson, L., Petrella, R.J., Roux, B., Won, Y.,
125
Archontis, G., Bartels, C., Boresch, S. & Caflisch, A. (2009). CHARMM: The biomolecular
[110]. Case, D.A., Cheatham, T.E., Darden, T., Gohlke, H., Luo, R., Merz, K.M., Onufriev, A.,
Simmerling, C., Wang, B. & Woods, R. J. (2005). The Amber biomolecular simulation programs.
[111]. Marrink, S. J., & Tieleman, D. P. (2013). Perspective on the Martini model. Chemical Society
[112]. Karplus, M., & McCammon, J. A. (2002). Molecular dynamics simulations of biomolecules.
[113]. Van Gunsteren, W. F., Dolenc, J., & Mark, A. E. (2008). Molecular simulation as an aid to
[114]. Velinova, M., Sengupta, D., Tadjer, A. V., & Marrink, S. J. (2011). Sphere-to-rod transitions of
[115]. Shinoda, W., DeVane, R., & Klein, M. L. (2012). Computer simulation studies of self-assembling
[116]. Messina, P. V., Miguel Besada-Porto, J., & M Ruso, J. (2014). Self-assembly drugs: From
[117]. Aoun, B., Sharma, V. K., Pellegrini, E., Mitra, S., Johnson, M., & Mukhopadhyay, R. (2015).
Structure and dynamics of ionic micelles: MD simulation and neutron scattering study. The
[118]. Vukovic, L., Khatib, F. A., Drake, S. P., Madriaga, A., Brandenburg, K. S., Král, P., & Onyuksel,
H. (2011). Structure and dynamics of highly PEG-ylated sterically stabilized micelles in aqueous
126
[119]. Gupta, J., Nunes, C., Vyas, S., & Jonnalagadda, S. (2011). Prediction of solubility parameters and
[120]. Loverde, S. M., Klein, M. L., & Discher, D. E. (2012). Nanoparticle shape improves delivery:
Rational coarse grain molecular dynamics (rCG‐MD) of taxol in worm‐like PEG‐PCL micelles.
[121]. Vuković, L., Madriaga, A., Kuzmis, A., Banerjee, A., Tang, A., Tao, K., Shah, N., Král, P. &
29(51), 15747-15754.
[122]. Huynh, L., Neale, C., Pomès, R., & Allen, C. (2010). Systematic design of unimolecular star
copolymer micelles using molecular dynamics simulations. Soft Matter, 6(21), 5491-5501.
[123]. Vácha, R., Martinez-Veracoechea, F. J., & Frenkel, D. (2012). Intracellular release of
endocytosed nanoparticles upon a change of ligand–receptor interaction. ACS nano, 6(12), 10598-
10605.
[124]. Tian, W. D., & Ma, Y. Q. (2012). Insights into the endosomal escape mechanism via
[125]. Arai, N., Yasuoka, K., & Zeng, X. C. (2013). A vesicle cell under collision with a Janus or
5(19), 9089-9100.
[126]. Makarucha, A. J., Todorova, N., & Yarovsky, I. (2011). Nanomaterials in biological
environment: A review of computer modelling studies. European Biophysics Journal, 40(2), 103-
115.
[127]. Li, Y.C., Rissanen, S., Stepniewski, M., Cramariuc, O., Róg, T., Mirza, S., Xhaard, H., Wytrwal,
M., Kepczynski, M. & Bunker, A. (2012). Study of interaction between PEG carrier and three
127
relevant drug molecules: Piroxicam, paclitaxel, and hematoporphyrin. The Journal of Physical
[128]. Martinez-Veracoechea, F. J., & Frenkel, D. (2011). Designing super selectivity in multivalent
[129]. Wang, S., & Dormidontova, E. E. (2010). Nanoparticle design optimization for enhanced
[130]. Liu, L., Parameswaran, S., Sharma, A., Grayson, S. M., Ashbaugh, H. S., & Rick, S. W. (2014).
Molecular dynamics simulations of linear and cyclic amphiphilic polymers in aqueous and
[131]. Lai, L., & Barnard, A. S. (2015). Functionalized nanodiamonds for biological and medical
[132]. Panczyk, T., Warzocha, T. P., & Camp, P. J. (2010). A magnetically controlled molecular
nanocontainer as a drug delivery system: The effects of carbon nanotube and magnetic
nanoparticle parameters from Monte Carlo simulations. The Journal of Physical Chemistry C,
114(49), 21299-21308.
[133]. Panczyk, T., Jagusiak, A., Pastorin, G., Ang, W. H., & Narkiewicz-Michalek, J. (2013).
Molecular dynamics study of cisplatin release from carbon nanotubes capped by magnetic
[134]. Ding, H. M., & Ma, Y. Q. (2014). Computer simulation of the role of protein corona in cellular
[135]. Brancolini, G., Kokh, D. B., Calzolai, L., Wade, R. C., & Corni, S. (2012). Docking of ubiquitin
[136]. Meneksedag-Erol, D., Tang, T., & Uludağ, H. (2015). Probing the effect of miRNA on siRNA–
128
[137]. Jorge, A. F., Dias, R. S., & Pais, A. A. (2012). Enhanced condensation and facilitated release of
DNA using mixed cationic agents: A combined experimental and Monte Carlo study.
[138]. Tian, F., Yue, T., Li, Y., & Zhang, X. (2014). Computer simulation studies on the interactions
between nanoparticles and cell membrane. Science China Chemistry, 57(12), 1662-1671.
[139]. Ding, H. M., & Ma, Y. Q. (2015). Theoretical and computational investigations of nanoparticle–
[140]. Li, Y., Zhang, X., & Cao, D. (2015). Nanoparticle hardness controls the internalization pathway
[141]. Li, Y., Yue, T., Yang, K., & Zhang, X. (2012). Molecular modeling of the relationship between
[142]. Vácha, R., Martinez-Veracoechea, F. J., & Frenkel, D. (2011). Receptor-mediated endocytosis of
[143]. Lin, X., Li, Y., & Gu, N. (2010). Nanoparticle's size effect on its translocation across a lipid
[144]. da Rocha, E. L., Caramori, G. F., & Rambo, C. R. (2013). Nanoparticle translocation through a
lipid bilayer tuned by surface chemistry. Physical Chemistry Chemical Physics, 15(7), 2282-
2290.
[145]. Van Lehn, R. C., & Alexander-Katz, A. (2011). Penetration of lipid bilayers by nanoparticles
with environmentally-responsive surfaces: Simulations and theory. Soft Matter, 7(24), 11392-
11404.
[146]. Lin, J., Zhang, H., Chen, Z., & Zheng, Y. (2010). Penetration of lipid membranes by gold
nanoparticles: Insights into cellular uptake, cytotoxicity, and their relationship. ACS nano, 4(9),
129
5421-5429.
[147]. Ding, H. M., & Ma, Y. Q. (2012). Role of physicochemical properties of coating ligands in
[148]. Li, Y. (2015). Computer investigations of influences of molar fraction and acyl chain length of
[149]. Ding, H. M., Tian, W. D., & Ma, Y. Q. (2012). Designing nanoparticle translocation through
[150]. Wang, S., & Dormidontova, E. E. (2012). Selectivity of ligand-receptor interactions between
[151]. Zhang, L., Becton, M., & Wang, X. (2015). Designing nanoparticle translocation through cell
119(9), 3786-3794.
[152]. Li, Y., Zhang, X., & Cao, D. (2014). A spontaneous penetration mechanism of patterned
[153]. Yue, T., & Zhang, X. (2011). Molecular understanding of receptor-mediated membrane responses
[154]. Barraza, L. F., Jiménez, V. A., & Alderete, J. B. (2015). Effect of PEGylation on the structure
and drug loading capacity of PAMAM‐G4 dendrimers: A molecular modeling approach on the
[155]. Lee, O. S., & Schatz, G. C. (2011). Computational simulations of the interaction of lipid
[156]. Mu, Q., Hu, T., & Yu, J. (2013). Molecular insight into the steric shielding effect of PEG on the
130
conjugated staphylokinase: Biochemical characterization and molecular dynamics simulation.
[157]. Lee, H. (2014). Molecular modeling of PEGylated peptides, dendrimers, and single-walled
[158]. Bunker, A. (2015). Molecular modeling as a tool to understand the role of poly (ethylene) glycol
[159]. Li, Y., Kröger, M., & Liu, W. K. (2014). Endocytosis of PEGylated nanoparticles accompanied
by structural and free energy changes of the grafted polyethylene glycol. Biomaterials, 35(30),
8467-8478.
[160]. Knop, K., Hoogenboom, R., Fischer, D., & Schubert, U. S. (2010). Poly (ethylene glycol) in drug
delivery: Pros and cons as well as potential alternatives. Angewandte Chemie International
[161]. Armstrong, J. K., Hempel, G., Koling, S., Chan, L. S., Fisher, T., Meiselman, H. J., & Garratty,
G. (2007). Antibody against poly (ethylene glycol) adversely affects PEG‐asparaginase therapy in
[162]. Zhang, L., Wang, Z., Lu, Z., Shen, H., Huang, J., Zhao, Q., Liu, M., He, N. & Zhang, Z. (2013).
PEGylated reduced graphene oxide as a superior ssRNA delivery system. Journal of Materials
[163]. Sun, X., Feng, Z., Zhang, L., Hou, T., & Li, Y. (2014). The selective interaction between silica
nanoparticles and enzymes from molecular dynamics simulations. PLoS ONE, 9(9), e107696.
[164]. Adnan, A., Lam, R., Chen, H., Lee, J., Schaffer, D.J., Barnard, A.S., Schatz, G.C., Ho, D. & Liu,
131
[165]. Lai, L., & Barnard, A. S. (2015). Molecular and analytical modeling of nanodiamond for drug
[166]. Bello, M.L., Junior, A.M., Vieira, B.A., Dias, L.R., de Sousa, V.P., Castro, H.C., Rodrigues, C.R.
insights on intercalated drugs arrangement into layered carrier material. PLoS ONE, 10(3),
e0121110.
[167]. Murthy, V., Xu, Z. P., & Smith, S. C. (2015). Molecular modeling of layered double hydroxide
[168]. Cox, P. (2012). Atomistic modelling of drug delivery systems. In Drug Design Strategies:
[169]. Fatouros, D.G., Douroumis, D., Nikolakis, V., Ntais, S., Moschovi, A.M., Trivedi, V., Khima, B.,
Roldo, M., Nazar, H. & Cox, P. A. (2011). In vitro and in silico investigations of drug delivery
[170]. Spanakis, M., Bouropoulos, N., Theodoropoulos, D., Sygellou, L., Ewart, S., Moschovi, A.M.,
Siokou, A., Niopas, I., Kachrimanis, K., Nikolakis, V. & Cox, P. A. (2014). Controlled release of
[171]. Bernini, M. C., Fairen-Jimenez, D., Pasinetti, M., Ramirez-Pastor, A. J., & Snurr, R. Q. (2014).
[172]. Meek, S. T., Greathouse, J. A., & Allendorf, M. D. (2011). Metal‐organic frameworks: A rapidly
132
[173]. Pavan, G. M. (2014). Modeling the interaction between dendrimers and nucleic acids: A
[174]. Jain, V., & Bharatam, P. V. (2014). Pharmacoinformatic approaches to understand complexation
[175]. Shcharbin, D., Shakhbazau, A., & Bryszewska, M. (2013). Poly (amidoamine) dendrimer
complexes as a platform for gene delivery. Expert Opinion on Drug Delivery, 10(12), 1687-1698.
[176]. Posocco, P., Laurini, E., Dal Col, V., Marson, D., Karatasos, K., Fermeglia, M., & Pricl, S.
organization and self-assembly in gene therapy. Current Medicinal Chemistry, 19(29), 5062-
5087.
[177]. Meneksedag-Erol, D., Tang, T., & Uludağ, H. (2014). Molecular modeling of polynucleotide
[178]. Molla, M. R., Rangadurai, P., Pavan, G. M., & Thayumanavan, S. (2015). Experimental and
3817-3837.
[179]. Ficici, E., Andricioaei, I., & Howorka, S. (2015). Dendrimers in nanoscale confinement: The
interplay between conformational change and nanopore entrance. Nano Letters, 15(7), 4822-
4828.
[180]. Wang, Y. L., Lu, Z. Y., & Laaksonen, A. (2012). Specific binding structures of dendrimers on
[181]. Nandy, B., Maiti, P. K., & Bunker, A. (2012). Force biased molecular dynamics simulation study
of effect of dendrimer generation on interaction with DNA. Journal of Chemical Theory and
[182]. Pavan, G. M., & Danani, A. (2010). The influence of dendron's architecture on the “rigid” and
133
“flexible” behaviour in binding DNA—a modelling study. Physical Chemistry Chemical Physics,
12(42), 13914-13917.
[183]. Pavan, G. M., Albertazzi, L., & Danani, A. (2010). Ability to adapt: Different generations of
PAMAM dendrimers show different behaviors in binding siRNA. The Journal of Physical
[184]. Posocco, P., Pricl, S., Jones, S., Barnard, A., & Smith, D. K. (2010). Less is more–multiscale
modelling of self-assembling multivalency and its impact on DNA binding and gene delivery.
[185]. Jones, S.P., Gabrielson, N.P., Wong, C.H., Chow, H.F., Pack, D.W., Posocco, P., Fermeglia, M.,
activity relationships for DNA binding and gene transfection. Molecular Pharmaceutics, 8(2),
416-429.
[186]. Jones, S. P., Pavan, G. M., Danani, A., Pricl, S., & Smith, D. K. (2010). Quantifying the effect of
[187]. Percec, V., Wilson, D.A., Leowanawat, P., Wilson, C.J., Hughes, A.D., Kaucher, M.S., Hammer,
D.A., Levine, D.H., Kim, A.J., Bates, F.S. & Davis, K. P. (2010). Self-assembly of Janus
dendrimers into uniform dendrimersomes and other complex architectures. Science, 328(5981),
1009-1014.
[188]. Zhang, S., Sun, H.J., Hughes, A.D., Moussodia, R.O., Bertin, A., Chen, Y., Pochan, D.J., Heiney,
P.A., Klein, M.L. & Percec, V. (2014). Self-assembly of amphiphilic Janus dendrimers into
uniform onion-like dendrimersomes with predictable size and number of bilayers. Proceedings of
[189]. Kavyani, S., Amjad-Iranagh, S., & Modarress, H. (2014). Aqueous poly (amidoamine) dendrimer
134
G3 and G4 generations with several interior cores at pHs 5 and 7: A molecular dynamics
[190]. Barnard, A., Posocco, P., Pricl, S., Calderon, M., Haag, R., Hwang, M.E., Shum, V.W., Pack,
D.W. & Smith, D. K. (2011). Degradable self-assembling dendrons for gene delivery:
Experimental and theoretical insights into the barriers to cellular uptake. Journal of the American
[191]. Carrasco-Sánchez, V., Vergara-Jaque, A., Zuñiga, M., Comer, J., John, A., Nachtigall, F.M.,
Valdes, O., Duran-Lara, E.F., Sandoval, C. & Santos, L. S. (2014). In situ and in silico evaluation
[192]. Shi, C., Yuan, D., Nangia, S., Xu, G., Lam, K. S., & Luo, J. (2014). A structure–property
[193]. Pearson, R. M., Patra, N., Hsu, H. J., Uddin, S., Král, P., & Hong, S. (2012). Positively charged
dendron micelles display negligible cellular interactions. ACS Macro Letters, 2(1), 77-81.
[194]. Lee, H., Choi, J. S., & Larson, R. G. (2011). Molecular dynamics studies of the size and internal
structure of the PAMAM dendrimer grafted with arginine and histidine. Macromolecules, 44(21),
8681-8686.
[195]. Lee, H., & Larson, R. G. (2011). Membrane pore formation induced by acetylated and
C, 115(13), 5316-5322.
[196]. Yang, L., & da Rocha, S. R. (2014). PEGylated, NH2-terminated PAMAM dendrimers: A
microscopic view from atomistic computer simulations. Molecular Pharmaceutics, 11(5), 1459-
1470.
135
[197]. Albertazzi, L., Mickler, F.M., Pavan, G.M., Salomone, F., Bardi, G., Panniello, M., Amir, E.,
Kang, T., Killops, K.L., Bräuchle, C. & Amir, R. J. (2012). Enhanced bioactivity of internally
[198]. Lee, H., & Larson, R. G. (2011). Effects of PEGylation on the size and internal structure of
dendrimers: Self-penetration of long PEG chains into the dendrimer core. Macromolecules, 44(7),
2291-2298.
[199]. Bhattacharya, R., Kanchi, S., Roobala, C., Lakshminarayanan, A., Seeck, O.H., Maiti, P.K.,
Ayappa, K.G., Jayaraman, N. & Basu, J. K. (2014). A new microscopic insight into membrane
[200]. Guo, R., Mao, J., & Yan, L. T. (2013). Unique dynamical approach of fully wrapping dendrimer-
like soft nanoparticles by lipid bilayer membrane. ACS nano, 7(12), 10646-10653.
[201]. He, X., Qu, Z., Xu, F., Lin, M., Wang, J., Shi, X., & Lu, T. (2014). Molecular analysis of
interactions between dendrimers and asymmetric membranes at different transport stages. Soft
[202]. Tu, C. K., Chen, K., Tian, W. D., & Ma, Y. Q. (2013). Computational investigations of a peptide-
34(15), 1237-1242.
[203]. Ma, Y. Q. (2012). pH-responsive dendrimers interacting with lipid membranes. Soft Matter, 8(9),
2627-2632.
[204]. Xie, L. Q., Feng, J. W., Tian, W. D., & Ma, Y. Q. (2014). Generation-dependent gel-fluid phase
[205]. Bellini, R. G., Guimarães, A. P., Pacheco, M. A., Dias, D. M., Furtado, V. R., de Alencastro, R.
B., & Horta, B. A. (2015). Association of the anti-tuberculosis drug rifampicin with a PAMAM
136
dendrimer. Journal of Molecular Graphics and Modelling, 60, 34-42.
[206]. Jain, V., Maingi, V., Maiti, P. K., & Bharatam, P. V. (2013). Molecular dynamics simulations of
[207]. Jiang, Y., Zhang, D., Zhang, Y., Deng, Z., & Zhang, L. (2014). The adsorption-desorption
[208]. Wen, X.F., Lan, J.L., Cai, Z.Q., Pi, P.H., Xu, S.P., Zhang, L.J., Qian, Y. & Wang, S. N. (2014).
[209]. Maingi, V., Kumar, M. V. S., & Maiti, P. K. (2012). PAMAM dendrimer–drug interactions:
Effect of pH on the binding and release pattern. The Journal of Physical Chemistry B, 116(14),
4370-4376.
[210]. Avila-Salas, F., Sandoval, C., Caballero, J., Guiñez-Molinos, S., Santos, L. S., Cachau, R. E., &
González-Nilo, F. D. (2012). Study of interaction energies between the PAMAM dendrimer and
[211]. Lim, J., Pavan, G. M., Annunziata, O., & Simanek, E. E. (2012). Experimental and computational
evidence for an inversion in guest capacity in high-generation triazine dendrimer hosts. Journal of
[212]. Ouyang, D., Zhang, H., Parekh, H. S., & Smith, S. C. (2011). The effect of pH on PAMAM
[213]. Nandy, B., & Maiti, P. K. (2010). DNA compaction by a dendrimer. The Journal of Physical
137
[214]. Amado Torres, D., Garzoni, M., Subrahmanyam, A. V., Pavan, G. M., & Thayumanavan, S.
location in amphiphilic dendrons. Journal of the American Chemical Society, 136(14), 5385-
5399.
[215]. Jensen, L. B., Pavan, G. M., Kasimova, M. R., Rutherford, S., Danani, A., Nielsen, H. M., &
410-418.
[216]. Karatasos, K., Posocco, P., Laurini, E., & Pricl, S. (2012). Poly (amidoamine)‐based
Generation on Dendrimer Structure and siRNA Binding. Macromolecular Bioscience, 12(2), 225-
240.
[217]. Pavan, G. M., Monteagudo, S., Guerra, J., Carrion, B., Ocana, V., Rodriguez-Lopez, J., Danani,
A., C Perez-Martinez, F. & Cena, V. (2012). Role of generation, architecture, pH and ionic
[218]. Pavan, G. M., Mintzer, M. A., Simanek, E. E., Merkel, O. M., Kissel, T., & Danani, A. (2010).
Computational insights into the interactions between DNA and siRNA with “rigid” and “flexible”
[219]. Pavan, G.M., Posocco, P., Tagliabue, A., Maly, M., Malek, A., Danani, A., Ragg, E., Catapano,
C.V. & Pricl, S. (2010). PAMAM dendrimers for siRNA delivery: Computational and
[220]. Shi, C., Guo, D., Xiao, K., Wang, X., Wang, L., & Luo, J. (2015). A drug-specific nanocarrier
138
[221]. Kang, M., Lam, D., Discher, D. E., & Loverde, S. M. (2015). Molecular modeling of block
[222]. Rösler, A., Vandermeulen, G. W., & Klok, H. A. (2012). Advanced drug delivery devices via
self-assembly of amphiphilic block copolymers. Advanced Drug Delivery Reviews, 64, 270-279.
[223]. Sun, C., & Tang, T. (2014). Study on the role of polyethylenimine as gene delivery carrier using
molecular dynamics simulations. Journal of Adhesion Science and Technology, 28(3-4), 399-416.
[224]. Loverde, S. M. (2014). Computer simulation of polymer and biopolymer self-assembly for drug
[225]. Meneksedag-Erol, D., Sun, C., Tang, T., & Uludag, H. (2014). Molecular dynamics simulations
of polyplexes and lipoplexes employed in gene delivery. In Intracellular Delivery II (pp. 277-
[226]. Guo, H., Qiu, X., & Zhou, J. (2013). Self-assembled core-shell and Janus microphase separated
structures of polymer blends in aqueous solution. The Journal of Chemical Physics, 139(8),
084907.
[227]. Cai, C., Wang, L., Lin, J., & Zhang, X. (2012). Morphology transformation of hybrid micelles
self-assembled from rod–coil block copolymer and nanoparticles. Langmuir, 28(9), 4515-4524.
[228]. Chang, H. Y., Lin, Y. L., Sheng, Y. J., & Tsao, H. K. (2013). Structural characteristics and fusion
[229]. Sheng, Y., An, J., & Zhu, Y. (2015). Self-assembly of ABA triblock copolymers under soft
[230]. Chen, H., & Ruckenstein, E. (2013). Formation and degradation of multicomponent multicore
micelles: Insights from dissipative particle dynamics simulations. Langmuir, 29(18), 5428-5434.
[231]. Taresco, V., Gontrani, L., Crisante, F., Francolini, I., Martinelli, A., D’Ilario, L., Bordi, F. &
139
Piozzi, A. (2015). Self-assembly of catecholic moiety-containing cationic random acrylic
[232]. Nie, S. Y., Sun, Y., Lin, W. J., Wu, W. S., Guo, X. D., Qian, Y., & Zhang, L. J. (2013).
[233]. Wang, T., Chipot, C., Shao, X., & Cai, W. (2010). Structural characterization of micelles formed
[234]. Peng, L.X., Ivetac, A., Chaudhari, A.S., Van, S., Zhao, G., Yu, L., Howell, S.B., McCammon,
[235]. Pi, P., Qin, D., Lan, J.L., Cai, Z., Yuan, X., Xu, S.P., Zhang, L., Qian, Y. & Wen, X. (2015).
Dissipative particle dynamics simulation on the nanocomposite delivery system of quantum dots
and poly (styrene-b-ethylene oxide) copolymer. Industrial & Engineering Chemistry Research,
54(23), 6123-6134.
[236]. Li, J., Ouyang, Y., Kong, X., Zhu, J., Lu, D., & Liu, Z. (2015). A multi-scale molecular dynamics
[237]. Srinivas, G., Mohan, R. V., & Kelkar, A. D. (2013). Polymer micelle assisted transport and
[238]. Ding, H. M., & Ma, Y. Q. (2013). Controlling cellular uptake of nanoparticles with pH-sensitive
[239]. Sun, C., Tang, T., & Uludag, H. (2013). A molecular dynamics simulation study on the effect of
140
2822-2833.
[240]. Liao, Z. X., Peng, S. F., Ho, Y. C., Mi, F. L., Maiti, B., & Sung, H. W. (2012). Mechanistic study
[241]. Peng, S. F., Tseng, M. T., Ho, Y. C., Wei, M. C., Liao, Z. X., & Sung, H. W. (2011).
[242]. Kepczynski, M., Jamróz, D., Wytrwal, M., Bednar, J., Rzad, E., & Nowakowska, M. (2011).
[243]. Chiu, Y.L., Ho, Y.C., Chen, Y.M., Peng, S.F., Ke, C.J., Chen, K.J., Mi, F.L. & Sung, H. W.
(2010). The characteristics, cellular uptake and intracellular trafficking of nanoparticles made of
[244]. Sun, C., Tang, T., & Uludağ, H. (2012). Probing the effects of lipid substitution on polycation
2982-2988.
[245]. Samanta, S., & Roccatano, D. (2013). Interaction of curcumin with PEO–PPO–PEO block
copolymers: A molecular dynamics study. The Journal of Physical Chemistry B, 117(11), 3250-
3257.
[246]. He, J., Chipot, C., Shao, X., & Cai, W. (2013). Cyclodextrin-mediated recruitment and delivery
[247]. Shan, P., Shen, J.W., Xu, D.H., Shi, L.Y., Gao, J., Lan, Y.W., Wang, Q. & Wei, X. H. (2014).
Molecular dynamics study on the interaction between doxorubicin and hydrophobically modified
141
[248]. Subashini, M., Devarajan, P. V., Sonavane, G. S., & Doble, M. (2011). Molecular dynamics
[249]. Anand, R., Ottani, S., Manoli, F., Manet, I., & Monti, S. (2012). A close-up on doxorubicin
[250]. Namgung, R., Lee, Y.M., Kim, J., Jang, Y., Lee, B.H., Kim, I.S., Sokkar, P., Rhee, Y.M.,
Hoffman, A.S. & Kim, W. J. (2014). Poly-cyclodextrin and poly-paclitaxel nano-assembly for
[251]. Wang, Z., & Jiang, J. (2014). Dissipative particle dynamics simulation on paclitaxel loaded PEO–
PPO–PEO block copolymer micelles. Journal of Nanoscience and Nanotechnology, 14(3), 2644-
2647.
[252]. Wang, Z., Wang, F., Su, C., & Zhang, Y. (2013). Computer simulation of polymer delivery
10(10), 2323-2327.
[253]. Razmimanesh, F., Amjad-Iranagh, S., & Modarress, H. (2015). Molecular dynamics simulation
study of chitosan and gemcitabine as a drug delivery system. Journal of Molecular Modeling,
21(7), 165.
[254]. Nie, S. Y., Lin, W. J., Yao, N., Guo, X. D., & Zhang, L. J. (2014). Drug release from pH-
sensitive polymeric micelles with different drug distributions: Insight from coarse-grained
[255]. Luo, Z., & Jiang, J. (2012). pH-sensitive drug loading/releasing in amphiphilic copolymer PAE–
PEG: Integrating molecular dynamics and dissipative particle dynamics simulations. Journal of
[256]. Hao, J., Cheng, Y., Ranatunga, R.U., Senevirathne, S., Biewer, M.C., Nielsen, S.O., Wang, Q. &
142
Stefan, M. C. (2013). A combined experimental and computational study of the substituent effect
[257]. Zheng, L. S., Yang, Y. Q., Guo, X. D., Sun, Y., Qian, Y., & Zhang, L. J. (2011). Mesoscopic
simulations on the aggregation behavior of pH-responsive polymeric micelles for drug delivery.
[258]. Rodríguez-Hidalgo, M.R., Soto-Figueroa, C., & Vicente, L. (2011). Mesoscopic simulation of the
drug release mechanism on the polymeric vehicle P(ST-DVB) in an acid environment. Soft
[259]. Malik, R., Genzer, J., & Hall, C. K. (2015). Proteinlike copolymers as encapsulating agents for
[260]. Patel, S. K., Lavasanifar, A., & Choi, P. (2010). Molecular dynamics study of the encapsulation
capability of a PCL–PEO based block copolymer for hydrophobic drugs with different spatial
[261]. Thakur, S. S., Parekh, H. S., Schwable, C. H., Gan, Y., & Ouyang, D. (2015). Solubilization of
[262]. Bagai, S., Sun, C., & Tang, T. (2012). Potential of mean force of polyethylenimine-mediated
[263]. Antila, H. S., Härkönen, M., & Sammalkorpi, M. (2015). Chemistry specificity of DNA–
polycation complex salt response: A simulation study of DNA, polylysine and polyethyleneimine.
[264]. Pereira, P., Jorge, A. F., Martins, R., Pais, A. A., Sousa, F., & Figueiras, A. (2012).
Characterization of polyplexes involving small RNA. Journal of Colloid and Interface science,
143
387(1), 84-94.
[265]. Sun, C., Tang, T., & Uludaǧ, H. (2012). Molecular dynamics simulations for complexation of
DNA with 2 kDa PEI reveal profound effect of PEI architecture on complexation. The Journal of
[266]. Ouyang, D., Zhang, H., Herten, D. P., Parekh, H. S., & Smith, S. C. (2010). Structure, dynamics
and energetics of siRNA - Cationic vector complexation: A molecular dynamics study. The
[267]. Ouyang, D., Zhang, H., Parekh, H. S., & Smith, S. C. (2010). Structure and dynamics of multiple
[268]. Jorge, A. F., Pereira, R. F., Nunes, S. C., Valente, A. J., Dias, R. S., & Pais, A. A. (2014).
15(2), 478-491.
[269]. Zhan, B., Shi, K., Dong, Z., Lv, W., Zhao, S., Han, X., Wang, H. & Liu, H. (2015). Coarse-
[270]. Zhao, M., Zhou, J., Su, C., Niu, L., Liang, D., & Li, B. (2015). Complexation behavior of
[271]. Ziebarth, J., & Wang, Y. (2010). Coarse-grained molecular dynamics simulations of DNA
[272]. Luo, Q., Wang, Y., Yang, H., Liu, C., Ding, Y., Xu, H., Wang, Q., Liu, Y. & Xie, Y. (2014).
Modeling the interaction of interferon α-1b to bovine serum albumin as a drug delivery system.
144
The Journal of Physical Chemistry B, 118(29), 8566-8574.
[273]. Vijayaraj, R., Van Damme, S., Bultinck, P., & Subramanian, V. (2013). Theoretical studies on the
[274]. Liu, H., Chen, J., Shen, Q., Fu, W., & Wu, W. (2010). Molecular insights on the cyclic peptide
7(6), 1985-1994.
[275]. Durzyńska, J., Przysiecka, Ł., Nawrot, R., Barylski, J., Nowicki, G., Warowicka, A., Musidlak,
32-42.
[276]. Krishnan, Y., & Simmel, F. C. (2011). Nucleic acid based molecular devices. Angewandte
[277]. McLaughlin, C. K., Hamblin, G. D., & Sleiman, H. F. (2011). Supramolecular DNA assembly.
[278]. Cutler, J. I., Auyeung, E., & Mirkin, C. A. (2012). Spherical nucleic acids. Journal of the
[279]. Barnaby, S. N., Sita, T. L., Petrosko, S. H., Stegh, A. H., & Mirkin, C. A. (2015). Therapeutic
[280]. Afonin, K. A., Kasprzak, W. K., Bindewald, E., Kireeva, M., Viard, M., Kashlev, M., & Shapiro,
B. A. (2014). In silico design and enzymatic synthesis of functional RNA nanoparticles. Accounts
[281]. Badu, S. R., Melnik, R., Paliy, M., Prabhakar, S., Sebetci, A., & Shapiro, B. A. (2014). Modeling
145
of RNA nanotubes using molecular dynamics simulation. European Biophysics Journal, 43(10-
11), 555-564.
[282]. Juul, S., Iacovelli, F., Falconi, M., Kragh, S.L., Christensen, B., Frøhlich, R., Franch, O.,
Kristoffersen, E.L., Stougaard, M., Leong, K.W. & Ho, Y. P. (2013). Temperature-controlled
encapsulation and release of an active enzyme in the cavity of a self-assembled DNA nanocage.
[283]. Elzoghby, A. O., Samy, W. M., & Elgindy, N. A. (2012). Protein-based nanocarriers as promising
drug and gene delivery systems. Journal of Controlled Release, 161(1), 38-49.
[284]. Elzoghby, A. O., Samy, W. M., & Elgindy, N. A. (2012). Albumin-based nanoparticles as
potential controlled release drug delivery systems. Journal of Controlled Release, 157(2), 168-
182.
[285]. Pourmousa, M., & Karttunen, M. (2013). Early stages of interactions of cell-penetrating peptide
penetratin with a DPPC bilayer. Chemistry and Physics of Lipids, 169, 85-94.
[286]. Pourmousa, M., Wong-ekkabut, J., Patra, M., & Karttunen, M. (2012). Molecular dynamic
studies of transportan interacting with a DPPC lipid bilayer. The Journal of Physical Chemistry B,
117(1), 230-241.
[287]. He, X., Lin, M., Sha, B., Feng, S., Shi, X., Qu, Z., & Xu, F. (2015). Coarse-grained molecular
[288]. Islami, M., Mehrnejad, F., Doustdar, F., Alimohammadi, M., Khadem‐Maaref, M., Mir‐
Derikvand, M., & Taghdir, M. (2014). Study of orientation and penetration of LAH4 into lipid
bilayer membranes: pH and composition dependence. Chemical Biology & Drug Design, 84(2),
242-252.
[289]. Li, Z. L., Ding, H. M., & Ma, Y. Q. (2012). Translocation of polyarginines and conjugated
146
nanoparticles across asymmetric membranes. Soft Matter, 9(4), 1281-1286.
[290]. Teixeira, R.S., Cova, T.F., Silva, S.M., Oliveira, R., Araújo, M.J., Marques, E.F., Pais, A.A. &
[291]. Todorova, N., Chiappini, C., Mager, M., Simona, B., Patel, I. I., Stevens, M. M., & Yarovsky, I.
[292]. Li, Y., Feng, D., Zhang, X., & Cao, D. (2015). Design strategy of cell-penetrating copolymers for
[293]. Sanchez‐Sanchez, A., Akbari, S., Moreno, A. J., Verso, F. L., Arbe, A., Colmenero, J., &
disordered proteins for combined delivery of dermal bioactive cargos. Macromolecular Rapid
[294]. Chen, L., Jiang, T., Cai, C., Wang, L., Lin, J., & Cao, X. (2014). Polypeptide-based “smart”
micelles for dual-drug delivery: A combination study of experiments and simulations. Advanced
[295]. Jabbari, E., Yang, X., Moeinzadeh, S., & He, X. (2013). Drug release kinetics, cell uptake, and
[296]. Guo, X. D., Zhang, L. J., Wu, Z. M., & Qian, Y. (2010). Dissipative particle dynamics studies on
7839-7844.
[297]. Tian, X., Sun, F., Zhou, X. R., Luo, S. Z., & Chen, L. (2015). Role of peptide self‐assembly in
147
[298]. Fu, I. W., Markegard, C. B., & Nguyen, H. D. (2014). Solvent effects on kinetic mechanisms of
self-assembly by peptide amphiphiles via molecular dynamics simulations. Langmuir, 31(1), 315-
324.
[299]. Fu, I. W., Markegard, C. B., Chu, B. K., & Nguyen, H. D. (2013). The role of electrostatics and
amphiphiles via molecular dynamics simulations. Advanced Healthcare Materials, 2(10), 1388-
1400.
[300]. Thota, N., Luo, Z., Hu, Z., & Jiang, J. (2013). Self-assembly of amphiphilic peptide (AF)
117(33), 9690-9698.
[301]. García-Fandiño, R., & Granja, J. R. (2013). Effect of organochloride guest molecules on the
toward the control of nanotube length. The Journal of Physical Chemistry C, 117(19), 10143-
10162.
[302]. Bakry, R., Vallant, R. M., Najam-ul-Haq, M., Rainer, M., Szabo, Z., Huck, C. W., & Bonn, G. K.
649.
[303]. Martincic, M., & Tobias, G. (2015). Filled carbon nanotubes in biomedical imaging and drug
[304]. Chaban, V. V., Savchenko, T. I., Kovalenko, S. M., & Prezhdo, O. V. (2010). Heat-driven release
of a drug molecule from carbon nanotubes: A molecular dynamics study. The Journal of Physical
[305]. Saikia, N., Jha, A. N., & Deka, R. C. (2013). Dynamics of fullerene-mediated heat-driven release
of drug molecules from carbon nanotubes. The Journal of Physical Chemistry Letters, 4(23),
148
4126-4132.
[306]. Arsawang, U., Saengsawang, O., Rungrotmongkol, T., Sornmee, P., Wittayanarakul, K.,
Remsungnen, T., & Hannongbua, S. (2011). How do carbon nanotubes serve as carriers for
gemcitabine transport in a drug delivery system?. Journal of Molecular Graphics and Modelling,
29(5), 591-596.
[307]. Li, Y., & Hou, T. (2010). Computational simulation of drug delivery at molecular level. Current
[308]. Chehel Amirani, M., & Tang, T. (2015). Binding of nucleobases with graphene and carbon
33(7), 1567-1597.
[309]. Santosh, M., Panigrahi, S., Bhattacharyya, D., Sood, A. K., & Maiti, P. K. (2012). Unzipping and
binding of small interfering RNA with single walled carbon nanotube: A platform for small
[310]. Nandy, B., Santosh, M., & Maiti, P. K. (2012). Interaction of nucleic acids with carbon nanotubes
[311]. Chen, B. D., Yang, C. L., Yang, J. S., Wang, M. S., & Ma, X. G. (2013). Dynamic mechanism of
HIV replication inhibitor peptide encapsulated into carbon nanotubes. Current Applied Physics,
13(6), 1001-1007.
[312]. Shen, J.W., Tang, T., Wei, X.H., Zheng, W., Sun, T.Y., Zhang, Z., Liang, L. & Wang, Q. (2015).
On the loading mechanism of ssDNA into carbon nanotubes. RSC Advances, 5(70), 56896-56903.
[313]. Wu, N., Wang, Q., & Arash, B. (2012). Ejection of DNA molecules from carbon nanotubes.
[314]. Li, J., Jiang, L., & Zhu, X. (2015). Computational studies of the binding mechanisms of
149
[315]. Wong-Ekkabut, J., Baoukina, S., Triampo, W., Tang, I. M., Tieleman, D. P., & Monticelli, L.
(2008). Computer simulation study of fullerene translocation through lipid membranes. Nature
[316]. Barnoud, J., Urbini, L., & Monticelli, L. (2015). C60 fullerene promotes lung monolayer
[317]. Nisoh, N., Karttunen, M., Monticelli, L., & Wong-Ekkabut, J. (2015). Lipid monolayer disruption
[318]. Skandani, A. A., Zeineldin, R., & Al-Haik, M. (2012). Effect of chirality and length on the
penetrability of single-walled carbon nanotubes into lipid bilayer cell membranes. Langmuir,
28(20), 7872-7879.
[319]. Lee, H. (2013). Membrane penetration and curvature induced by single-walled carbon nanotubes:
The effect of diameter, length, and concentration. Physical Chemistry Chemical Physics, 15(38),
16334-16340.
[320]. Baoukina, S., Monticelli, L., & Tieleman, D. P. (2013). Interaction of pristine and functionalized
carbon nanotubes with lipid membranes. The Journal of Physical Chemistry B, 117(40), 12113-
12123.
[321]. Chiu, C. C., Shinoda, W., DeVane, R. H., & Nielsen, S. O. (2012). Effects of spherical fullerene
[322]. Jusufi, A., DeVane, R. H., Shinoda, W., & Klein, M. L. (2011). Nanoscale carbon particles and
[323]. Lee, H. (2013). Interparticle dispersion, membrane curvature, and penetration induced by single-
walled carbon nanotubes wrapped with lipids and PEGylated lipids. The Journal of Physical
150
[324]. Kraszewski, S., Bianco, A., Tarek, M., & Ramseyer, C. (2012). Insertion of short amino-
[325]. Kraszewski, S., Picaud, F., Elhechmi, I., Gharbi, T., & Ramseyer, C. (2012). How long a
functionalized carbon nanotube can passively penetrate a lipid membrane?. Carbon, 50(14),
5301-5308.
[326]. Lee, H. (2015). Dispersion and bilayer interaction of single-walled carbon nanotubes modulated
[327]. Di Crescenzo, A., Aschi, M., & Fontana, A. (2012). Toward a better understanding of steric
[328]. Sridhar, A., Srikanth, B., Kumar, A., & Dasmahapatra, A. K. (2015). Coarse-grain molecular
dynamics study of fullerene transport across a cell membrane. The Journal of Chemical Physics,
143(2), 024907.
[329]. Sarukhanyan, E., Milano, G., & Roccatano, D. (2014). Coating mechanisms of single-walled
carbon nanotube by linear polyether surfactants: Insights from computer simulations. The Journal
[330]. Lee, H. (2013). Molecular dynamics studies of PEGylated single-walled carbon nanotubes: The
effect of PEG size and grafting density. The Journal of Physical Chemistry C, 117(49), 26334-
26341.
[331]. Shrestha, H., Bala, R., & Arora, S. (2014). Lipid-based drug delivery systems. Journal of
Pharmaceutics, 2014.
[332]. Mashaghi, S., Jadidi, T., Koenderink, G., & Mashaghi, A. (2013). Lipid nanotechnology.
151
[333]. Klibanov, A. L., Maruyama, K., Torchilin, V. P., & Huang, L. (1990). Amphipathic
polyethyleneglycols effectively prolong the circulation time of liposomes. FEBS Letters, 268(1),
235-237.
[334]. Dzieciuch, M., Rissanen, S., Szydłowska, N., Bunker, A., Kumorek, M., Jamróz, D., Vattulainen,
I., Nowakowska, M., Rog, T. & Kepczynski, M. (2015). PEGylated liposomes as carriers of
[335]. Lehtinen, J., Magarkar, A., Stepniewski, M., Hakola, S., Bergman, M., Róg, T., Yliperttula, M.,
Urtti, A. & Bunker, A. (2012). Analysis of cause of failure of new targeting peptide in PEGylated
liposome: Molecular modeling as rational design tool for nanomedicine. European Journal of
[336]. Magarkar, A., Róg, T., & Bunker, A. (2014). Molecular dynamics simulation of PEGylated
membranes with cholesterol: Building toward the DOXIL formulation. The Journal of Physical
[337]. Laasonen, K., & Klein, M. L. (1995). Molecular dynamics simulations of the structure and ion
diffusion in poly (ethylene oxide). Journal of the Chemical Society, Faraday Transactions,
91(16), 2633-2638.
[338]. Müller‐Plathe, F., & van Gunsteren, W. F. (1995). Computer simulation of a polymer electrolyte:
Lithium iodide in amorphous poly (ethylene oxide). The Journal of Chemical Physics, 103(11),
4745-4756.
[339]. Di̇ nç, C. Ö., Ki̇ barer, G., & Güner, A. (2010). Solubility profiles of poly (ethylene glycol)/solvent
[340]. Borodin, O., Smith, G. D., & Jaffe, R. L. (2001). Ab initio quantum chemistry and molecular
152
Computational Chemistry, 22(6), 641-654.
[341]. Nowakowska, M., Nawalany, K., Kępczyński, M., & Krawczyk, Z. (2009). Novel nanostructural
hybride materials for photodynamic theraphy. In Macromolecular Symposia (Vol. 279, No. 1, pp.
132-137).
[342]. Hakem, I. F., Lal, J., & Bockstaller, M. R. (2004). Binding of monovalent ions to PEO in
[343]. Magarkar, A., Karakas, E., Stepniewski, M., Róg, T., & Bunker, A. (2012). Molecular dynamics
simulation of PEGylated bilayer interacting with salt ions: A model of the liposome surface in the
[344]. Stepniewski, M., Pasenkiewicz-Gierula, M., Róg, T., Danne, R., Orlowski, A., Karttunen, M.,
Urtti, A., Yliperttula, M., Vuorimaa, E. & Bunker, A. (2011). Study of PEGylated lipid layers as
a model for PEGylated liposome surfaces: Molecular dynamics simulation and Langmuir
[345]. Pannuzzo, M., De Jong, D. H., Raudino, A., & Marrink, S. J. (2014). Simulation of polyethylene
glycol and calcium-mediated membrane fusion. The Journal of Chemical Physics, 140(12),
124905.
[346]. Magarkar, A., Róg, T., & Bunker, A. (2014). Molecular dynamics simulation of inverse-
[347]. Winter, N. D., Murphy, R. K., O’Halloran, T. V., & Schatz, G. C. (2011). Development and
[348]. Perttu, E. K., Kohli, A. G., & Szoka Jr, F. C. (2012). Inverse-phosphocholine lipids: A remix of a
[349]. Lee, H., & Pastor, R. W. (2011). Coarse-grained model for PEGylated lipids: Effect of
153
PEGylation on the size and shape of self-assembled structures. The Journal of Physical Chemistry
B, 115(24), 7830-7837.
[351]. Shinoda, W., Discher, D. E., Klein, M. L., & Loverde, S. M. (2013). Probing the structure of
11556.
[352]. Janke, J. J., Bennett, W. D., & Tieleman, D. P. (2014). Oleic acid phase behavior from molecular
[353]. Dan, N. (2014). Nanostructured lipid carriers: Effect of solid phase fraction and distribution on
[354]. Dan, N. (2015). Drug release through liposome pores. Colloids and Surfaces B: Biointerfaces,
126, 80-86.
[355]. Vitorino, C., Almeida, J., Gonçalves, L. M., Almeida, A. J., Sousa, J. J., & Pais, A. A. C. C.
experimental design to the molecular detail. Journal of Controlled Release, 167(3), 301-314.
[356]. Jämbeck, J. P., Eriksson, E. S., Laaksonen, A., Lyubartsev, A. P., & Eriksson, L. A. (2014).
validation, and simulations with a coarse-grained model. Journal of Chemical Theory and
[357]. Van Hoof, B., Markvoort, A. J., van Santen, R. A., & Hilbers, P. A. (2014). Molecular simulation
3346-3354.
[358]. Novina, C. D., & Sharp, P. A. (2004). The RNAi revolution. Nature, 430(6996), 161–164.
154
[359]. Wang, S. L. (2013). Optimizing drug delivery: Characterization of DLin-KC2-DMA /
[360]. Rozmanov, D., Baoukina, S., & Tieleman, D. P. (2014). Density based visualization for
[361]. Martin-Molina, A., Luque-Caballero, G., Faraudo, J., Quesada-Perez, M., & Maldonado-
Valderrama, J. (2014). Adsorption of DNA onto anionic lipid surfaces. Advances in Colloid and
[362]. Ajnai, G., Chiu, A., Kan, T., Cheng, C. C., Tsai, T. H., & Chang, J. (2014). Trends of gold
nanoparticle-based drug delivery system in cancer therapy. Journal of Experimental & Clinical
[363]. Yang, A. C., & Weng, C. I. (2010). Structural and dynamic properties of water near monolayer-
protected gold clusters with various alkanethiol tail groups. The Journal of Physical Chemistry C,
114(19), 8697-8709.
[364]. Milano, G., Santangelo, G., Ragone, F., Cavallo, L., & Di Matteo, A. (2011). Gold
nanoparticle/polymer interfaces: All atom structures from molecular dynamics simulations. The
[365]. Yu, J., Becker, M. L., & Carri, G. A. (2011). The influence of amino acid sequence and
functionality on the binding process of peptides onto gold surfaces. Langmuir, 28(2), 1408-1417.
[366]. Lee, K. H., & Ytreberg, F. M. (2012). Effect of gold nanoparticle conjugation on peptide
[367]. Van Lehn, R.C., Ricci, M., Silva, P.H., Andreozzi, P., Reguera, J., Voïtchovsky, K., Stellacci, F.
& Alexander-Katz, A. (2014). Lipid tail protrusions mediate the insertion of nanoparticles into
155
[368]. Van Lehn, R. C., & Alexander-Katz, A. (2014). Free energy change for insertion of charged,
nanoparticles in aqueous salt solution from atomistic molecular dynamics simulations. The
[370]. Van Lehn, R. C., & Alexander-Katz, A. (2013). Ligand-mediated short-range attraction drives
[371]. Marrink, S. J., De Vries, A. H., & Tieleman, D. P. (2009). Lipids on the move: Simulations of
membrane pores, domains, stalks and curves. Biochimica et Biophysica Acta (BBA)-
156
Chapter Four: Ionizable Amino Lipid Interactions with POPC: Implications for Lipid
Nanoparticle Function
Contributions:
I ran and analyzed all the simulations and wrote the major part of this chapter. I had major contributions
in methodology, data interpretation, hypothesizing, model development and validation, and visualization.
The experimental data and methods sections were provided by Dr. Jenifer L. Thewalt, Dr. Sherry S.W.
Leung, Dr. Miranda L. Schmidt, and Iulia Bodnariuc at the Simon Fraser University, and by Dr. Pieter R.
Abbreviations:
LNP, lipid nanoparticle; ICL, ionizable cationic lipid; 2H NMR, deuterium nuclear magnetic resonance;
MD, molecular dynamics; 𝑆𝐶𝐷 , deuterium order parameter for C-D bond; POPC, 1-palmitoyl-2-oleoyl-sn-
cationic lipid; KC2H, KC2-protonated; KC2(H), KC2 or KC2H; Cryo-TEM, cryogenic transmission
4.1 Abstract
Lipid nanoparticles (LNPs) based on ionizable cationic lipids are currently the leading systems for siRNA
delivery in liver disease, with a major limitation of low release efficacy in the cytoplasm. Ionizable
cationic lipids are known to be of critical importance in LNP structure and stability, siRNA loading, and
siRNA release stage. However, their distribution inside the LNPs and their exact role in delivery efficacy
are not known. A recent study [Kulkarni et al, ACS nano, 2018] on siRNA-LNPs containing cationic lipid
157
DLin-KC2-DMA (also known as KC2 with an apparent pKa of ca. 6.7) suggested that neutral KC2
segregates from other components and form an amorphous oil droplet in the core of LNPs. However, no
direct evidence for the presence of KC2 in the core was provided.
We further studied KC2 segregation in the presence of POPC using molecular dynamics
simulation, deuterium NMR, SAXS, and cryo-TEM experiments. Our findings strongly suggest that
neutral KC2 has a high tendency to separate from POPC dispersions, further supporting the model
proposed in [Kulkarni et al, ACS nano, 2018]. KC2 confinement, upon raising the pH during the
formulation process, could rearrange the internal structure of LNPs and consequently affect the LNP
transfection efficacy. As interactions between cationic KC2 and anionic endosomal lipids are thought to
be a key factor in cargo release, KC2 confinement inside the LNP may lower the efficacy of release.
4.2 Introduction
Lipid nanoparticles (LNPs) containing ionizable cationic lipids (ICLs) are among the leading drug
delivery systems with promising applications for siRNA and mRNA delivery [1-4]. One current
limitation of these systems is the low release efficacy of siRNA, estimated to be 1-2% [5]. The size,
surface composition, and lipid distribution inside the LNP, as well as the internal structure of these
systems are known to affect their transfection efficacy to a considerable extent [2, 6-8]. The ionizable
cationic lipid (ICL) component of these siRNA-LNPs, with an apparent pKa of less than 7.0 [9], plays a
critical role in siRNA loading, internal arrangement of biomolecules in the LNPs, and endosomal escape
of the therapeutics.
DLin-KC2-DMA, also known as KC2, is one of the recently developed and optimized ICLs with
a reported apparent pKa of ca. 6.7 [9, 10]. This pKa ensures a neutral or low cationic surface charge for
the LNP in physiological pH and a high positive surface charge in endosomal pH. Electrostatic
interactions between cationic lipids and naturally occurring anionic lipids in endosomal membranes have
158
been proposed as the underlying mechanism of drug release for RNAi-LNPs containing (ionizable)
cationic lipids [2, 9, 11]. Thus, the presence of protonated KC2 on the LNP surface is assumed to be
required for destabilizing the endosomal membrane and releasing siRNA. However, there are still
discrepancies regarding the detailed internal structure and lipid distribution across this class of LNPs [2,
12, 13]. Previous work, involving cryo-TEM, has shown that LNP-siRNA systems containing KC2 form
structures with electron dense cores [6, 12-15]. This electron-dense core was hypothesized to be the result
of inverted micellar structures generated through rapid-mixing procedures [12, 14]. More recently, it was
determined that LNPs containing KC2 generated liposomal structures when the KC2 is protonated at pH
4, while forming electron-dense structures at pH 7.4 [13]. This suggests that KC2 might adopt an oil-
phase when neutralized. However, no direct evidence for localization of neutral KC2 in the LNP’s core
was found.
Understanding the structural properties of LNPs is difficult due to the inherently complex nature
of LNP manufacturing. To what extent are the fundamental interactions between ionizable cationic lipids
such as KC2 and phospholipids responsible for the sequestration of neutral KC2 away from other typical
LNP constituents such as phosphatidylcholine? In this study we used molecular dynamics (MD)
simulation, solid state deuterium NMR (2H NMR) and small-angle X-ray scattering (SAXS) experiments
to investigate the interactions between KC2 and POPC – lipid bilayers in simulations and multilamellar
dispersions in experiments - as a function of pH, temperature and mixing ratio. To further investigate
KC2 localization in LNPs we performed cryo-TEM experiments on LNPs composed of POPC, KC2 and
cholesterol.
159
4.3 Methods
4.3.1 Approach.
We used 2H NMR to measure the deuterium order parameter (𝑆𝐶𝐷 ) for the POPC palmitoyl chain in
POPC/KC2 and POPC/KC2H bilayers, to study the effect of KC2 and its protonated state (KC2H) on the
POPC bilayer structure. POPC/KC2 systems of four different mixing ratios (100/00, 90/10, 80/20, and
70/30) were studied at two pH levels of 4.4 and 8.1, and at several temperatures in a range of 288 to 313
K (section 4.3.3).
The 𝑆𝐶𝐷 profiles at 313 K were further used to develop force field parameters for KC2. Both KC2
and KC2H were parameterized using the CHARMM36 force field (C36 FF) [16]. Bilayers composed of
POPC and KC2 or KC2H were simulated, and the 𝑆𝐶𝐷 profiles for the POPC palmitoyl acyl chain
obtained from simulations were compared with the corresponding profiles extracted from 2H NMR
Simulations using the validated models were used to study the POPC/KC2 binary mixtures as a
function of pH and mixing ratio [Table 4-1] (section 4.3.2). Herein again, systems with all the above-
mentioned mixing ratios were simulated at acidic, neutral and basic pH levels. Considering the pKa of ~
6.7 for KC2, the POPC/KC2 systems, where all the KC2 were taken as neutral, correspond to pH levels of
7.4 to 8.1. Also, the POPC/KC2H bilayers, with all the KC2 taken as protonated (KC2H), correspond to
experimental systems at pH of ~ 4.
Based on the results from MD simulations, additional SAXS and cryo-TEM experiments were
designed and conducted. SAXS experiments were used to measure the bilayer repeat spacing (d-spacing)
in the POPC/KC2 lamellar phase for several mixing ratios at basic pH values, whereas cryo-TEM
experiments were conducted to detect possible KC2 segregation from POPC. In the cryo-TEM
experiments, cholesterol was added to the POPC/KC2 and POPC/KC2H systems (section 4.3.3).
160
4.3.2 Computational method.
All the systems (binary systems and pure POPC bilayers) were constructed using an in-house developed
script which is based on GROMACS package [17] tools as described in section 2.5.1 of Chapter 2.
Briefly, for the binary mixtures, two monolayers each composed of only one lipid type were built first.
Each monolayer was composed of 100 lipids distributed on a two-dimensional grid (10 by 10). Two
GROMACS modules, gmx editconf and gmx genconf, were used to translate and rotate the lipids on each
grid point. Next, a set of random numbers from 1 to 100 was generated to satisfy the desired mixing ratio.
For instance, if POPC/KC2H (80/20) was of interest, 20 random numbers were generated. These 20
numbers were used to both keep the KC2H in 20 grid points, and to delete the POPC in the same grid
points. Finally, the coordinate files from two modified monolayers were merged resulted in one mixed
monolayer with a desired mixing ratio; which was taken as the upper leaflet. Assuming this monolayer is
in the XY-plane, its rotation (using gmx editconf module) around either the X or Y axis gives the lower
leaflet. Adjusting the center of both leaflets to (0,0,0) and subsequently merging the coordinate files for
these two leaflets resulted in the POPC/KC2H (80/20) bilayer. Waters were subsequently added on the
top and bottom of this system. Finally, counter ions (Cl-) were added to neutralize the systems composed
of POPC and KC2H. For the neutral POPC/KC2 systems no ions were added. When excess ion
concentration was of interest, structures at 200 ns from pure POPC simulations and 500 ns from
POPC/KC2H simulations were extracted, and Na+ and Cl- ions were added to these structures to reach the
161
4.3.2.2 Molecular dynamics simulations.
All the simulations were conducted using the GROMACS package (v. 2016.3) (see Table 4-1 for a full
list of systems simulated in this study). The CHARMM36 FF [16] was used to model the molecules in the
system, and about 70 TIP3P water molecules [18, 19] were added to each system to fully hydrate the
bilayer. Periodic boundary conditions were applied in all directions. A time step of 2 fs was used, and the
coordinates were saved every 50000 steps. A semi-isotropic pressure coupling was applied to the system,
with Z direction (the normal direction to the bilayer surface) treated independently from the XY-plane.
The Parrinello-Rahman pressure coupling [20, 21] with a time constant of 5.0 ps was used to keep
the pressure at 1 bar. The temperature was maintained at either 288 K or 313 K using Nose-Hoover
extended ensemble [22, 23]. The temperature for two groups of lipids and water plus ions were coupled
separately using a time constant of 1 ps, and the center of mass translation of these groups were removed
The nonbonded interactions were treated using Verlet cut-off scheme. The van der Waals and
Coulomb interactions were switched at 0.8 nm and 0.0 nm, respectively, whereas both were cut-off at 1.2
nm. No dispersion correction was applied, and PME [24, 25] was used to treat the long-range electrostatic
interactions after the cut-off. The force-switch and potential-shift functions were used as van der Waals
and the coulomb modifier functions, respectively. All bonds involving hydrogen atoms were constrained
The other simulation parameters were mainly the ones recommended by CHARMM-GUI [27] for
using C36 FF in GROMACS package for simulation of lipid bilayers, except that the van der Waals
interactions were switched at 0.8 instead of 1.0 nm [28]. The rest of the parameters were kept as default
Pure POPC, POPC/KC2, and POPC/KC2H systems were simulated for 200 ns, 700 ns, and 700
ns, respectively. The systems with 150 mM salt concentration were simulated for 300 ns [Table 4-1].
162
Table 4-1: Summary of simulations.
(mM) (ns)
80/20
70/30
80/20
70/30
80/20
70/30
All analyses were performed using GROMACS (v. 2016.2) modules, except otherwise mentioned. The
last 100 ns of each simulation was used for the analysis. VMD [29] and Xmgrace were used for
163
Deuterium order parameters: The order parameter is defined as:
where θ is the angle between the C-D bond in the methylene group and the vector normal to the bilayer,
and the angular brackets represent the ensemble average. The SCD parameters for the POPC palmitoyl
chain in each system were calculated from the simulations. The script to calculate the SCD from
simulations was provided by Dr. Thomas Piggot [30]. Given the all-atom nature of the C36 FF, the actual
angle between the C-H bond in each methylene group in the POPC palmitoyl chain and the Z axis (the
normal direction to the membrane surface) was used for the SCD calculation.
Electron and number densities: All the electron/number densities were calculated using the gmx density
module of the GROMACS package. The box in the Z direction was divided into 100 slices and the
POPC P→N vector angle distribution: The distribution of the angle between the POPC P→N vector
and the vector normal to each bilayer leaflet pointing in the direction of the water phase was calculated
using the gmx gangle module of the GROMACS package. The bin width was 0.5 degrees, and graphs
show a running average over 20 data points (10 degrees) to highlight the observed trend in tilt angle
164
4.3.3 Experimental methods.
Lipids:
Polar Lipids Inc. (Alabaster, AL). The POPC-d31 was dissolved in 80/20 benzene/methanol (v/v) and then
the solvent was removed under vacuum to obtain a completely dry powder. The ionizable cationic lipid
Inc., (Vancouver BC). KC2 came in an oil form and was not lyophilized prior to use. All lipids were
stored at -20°C.
Buffers:
HEPES buffer was prepared by dissolving the salts in deuterium depleted water (reagents from Bioshop
Canada Inc., (Burlington, ON) and Sigma-Aldrich Canada (Oakville, ON) respectively). The pH was
adjusted to the desired value by adding NaOH (EM Science, affiliates of Merck KGaA (Darmstadt,
Germany)).
Appropriate amounts of each lipid were weighed out and the lipids were co-dissolved in 1-2 mL of a
mixture of 80/20 (v/v) benzene/methanol. The solvent was removed in two steps: first N2 gas was used to
evaporate the bulk of the solvent until only a thin lipid film was left, then any residual solvent was
eliminated by lyophilization of the sample using a strong vacuum pump for several hours or overnight
165
Samples contained approximately 50 – 60 mg total lipids and were hydrated in an excess of 10
mM HEPES buffer (~625-650 μL) at the desired pH. A series of five freeze (in liquid N2) – thaw (room
temperature) – vortex cycles were performed to ensure sample homogeneity. The sample was then
transferred into the NMR tube and/or a quartz capillary tube (Charles Supper Company Inc. (Natick,
MA)) for X-ray analysis. Samples in NMR tubes were stored at -20° C. However, since the quartz
capillary tube were sealed with wax, samples for X-ray analysis were put into the capillary tube either
After the experiments under basic pH conditions were completed, the pH of the sample was
changed inside the sample tube by adding an appropriate amount of glacial acidic acid (Anachemia
Science Canada Inc. (Vancouver, BC)), and then vortexing the sample to ensure homogeneous mixing.
LNPs were prepared as previously described [13]. Briefly, lipid components (POPC, cholesterol and
KC2) dissolved in ethanol were mixed at the appropriate ratios to a final lipid concentration of 20 mM.
The lipids in ethanol were mixed with 25 mM Na Acetate pH 4 buffer using a T-junction mixer [31-33].
The resulting suspension was dialysed against 1000-fold volumes of pH 4 buffer or phosphate buffered
saline (PBS) overnight. The LNPs were then concentrated using 10 kDa NWCO Amicon centrifugal units
MHz and a TecMag Scout (TecMag, Inc. (Houston, TX)) spectrometer. The standard quadrupolar echo
pulse sequence was used. The two out of phase 90° pulses were 3.95 μs long, the interpulse delay was 40
166
μs and the recycle delay was 300 ms. Data was collected using 8-cycle CYCLOPS phase cycling. All
temperatures investigated were well above the gel-fluid transition of POPC, and an equilibration time of
45 minutes was used for each 5 degree temperature increment. 50 000 scans were collected to ensure an
adequate signal-to-noise ratio in the spectra that were de-Paked using the iterative method described by
The first moment of a deuterium NMR spectrum is proportional to the average quadrupolar
𝑥
1 Equation 4-2
𝑀1 = ∑ |𝜔|𝑓(𝜔)
𝐴
𝜔=−𝑥
where ω is the frequency, f(ω) describes the spectrum centered at 0, A is the spectral area ( A =
∑xω=−x f(ω)), and ±x is chosen such that the entire spectral area is contained within the limits [35]. The
where θ is the angle between the C-D bond vector and the bilayer normal, and can be obtained from the
3 Equation 4-4
𝛥𝜈𝑄 = (167𝑘𝐻𝑧)|𝑆𝐶𝐷 |
4
167
where ΔνQ is the quadrupolar splitting and 167 kHz is the quadrupolar coupling constant. The smoothed
order parameter profiles were obtained experimentally by measuring the quadrupolar splittings of the de-
Small-angle X-ray scattering (SAXS) curves were obtained using a SAXSLAB Ganesha 300XL
(Skovlunde, Denmark) setup with a Linkam variable temperature sample stage. The X-ray source was Cu-
Kα (λ= 1.54 Å). The Linkam stage sample to detector distance was calibrated using AgBeh and the
temperature was calibrated using the pre-transition and chain melting transition of both DPPC and DMPC
bilayers. All temperatures reported are the corrected temperatures. Disposable thin-walled quartz capillary
tubes (80 mm long, 1.5 mm (outer diameter), 0.01 mm thickness) were used to hold the samples. To
ensure that the sample was equilibrated and stable, two sets of data were typically collected at each
temperature: one after 30 minutes, and the second 20 minutes after the completion of the first run (an hour
after the setpoint temperature was reached). Each data collection run was 600 s. Initial and final data were
collected at the same temperature to check for any changes in the sample over the course of the
temperature run.
The bilayer repeat spacing (d-spacing) was measured from the SAXS scattering curves using the
following relationship
where qpeak is the position of the peak and n is its order number. The peak order number depends on the
168
4.3.3.4 Cryo-TEM of LNPs.
Cryo-TEM was performed as previously described [13]. Briefly, 3-5 L of concentrated LNP suspensions
were added to glow-discharged copper grids. A FEI Mark IV Vitrobot (FEI, Hillsboro, OR) was used to
plunge-freeze blotted grids into liquid ethane. Vitrified samples were stored under liquid nitrogen. Frozen
grids were imaged using a FEI Tecnai G2 (FEI, Hillsboro, OR) operating at 200 kV in low-dose mode
and images were obtained using an FEI Eagle 4K CCD camera (FEI, Hillsboro, OR). All sample
preparation and imaging were performed at the UBC BioImaging Facility (Vancouver, BC).
Using the present technique, we have shown that POPC liposomes produced by rapid-mixing are
extremely small and poorly resolved by cryo-TEM. However, systems containing cholesterol are larger
and thus can be imaged easily [38, 39]. Additionally, it was shown that replacement of PC-lipid with
ICLs resulted in decreased particle sizes [13]. Thus, in order to study the effect of KC2(H) on
nanoparticles (at sizes that are readily resolved by the present technique) cholesterol was included in the
formulation.
In simulations, starting from a binary lipid mixture composed of POPC and neutral KC2 in a bilayer form,
almost all the neutral KC2 segregated from POPC and incorporated into the hydrophobic interior between
two POPC monolayers [Figure 4.1-a, pH 7.4]. In contrast, KC2H, with a net charge of +e, stays in the
lipid-water interface and interacts with water molecules, ions, and POPC headgroups [Figure 4.1-a, pH
4], at both temperatures and all mixing ratios studied [Table 4-1]. These observations in simulations were
169
The cryo-TEM micrographs of POPC/KC2/cholesterol dispersions are shown in Figure 4.1-b.
These micrographs suggest that all particles (regardless of KC2 content) displayed bilayer structures at
pH 4.0, but not at pH 7.4. For systems with no KC2, raising the pH from 4.0 to 7.4 changed the particle
structures from unilamellar vesicles to bi-/oligo-lamellar structures. When 30 mol % KC2 was added,
unilamellar vesicles but with smaller average sizes compared to the POPC/cholesterol dispersions were
formed at low pH. However, at pH 7.4 larger multi-lamellar vesicles with electron-dense cores were
formed. The appearance of this electron dense core upon increasing pH level supports the KC2 separation
from POPC and cholesterol, and formation of the oily cores. Furthermore, systems with 30 mol% KC2 at
pH 7.4 were larger than the same systems at pH 4.0, implying that particle fusion has occurred [13].
Taken together, the separation of POPC and KC2 is supported by the cryo-TEM of
cholesterol which makes them different from POPC/KC2(H) dispersions studied in simulations. As
mentioned in the Methods sections, addition of cholesterol to POPC/KC2(H) was required to form
nanoparticles which could be imaged easily. On the other hand, these systems are more directly related to
170
Figure 4.1: KC2 segregation from POPC as a function of pH.
a) Bilayers are composed of POPC and KC2 or KC2H. Snapshots are taken from systems with 70 mol%
POPC. Green and red spheres represent the nitrogen atoms of KC2 and phosphorus atoms of POPC,
whereas the blue and orange lines are representative of KC2 and POPC acyl chains, respectively.
171
Hydrogen atoms, ions, and water molecules are not shown for clarity. The simulation box is shown as a
blue rectangle. The vertical line represents the phosphate-phosphate bilayer thickness defined as the
ensemble averaged distance between the red spheres (phosphorus atoms of POPC) in two POPC leaflets.
Order parameter ( 𝑆𝐶𝐷 ) profiles from 2H NMR can also be used for characterizing lipid
organization and interactions in lipid mixtures. 2H NMR 𝑆𝐶𝐷 profiles of POPC-d31/KC2(H) systems are
shown in Figure B.2 and Figure B.3. At 288 K, adding KC2H to POPC-d31 increases palmitoyl chain
order [Figure B.2], implying that KC2H intercalates between POPC-d31 molecules as predicted by the
MD simulations. In contrast, adding even 30 mol% KC2 to POPC-d31 has little to no effect on the
conformational freedom of the palmitoyl chain. This suggests that KC2 does not interact with POPC-d31
chains to a significant extent, since KC2H does affect POPC-d31 palmitoyl chain order. Note, though,
that it is possible for amphiphiles to intercalate among phospholipid chains without affecting chain order.
For example, 25 mol% 1-decanol does not affect DPPC palmitoyl chain order parameters [40, 41]. But
given that KC2H significantly orders POPC-d31, despite its charged dimethylamine, it is likely that
KC2’s lack of effect on POPC-d31 chain order stems from a lack of interaction. The observed differences
between POPC/KC2(H) order parameter profiles at low and high pH become less significant with heating
and are insignificant at 313 K [Figure B.3], likely because thermal energy enhances chain fluctuations
Our SAXS results can also be interpreted to suggest that KC2 and POPC are segregated [Figure
B.4]. The broadening of scattering peaks upon the incorporation of KC2 implies that the lamellae are
significantly less correlated than in pure POPC, as would be the case if they were interrupted by pools of
172
KC2. KC2 is silent in the SAXS patterns, and is likely separating into oily droplets associated with the
POPC lamellae and thereby disrupting their stacking regularity. As well, the increases in lamellar repeat
spacing observed for POPC/KC2 multilamellar dispersions at pH 8.1 or 8.5 (section B.3) are consistent
with the interbilayer repulsion expected for the small amount of KC2H at these pH values [42].
Both simulations and experiments suggest that KC2 separates from POPC, but the distribution of
KC2 in the system is not the same between two approaches. In simulations, the segregated KC2 lipids are
confined within the bilayer center. This confinement results in a systematic increase in the POPC inter-
leaflet distance [Figure 4.2]; from 3.72 nm to 4.18 nm, 4.44 nm, and 5.32 nm upon raising the KC2 molar
fraction from 0 % to 10 %, 20 %, and 30 %, respectively. The KC2 incorporation in the bilayer center was
not, however, supported by the SAXS experiments. In the cryo-TEM experiments, furthermore, the
neutral KC2 goes to the nano-formulation core [Figure 4.1-b]. This difference in KC2 distribution
between simulations and experiments might be a direct effect of periodic boundary condition in the
simulation. The complete separation of KC2 to the aqueous phase is not energetically favorable (or
allowed) due to the applied boundary condition. Indeed, in the simulations confinement into the bilayer
interior is the only possible scenario, unless a massive system is simulated that allows (energetically
173
Figure 4.2: Systematic increase in POPC inter-leaflet distance upon rising the KC2 concentration at
basic pH level.
Symmetrized electron densities for the phosphorus atoms of POPC were calculated from the simulated
systems with different mixing ratios at neutral/basic pH and T = 313 K. The X axis shows the relative
position from the bilayer center in the direction normal to the bilayer surface. Each double-headed arrow
represents the peak-to-peak distance for a mixing ratio, corresponding to the phosphate-to-phosphate (P-
P) inter-leaflet distance. For the pure POPC at 313 K, the P-P distance is about 3.72 nm.
174
The KC2 incorporation into the POPC bilayer center is not, however, impossible, and might be
the case for systems with small content of KC2. There are experimental reports on the alkanes segregating
into the middle of bilayers [43], or into the voids in the inverted hexagonal structures [44]. Furthermore,
there are reports on the confinement of water-soluble polymers into the center of bilayers composed of
non-ionic surfactants [45]. There are also computational studies where partitioning of hydrophobic
alkanes and polymers in the bilayer interior has been reported [46]. Although there are numerous
examples on partitioning of both small and large drug molecules into the lipid membranes, there is no
Simulation also suggests that the level of KC2 confinement was a function of KC2 concentration
[Figure 4.3]. That being said the KC2 percentage confined in the hydrophobic section is higher for the
system with a higher concentration of KC2. For the 10 mol% KC2, a few percent of the lipids could stay
in the lipid-water interface but as the concentration of KC2 increased to 20 % and 30 %, more and more
175
Figure 4.3: KC2 confinement as a function of mixing ratio.
The symmetrized number densities for the KC2 nitrogen atoms and POPC phosphorus atoms are shown at
neutral/basic pH and T =313 K for two mixing ratios: Left) POPC/KC2 (90/10), and Right) POPC/KC2
(70/30). Assuming the bilayers are centered at zero, the densities for the positive positions are only
shown. The X axis shows the direction normal to the bilayer surface. The insets are snapshots taken from
the corresponding systems. The red and green spheres represent the POPC phosphorus atoms and KC2
nitrogen atoms, respectively. The blue box represents the simulation box. Water, ions, and other atoms in
POPC and KC2 are not shown for clarity. System with 30 mol% KC2 seems to have less KC2 headgroups
in the lipid-water interface (shown as the POPC phosphorus atom densities) compared to the system with
10 mol% KC2.
These results suggest that an increase in pH from 4.4 to 7.4/8.1 causes the KC2 to separate from
POPC. Results from POPC/KC2(H) and POPC/KC2(H)/cholesterol systems can be further related to the
LNPs because phosphatidylcholine, KC2(H) and cholesterol are three major components of LNPs.
Therefore, interactions between these molecules play critical roles on the internal structure and
176
consequently the drug delivery efficacy of LNPs. However, LNPs are more complex and contain other
molecules in their formulations. For instance, LNPs contain distearoylphosphatidylcholine (DSPC) which
is different from POPC structurally, and PEG-lipids which assist in stabilizing the LNPs [3]. These
differences could result in different molecular behavior than what was observed in the simple model
Considering these limitations, our findings suggest that KC2 is the major component forming the
oil droplets observed in the core of LNPs [12, 13]. As proposed elsewhere [13], these oil droplets are
expected to be stabilized by a lipid monolayer on the surface. The material for this covering layer would
be drawn from the rest of the LNP particle, which consequently rearranges the structural components of
the LNP membranes. This rearrangement of the molecules in the LNP could affect the internal structure
of LNP and destabilizes it. Given a certain number of phospholipids present in the formulation, these
destabilized LNPs are fused together and form larger LNPs which are more stable. The underlying reason
for this stability is that, assuming a spherical shape for the LNP, the surface to volume ratio for the
spherical nanoparticles decreases as the radius increases. Thus, larger LNPs will be more stable because
The confinement of KC2 inside the LNP might also cause a delay in the detection of endosomal
acidification, and a consequent delay in KC2 protonation and their transfer to the LNP surface. Lack of
sufficient KC2H on the LNP’s surface causes a delay in the LNP’s attachment to and effective
interactions with negatively charged endosomal membrane. Assuming that LNPs attached to the internal
side of the endosomal membrane, the destabilizing step of endosomal membrane will cause another level
Despite the neutral KC2 which were segregated from POPC, KC2H stays in the bilayer along
with POPC and interact with other molecules in the lipid-water interface. Both 2H NMR experiments and
simulations suggest that including KC2H in the system slightly induces order in the POPC palmitoyl
177
chain in a concentration-dependent way [Figure B.2 and Figure B.3]. In the simulations, KC2H were
shown to leave the P-P distance almost unaffected [Figure B.5]. Moreover, adding KC2H to the systems
was shown to attract Cl- ions to the membrane surface, and push the Na+ ions far from the lipid-water
interface, as expected [Figure B.6]. It also affects the POPC choline group distribution which
consequently affects the POPC P→N vector orientation [Figure B.7]. These observations are consistent
4.5 Conclusions
Our results suggest neutral KC2 is segregated from POPC, while charged KC2H mixes with POPC in the
lipid bilayer. KC2H affects the POPC head group orientation but had no effect on the bilayer thickness.
The segregation of KC2 supports the recently proposed model for LNPs containing KC2 [13]. The
sequestration of KC2 from POPC might also explain why this class of LNPs suffers from a low efficacy
4.6 Bibliography
[1]. Zimmermann, T.S., Lee, A.C., Akinc, A., Bramlage, B., Bumcrot, D., Fedoruk, M.N., Harborth,
J., Heyes, J.A., Jeffs, L.B., John, M., Judge, A. D. et al. (2006). RNAi-mediated gene silencing in
[2]. Arteta, M.Y., Kjellman, T., Bartesaghi, S., Wallin, S., Wu, X., Kvist, A.J., Dabkowska, A.,
Székely, N., Radulescu, A., Bergenholtz, J. & Lindfors, L. (2018). Successful reprogramming of
178
[3]. Tam, Y., Chen, S., & Cullis, P. R. (2013). Advances in lipid nanoparticles for siRNA delivery.
[4]. Wan, C., Allen, T. M., & Cullis, P. R. (2014). Lipid nanoparticle delivery systems for siRNA-
[5]. Gilleron, J., Querbes, W., Zeigerer, A., Borodovsky, A., Marsico, G., Schubert, U., Manygoats,
K., Seifert, S., Andree, C., Stöter, M. & Epstein-Barash, H. (2013). Image-based analysis of lipid
[6]. Chen, S., Tam, Y. Y. C., Lin, P. J., Sung, M. M., Tam, Y. K., & Cullis, P. R. (2016). Influence of
particle size on the in vivo potency of lipid nanoparticle formulations of siRNA. Journal of
[7]. Koltover, I., Salditt, T., Rädler, J. O., & Safinya, C. R. (1998). An inverted hexagonal phase of
cationic liposome-DNA complexes related to DNA release and delivery. Science, 281(5373), 78-
81.
[8]. Kim, H., & Leal, C. (2015). Cuboplexes: Topologically active siRNA delivery. ACS nano, 9(10),
10214-10226.
[9]. Semple, S.C., Akinc, A., Chen, J., Sandhu, A.P., Mui, B.L., Cho, C.K., Sah, D.W., Stebbing, D.,
Crosley, E.J., Yaworski, E., Hafez, I. M., et al. (2010). Rational design of cationic lipids for
[10]. Jayaraman, M., Ansell, S.M., Mui, B.L., Tam, Y.K., Chen, J., Du, X., Butler, D., Eltepu, L.,
Matsuda, S., Narayanannair, J.K. & Rajeev, K. G. (2012). Maximizing the potency of siRNA
lipid nanoparticles for hepatic gene silencing in vivo. Angewandte Chemie International Edition,
51(34), 8529-8533.
179
[11]. Chan, C. L., Majzoub, R. N., Shirazi, R. S., Ewert, K. K., Chen, Y. J., Liang, K. S., & Safinya, C.
[12]. Leung, A.K., Hafez, I.M., Baoukina, S., Belliveau, N.M., Zhigaltsev, I.V., Afshinmanesh, E.,
Tieleman, D.P., Hansen, C.L., Hope, M.J. & Cullis, P. R. (2012). Lipid nanoparticles containing
[13]. Kulkarni, J.A., Darjuan, M.M., Mercer, J.E., Chen, S., van der Meel, R., Thewalt, J.L., Tam,
Y.Y.C. & Cullis, P. R. (2018). On the formation and morphology of lipid nanoparticles
containing ionizable cationic lipids and siRNA. ACS nano, 12(5), 4787-4795.
[14]. Leung, A. K., Tam, Y. Y. C., Chen, S., Hafez, I. M., & Cullis, P. R. (2015). Microfluidic mixing:
A general method for encapsulating macromolecules in lipid nanoparticle systems. The Journal of
[15]. Belliveau, N.M., Huft, J., Lin, P.J., Chen, S., Leung, A.K., Leaver, T.J., Wild, A.W., Lee, J.B.,
Taylor, R.J., Tam, Y.K. & Hansen, C. L. (2012). Microfluidic synthesis of highly potent limit-
size lipid nanoparticles for in vivo delivery of siRNA. Molecular Therapy-Nucleic Acids, 1, e37.
[16]. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez,
C., Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom
additive force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B,
114(23), 7830-7843.
[17]. Hess, B., Kutzner, C., Van Der Spoel, D., & Lindahl, E. (2008). GROMACS 4: Algorithms for
highly efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory
180
[18]. Durell, S. R., Brooks, B. R., & Ben-Naim, A. (1994). Solvent-induced forces between two
[19]. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983).
Comparison of simple potential functions for simulating liquid water. The Journal of Chemical
[20]. Parrinello, M., & Rahman, A. (1981). Polymorphic transitions in single crystals: A new
[21]. Nosé, S., & Klein, M. L. (1983). Constant pressure molecular dynamics for molecular systems.
[22]. Nosé, S. (1984). A molecular dynamics method for simulations in the canonical ensemble.
[24]. Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N⋅log(N) method for
Ewald sums in large systems. The Journal of Chemical Physics, 98(12), 10089-10092.
[25]. Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., & Pedersen, L. G. (1995). A
smooth particle mesh Ewald method. The Journal of Chemical Physics, 103(19), 8577-8593.
[26]. Hess, B., Bekker, H., Berendsen, H. J., & Fraaije, J. G. (1997). LINCS: A linear constraint solver
[27]. Lee, J., Cheng, X., Swails, J.M., Yeom, M.S., Eastman, P.K., Lemkul, J.A., Wei, S., Buckner, J.,
Jeong, J.C., Qi, Y. & Jo, S. (2015). CHARMM-GUI input generator for NAMD, GROMACS,
181
[28]. Pluhackova, K., Kirsch, S. A., Han, J., Sun, L., Jiang, Z., Unruh, T., & Böckmann, R. A. (2016).
A critical comparison of biomembrane force fields: Structure and dynamics of model DMPC,
POPC, and POPE bilayers. The Journal of Physical Chemistry B, 120(16), 3888-3903.
[29]. Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: Visual molecular dynamics. Journal of
[30]. Piggot, T. J., Allison, J. R., Sessions, R. B., & Essex, J. W. (2017). On the calculation of acyl
chain order parameters from lipid simulations. Journal of Chemical Theory and Computation,
13(11), 5683-5696.
[31]. Hirota, S., de Ilarduya, C. T., Barron, L. G., & Szoka Jr, F. C. (1999). Simple mixing device to
[32]. Jeffs, L. B., Palmer, L. R., Ambegia, E. G., Giesbrecht, C., Ewanick, S., & MacLachlan, I.
(2005). A scalable, extrusion-free method for efficient liposomal encapsulation of plasmid DNA.
[33]. Kulkarni, J. A., Tam, Y. Y. C., Chen, S., Tam, Y. K., Zaifman, J., Cullis, P. R., & Biswas, S.
[34]. Sternin, E., Bloom, M., & Mackay, A. L. (1983). De-Pake-ing of NMR spectra. Journal of
[35]. Davis, J. H. (1979). Deuterium magnetic resonance study of the gel and liquid crystalline phases
[36]. Lafleur, M., Fine, B., Sternin, E., Cullis, P. R., & Bloom, M. (1989). Smoothed orientational
order profile of lipid bilayers by 2H-nuclear magnetic resonance. Biophysical Journal, 56(5),
1037-1041.
[37]. Schnablegger, H., & Singh, Y. (2013). The SAXS guide. Anton Paar GmbH.
182
[38]. Zhigaltsev, I. V., Belliveau, N., Hafez, I., Leung, A. K., Huft, J., Hansen, C., & Cullis, P. R.
(2012). Bottom-up design and synthesis of limit size lipid nanoparticle systems with aqueous and
[39]. Zhigaltsev, I. V., Tam, Y. K., Leung, A. K., & Cullis, P. R. (2016). Production of limit size
nanoliposomal systems with potential utility as ultra-small drug delivery agents. Journal of
[40]. Thewalt, J. L., Wassall, S. R., Gorrissen, H., & Cushley, R. J. (1985). Deuterium NMR study of
the effect of n-alkanol anesthetics on a model membrane system. Biochimica et Biophysica Acta
[41]. Thewalt, J. L., Tulloch, A. P., & Cushley, R. J. (1986). A deuterium NMR study of labelled n-
alkanol anesthetics in a model membrane. Chemistry and Physics of Lipids, 39(1-2), 93-107.
[42]. Navas, B.P., Lohner, K., Deutsch, G., Sevcsik, E., Riske, K.A., Dimova, R., Garidel, P. & Pabst,
[43]. White, S. H., King, G. I., & Cain, J. E. (1981). Location of hexane in lipid bilayers determined by
[44]. Chen, Z., & Rand, R. P. (1998). Comparative study of the effects of several n-alkanes on
[45]. Radlinska, E. Z., Gulik-Krzywicki, T., Lafuma, F., Langevin, D., Urbach, W., Williams, C. E., &
[46]. Bochicchio, D., Panizon, E., Monticelli, L., & Rossi, G. (2017). Interaction of hydrophobic
183
[47]. Gurtovenko, A. A., Patra, M., Karttunen, M., & Vattulainen, I. (2004). Cationic DMPC/DMTAP
184
Chapter Five: Structural Properties of Inverted Hexagonal Phase: A Hybrid
Contributions:
I ran and analyzed all the simulations and wrote the major part of this chapter. I had major contributions
in methodology, data interpretation, hypothesizing, model development and validation, and visualization.
The experimental data and methods sections were provided by Dr. Jenifer L. Thewalt, Dr. Miranda L.
Schmidt, and Bashe Y. M. Bashe at the Simon Fraser University, and by Dr. Paul E. Harper, Jason R.
Abbreviations:
HII, reverse/inverted hexagonal phase; APL, area per lipid; MD, molecular dynamics; DOPE, 1,2-
oleoyl-sn-glycero-3-phosphoethanolamine; SCD, deuterium order parameter for C-D bond; SAXS, small-
angle X-ray scattering; dhex, lattice distance; Rw, water core radius; nw, water molecules per lipid inside
the lipid cylinder (or equivalently hydration level); nc, number of lipids per slice along the lipid cylinder;
2
H NMR, deuterium nuclear magnetic resonance
5.1 Abstract
Inverted/reverse hexagonal (HII) phases are of special interest in several fields of research, including
nanomedicine. Study of the structural properties of HII phase at low hydration levels, as well as
determination of the internal dimensions of HII structure in presence of excess water could be challenging
experimentally.
185
We used molecular dynamics (MD) simulation to study HII systems composed of DOPE and
POPE at several hydration levels and temperatures. The effect of hydration level on several HII structural
parameters, including deuterium order parameters, was investigated. We further used MD simulations to
estimate the maximum hydrations of DOPE and POPE HII lattices at several given temperatures. Finally,
the effect of acyl chains on HII structure was studied via comparing the DOPE with POPE HII systems.
In addition to MD simulations, we used deuterium NMR (2H NMR) and small-angle X-ray scattering
(SAXS) experiments to measure the DOPE acyl chain order parameters, lattice spacings, and the water
core radius in the DOPE HII systems at several temperatures in the presence of excess water.
Structural parameters calculated from MD simulations are in excellent agreement with the
experimental data. In addition, the water core radius calculated directly from MD simulations agrees with
the estimated value based on electron density profiles in SAXS. Dehydration is shown to increase the
lipid cylinder length and decrease the radius of water core. An increase in hydration level slightly
increased the deuterium order parameter of lipids acyl chains, whereas an increase in temperature
decreased it. Lipid cylinders were shown to be undulated along the cylinder axis as a function of
hydration level. Provided that a single experimental measurement for the lattice distance be available in
the presence of excess water, the maximum hydration of PE HII phases at the corresponding temperature
were successfully predicted by MD simulations. An increase in temperature were shown to decrease the
maximum hydration, and consequently the radius of water core and lattice distances. Finally, DOPE was
shown to form HII structures with a higher curvature compared to POPE, as expected. We propose a
general protocol for constructing computational HII systems corresponding to the experimental systems.
This protocol could be used to study HII systems composed of other molecules than PEs, for which
186
5.2 Introduction
amphiphiles (e.g. lipids, peptides, co-polymers) dispersed in polar solvents, e.g. water. Lipid tubules are
filled with water and ions, and are arranged in a hexagonal geometry [1] [Figure 5.1]. Given its special
geometry, HII phase has been of special interest for its (potential) use in several research areas [1-7],
including drug/gene delivery systems [8-10] and drug release mechanism [11-14].
Structural properties of HII has been the subject of many experimental studies. Lattice distance
(dhex), radius of water core (Rw), area per lipid (APL), deuterium order parameter (SCD), and shape of the
water core are a few examples of the parameters studied using experimental techniques [15-20]. Several
factors have been shown to affect the HII structure significantly, including hydration level, temperature,
Experimental studies on structural characterization of HII are usually conducted in the excess
water regime [Figure 5.1], where two phases coexist. The first phase is the HII lattice formed from lipid
cylinders filled with a certain number of water molecules (referred as the maximum hydration) allowed.
The excess water forms the second phase; a bulk of free water molecules outside of the HII lattice.
Estimation of volume fraction of water in each phase is not readily doable [15]. However, volume fraction
of water is required to determine internal dimensions of the HII phase. There are several experimental
approaches and methodologies developed to assist in determination of maximum hydration and internal
dimensions of HII phase. One traditional approach, developed by Tate and Gruner [15], is to conduct a
series of SAXS experiments in low hydration levels and measure the dhex at each hydration level. The
hydration level which dhex saturates is considered as the maximum hydration taken up by the HII lattice.
An alternative approach is based on the diffraction pattern in the SAXS scattering curves [16, 17, 21].
Using the methods developed by Turner and Gruner [16] and Harper et al. [17, 21], the peak intensities in
a scattering curve can be used to reconstruct the electron density profile for the HII structure, from which
187
the Rw can be estimated. Given these values, the internal dimensions of the HII, including the volume
fraction of water in the HII lattice, can be calculated using geometric considerations and the symmetry of
HII phase. In this approach, a single sample prepared in presence of excess water is enough to extract the
HII structural dimensions. This method was shown to result in structural parameters in close agreement
with the traditional method by Tate and Gruner [15], although in a relatively easier way.
Experimental approaches are usually limited in terms of resolution. Regarding the HII phase,
conducting experiments at low hydration levels are usually challenging. It is also challenging to study the
effect of several factors independently. For instance, an increase in temperature results in a decrease in
maximum hydration of HII phase, and therefore the lattice distance [15, 17]. Therefore, to what extent the
structure of HII system and its comprising molecules are affected by hydration level or temperature is
188
Figure 5.1: HII phase in experiments and simulation.
Right) A schematic view for a test tube containing lipid/water dispersions in experiments. HII phase
containing lipids and water molecules form at the bottom of the tube. The excess waters are located on the
top of lipid dispersions, outside of the HII lattice. Left) A schematic view of the part of lipid dispersion
which is modeled and used for the simulations. The blue box represents the simulation box containing one
single lipid tubule and its corresponding water channel. HII phase is a group of these lipid tubules
arranged in hexagonal geometry. The tan colored spheres represent the phosphorus atom of phospholipids
whereas the red spheres are the carbon atoms of CH3 groups at the end of each acyl chain. Water
molecules inside the central lipid tubule is represented as the cyan dots. The rest of the atoms, as well as
the water molecules in the neighbor cells are not shown for clarity. The pink, yellow, and white lines are
the radius of water core (𝑅𝑤 ), lattice distance (𝑑ℎ𝑒𝑥 ), and unit cell spacing (𝑎), respectively. The white
and orange arrows represent the interstitial and interaxial directions, respectively. VMD [22] was used to
189
Computer simulation is a complementary approach to experimental techniques, which can
provide additional details in a molecular level and be used to study each effect independently [23].
Molecular dynamics simulations have been previously used to study the structural and dynamical
properties of HII phase [24-26] and lamellar-to-HII phase transition process as a function of hydration,
temperature, and mixing ratio [27, 28]. There are also computational studies of HII phase used to
calculate the bending free energies, spontaneous curvatures of lipid monolayers, and bending modulus
[29, 30]. It has also been used to determine the pivotal plane - the plane at which the area per molecule at
the given curvature is the same with the flat bilayer - for lipid molecules [31].
Ideally and in most cases, comparing the results from simulations with experiments requires
equivalent molecular systems in both approaches. However, construction of such equivalent systems in
HII phase are not straightforward, especially for systems for which experimental data are not available.
Herein again, one of the key challenges is that the maximum hydration at a certain temperature is not
always known for an arbitrary molecule/mixing ratio. Assuming the hydration level is known, it is still
challenging to make reasonable initial guesses for the water core radius and other HII internal dimensions
to construct the HII structure with. Assuming that both hydration level and system dimensions can be
determined, construction of stable HII systems with the desired mixing ratio is another challenge to
overcome. There are several computational tools designed to assist in overcoming the last challenge [32-
35]. CHARMM-GUI is a popular online toolkit for construction of molecular systems to the pre-
production run step for a variety of systems. However, there are shortcomings when simulation of HII
phase is of interest. For instance, not all the lipids or molecules (e.g. copolymers, detergents, synthetic
lipids, and peptides) are available in the CHARMM-GUI library. In addition, CHARMM-GUI requires
alkanes to be added to the corners of the hexagonal box, does not provide enough control of the number
of waters inside the HII channels, and does not support adding other solvents than water to the HII phase.
190
The output files given by CHARMM-GUI are only for all-atom CHARMM36 force field (C36 FF) but
not other force fields (FFs). These output files are for use in the NAMD package and can not be used in
the GROMACS package directly. Finally, the hexagonal box shape given by CHARMM-GUI is not yet
Therefore, the first objective of this study was to develop a general protocol for construction and
simulation of HII systems composed of amphiphilic molecules using the minimum amount of
experimental data. The second goal was to evaluate the level of agreement between the structural
parameters, including the maximum hydration at a given temperature, predicted by this protocol and the
known experimental values. Finally, it was interesting to see whether this protocol could reproduce the
expected structural differences between structurally similar molecules in HII phase (i.e. sensitivity of the
In this study, we used MD simulations to investigate the structural properties of DOPE and POPE
HII systems as a function of hydration and temperature. Phosphatidylethanolamines (PEs) such as DOPE
and POPE are known to form HII phases at high temperatures and low hydration levels [36]. Inverted
hexagonal systems composed of DOPE and POPE have been extensively studied experimentally, which
could be used for both validation of our protocol and comparing the structural parameters with
simulations. Small angle X-ray scattering experiment was used to measure the lattice distance and water
core radius in DOPE HII phase at several temperatures. Also, deuterium NMR ( 2H NMR) was used to
measure the deuterium order parameter (SCD) for the last carbon atoms in DOPE acyl chains. Several
structural parameters of DOPE and POPE HII systems, including water core radius and SCD, were
calculated from MD simulations and compared with experimental data. The results from DOPE and
POPE simulations at 323 K were further compared to investigate the effect of different acyl chains and to
evaluate the sensitivity of our protocol. Molecular dynamics simulations were also used to estimate the
maximum hydration for both DOPE and POPE HII systems at several temperatures, and to investigate the
191
effect of temperature on both lattice distance and maximum hydration. Finally, a protocol for constructing
5.3 Methods
Although experimental and computational data on both DOPE and POPE HII systems are available, none
of these data were used in construction of the HII molecular systems. This way, we could validate our
protocol and see whether MD simulations could blindly reproduce the experimental data.
For each desired water per lipid ratio inside the cylinder (nw), an initial lipid cylinder filled with
required amount of water molecules was constructed (see below for details). Next, each system was
relaxed during the simulation to adjust its radius and length. Based on temperature, lipid type, and n w,
lipid cylinders were either stretched or shrunk along the cylinder axis during the equilibration step, as a
direct result of simulations. The structural parameters were then calculated from these equilibrated
As mentioned in the caption of Figure 5.1, the excess water in the experimental setups is not
modeled in simulations. Therefore, any water exchange between the excess water and waters inside the
HII lattice is prohibited in simulations. This water exchange, however, is possible in the experiments, for
A single tube composed of either DOPE or POPE lipids was constructed and oriented in the Z direction
using GROMACS [37] modules, as described in section 2.5.2 in Chapter 2 [38]. Next, CHARMM-GUI
Solvator module [39] was used to build a water column with the same dimensions and orientation. The
192
atomic coordinates from these two systems were subsequently merged to give a water channel covered by
lipids. Water molecules were gradually deleted to give the system with the desired n w. A triclinic box was
employed for the simulations and periodic boundary conditions (PBC) were applied in all the three
directions. A few lipid and water molecules were deleted from several DOPE HII systems due to bad
contacts. This resulted in slightly higher hydration for DOPE HII systems than POPE (see Table 5-1 for
details). Each system was subsequently energy minimized using steepest descent algorithm and used for
Preliminary simulation suggested that system with 4 nw tends to deform from its cylindrical
geometry and might not be stable in HII phase. For this system, therefore, a larger system composed of
four lipid cylinders in a rectangular box [40] was simulated instead to allow any possible phase transition
DOPE Anisotropic 283, 303, 323 10.26, 14.35, 16.41, 20.26, 500
193
lipids (351 lipids per cylinder)
POPE Anisotropic 323, 348, 358 10, 14, 16, 20, 22, 24, 26, 30 500
GROMACS version 2016.3 [37] was used for all the simulations. The force field parameters for the
molecules in the system were taken from CHARMM36 FF version Nov.2016 [41], and the standard
TIP3P water model [42, 43] was used for these simulations.
Each energy minimized system was first simulated under NPT ensemble. An anisotropic pressure
coupling was applied, to control the pressure in each direction independently, using the Berendsen [44]
weak coupling algorithm. Next, the output structures were used for the longer equilibration and
For the production runs, a time-step of 2 fs was employed, and atom coordinates were saved
every 100 ps. The van der Waals interactions were switched off and cut-off at 0.8 and 1.2 nm,
respectively, and force-switch function was used to smoothly switch off the interactions. The short-range
electrostatic interactions were cut-off at 1.2 nm and PME [45, 46] was used for treatment of the long-
range electrostatic interactions. Lipid cylinders and solvent were coupled to the heat bath separately.
Temperature was controlled using Nose-Hoover coupling [47, 48] with a time-constant of 1.0 ps.
Parrinello-Rahman extended-ensemble pressure coupling [49, 50] with a time-constant of 5.0 ps was used
to control the pressure. For most of the simulations in this study, an anisotropic pressure coupling was
used, which requires 6 values for each compressibility and reference pressure. A compressibility of 4.5e-5
bar -1 and a reference pressure of 1.0 bar were used for all the diagonal components (i.e. XX, YY, and ZZ)
of compressibility and reference pressure matrices, respectively, whereas all the off-diagonal components
(i.e. XY/YX, XZ/ZX, and YZ/ZY) in these matrices were set to zero. This combination of parameters
194
maintained the box shape through the simulation course while still allowing the box sizes to scale in each
direction independently. In cases, where the effect of pressure coupling was of interest, several systems
were also simulated using the semi-isotropic coupling [Table 5-1]. Assuming the lipid cylinders are along
the Z axis, a semi-isotropic coupling allows the Z direction and XY-plane to be scaled independently.
Herein again, all the off-diagonal components in both compressibility and pressure matrices were set to
-1
zero, whereas a compressibility of 4.5e-5 bar and a reference pressure of 1.0 bar were used for the
diagonal components. LINCS [51] was employed to constrain all the bonds involving hydrogen atoms.
Visual Molecular Dynamics (VMD) [22] and Xmgrace were used to generate the figures and graphs,
respectively, while Python 3.6.0 and Excel was used for some of the analysis.
Several structural parameters of HII systems were calculated from the last 100 ns of each simulation. The
data corresponding to every 100 ps in both structure and energy files were used for the analysis, resulting
in a total of 1000 data points for calculation of averages and standard deviation of means. Each of the
following structural parameters was first calculated for each frame. Then the average and standard
deviations of mean were calculated from these data and shown as the error bars on the corresponding
graphs. The following sections explain the process (in order) by which these parameters were calculated.
First, the PBC effects in simulations were fixed using gmx trjconv module of GROMACS
package; all the broken molecules were made as whole (using the -pbc whole option), and all the atoms in
the system were put at the closest distance from the simulation box center (using the -ur compact option).
Next, the system was centralized in the simulation box (using -center option).
195
Deuterium order parameter:
Given the PBC effects being fixed, the SCD parameters were calculated according to the formula in
Equation 5-1.
Where θ is the angle between the C-D bond in the methylene group and the Z axis (i.e. the cylinder axis
which is the lipid’s symmetry axis in HII phase). The angular brackets represent the ensemble average. A
script provided by Dr. Thomas Piggot [52] was used to calculate the 𝑆𝐶𝐷 from simulations. This script is
based on the actual angle between the C-H bond and the Z axis. Since the cylinder axis is aligned with the
Z axis, this angle will be the angle between the C-D bond and the cylinder axis (see next paragraph).
Experimentally, the SCD parameters are calculated from the quadrupolar splittings in the de-Paked
2
H NMR spectra [53, 54] (see section 5.3.2.4 and Equation 5-9). These de-Paked spectra correspond to
the sample orientations which lipid’s symmetry axis is parallel to the magnetic field direction. For
lamellar phase, this means that the bilayer normal is aligned with the magnetic filed, therefore the θ is the
angle between the C-D bond vector and the bilayer normal. For the HII phase, the lipid’s symmetry axis is
the cylinder axis which is oriented parallel to the magnetic field direction after de-Pake-ing. Therefore, if
the same formulae in Equation 5-1 and Equation 5-9 are going to be used for HII phase, the angle θ will
represent the angle between the C-D bond vector and the cylinder axis [53, 54].
Assuming a cylindrical geometry for the lipid tubules in the HII phase, Figure 5.2 illustrates several
structural parameters calculated from MD simulations in this study. As it is often assumed in HII phase
196
studies, lipids are assumed to be distributed radially around the cylinder axis, and their long axis is on a
This cylinder is covered by Nlipids of lipid molecules, filled with Nwater water molecules. This
molecular composition result in a HII system with Nwater/Nlipids of water per lipid inside the cylinder
(referred as nw in this study), a radius of Rw, and a length of Z along the cylinder axis. Each lipid, with
headgroup pointing towards the water core is taking an average area per lipid of APL at the lipid-water
197
Figure 5.2: Definitions for several structural parameters calculated from MD simulations.
A single lipid tubule in the HII phase. Rw, APL, and Z represent the radius of water core, area per lipid at
the lipid-water interface, and the instantaneous length of lipid cylinder along the cylinder axis,
respectively. The solid blue cylinder represents the water core, whereas the dotted green cylinder
represents the pivotal plane for this lipid at this curvature. The pivotal area per lipid for DOPE (at 275 K)
and POPE (at 348 K) is estimated to be ca. 0.65 nm2 [19, 55], from which its square root (~ 0.806 nm) can
be taken as the approximate average length for a PE lipid along the lipid cylinder [18]. In a study by Rand
and Fuller [18], the average number of DOPE were counted for a length of L = 0.806 nm slice along the
cylinder axis (slice colored in green). The same value (referred to as nc in this study) was calculated from
simulations to compare.
198
The gmx energy module of GROMACS was used to extract the length of simulation box in X, Y,
and Z directions from the energy file (.edr). The average length of the simulation box in Z direction was
calculated and taken as the average length of lipid cylinder <Z> [Figure 5.2].
The number of lipids per slice of length L = 0.806 nm along the lipid cylinder (nc) were obtained
Where Nlipids is the number of lipids forming the lipid cylinder, and Z is the instantaneous length of lipid
The lattice distance (dhex) for each system was calculated using simulation box sizes in X and Y
Equation 5-3
𝑑ℎ𝑒𝑥 = ((√3⁄2) 𝑋 + 𝑌) /2
This was done for each time frame and the average was reported in this study. The standard deviations of
means were calculated from this data and shown as the error bars in the corresponding figures.
For the rest of structural parameters, the volume fraction of lipids (𝜑𝑙 = 𝑉𝑙𝑖𝑝𝑖𝑑𝑠 / 𝑉𝑡𝑜𝑡𝑎𝑙 ) and
waters (𝜑𝑤 = 1 − 𝜑𝑙 ) were required. In this equation, 𝑉𝑡𝑜𝑡𝑎𝑙 is the total volume of the simulation box,
and 𝑉𝑙𝑖𝑝𝑖𝑑𝑠 is the volume taken by the lipids in the system. To calculate 𝜑𝑙 and 𝜑𝑤 , the following
approach was taken. First, a system composed of 15000 water molecules was simulated at four
temperatures (288 K, 303 K, 313 K, 348 K) and the volume for a single water molecule at each
199
temperature was calculated. A linear fit to these four volumes was made using Xmgrace, from which the
volume for a single water molecule at all the temperatures of interest was extracted. Given the total
number of water molecules inside the lipid cylinder, the volume taken by the water molecules inside a
cylinder was calculated. Next, the total volume of the simulation box (𝑉𝑡𝑜𝑡𝑎𝑙 ) for each HII system was
calculated using gmx energy module of GROMACS. The difference between the total volume and volume
taken by the water molecules was taken as 𝑉𝑙𝑖𝑝𝑖𝑑𝑠 . Given all this data, the 𝜑𝑙 and 𝜑𝑤 was calculated for
each system and were used for calculation of water core radius (Rw) and area per lipid (APL) at the lipid-
Assuming the water cores are cylindrical in shapes, the radius of water core was calculated
Equation 5-4
𝑅𝑤 = 𝑎 √𝜑𝑤 ( √3 / 2 𝜋)
Where 𝑎 is the unit cell spacing (spacing between adjacent cylinders) [Figure 5.1] and is equal to
2
( 3) dhex.
√
Finally, the area per lipid (APL) at the lipid-water interface (corresponding to Rw) was calculated
In both simulations and experiments [Figure 5.3], Rw and APL parameters were calculated based
on geometrical assumption that the lipid tubules are cylindrical in shape and have an approximate uniform
200
circular cross section along the Z direction [18, 56]. In this study, the SAXS scattering curves were
further analyzed to reconstruct the electron density profiles for the HII systems (see Section 5.3.2.3) [17].
From these electron densities, the water core radius at the water-lipid interface was estimated. The peaks
in these electron density profiles correspond to the DOPE phosphate groups [Figure 5.4-left]. The water
core radius for one of the simulated systems was also calculated from simulations as described below and
compared with the experimental data. We chose DOPE HII system at 303 K for comparison because the
maximum hydration for DOPE at 303 K is about 16 nw [15] and simulation for this specific system was
conducted in this study. Herein again, this analysis was done on the last 100 ns of the simulation with
frames saved every 100 ps. Considering that the lipid tubules in simulations are not perfectly cylindrical
and straight along the Z direction, the following approach was taken to calculate the water core radius
First, the coordinates of DOPE P atoms for each frame was saved separately resulted in 1001
structure files; each structure was centered in a cubic box. For this specific system, the average length of
lipid cylinder was about 14 nm. Thus, each structure was divided into 14 slices/rings; each ring had an
approximate length of 1 nm, resulted in 14014 rings in total. Each ring was subsequently centered in the
box so that the P atoms were radially distributed around the Z axis in the XY-plane. The radial distance
for each P atom from the Z axis was then calculated based on its coordinates:
Next, the average radial distance of P atoms from the center was calculated for each ring. Finally, a
histogram was made from these 14014 average values as shown in [Figure 5.4-right].
201
5.3.2 Experimental methods.
Avanti Polar Lipids Inc. (Alabaster, AL). The DOPE was transferred from the sealed glass ampule into a
scintillation vial using an 80/20 (v/v) benzene/methanol mixture. The lipid/solvent mixture was frozen in
liquid nitrogen for approximately ten minutes and then the solvent was removed by lyophilization over a
period of about 12 hours (until the weight had stabilized). The result was a condensed, powdered lipid
that could be weighed for sample preparation. When not in use, lipids were stored at -20 °C. In order to
prepare the sample for SAXS analysis, 100 μL of deuterium depleted water (ddw) (Sigma-Aldrich
Canada (Oakville, ON)) was added to 10 mg of dry, powdered DOPE. A series of five freeze (in liquid
N2) – thaw (room temperature) – vortex cycles were performed to ensure full hydration of the hexagonal
phase. The sample was carefully transferred into to a disposable thin-walled quartz capillary tube (80 mm
long, 1.5 mm (outer diameter), 0.01 mm thickness) which was then sealed with wax and analyzed
immediately.
For the 2H NMR experiments, partially deuterated DOPE-d18 was purchased in chloroform from
Avanti Polar Lipids Inc. (Alabaster, AL). In DOPE-d18, the protons on the last four carbons of each chain
(carbons 15-18) were replaced with deuterons. The DOPE-d18/chloroform mixture was transferred to a
pre-weighed scintillation vial and the bulk of the solvent was evaporated using a gentle flow of nitrogen
gas. Any residual solvent was then removed under vacuum over the course of 6 hours (until the weight
had stabilized). The resulting lipid film was used in its entirety to prepare the 2H NMR sample. The
approximately 41 mg of DOPE-d18 lipid were mixed with 700 μL ddw and a series of five freeze (in
liquid N2) – thaw (room temperature) – vortex cycles were performed. The hydrated lipid mixture was
202
5.3.2.2 Small angle X-ray scattering (SAXS).
A SAXSLAB Ganesha 300XL (Skovlunde, Denmark) setup with a Linkam variable temperature sample
stage was used to obtain the SAXS scattering curves. The Ganesha 300XL system has a Cu-Kα X-ray
source with a wavelength λ= 1.54 Å. The sample-to-detector distance for the Linkam stage setup was
calibrated using AgBeh and the sample temperature was calibrated using the pre-transition and chain
melting transition of both DPPC and DMPC bilayers. At each temperature, two sets of data were
collected: one after 30 minutes of equilibration, and a second 20 minutes after the completion of the first
acquisition (about an hour after the setpoint temperature was reached). Each data collection run was 600
s. Both the initial and final runs were collected at the same temperature to monitor sample stability across
the experiment.
The lattice distance (dhex) can be measured directly from the SAXS scattering curves via
where qpeak is the position of the peak in the scattering curve and n is its order number which is dictated
by the symmetry of the system. The Miller indices to define the hexagonal phase peaks can be written in
terms of h, and k, as (h,k). The corresponding order number for each peak is 𝑛 = √ℎ2 + ℎ𝑘 + 𝑘 2 . For the
hexagonal phase the order numbers for the peaks follow the sequence 𝑛 = 1, √3, 2, √7, 3, √12, … [57].
Diffraction data was used to build electron density reconstructions by established techniques, as
detailed in Harper et al. [17, 21]. Briefly, X-ray diffraction patterns were radially integrated, and
203
the peaks fitted to Gaussians. After applying the Lorentz and multiplicity corrections to the peak
intensities, the peak amplitudes were extracted. In order to complete the electron density
reconstructions, the proper phasing or sign for each of the peak amplitudes must be obtained. For
DOPE in excess water, it is well established that the proper phasing is “+,-,-,+,+,+,+” [16, 17].
ℎ,𝑘 𝑚𝑎𝑥
where ρ(r) is the electron density relative to the mean electron density as a function of position r,
h and k are the Miller indices, αhk and Fhk are, respectively, the phase and amplitude for each set
The electron density profiles along the (1,0) and (1,1) directions, as well as the 3-D
electron density profile in the unit cell were then reconstructed. It has been shown that the peak
maxima fortuitously correspond to the location of the Luzatti interface [16], which is the
boundary between lipid and water when they are modeled as separate, non-overlapping
components. Hence, the average radius of the peak maxima of the electron density reconstruction
frequency of 46.8 MHz and a TecMag Scout (TecMag, Inc. (Houston, TX)) spectrometer. The standard
204
quadrupolar echo pulse sequence with two out of phase 90° pulses was used. The 90° pulses were 3.95 μs
long with an interpulse delay of 40 μs and a recycle delay of 2 s. To eliminate artefacts, data was
collected using 8-cycle CYCLOPS phase cycling. These 2H NMR experiments were performed at a series
of temperature ranging from 20 °C to 40 °C where DOPE is in the hexagonal phase. Between each 5
degrees temperature increment the sample was allowed to equilibrate for 45 minutes before the 20,000
scan experiment was run. Again, the first and last experiments were run at the same temperature to ensure
that there was no sample degradation. 20,000 scans provided a sufficient signal-to-noise ratio in the
resulting de-Paked spectra for the order parameters to be determined. De-Pake-ing was performed using
the iterative method as described by Sternin et al. [53]. Smoothed order parameter profiles for the C-D
bonds were obtained using the method outlined by Lafleur et al. [54]. The order parameter, SCD can be
4 ℎ Equation 5-9
|𝑆𝐶𝐷 | = 𝛥𝑣𝑄
3 𝑒 2 𝑞𝑄
𝑒 2 𝑞𝑄
where ℎ
≈ 167 kHz and Δ𝑣𝑄 is the quadrupolar splitting of the C-D bond. The quadrupolar splittings
for carbons 15-17 on the sn-1 and sn-2 chains were calculated using a fitting function for the 12
deuterons. The left- and right-side data were averaged at each temperature. The twelve ΔνQ values
obtained, which were organized from largest to smallest, were assigned pairwise to the carbons from the
sn-1 and sn-2 chains. The first two were assigned to the carbon 17 sn-2- deuterons, the next two were
assigned to the same carbon on the sn-1 chain, this pattern was continued alternating between the sn-2 and
sn-1 chains and decreasing carbon number until the peaks for the carbon 15 sn-1 chain were assigned.
205
5.4 Results
The results presented in the following sections are calculated from simulations conducted using
anisotropic pressure coupling. Effects of using semi-isotropic pressure coupling on HII structures and
their comparison with anisotropic simulations are discussed in Appendix C. Briefly, for hydration levels
less than 20 nw, the dhex values and therefore other structural properties calculated from dhex agreed
One of the main objectives in this study was to see whether the constructed HII systems were correct and
MD simulations could reproduce the experimentally known structural parameters of HII system. To do so,
several structural parameters were calculated from DOPE HII simulations and are shown as a function of
hydration level (nw) in Figure 5.3. Experimental data from Rand and Fuller [18] are also shown for
comparison.
Structural parameters calculated from simulations are in excellent agreement with experimental
data [Figure 5.3]. In both simulations and experiments, there is a linear increase in dhex, Rw, and nc
parameters with hydration level [Figure 5.3, a-c]. For the APL, a logarithmic fit described the observed
trend in Figure 5.3-d better according to the chi-squared values for the fitted lines. APL increases as the
hydration increase and reaches a plateau at approximately 20 nw, which the APL is estimated to be about
50 A2 at. Among these four parameters, the Rw was the most affected by hydration level. An increase of
18 nw in the hydration level caused about 400 %, 75 %, 100 % and 100 % increase in R w, dhex, nc, and
APL, respectively.
206
Figure 5.3: Structural parameters for DOPE HII system as a function of hydration.
Data represent the average a) number of PE lipids per 0.806 nm slice along the cylinder axis (see Figure
5.2 for definitions), b) area per lipid, c) radius of water core, and d) lattice distance. Numbers on the X
axis are the number of water molecules inside the lipid cylinder per lipid. The maximum hydration for
207
DOPE at 303 K is about 16 nw [15]. Black circles and green diamonds represent data calculated from MD
simulations (T = 303 K) and the experimental data (T = 295 K) from Rand and Fuller [18], respectively.
Lines are linear (a, c, d) or logarithmic (b) fits to each data sets. Error bars are the standard deviation of
means for simulation data, and their size is comparable with the symbol sizes. See Figure 5.5 for POPE
The Rw for DOPE HII systems were also calculated from the electron density profiles
reconstructed from SAXS scattering curves. The three- and two-dimensional electron density profiles for
DOPE system at 303 K are shown in Figure 5.4-left. The peaks in these profiles correspond to the
phosphate groups of DOPE. The radial distance of these peaks from the center of water core are 19.91 and
20.22 angstroms in (1,0) and (1,1) directions, respectively. The average, 20.07 angstroms, was taken as
The Rw for the corresponding system was also calculated from MD simulations via calculating
the probability distribution of DOPE P atoms with respect to the lipid cylinder axis [Figure 5.4-right].
System with 16 nw at 303 K was chosen for this analysis [15]. The distribution reaches its maximum
around 20.01 angstroms, in an excellent agreement with the experimentally determined value. Both the
Rw from electron density profiles and the one calculated directly from MD simulations are also in good
agreement with Rw calculated from geometrical assumptions (~ 20 angstroms) shown in Figure 5.3-c.
208
Figure 5.4: Radius of water core from reconstructed electron density profiles and MD simulation.
Left) The two-dimensional electron density profile for DOPE HII system at 303 K is shown in (1,0) and
(1,1) directions. The inset shows the three-dimensional profile for the same system. Right) The
probability distribution of DOPE P atoms calculated from MD simulations from DOPE HII systems with
16 nw at 303 K.
DOPE has a lower bilayer to HII phase transition temperature than POPE [36, 58], which is due to
structural differences between the two lipid types. While DOPE has two oleoyl chains, POPE has one
oleoyl and one palmitoyl chain. Therefore, at a given temperature (e.g. T = 323 K), DOPE is expected to
form HII structures with higher curvature and smaller Rw compared to POPE. To see whether these
structural differences and expectations in HII phase can be reproduced by MD simulations and the
protocol used, simulations for DOPE and POPE HII systems were compared at 323 K [Figure 5.5].
The overall trends for all the structural properties are similar between DOPE and POPE. POPE
HII systems have a slightly higher nc, Rw, and dhex but lower APL than DOPE at each hydration level.
209
POPE HII systems contain approximately one more lipid per slice than DOPE for low hydration levels.
As the hydration level increases, this difference also increases from one to two molecules.
Figure 5.5: Structural parameters of DOPE and POPE HII systems as a function of hydration.
210
Data represent the average a) number of PE lipids per 0.806 nm slice along the cylinder axis (see Figure
5.2 for definitions), b) area per lipid, c) radius of water core, and d) lattice distance. Numbers on the X
axis are the number of water molecules inside the lipid cylinder per lipid. Black circles and red squares
represent data for DOPE and POPE, respectively. Lines are linear (a, c, d) or logarithmic (b) fits to each
data sets. Error bars are the standard deviation of means for simulation data, and their size is comparable
with the symbol sizes. Data corresponds to the temperature of 323 K for both DOPE and POPE HII
systems.
length of lipid cylinders [18]. To study the effect of dehydration on lipid cylinder elongation, we further
calculated the average length of lipid cylinders for DOPE and POPE HII systems from simulations for
several hydration levels and compared [Figure 5.6]. For both lipid types, dehydration caused an increase
in the average length of lipid tubules along the cylinder axis. For instance, removing water molecules
from 20 nw to 10 nw resulted in ~ 61 % increase in the DOPE cylinder length. Moreover, POPE cylinders
211
Figure 5.6: Cylinder length in DOPE and POPE HII systems as a function of hydration.
Average size of simulated lipid cylinder in Z direction (see Figure 5.2 for definitions) is calculated from
simulations and plotted versus hydration level (nw). Lines are the inverse (1/(aX+b)) curves fitted to the
data, where a and b are two fitting parameters. Error bars represent the standard deviation of means. Data
corresponds to the temperature of 323 K for both DOPE and POPE HII systems.
The water core of HII systems are usually assumed to be cylindrical in geometry for systems with a dhex
of ~ 65 angstrom or less [16]. There are also experimental studies that suggest the lipid tubules in HII
phase are not perfectly straight [59]. This further motivated us to look at the HII structures in more
details.
212
Figure 5.7 shows snapshots from DOPE HII systems with ca. 4 nw (a, b), 10 nw (c, d), and 16 nw
(e, f) simulated at 303 K. Among the low hydration systems (i.e. up to 16 n w) studied, systems with 4 nw
were the most deviated systems from a cylindrical geometry and were highly curved/undulated along the
cylinder axis. In addition, the water cores in these systems had a strip-like shape instead of a cylindrical
geometry. Nevertheless, the water channels maintained their hexagonal packing in the system [Figure
5.7-a]. As the hydration level increased, both the deviations from a cylindrical geometry and the
213
Figure 5.7: Effect of hydration level on lipid tubules shape in DOPE HII systems.
The snapshots corresponding to the last frame of each simulation for DOPE HII systems with different
hydration levels at 303 K are shown in both XY-plane (Left) and along the Z direction (Right). Systems
with ca. 4 nw (a, b), 10 nw (c, d), and 16 nw (e, f) are shown. Yellow and red spheres represent the DOPE
214
phosphorus atoms and the methyl group’s carbon atoms in acyl chains, respectively. Orange lines
represent the lipid acyl chains, and water is shown as the cyan surface. The blue box is the simulation box
containing four lipid cylinders in a rectangular box for system with 4 n w, and one lipid cylinder in a
triclinic box for systems with 10 nw and 16 nw. The red spheres are not shown in the right column for
Structural changes in a molecular system are associated with the changes in structure, orientation and
organization of its molecular components. Deuterium order parameter (SCD), a parameter which can be
both measured by deuterium NMR (2H NMR) technique and be calculated directly from simulations, can
Deuterium order parameter (SCD) for the last carbon atoms (carbon atom numbers of 15-17) of
DOPE in sn-1 and sn-2 chains were measured in DOPE systems at 303 K using 2H NMR experiments [
Figure 5.8-a]. The same parameters were calculated from MD simulations of DOPE system with 16 nw at
the same temperature to compare. Systems with 16 nw were used for comparison because the maximum
hydration for DOPE lipids at 303 K (and for POPE at 348 K) has been experimentally estimated to be ca.
16 nw [15, 19]. Simulations and experiments agree on the general trend observed in the SCD profiles. sn-1
is shown to be more disordered than sn-2, and SCD decreases along the acyl chains from carbon number
15 to 17. However, SCD parameters from simulations are higher than experimental data by ~ 0.01.
Deuterium order parameters for sn-1 of POPE HII system with 16 nw at 348 K were also
calculated from simulations, sorted descending, and compared with experimental data from Lafleur et al.
[60] [Figure 5.8-b]. In both simulations and experiments, the SCD for POPE palmitoyl chain decreases
along the acyl chain towards the methyl terminus. The computational SCD are higher than experimental
215
values, and the difference between the two data set is highest for the middle segment of sn-1. Overall, the
computational SCD for both DOPE and POPE are in good agreement with the corresponding experimental
values.
We further investigated and compared the effect of hydration on SCD parameters in sn-2 chain of
DOPE and POPE HII systems at 323 K [Figure 5.8, c and d]. In DOPE systems, a change of nw from 4 to
10 increased the SCD significantly [Figure 5.8-c], and the segment containing the unsaturated bond was
the least affected segment upon hydration. For nw higher than 10, hydration level had little to no effect on
SCD of DOPE and POPE [Figure 5.8, c and d], although dehydration tends to decrease the SCD. For
example, in both DOPE and POPE systems, systems with 20 nw has the highest SCD values.
Effect of temperature on SCD parameters in sn-2 chain of DOPE and POPE were also studied from
simulations [Figure 5.8, e and f]. The results for three temperatures for each lipid type (283, 303, and 323
K for DOPE, and 323, 348, and 358 K for POPE) are shown to compare. Increasing temperature induced
disorder in both DOPE and POPE systems. Finally, the SCD for sn-2 is similar between DOPE (shown in
green) and POPE (shown in black) at 323 K, although the terminal segment (i.e. carbon numbers 12 to
17) of POPE is more ordered than DOPE. This might be due to an increase in van der Waals interactions
of sn-2 chain with fully saturated and more ordered palmitoyl chain in POPE than unsaturated oleoyl
216
Figure 5.8: Deuterium order parameter (SCD) for DOPE and POPE in HII phase.
SCD parameters calculated from simulations of DOPE (Left) and POPE (Right) HII systems. a) SCD
parameters for the last carbon atoms in both DOPE sn-1 and sn-2 acyl chains calculated from system with
16 nw at 303 K, and compared with data measured using 2H NMR technique at 303 K. b) SCD parameters
for POPE sn-1 chain calculated from system with 16 nw at 348 K, sorted descending, and compared with
217
the experimental data from Lafleur et al. [60] conducted at T = 348 K. c, d) SCD parameters for DOPE and
POPE sn-2 acyl chains calculated from simulations with different levels of hydrations at 323 K. e, f) SCD
for DOPE and POPE sn-2 acyl chains calculated from simulations with 16 nw at three different
temperatures.
The method used in Tate and Gruner [15] was used to estimate the maximum hydrations inside the lipid
cylinders (see Figure 5.9 caption). However, in our hybrid approach, MD simulations were used to obtain
For DOPE systems, the experimental dhex for each temperature were taken from Tate and Gruner
[15]. These dhex values correspond to experiments conducted in the presence of excess water and were
served as horizontal lines in Figure 5.9-left. The estimated maximum hydrations using this hybrid model
are compared with the experimentally measured values [15] in Table 5-2. Assuming a 0.5 uncertainty in
experimental dhex measurements [17], the uncertainty in maximum hydration calculated using the current
hybrid approach is ca. 0.3 water molecules per lipid. The predicted values for T = 303 K and 323 K are in
good agreement with corresponding experimental data. For the lower temperature of 283 K, the hybrid
model gave higher hydration than experiment. Although no simulation was conducted at 295 K,
interpolation of simulation data in Table 5-2 suggest that the maximum hydration of DOPE HII systems
at 295 K should be between 19.7 (for 283 K) and 16.6 (303 K). This agrees with the reported values of
18, 18±1, and 19 for DOPE at 295 K [18, 62, 63]. Finally, in both simulations and experiments, an
increase in the temperature resulted in lower maximum hydration [Table 5-2] as expected [15].
For POPE, the experimental dhex at the excess water regime are taken from Rappolt et al. [19]
[Figure 5.9-right]. Herein again, the maximum hydrations estimated computationally are in a good
218
agreement with experimental values. In experiments, increasing temperature from 348 K to 358 K
increased the maximum hydration by 1 nw. However, in simulations, the same increase in temperature
caused a slight decrease in maximum nw from 16.8 to 16.2, which is more in line with experimental
Left) DOPE, Right) POPE HII systems. Simulations were used to calculate the dhex for HII systems at
each hydration level and temperature. These data are plotted for each nw and the linear fit to each data set
is shown. The dhex values corresponding to the horizontal lines are taken from experiments conducted at
excess water regime by Tate and Gruner [15] (for DOPE), and Rappolt et al. [19] (for POPE). Each
horizontal line corresponds to one temperature. The nw where these horizontal and diagonal lines cross is
taken as the maximum hydration for that system at that specific temperature [15].
Table 5-2 can also be used to compare the maximum hydration of DOPE and POPE HII systems
at 348 K. At this temperature, POPE systems were estimated to have ~ 16 nw (experiments) or 16.8 nw
219
(simulation) [Table 5-2]. The maximum hydration for DOPE HII systems at 348 K is not known, but it
can be extrapolated from the data. According to simulation data and the trend observed in Table 5-2, the
maximum hydration for DOPE should be less than 15.1 nw. This is, again, in agreement with the
expectations that at a given temperature, DOPE form HII structures with higher curvature and therefore,
with a smaller Rw and number of trapped water molecules in their core than POPE.
Table 5-2: Maximum hydrations for DOPE and POPE HII systems at several temperatures.
DOPE POPE
T (K)
5.4.6 Temperature effect on DOPE HII lattice distance and maximum hydration.
The effect of hydration level and temperature are coupled in HII systems [15, 17]. An increase in
temperature results in a decrease in the maximum number of water molecules taken up by the HII lattice.
The change in dhex has been explained to affect the effective shape of the comprising lipids, and
consequently the hydration level and Rw. We further studied this experimental observation as follows.
Lattice distances for DOPE HII systems with different hydration levels (nw = 10, 14, 16 and 20)
were calculated for three temperatures (T = 283 K, 303 K, and 323 K) from simulations and are shown in
220
Figure 5.10. The experimental data measured in the present study, and the data from Harper et al. [17] are
also shown for comparison (Our experimentally measured dhex values are in excellent agreement with data
angstroms decrease in dhex [Figure 5.10]. In simulations, the lines connecting the values for each nw are
approximately horizontal meaning that for a given hydration level, increasing temperature has little to no
effect on dhex. However, a systematic increase in dhex was observed upon hydration of HII system at each
temperature. Adding enough water inside the cylinders to double the nw from 10 to 20 resulted in an
221
Figure 5.10: Lattice distance as a function of temperature for DOPE HII systems.
Lattice distances for DOPE HII systems were calculated from MD simulations and are shown for each n w
and temperature. Data are the means and error bars represent standard deviations of means. Experimental
dhex measured in this study are shown as squares colored in magnet, whereas the experimental data from
Harper et al. (2001) [17] are shown as orange circles. Experimental uncertainties in dhex measurements are
The number of water molecules inside the lipid cylinders at each temperature could also be
estimated from Figure 5.10. The diagonal line (colored in orange) corresponding to the experimental data
crosses the green line (nw = 16) at 303 K, suggesting the experimental systems, most likely, have had
222
about 16 nw inside their lipid cylinders. This agrees with 16.2 nw reported by Tate and Gruner (1989) [15]
[Table 5-2]. By extrapolations of each line in Figure 5.10, the orange line seems to have crossed the red
line (nw =14) at 333 K. Simulation data for other hydration levels are not available. However, according
to the observed trend, the lines corresponding to 15 nw are expected (by interpolation) to be located
somewhere between the red (nw =14) and green (nw =16) lines. This further suggests that the orange line
would have, most likely, crossed the line corresponding to 15 nw at 323 K. Therefore, simulations suggest
that experimental DOPE HII systems studied by Harper et al. (2001) [17] have had about 16, 15, and 14
5.5 Discussion
In a study by Rand and Fuller [18], X-ray diffraction technique was used to study the structural
dimensions of DOPE dispersions in HII phase as a function of hydration. Dehydration was shown to
reduce the dhex, Rw and APL, and to cause formation of longer lipid cylinders with less lipid molecules per
nm slice along the cylinder axis [18]. The same parameters calculated from MD simulations using both
anisotropic and semi-isotropic pressure coupling were in excellent agreement with experimental data
The structural parameters of this study were calculated using the formulae described in Rand and
Fuller [18], which are based on the geometrical assumption that the cross-section area of water core is
cylindrical. This was done to illustrate how MD simulation could reproduce the experimental data using
the same assumptions used in experiments. However, structural parameters could also be calculated
directly from MD simulations, without any assumption on cylindrical geometry. The Rw calculated from
223
the positions of DOPE P atoms for a test system was in excellent agreement with the R w measured from
Deuterium order parameter (SCD) calculated from MD simulations were also in good level of
agreement with experimental data measured by 2H NMR experiments [Figure 5.8, a and b]. This further
validates the methods used to simulate HII systems and to calculate the SCD parameters from simulations
in this phase (see Methods). SCD parameters in simulations were, however, higher than experimental data,
especially for the last carbon atoms of DOPE acyl chains [Figure 5.8-a]. The most likely reason for such
a difference could be that the hydration level and actual temperature in the experimental setup might be
slightly different than the computational systems. It might also be partially because of using
CHARMM36 FF in GROMACS package [64, 65]. Moreover, the force field parameters for DOPE are
not perfect and could affect the structural properties in HII phase. However, DOPE parameters in
CHARMM36 FF have been shown to be transferable to the simulation of HII systems [29]. Considering
the algorithm used to calculate the SCD (see Methods), lipid cylinder undulations along the cylinder axis
and deviation of the water core from a cylindrical geometry in computational systems might also cause
those differences in SCD. However, systems with 16 nw were almost cylindrical and straight in shape [
There are other computational studies on HII systems using both atomistic and coarse-grained
models [24-30]. In a study by Marrink and Mark [24], the dhex values for DOPE HII systems predicted by
simulations were larger than experimental values by ca. 10 percent. This increase was originated from
using coarser models and consequently an increase in the length of DOPE acyl chains. Therefore, the all-
atom nature of the simulations in the present study could be one of the reasons for a better agreement of
224
5.5.2 HII water channels are not straight and cylindrical at low hydrations.
The assumption of circular cross-section for the water core might not be necessarily true for all hydration
levels. Systems with high hydration levels corresponding to the dhex of ~ 65 angstroms and higher are
known to deviate from circularity significantly [16, 24], whereas for systems with lower hydration levels
the water core was circular within 5 % of Rw [16]. The systems studied in [16] were all in presence of
excess water and changes in the hydration were due to the changes in temperature. Therefore, these
experimental findings might not be applicable to very low hydration levels such as 4 nw simulated at a
In our simulations, the water core in systems with 4 nw deviated from circularity and had a strip-
like shape [Figure 5.7]. However, coarse-grained simulations using MARTINI model suggested that this
system had a circular water core [24]. This discrepancy with the coarse-grained simulation might be due
to the coarser models for DOPE and water molecules, or different number of water channels simulated in
two studies. It might also be because of using the fully anisotropic pressure coupling in [24], which
allowed the simulation box to deform. In our simulations, although an anisotropic pressure coupling was
used, the box shape was fixed during the simulation course while the box sizes were allowed to vary. The
HII formation process as well as equilibration process and different time scales between two studies could
be among other reasons for such a difference in water core shapes. Disregard of water core shape, the dhex
and other structural properties of system with 4 nw followed the overall trend observed for higher
HII structures were also shown to be curved/undulated along the cylinder axis [Figure 5.7],
which is consistent with the experimental finding that HII cylinders are not perfectly straight [59, 62]. The
level of curvature/undulation was further a function of hydration level. Systems with lower hydration
experienced higher levels of undulations [Figure 5.7]. This might be because of the change in mechanical
properties and bending modulus of lipid tubules as a function of hydration level. Alternatively, it could be
225
because of the larger size of the equilibrated systems with lower hydrations along the cylinder axis.
Undulations have been previously reported for lipid bilayers [66]. These undulations could be supressed
MD simulation could successfully reproduce the structural differences between DOPE and POPE in HII
as expected [4]. At a given temperature, DOPE was shown to form HII systems with higher curvature
than POPE [Figure 5.5 and Figure 5.6]. Due to the two oleoyl acyl chains in DOPE, the lipid tails splay
apart more than they do in POPE. This makes DOPE less ordered and more cone-shaped which support
Increasing temperature is known to decrease the acyl chain order in both lipid bilayers [68, 69] and HII
phases [60, 70]. In HII experiments, increasing temperature is associated with a decrease in the number of
water molecules inside the lipid cylinders [15] [Table 5-2]. Thus, the observed decrease in SCD profiles
[60, 70] in HII phase is indeed a sum of two effects: temperature and dehydration. Each effect was
the SCD profiles in HII phase; both resulted in smaller SCD parameters [Figure 5.8]. The reduction of SCD
by an increase in temperature was expected due to the entropic effects [71], but decrease of SCD with
Experimentally, it has been observed that the HII dehydration decreases the average length of
DOPE lipid hydrocarbons in both interstitial and interaxial directions [Figure 5.1] [18]. This shorter
hydrocarbon lengths might mean that the lipid tails are more splayed in lower hydration levels or are
226
tilted more towards the regions formed by three neighboring lipid cylinders [18, 72]. This observed
shortening in lipid hydrocarbon chains can explain the observed decrease in SCD parameters upon
dehydration, or vice versa. In bilayers, also, there is a correlation between the bilayer hydrophobic
thickness and SCD profiles [68]. Lipid tails with smaller SCD parameters are less ordered and are more
splayed. This results in a shorter length for the projection of acyl chains along the bilayer normal, and
5.5.5 Temperature affects the maximum hydration and lattice distances in HII.
The maximum hydrations estimated from our hybrid model were in an excellent agreement with the
estimated values from experiments [Table 5-2]. Increasing temperature caused a decrease in the
maximum hydration in HII phase [Figure 5.9]. This was further shown to be the main factor for the
experimentally observed temperature-dependent reductions in dhex [15, 17] [Figure 5.10]. As explained
elsewhere, an increase in temperature affects the molecular shapes, and consequently decreases the lipid
monolayer spontaneous radius of curvatures (Ro) [15]. In the presence of excess water, the Rw is near Ro
so that the system is in its minimum free energy state. Therefore, HII systems with smaller R w and less
amount of water molecules inside the lipid cylinder are more energetically favorable systems at higher
temperatures. Radius of water core and dhex are also directly correlated [Figure 5.3], meaning that a
decrease in Rw decreases dhex as well. Furthermore, temperature is known to have little effect (~ 1
angstrom for 80 degrees increase in temperature) on the hydrophobic length of DOPE tails in HII phase
[15]. Therefore, most of the changes in dhex as a function of temperature is due to change in maximum
hydration.
227
5.5.6 Protocol for construction of equivalent HII systems to experiments.
As demonstrated in Figure 5.9, MD simulations for low hydration levels can be used in combination with
a single SAXS measurement for dhex (in presence of excess water) to estimate the maximum hydration for
the HII systems at a certain temperature. This maximum hydration can be further used for construction of
This protocol should be applicable to other amphiphilic molecules, e.g. polymers, in HII phase.
However, like other protocols it has a few limitations which should be considered. First, at least one d hex
value measured in presence of excess water at the temperature of interest is required. Second, at least two
hydration levels need to be simulated to obtain the diagonal line. Simulation of systems with higher
hydration levels (i.e. more than 10 nw) is recommended because systems are straighter and more
cylindrical in geometry. Third, equilibration of lipid tubules in simulations might be in the order of 200 ns
for systems which the initial dimensions are far from equilibrium state. However, considering the recent
advances in computing resources, this would not be a problem. Fourth, although coarse-grained models
could also be used to measure the dhex and estimate the maximum hydrations, such coarser models might
result in larger dhex values for some molecules [24]. This will result in non-accurate estimations for the
maximum hydrations. Finally, the force field parameters are critical to result in correct structural
parameters such as dhex. Non-accurate parameters could result in inaccurate APL and consequently dhex
As an application example, this protocol can be used for parameterizing new molecules, e.g.
synthetic lipids, using the target data in HII phase [Figure 5.11]. Force field development and validation
usually requires comparing data between equivalent systems in simulations and experiments. Most of the
structural parameters of HII systems were shown to be highly affected by the hydration level. Therefore, a
reasonable estimation of maximum hydration is a necessity for such applications. Assuming one single
measurement for dhex in excess water regime at the desired temperature (red dotted horizontal line in
228
Figure 5.11-left), and at least one target data for the corresponding systems are available, the
parameterization process can be done as follows. In this example, SCD parameters are used as the target
First, reasonable initial parameters (parameter set 1) are assigned to each atom in the molecule
according to the parent force field, e.g. CHARMM36 force field. Using parameter set 1, two hydration
levels (e.g. with nw of 10 and 22) are constructed and simulated using semi-isotropic pressure coupling to
obtain the first diagonal line (the solid pink line in Figure 5.11-left). The nw where this line crosses with
the red dotted line is extracted (i.e. 20 nw in this example). Next, the HII system with 20 nw is simulated
and SCD parameters are compared with the experimental SCD data. According to the differences between
computational and experimental SCD profiles, force field parameters are further adjusted resulting in
parameter set 2. Parameter set 2 are used to simulate HII systems with 10 and 22 n w again. This results in
a new diagonal line (the solid blue line in Figure 5.11-left) and estimation for the maximum hydration
(i.e. 18 nw in this example). Next, HII system with 18 nw is simulated and SCD are compared between
simulation and experiment. As we iterate through this process, the simulation data are expected to
approach the experimental data. This method might require several iterations till estimated n w or the SCD
converge.
229
Figure 5.11: A proposed protocol for molecule parameterization in HII phase.
Left) Hypothetical dhex values as a function of hydration calculated using parameter sets 1 and 2. The
parameter set 1 is the initial set of parameters assigned to the molecule, whereas the parameter set 2 refers
to the modified parameters after one iteration. The black line represents the experimental dhex data for low
hydration levels (which is not known!). The dotted line represents the experimentally measured dhex in
presence of excess water at the desired temperature. The dashed vertical lines point to the n w where the
horizontal and diagonal lines cross at. Right) Typical SCD profiles obtained by parameter sets 1 and 2
from systems constructed with 20 nw and 18 nw, respectively. These nw were estimated from the left panel.
The experimental SCD are shown in solid black line corresponding to system with 16 nw (This is known
5.6 Conclusions
We show that MD simulations of DOPE and POPE HII phase successfully reproduce d hex, Rw, APL, SCD,
nc, and length of the lipid cylinders in the HII phase at different hydration levels, including low hydration
230
levels where conducting experiments is usually challenging. Dehydration was shown to result in longer
lipid cylinders with smaller Rw, APL, nc, and SCD parameters. HII structures were shown to be undulated
along the cylinder axis and deviate from a perfect cylindrical geometry at 4 nw. Hydration was further
shown to reduce the level of undulations. We obtained reasonable estimates of the maximum hydration
for PE HII structures at several studied temperatures, using only a single experimental d hex measured from
a single sample in the excess water regime. In agreement with experimental data, increasing the
temperature decreased the maximum hydration in HII systems, accounting for the experimentally
observed temperature-dependent reductions in dhex. The POPE HII systems studied followed the same
trends as DOPE, although as expected DOPE formed HII structures with a higher curvature than POPE.
We proposed a general protocol for constructing HII phases and molecular parametrization using the
target data in HII phase using MD simulations, 2H NMR and SAXS experiments.
5.7 Bibliography
[1]. Seddon, J. M. (1990). Structure of the inverted hexagonal (HII) phase, and non-lamellar phase
1-69.
[2]. Gin, D. L., Gu, W., Pindzola, B. A., & Zhou, W. J. (2001). Polymerized lyotropic liquid crystal
[3]. Jouhet, J. (2013). Importance of the hexagonal lipid phase in biological membrane organization.
[4]. Cullis, P. R., Hope, M. J., & Tilcock, C. P. (1986). Lipid polymorphism and the roles of lipids in
[5]. Ellens, H., Siegel, D.P., Alford, D., Yeagle, P.L., Boni, L., Lis, L.J., Quinn, P.J. & Bentz, J.
231
[6]. Siegel, D. P., & Epand, R. M. (1997). The mechanism of lamellar-to-inverted hexagonal phase
[7]. Caffrey, M. (2003). Membrane protein crystallization. Journal of Structural Biology, 142(1), 108-
132.
[8]. Hirlekar, R., Jain, S., Patel, M., Garse, H., & Kadam, V. (2010). Hexosomes: A novel drug
[9]. Chen, Y., Ma, P., & Gui, S. (2014). Cubic and hexagonal liquid crystals as drug delivery systems.
[10]. Barauskas, J., Johnsson, M., & Tiberg, F. (2005). Self-assembled lipid superstructures: Beyond
[11]. Hafez, I. M., Maurer, N., & Cullis, P. R. (2001). On the mechanism whereby cationic lipids
[12]. Hafez, I. M., & Cullis, P. R. (2001). Roles of lipid polymorphism in intracellular delivery.
[13]. Koltover, I., Salditt, T., Rädler, J. O., & Safinya, C. R. (1998). An inverted hexagonal phase of
cationic liposome-DNA complexes related to DNA release and delivery. Science, 281(5373), 78-
81.
[14]. Semple, S.C., Akinc, A., Chen, J., Sandhu, A.P., Mui, B.L., Cho, C.K., Sah, D.W., Stebbing, D.,
Crosley, E.J., Yaworski, E., Hafez, I. M., et al. (2010). Rational design of cationic lipids for
[15]. Tate, M. W., & Gruner, S. M. (1989). Temperature dependence of the structural dimensions of
232
[16]. Turner, D. C., & Gruner, S. M. (1992). X-ray diffraction reconstruction of the inverted hexagonal
[17]. Harper, P. E., Mannock, D. A., Lewis, R. N., McElhaney, R. N., & Gruner, S. M. (2001). X-ray
[18]. Rand, R. P., & Fuller, N. L. (1994). Structural dimensions and their changes in a reentrant
[19]. Rappolt, M., Hickel, A., Bringezu, F., & Lohner, K. (2003). Mechanism of the lamellar/inverse
hexagonal phase transition examined by high resolution X-ray diffraction. Biophysical Journal,
84(5), 3111-3122.
[20]. Lafleur, M., Cullis, P. R., Fine, B., & Bloom, M. (1990). Comparison of the orientational order of
lipid chains in the L.alpha. and HII phases. Biochemistry, 29(36), 8325-8333.
[21]. Harper, P. E., Mannock, D. A., Lewis, R. N. A. H., McElhaney, R. N., & Gruner, S. M. (2012).
phases (vol 81, page 2693, 2001). Biophysical Journal, 102(5), 1236-1236.
[22]. Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: Visual molecular dynamics. Journal of
[23]. Van Gunsteren, W. F., Dolenc, J., & Mark, A. E. (2008). Molecular simulation as an aid to
[24]. Marrink, S. J., & Mark, A. E. (2004). Molecular view of hexagonal phase formation in
233
[25]. Marrink, S. J., & Tieleman, D. P. (2002). Molecular dynamics simulation of spontaneous
membrane fusion during a cubic-hexagonal phase transition. Biophysical Journal, 83(5), 2386-
2392.
[26]. Bandyopadhyay, S., Klein, M. L., Martyna, G. J., & Tarek, M. (1998). Molecular dynamics
[27]. Knecht, V., Mark, A. E., & Marrink, S. J. (2006). Phase behavior of a phospholipid/fatty
acid/water mixture studied in atomic detail. Journal of the American Chemical Society, 128(6),
2030-2034.
[28]. Marrink, S. J., De Vries, A. H., & Mark, A. E. (2004). Coarse grained model for semiquantitative
[29]. Sodt, A. J., & Pastor, R. W. (2013). Bending free energy from simulation: Correspondence of
planar and inverse hexagonal lipid phases. Biophysical Journal, 104(10), 2202-2211.
[30]. Johner, N., Harries, D., & Khelashvili, G. (2014). Curvature and lipid packing modulate the
elastic properties of lipid assemblies: Comparing HII and lamellar phases. The Journal of
[31]. Wang, X., & Deserno, M. (2015). Determining the pivotal plane of fluid lipid membranes in
[32]. Boyd, K. J., & May, E. R. (2018). BUMPy: A model-independent tool for constructing lipid
bilayers of varying curvature and composition. Journal of Chemical Theory and Computation,
14(12), 6642-6652.
[33]. Durrant, J. D., & Amaro, R. E. (2014). LipidWrapper: An algorithm for generating large-scale
234
[34]. Martínez, L., Andrade, R., Birgin, E. G., & Martínez, J. M. (2009). PACKMOL: A package for
[35]. Jo, S., Cheng, X., Lee, J., Kim, S., Park, S.J., Patel, D.S., Beaven, A.H., Lee, K.I., Rui, H., Park,
S. & Lee, H. S. (2017). CHARMM‐GUI 10 years for biomolecular modeling and simulation.
[37]. Berendsen, H. J., van der Spoel, D., & van Drunen, R. (1995). GROMACS: A message-passing
[38]. Kolev, V., Ivanova, A., Madjarova, G., Aserin, A., & Garti, N. (2012). Molecular dynamics
approach to water structure of HII mesophase of monoolein. The Journal of Chemical Physics,
136(7), 074509.
[39]. Jo, S., Kim, T., Iyer, V. G., & Im, W. (2008). CHARMM‐GUI: A web‐based graphical user
[40]. Nguan, H., Ahmadi, S., & Hashim, R. (2014). Molecular dynamics simulations of the lyotropic
[41]. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez,
C., Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom
additive force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B,
114(23), 7830-7843.
235
[42]. Durell, S. R., Brooks, B. R., & Ben-Naim, A. (1994). Solvent-induced forces between two
[43]. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983).
Comparison of simple potential functions for simulating liquid water. The Journal of Chemical
[44]. Berendsen, H. J., Postma, J. V., van Gunsteren, W. F., DiNola, A. R. H. J., & Haak, J. R. (1984).
Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics, 81(8),
3684-3690.
[45]. Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N⋅log(N) method for
Ewald sums in large systems. The Journal of Chemical Physics, 98(12), 10089-10092.
[46]. Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., & Pedersen, L. G. (1995). A
smooth particle mesh Ewald method. The Journal of Chemical Physics, 103(19), 8577-8593.
[47]. Nosé, S. (1984). A molecular dynamics method for simulations in the canonical ensemble.
[49]. Parrinello, M., & Rahman, A. (1981). Polymorphic transitions in single crystals: A new
[50]. Nosé, S., & Klein, M. L. (1983). Constant pressure molecular dynamics for molecular systems.
[51]. Hess, B., Bekker, H., Berendsen, H. J., & Fraaije, J. G. (1997). LINCS: A linear constraint solver
236
[52]. Piggot, T. J., Allison, J. R., Sessions, R. B., & Essex, J. W. (2017). On the calculation of acyl
chain order parameters from lipid simulations. Journal of Chemical Theory and Computation,
13(11), 5683-5696.
[53]. Sternin, E., Bloom, M., & Mackay, A. L. (1983). De-Pake-ing of NMR spectra. Journal of
[54]. Lafleur, M., Fine, B., Sternin, E., Cullis, P. R., & Bloom, M. (1989). Smoothed orientational
order profile of lipid bilayers by 2H-nuclear magnetic resonance. Biophysical Journal, 56(5),
1037-1041.
[55]. Gruner, S.M., Tate, M.W., Kirk, G.L., So, P.T.C., Turner, D.C., Keane, D.T., Tilcock, C.P.S. &
[56]. Rand, R. P., Fuller, N. L., Gruner, S. M., & Parsegian, V. A. (1990). Membrane curvature, lipid
segregation, and structural transitions for phospholipids under dual-solvent stress. Biochemistry,
29(1), 76-87.
[57]. Schnablegger, H., & Singh, Y. (2013). The SAXS guide. Anton Paar GmbH.
[58]. Bryant, G., Pope, J. M., & Wolfe, J. (1992). Low hydration phase properties of phospholipid
[59]. Gruner, S. M., Rothschild, K. J., & Clark, N. A. (1982). X-ray diffraction and electron
microscope study of phase separation in rod outer segment photoreceptor membrane multilayers.
[60]. Lafleur, M., Bloom, M., Eikenberry, E. F., Gruner, S. M., Han, Y., & Cullis, P. R. (1996).
Correlation between lipid plane curvature and lipid chain order. Biophysical Journal, 70(6),
2747-2757.
237
[61]. Seu, K. J., Cambrea, L. R., Everly, R. M., & Hovis, J. S. (2006). Influence of lipid chemistry on
membrane fluidity: Tail and headgroup interactions. Biophysical Journal, 91(10), 3727-3735.
[62]. Gawrisch, K., Parsegian, V. A., Hajduk, D. A., Tate, M. W., Gruner, S. M., Fuller, N. L., & Rand,
[63]. Marsh, D. (2011). Pivotal surfaces in inverse hexagonal and cubic phases of phospholipids and
[64]. Lee, J., Cheng, X., Swails, J.M., Yeom, M.S., Eastman, P.K., Lemkul, J.A., Wei, S., Buckner, J.,
Jeong, J.C., Qi, Y. & Jo, S. (2015). CHARMM-GUI input generator for NAMD, GROMACS,
[65]. Pluhackova, K., Kirsch, S. A., Han, J., Sun, L., Jiang, Z., Unruh, T., & Böckmann, R. A. (2016).
A critical comparison of biomembrane force fields: structure and dynamics of model DMPC,
POPC, and POPE bilayers. The Journal of Physical Chemistry B, 120(16), 3888-3903.
[66]. Lindahl, E., & Edholm, O. (2000). Mesoscopic undulations and thickness fluctuations in lipid
[67]. Feller, S. E., & Pastor, R. W. (1996). On simulating lipid bilayers with an applied surface tension:
[68]. Petrache, H. I., Dodd, S. W., & Brown, M. F. (2000). Area per lipid and acyl length distributions
3172-3192.
[69]. Zhuang, X., Makover, J. R., Im, W., & Klauda, J. B. (2014). A systematic molecular dynamics
238
[70]. Thurmond, R. L., Lindblom, G., & Brown, M. F. (1993). Curvature, order, and dynamics of lipid
[71]. Petrache, H. I., Dodd, S. W., & Brown, M. F. (2000). Area per lipid and acyl length distributions
3172-3192.
[72]. Perutková, Š., Daniel, M., Dolinar, G., Rappolt, M., Kralj‐Iglič, V., & Iglič, A. (2009). Stability
of the inverted hexagonal phase. Advances in Planar Lipid Bilayers and Liposomes, 9, 237-278.
239
Chapter Six: An Auto-Inhibitory Helix in CTP:Phosphocholine Cytidylyltransferase
Hijacks the Catalytic Residue and Constrains a Pliable, Domain-Bridging Helix Pair
Copyrights:
This research was originally published in the Journal of Biological Chemistry. Ramezanpour, M.*, Lee,
J.*, Taneva, S.G., Tieleman, D.P., & Cornell R.B (2018). An auto-inhibitory helix in
CTP:phosphocholine cytidylyltransferase hijacks the catalytic residue and constrains a pliable, domain-
bridging helix pair. Journal of Biological Chemistry, 293(18), 7070 –7084. (* contributed to work
equally). Copyright © the American Society for Biochemistry and Molecular Biology.
Contributions:
I ran all the simulations and analyzed all the simulation data. I had contributions in methodology,
conceptualization, data interpretation, hypothesizing, validation, and experimental design. I wrote the
computational methods section and had contribution in editing the other sections in the manuscript. The
experimental data and methods section were provided by Dr. Rosemary B. Cornell, Dr. Jaeyong Lee, and
Abbreviations:
egg phosphatidylglycerol; PE, phosphatidylethanolamine; PDB, Protein Data Bank; MD, Molecular
Dynamics; RMSF, Root Mean Square Fluctuation; NOE, Nuclear Overhauser Enhancement; PR,
240
6.1 Abstract
membrane phospholipids. When dissociated from membranes, a portion of the M domain functions as an
auto-inhibitory (AI) element to suppress catalysis. The AI helix from each subunit binds to a pair of
helices (E) that extend from the base of the catalytic dimer to create a four-helix bundle. The bound AI
helices make intimate contact with loop L2, housing a key catalytic residue, Lys-122. The impacts of the
Extensive MD simulations with and without the AI helix revealed that backbone carbonyl
oxygens at the point of contact between the AI helix and loop L2 can entrap the Lys-122 side chain,
effectively competing with the substrate, CTP. In silico, removal of the AI helices dramatically increased
E helix dynamics at a predicted break in the middle of these helices, enabling them to splay apart and
forge new contacts with loop L2. In vitro cross-linking confirmed the reorganization of the E element
upon membrane binding of the AI helix. Moreover, when E bending was prevented by disulfide
engineering, CCT activation by membrane binding was thwarted. These findings suggest a novel two-part
auto-inhibitory mechanism for CCT involving capture of Lys-122 and restraint of the pliable E helices.
We propose that membrane binding enables bending of the E helices, bringing the active site closer to
6.2 Introduction
Many regulatory enzymes are silenced by inter-domain interactions that are broken by activating ligands.
The inhibitory interactions can involve a direct steric block of the active site by occupation of a
pseudosubstrate or ligand-binding domain in its apo form [1, 2] or by binding of the unoccupied ligand-
241
binding domain to an allosteric site that shifts the equilibrium toward unproductive configurations of an
element in the active site [3-5]. An X-ray structure of the catalytically silenced form of
novel allosteric silencing mechanism involving electrostatic re-direction of a key catalytic residue [6]. We
The enzyme CCT catalyzes a rate-limiting step in phosphatidylcholine (PC) synthesis and
controls PC homeostasis by being active only when the membrane PC content is low. Its conserved
catalytic domain is linked to a weakly conserved regulatory domain via a highly conserved short linker
segment [Figure 6.1-A]. CCT’s regulatory domain (domain M) is an inducible 60-70-residue membrane-
binding amphipathic helix [7-10], and the membrane is its “ligand”. The enzyme can thus interconvert
between a soluble, inactive form and a membrane-bound, active form by transformation of domain M
from mostly disordered into a long helix [Figure 6.1-B]. CCT membrane binding and hence its
catalytic power are regulated by physical properties of the membrane [11]. PC-rich membranes have low
net surface charge and tight packing between neighboring lipid molecules of the bilayer. PC-deficient
membranes have a higher surface negative charge, and increased packing stress [12]. CCT’s domain M
responds to the latter properties, binds and inserts partway into the membrane as an amphipathic helix,
and this binding event is communicated to the active site to enhance catalysis more than two orders of
magnitude by effects on both kcat and Km for CTP [13, 14]. This process accelerates PC synthesis to
restore PC compositional homeostasis. CCT mutations are linked causatively to three human diseases
that impair development of bone, cartilage, and/or retinal tissue [15-17] or lipid metabolism [18].
The catalytic domain of CCT derives from an ancient domain designed to catalyze nucleotide
transfers to a variety of metabolites, such as the transfer of AMP from ATP to amino acids and CMP from
242
(ECT)) catalysis is accomplished by charge-stabilization of the highly charged CTP with basic residues
residing in loops L1, L2, and L6, and in the junction of 4/L5. Residues at the N-terminus of helix E
make contacts with CTP to promote a contorted U-shape [21]. This enables an in-line attack of the
nucleophilic phosphate at the -phosphate of CTP to displace diphosphate [20]. The crystal structure of
the catalytically silenced form of CCT [6] as well as fluorescence anisotropy analyses [22] showed that
when not engaging membranes, the M domain is composed of a disordered leash segment of ~40 residues
followed by a ~22-residue auto-inhibitory (AI) helix that docks onto two elements of the catalytic
domain: the E helices, and loop L2 [Figure 6.1-C]. These two elements are special in that their
sequence differs between lipid-regulated CCTs and non-regulated cytidylyltransferases, such as GCT and
ECT. The E helices are much longer in the CCTs (~22 residues versus 10-12 residues in GCT or ECT)
and are predicted to be interrupted by a short disordered segment in the middle [Figure 6.1-A] [23]. Thus,
the long uninterrupted E helices observed in the structure of the silenced form may be stabilized by the
AI helices, which, together with the two E helices, form a four-helix bundle and may be destabilized
once the AI helices are dissociated by membrane binding. The L2 sequence in all cytidylyltransferases
contains one or more lysines that participate in catalysis based on mutagenesis [24, 25] and solved
structures with substrate and product [6, 20, 21, 25]. In rat CCT, a substitution with arginine at Lys-122 in
L2 results in a ~5-order of magnitude decrease in catalytic efficiency [24], the strongest mutagenic effect
for any single residue tested in CCT. Any impact on Lys-122 dynamics or alignment could have severe
We proposed, based on the structure of the silenced enzyme [6], that CCT’s mode of auto-
inhibition involves a partial occlusion of the active site entrance by the AI helix and clamping of the
active site in a non-productive configuration by the AI helix. Limited molecular dynamics simulations
suggested that the docking of the AI helix could restrict the dynamics of the E and L2 loops. Most
243
intriguingly, where the end of the AI helix contacts loop L2 (the AI turn), the backbone carbonyls of the
turn steered the key catalytic residue, Lys-122, away from its orientation into the active site. This
electrostatic steering involved a carbonyl on loop L2 at Phe-124 and a carbonyl at Phe-293 on the AI turn,
and we refer to this interaction as the “backbone trap” [6]. These single-replicate 200-ns simulations were
A key unanswered question arising from the previous work was whether the backbone trap can
operate when the active site is occupied with substrate. Can the backbone carbonyls compete with
substrate for the key catalytic residue? In the present work, we determined the impact of the AI helix on
active site dynamics and positioning of Lys-122 in the presence and absence of the substrate CTP. The
new data, involving a total of 40 separate atomistic 1-s simulations, suggest that the backbone trap can
effectively compete with CTP. We also tested loop L2 flexibility in vitro and its role in catalysis by
mutation of a conserved glycine, Gly-123. Furthermore, simulations in the absence of the AI helices
revealed a remarkable plasticity for the E helices once the AI helices are displaced, generating bends in
the E helices at the predicted helix break [Figure 6.1-A]. Cross-linking measurements in vitro
confirmed a reorganization of the E helices upon membrane binding. Straightjacketing the helices with a
disulfide prevented lipid activation, suggesting that the conformational changes observed upon removal of
the AI helices may reflect the active membrane-bound configuration of the E. We discuss various
hypotheses for how the bent E helices might accelerate the CCT-catalyzed reaction.
244
Figure 6.1: Structure of mammalian CCT and its active site.
A) domain map. The N region is non-conserved and disordered, and the catalytic (C) domain is a
conserved dinucleotide fold encompassing ∼150 residues. The locations of the two elements in domain C
where the AI helices dock (L2 and αE) are indicated. In lipid-regulated CTs, the αE helix is predicted to
be two helices (αEN and αEC) with a helical break at residues ∼212–214. When not membrane-bound, the
M domain consists of a disordered leash followed by the AI helix. The phosphorylation (P) region is
poorly conserved and disordered. B) schematic of the CCT fold for silenced and active forms. Left,
residues 40–223 of the catalytic dimer (cyan and green chains) with the AI helices docked onto the αE
helix pair to create a four-helix bundle. Elements in the crystallized protein that are not visible in the
solved structure (PDB code 4MVC) are represented as dotted lines. The box encompasses one active site,
enlarged in (C). Right, a proposed membrane-bound active form using the catalytic dimer from the
solved structure of CCT236 (PDB code 3HL4) and the M domain modeled as a simple unbroken helix
(not to scale). The αEC and linker are of unknown configuration in the active form and are represented as
dotted lines. CDP-choline occupies both active sites. C) close-up of the CCT active site showing
conserved residues that influence catalysis (stick representation of Lys-122, Tyr-173, Arg-196, and Thr-
245
202), the backbone-to-backbone contact between the AI turn and loop L2 (amide N at Phe-124 and
carbonyl oxygens at Met-292 and Phe-293; spheres with H-bonds shown as dashed lines), and the
hydrophobic pocket (side-chain spheres) that is created by the confluence of the AI helix, αEN, and loop
L2. This image is from the silenced form (PDB code 4MVC), with CDP-choline removed.
6.3 Results
Loop L2 has been identified as a key loop in catalysis based on studies with CCT and a related bacterial
cytidylyltransferase, GCT [24, 25], which is involved in cell-wall biosynthesis and is not regulated by
lipids. The loop L2 sequence of GCT differs from that of the lipid-regulated eukaryotic CCTs. Whereas
the latter contain a single lysine (Lys-122) followed by an invariant glycine, L2 in GCT has two
conserved lysines and no glycine [Figure 6.2-A]. Mutagenesis of B. subtilis GCT showed that both Lys-
44 and Lys-46 are required for catalysis [25]. In the crystal structure of B. subtilis GCT in complex with
CTP (PDB code 1COZ), Lys-46 contacts the -phosphate of CTP. In the structure of the same enzyme in
complex with the product CDP-glycerol, Lys-46 contacts the - and -phosphates of CDP-glycerol and
Lys-44 contacts the -phosphate (PDB code 1N1D) [Figure 6.2-B]. Thus, it appears that two lysines
participate in multiple alternative interactions with substrate and product during a catalytic cycle. The
backbone configurations of L2 at these lysines are similar in both crystal structures, and movement of the
lysines to engage the two phosphates of the product are rigid body translocations of 3 Å (21).
In CCT, however, loop L2 (residues 122-127) is 3 residues shorter and likely to be less mobile,
and Lys-122 is the sole lysine. In the crystal structures of CCT in complex with CDP-choline (PDB codes
3HL4 and 4MVC), Lys-122 occupies a position between that of Lys-44 and Lys-46 of GCT, and contacts
246
both the - and -phosphate of CDP-choline [Figure 6.2-B]. The backbone at Lys-122 in CCT occupies a
configuration in the -helix portion of a Ramachandran plot, whereas the , angles at the lysines in the
crystal structures of GCT occupy a configurational space within or near the β portion. We hypothesized
that the single lysine in L2 of CCT must be agile during a reaction cycle so that its -amino group can
partner with either the -, β- or -phosphate oxygens of CTP, or with phosphocholine, and that this
requires a very flexible backbone, enabled by Gly-123. This hypothesis was explored by mutagenesis to
alanine or proline, which would restrict the , angles and present a larger barrier opposing backbone
conformational sampling at Lys-122. The data in Figure 6.2-C show that neither Ala nor Pro can
effectively substitute for the glycine at residue 123. The Pro substitution was completely inactivating in
the context of CCT236, which lacks the membrane-binding domain. In a construct with domain M
(CCT312), the effect of the proline substitution on the lipid-activated enzyme was curiously less severe,
suggesting that when membrane-bound, additional factors may mitigate the detrimental loss of flexibility.
Further experiments will be needed to establish the basis for the less stringent L2 flexibility requirement
of membrane-bound CCT. These data do support the hypothesis that configurational lability of loop L2
promotes catalysis. In support of Lys-122 as a primary target of regulation in lipid-dependent CCTs, the
positions of four other functional active-site residues, Arg-196, Thr-202, His-89, and His-92, super-
impose within <1.5 Å with the analogous residues in the GCT structures with substrate or product.
247
Figure 6.2: CCT relies on a single lysine followed by glycine in loop L2 for catalysis.
and CCT–CDP-choline (3HL4) showing H-bonds between the lysines in loop L2 and CTP, CDP-glycerol,
or CDP-choline. C) impact of mutations at Gly-123 in L2 on the activity of CCT. Purified CCTs were
assayed for activity under conditions optimal for the CCT312 or CCT236 constructs in the absence (gray
bars) or presence (green bars) of 0.2 mM PC/PG (1:1) vesicles. Data are means ± average deviation (error
248
6.3.2 The AI helices selectively repress the dynamics of helix αE.
To explore the impact of the AI helix on the dynamics of loop L2 and other active site loops and catalytic
residues, we performed 1-µs MD simulations of the CCT dimer, residues 40–223, comprising the
complete catalytic domain, with and without the AI helices. A 3.0 Å structure of this segment with bound
AI helices [6] provided the starting coordinates for the simulations. The effect of the substrate CTP was
also probed in each of these sets. The final 8 residues of the E helices were restrained by either of two
methods (referred to as PR or NOE) to prevent their unwinding and migration deep into the active site,
which was observed in a preliminary 200-ns simulation. Five replicates of each of the four conditions
(protein alone; +AI; +CTP; +CTP and +AI) were simulated for each restraining method, for a total of 40
independent simulations (40 µs total). We discovered early on that the mobility and interactions of key
loops and residues were frequently dissimilar for the two active sites. For example, when Lys-122
interacted with CTP in chain A, it might be contacting a carbonyl of the backbone trap in chain B. Thus,
In the 40 chains containing CTP in the active site, CTP was maintained in the pocket except for
the last 150 ns of just one replicate, where it diffused away from most of its partners in the starting
structure (e.g. Arg-196, Thr-202). More common was a somewhat fixed position of the cytosine and
ribose while a rotation about the ribose C1 oxygen - phosphorus bond redirected the - and -
phosphates away from the starting contact with Thr-202 in E and toward Lys-122 and loop L2. This
transformed the CTP from its U-shaped starting conformation to a more stretched conformation.
The loops contributing to the active site are L1, L2, L5 and L6 [Figure 6.1]. The N terminus of
E (EN) also contributes Thr-202 and Ser-203 to the active site. The docking site for each AI helix is
formed from the E helices of both chains and loop L2 of one chain [Figure 6.1]. This prompted an
249
investigation of the impact of the AI helices on L2 and E dynamics by computing the RMSF for
backbone atoms and key catalytic residues housed in these loops. Loops L5 and L6, which are longer and
more dynamic, were also investigated. From a visual inspection of the many simulations, it was obvious
that L1 is not mobile, and neither CTP nor the AI reduced its mobility further. Figure 6.3 shows that the
dynamics of L2 was reduced a small and variable degree by the AI and by CTP in isolation, but the
combination of CTP and the AI helix reduced dynamics significantly by 20%. The same effect was
observed on the dynamics of Lys-122 in L2, where the combination of CTP and the AI resulted in a
significantly depressed RMSF (24%), but neither CTP nor the AI alone showed a significant effect on
mobility. The immobilization of EN and its resident active site residue, Thr-202, reflected the combined
impact of active site residues bonding with the ligand and the AI. The AI helix dampened the dynamics of
the EN backbone and Thr-202 side-chain by ~ 30%. The dynamics at Thr-202 was reduced 40% by
CTP, in keeping with a direct H-bonding contact. Together, the AI and CTP reduced Thr-202 RMSF by
The dynamics of loop L5 and its resident active site residue Tyr-173 were high and were not
suppressed by either CTP or the AI helix [Figure 6.3]. The dynamics of loop L6 and its active site
residue, Arg-196, were not impacted by the AI helix, but were reduced 40-50% by CTP, with which it
formed direct H-bonds. The dynamics of the loop/hinge region linking helix EN and helix EC was
profoundly reduced by the AI helix in the NOE-restrained simulations [Figure D.5-B]. The high mobility
of this region when the AI is not present is discussed below. In summary, the docking of the AI helices to
form the four-helix bundle has a strong immobilizing effect on the E helices and only weakly constrains
250
Figure 6.3: Effect of the AI helix and CTP on the dynamics of active-site loops and side chains.
A) RMSF of heavy backbone atoms of the indicated loops/structural elements and the heavy side-chain
atoms of the indicated residues were computed as described under “Materials and Methods.” The
195–Ser-201), and αEN (Thr-202–Val-210). The data are means ± average deviation (error bars) of five
replicates for each condition, indicated in the matrix at the top. n = 10 for each condition, as there are two
active sites for each CCT dimer. p values are provided with reference to the first set (white bar). *, p <
0.05; **, p < 0.005. These analyses used simulations with NOE restraints applied on helix αE. The same
analysis on 20 simulations using positional restraints is available in Figure D.1. B) dynamics of the
active-site loops and select side chains are shown, visualized by the overlay of protein coordinates from
simulations at 100-ns intervals at 0 (gray), 100 (red), 200 (orange), 300 (yellow), 400 (green), 500 (cyan),
251
600 (blue), 700 (magenta), 800 (wheat), 900 (pale green), and 1000 ns (black). The side chains of the
active site residues Lys-122, Tyr-173, Arg-196, and Thr-202 are shown as sticks. Each set is from a
representative simulation, and the presence/absence of the AI and CTP is indicated on the figure. For
To investigate L2 dynamics in vitro we examined the effect of a docked AI helix on the mobility
of a tryptophan substituted for Phe-124 in loop L2. The F124W substitution did not affect the activity in
the presence or absence of lipid. We also monitored an engineered F121W in the B helix adjacent to
loop L2, and a native Trp (Trp-151) located in the active site [20]. The location of these Trp residues is
shown in Figure D.2-A. The fluorescence anisotropy of Trp-124 was compared for this CCT construct in
its soluble form and when bound to a lipid micelle that induces dissociation of the AI helices from the
catalytic domain [22] and fully activates the enzyme [10]. The anisotropy measurements are sensitive to
motions in the 1-10 ns regime. The anisotropy values for each of the three Trp residues were 0.15 – 0.17,
typical values for a Trp that is constrained in a folded element of a protein [26]. There was only a
marginal decrease in the anisotropy of Trp-124 upon binding lipids (12 ± 3% decrease), and a slightly
stronger effect on Trp-121 (20 8 % decrease; Figure D.2-B). In keeping with these decreases in
dynamics, the RMSF value for Phe-124 computed from a set of 10 NOE simulations was reduced 18% by
the AI (P = 0.01). There was no change induced by lipids when Trp-124 was evaluated in the context of a
CCT lacking an M domain (CCT236; Figure D.2-B), confirming that the effect on dynamics is due to
lipid binding. Thus, in keeping with the simulation results, docking of the AI onto loop L2 has only a
small restraining effect on the dynamics of a hydrophobic side chain at residue 124 in L2. Phe-124
rotations are sustained within a hydrophobic pocket that is created from side chains of L2 (Val-126) and
252
6.3.3 The AI helix supplies H-bonding partners for Lys-122 that compete with CTP.
Previous published simulations without substrate suggested that the AI interactions could steer the Lys-
122 N into an H-bonding trap with backbone carbonyls contributed from the AI and the L2 loop [6].
When the active site is occupied by ligand, can the AI compete with the phosphates of the ligand for
contact with Lys-122 to inhibit catalysis? To further explore the likelihood of the backbone trap for Lys-
122, we mined the full set of 40 simulations with and without CTP and the AI helix for the frequency of
Lys-122 H-bonding with one or more oxygen atoms from 5 residues that were frequently found in shared
H-bonds with Lys-122. These are the Phe-124 carbonyl in loop L2, Asp-86 carboxyl and backbone
carbonyl in loop L1, and, in the presence of the AI helix, three backbone carbonyls at the C-terminal turn
of the AI at Phe-293, Gly-294, and Pro-295. Figure 6.4-A shows a snapshot of a common 3-way
interaction with Lys-122. In some of the simulations in the presence of the AI, contact between the AI C-
terminus and loop L2 was lost for prolonged time intervals during the simulations. The contact between
AI and L2 was monitored by the formation of an H-bonding interaction between Met-292 carbonyl and
Phe-124 backbone NH, a contact that was also observed in the crystal structure of the silenced CCT (PDB
code 4MVC). For analysis of the impact of the AI helix on Lys-122 H-bonding partners, we deleted time
intervals where the Met-292 – Phe-124 contact was disengaged for more than 50 ns. This step reduced
total simulation time associated with the production runs from 38.4 to 32.4 µs.
In both the PR and NOE-restrained simulations, there was a significant effect of the AI helix on
Lys-122 contact with the trapping atoms [Figure 6.4-B]. The frequency of contact with these atoms due
to the AI alone increased >2-fold, from 32% to 83% (PR simulations) and from 31 to 72% (NOE
simulations). When CTP occupied the active site, the AI helix increased the frequency of Lys-122 contact
with the backbone trap 3 – 4-fold, from 14 to 38% (PR) and from 8 to 36% (NOE). At the same time, the
H-bonding frequency with CTP decreased from 67.5 to 43% (PR) and from 67 to 39% (NOE). This
253
analysis strongly supports the hypothesis that the interaction of the AI helix with the L2 loop steers Lys-
122 away from CTP by forging direct contacts with backbone atoms. Figure 6.4-C shows an image of the
electrostatic surface surrounding the active site, with and without the docked AI helix. The C-terminus of
the AI helix bends at its juncture with loop L2, and that bend is highly electronegative, thus exerting an
electrostatic pull on the Lys-122 amino group. With no side-chain, Gly-123 enables close docking of the
AI-turn on loop L2, enhancing this pull. The AI helix did not affect the H-bonding of two other near-by
active site residues that are significant partners with CTP: Arg-196 in loop L6 and Thr-202 in the EN
(data not shown). These residues have no or little interaction with the atoms of the backbone trap, and
their stable interactions with CTP were not disturbed by the AI helix. Thus, it appears that the backbone
Figure 6.4-B also shows that CTP and/or the AI helix reduce the frequency of Lys-122 partnering
with other atoms and the time spent searching for partners. The other atoms that served as prominent
partners for Lys-122 in the absence of CTP or the AI include Asp-82 carboxyl and Gly-83 carbonyl in 1,
Glu-198 in loop L6, and Thr-202 in helix E. Partnering with side chain or backbone atoms of loop L5,
such as Tyr-173, Ser-174, and Ser-175, was also seen. Many of these contacts are not compatible with an
active site occupied by CTP, phosphocholine or CDP-choline, as the position of the side chains would
We noticed that the backbone of L2 at Lys-122 can flip between - and - dihedral
configurations, facilitated by Gly-123, and that the backbone configuration predominated over the
configuration when Lys-122 was contacting the trapping atoms. This motivated us to analyze whether the
backbone configuration could deter contact of Lys-122 with the substrate, CTP. The analyses of the
simulations (see supplemental information in Appendix D and Figure D.3) showed that CTP contact was
254
compatible with either the or configuration, and that even when the AI was present and L2 was in the
A) left, active site of chain A with CTP in its productive U-shape, interacting with Ser-203 in αE, and
Lys-122 H-bonding with CTP. Right, 10 ns later in the same simulation, Lys-122 is in a split H-bond with
Asp-86 in loop L1, Phe-124 in loop L2, and Phe-293 in the AI. These interactions compete with CTP for
Lys-122 for the duration of the 1-μs simulation. B) effects of AI and CTP on Lys-122 H-bonding
partners. The simulation conditions are indicated in the matrix at the top. Lys-122 H-bonding frequency
with the indicated residues or CTP was analysed using the GROMACS tool, g_hbond. A contact
involving 1–3 H-bonds to a partner was scored as one H-bond. Lys-122 H-bonded with the α-, β-, or γ-
255
phosphate oxygens of CTP. The group of five included the carboxyl of Asp-86 and the carbonyl of
residues 124, 293, 294, and/or 295. None, frequency of H-bonds with solvent; other, any other protein
atom. Data are the means ± average deviation (error bars) of 10 replicates for each condition (5
simulations × 2 chains), except 9 replicates for the condition PR-restrained, +CTP, +AI. p values for the
effect of the AI are indicated as follows: *, p < 0.05; **, p ≤ 0.005. C) electrostatic surface of active site
with and without the AI helix. PDB images shown are of 4MVC with the AI removed (left) and AI
present (right). The electrostatic surface was calculated using default settings in PyMOL.
In 17 of 20 simulations carried out in the presence of the AI helices the 4-helix bundle remained intact for
the large majority of the simulation time, with occasional unwinding of the N-terminal portion of one AI
helix at the base of the 4-helix bundle. The contacts between the 4 helices are mostly hydrophobic, and
the AI-helix-turn/loop L2 interaction involves an aromatic cluster [Figure D.4]. In these simulations, the
E segments remained unbroken -helices with minimal backbone fluctuation, even in the hinge region.
The E - E contact was maintained by the hydrophobic side chains of Ile-206, Ile-209, and Val-210 in
the N-terminal segment of the E [Figure D.4]. This hydrophobic interaction is also seen in the crystal
structure [6]. The two turns at the C-terminus of the E helices have fewer interactions but were fixed by
restraints. In 3 of the 20 simulations, one of the AI helices disengaged for hundreds of ns at the N
terminus [Figure D.4, image on right] and rotated on its axis to become nearly perpendicular to the E
helices. Despite this rotation, hydrophobic contacts between the middle of the AI (Phe-285, Ile286, Phe-
289) and the E helices (Val-210, Tyr-213, Tyr-216) were maintained. The two E helices in all
simulations remained stable in their conformation and interactions with each other [Figure D.4].
256
One of the most striking observations, which occurred in all 10 NOE-restrained simulations
without the AI helices, was the bending of helix E at the hinge (residues 211-215). We refer to this
helical break as a hinge following the definition of Wolfson and colleagues as a region about which
protein domains rotate [27]. In the equilibrated starting structures, the two E helices were unbroken
[Figure 6.5-A], and their interaction was mediated by the hydrophobic cluster described above. The two
helices cross-over at the Ile-209 –Val-210 contact site. In all 10 simulations, the helices drifted together
toward one active site for 10 – 200 ns, and then the segment between residues Arg-211- Val-215
unwound and/or bent in both helices, enabling contact of at least one EC with loop L2. In 6 out of 10
dimer simulations, the bend at this loop accompanied the complete dissociation of the EC portions from
each other (EC splay; configurations 4, 5, and 6; Figure 6.5). The EC in the splayed conformation
formed contacts with loop L2, principally via an aromatic cluster composed of Phe-124 in L2 and Tyr-
213 and Tyr-216 in the EC of the opposite chain [Figure 6.5, C-E]. In 15 of 20 simulated chains, this
aromatic contact (with a minimum distance of ~3Å) persisted for more than 550 ns. The EN retained its
tight inter-chain hydrophobic interaction for the duration of the simulations. There were at least 5
common configurations of the CCT dimer with bent E helices [Figure 6.5], and these were observed in
the absence or presence of CTP, but never in the presence of the AI helices. These bent configurations
could interconvert (i.e. 2 3 and 4 5 6) but did not return to the starting configuration. Once the
two EC helices had engaged separate L2 loops, they did not break away to re-forge contacts with each
other. In 6 replicates, the splayed configurations were maintained for more than 500 ns, for a cumulative
total of 4 µs of a total of 10 µs. In the splayed configurations 4-6, the frequency of H-bonding of Tyr-213
or Tyr-216 hydroxyl with Phe-124 carbonyl in L2 increased, especially when assuming a configuration
with the EC helix at an approximate right angle beneath L2 [Figure 6.5-E].
257
The simulations with CTP present revealed a frequent H-bond between Arg-223 at the C-
terminus of EC and the -phosphate of CTP [Figure D.5-A]. This occurred in 5 of 10 chains for a total
of ~2.7 µs of 5.0 µs. This contact was specifically associated with splayed EC configurations that
featured close packing of EC under and perpendicular to loop L2 [Figure 6.5, D and E]. The CTP
contact with Arg-223 did not preclude an interaction of Lys-122 with CTP [Figure D.5-A]. The impact of
the AI on the flexibility of the E hinge is shown in Figure D.5-B. The RMSF of residues 211-215 was
reduced by >2-fold, both in the presence and absence of CTP. For the most part, the AI helices constrain
258
Figure 6.5: αE conformational change triggered by removal of the AI helices.
The central panel shows a time plot of the interchain distance between Tyr-213 and Phe-124 (top), and
the interchain distance between the two Tyr-213 hydroxyls of the dimer (bottom) for one replicate of the
NOE simulations. The abrupt change after 180 ns correlates with the loss of αE–αE contact. A–E,
configuration 1 is the equilibrated starting structure. Configuration 2 occurs after a short period of drift.
The αE helices remain in contact with each other, and one αE forges contact with L2 via the aromatic
cluster composed of Tyr-213 and Tyr-216 in the αE and Phe-124 in L2 (gray arrow). Configuration 3 is
shown in Figure D.6. Configuration 4 has both αE helices splayed apart and interacting with L2 via the
aromatic cluster. In configuration 5, a sharp bending at the hinge in one of the αE helices positions the
αEC at right angles to loop L2; this orientation is stabilized by aromatic cluster and Arg-223–CTP
259
contacts. Configuration 6 features both αE helices adopting a close orientation under L2, with αE helices
Membrane binding dissociates the AI helices, which incorporate into the long M-helices. If the E helices
are malleable and splay apart upon dissociation of the AI helices, as suggested by the MD simulations,
then their average inter-chain distances should increase upon membrane binding [Figure 6.6-A]. We
substituted cysteine in place of a nonconserved, native alanine at residue 217 in the EC helix in a
cysteine-free CCT-312 background where five native cysteines were changed to serines [22, 28]. The
activity was lowered ~2-fold by the substitution from Ala to Cys at position 217, relative to the Cys-free
Inter-E helix cross-bridging will generate a covalent CCT dimer which can be detected on gels [Figure
6.6-B]. Because cross-bridging is in competition with reactions of each cysteine individually with
separate bis-maleimides, monomer-to-dimer conversion will necessarily be incomplete. All three bis-
maleimides were modestly effective at cross-linking cysteines at residue 217, which is buried within the
four-helix bundle in the silenced CCT form. Displacement of the AI helices by lipid vesicle binding
should increase their accessibility, but we detected much less covalent dimer in the presence of a large
molar excess of lipid [Figure 6.6-B and Figure D.7]. These results support an increase in average
distance between sites in the EC upon membrane binding and dissociation of the AI from the E helices.
260
Figure 6.6: Lipid vesicles reduce interchain cross-linking efficiency between the two αEc helices.
A) increased interchain distance between Ala-217 in the αEC accompanies αE hinge bending. Images
from simulations in the presence (left) or absence (right) of the AI helices are shown. This site was
substituted with cysteine in a Cys-free background for conjugation with the indicated cross-linkers. B)
lipid vesicles reduce covalent dimer formation. Sulfhydryl-directed cross-linking reactions with three bis-
261
maleimide reagents with the indicated optimal cross-linking distances for each [29]. Coomassie-stained
11% reducing gels show the cross-linking reaction as a conversion of CCT312 (A217C) monomers to
dimers in the absence (−) or presence (+) of 100-fold molar excess lipid vesicles (PC/PG (1:1)). Bottom
panels, quantification of band density using ImageQuant version 5.2: gray symbols, no lipid; green
symbols, with lipid. Data are means of two images ± range (error bars).
To assess whether restructuring of the E helices is required for enzyme activation upon binding
to membrane vesicles, we purified CCT312-A217C and its cysteine-free analog under mild oxidizing
conditions. In contrast to the cysteine-free control enzyme, the A217C variant was partially oxidized to
produce a covalent dimer at Cys-217, and 1 mM DTT was sufficient for its complete reduction [Figure
6.7-A]. Figure 6.7-B shows the impact of oxidation and reduction of this sulfhydryl on CCT activity in
the presence of excess lipid vesicles. DTT had a small (~ 20%) positive impact on the activity of the Cys-
free control, suggesting a minor oxidation effect unrelated to disulfide bond formation. However, the
activity of oxidized CCT-A217C was only 12 5 % of the activity of the reduced A217C, assayed in the
presence of 10 mM DTT, an ~8-fold effect of reduction. To ensure that the low activity of the disulfide-
bridged CCT was not due to an impairment of the membrane binding of domain M, we compared
oxidized and reduced CCT in an assay that monitors fluorescence resonance energy transfer (FRET)
between Trp-278 in the M domain and dansyl-phosphatidylethanolamin in lipid vesicles. The FRET
signal was the same for both oxidized and reduced CCT-A217C and was the same as FRET observed
with a Cys-free CCT resistant to disulfide formation [Figure D.8]. These data suggest that preventing
restructuring of the E by di-sulfide bonds (or alternatively by formation of a stable 4-helix bundle with
262
Figure 6.7: A disulfide bond linking the αE helices at Cys-217 inhibits CCT activity.
CCT312 (Cys-free) and CCT312 (A217C) with a single cysteine in the αE were purified under oxidizing
conditions. A) DTT reduction of oxidized CCT. Oxidized CCT preparations were incubated for 10 min at
37 °C in the presence of the indicated DTT concentration before analysis by nonreducing SDS-PAGE of
monomeric (36-kDa) and dimeric (72-kDa) species. B) reduction of the disulfide between αE helices is
263
required for lipid activation. In a separate experiment, enzyme activities were determined for A217C (▲),
and Cys-free (♦) using standard conditions, 0.2 mM egg PC/egg PG (1:1) sonicated vesicles, and the
indicated concentration of DTT. The data are means of two independent experiments with ranges (error
bars) and are expressed as units/mg of CCT normalized to peak activity at 10 mM DTT (10378 ± 1654
units/mg for the Cys-free CCT and 4460 ± 780 for CCT-A217C).
6.4 Discussion
In the MD simulations the docked AI helices did not cause any major rearrangements of the active site.
The AI helices reduced access to the active site from the underside of the catalytic domain, although the
lateral opening remained sufficiently wide for entry and escape of CTP and CDP-choline (~16 Å × 10 Å,
larger than the length of CDP-choline at 13.5 Å; Figure 6.4-C). The sequence of the AI helix-turn is
acidic and glycine-rich across species that show lipid regulation. The placement of the electronegative AI
helix-turn at the mouth of the active site might also deter an approaching CTP or phosphocholine
molecule from entering the active site. In addition to the restricted access of substrates and release of
products, we have considered three possible mechanisms for auto-inhibition by the AI helices: (i)
restricting the dynamics of active site loops and participating residues; (ii) hijacking the catalytic lysine;
and (iii) preventing a conformational transition in the E helices that enables an EC interaction with the
active site. The AI helices dock onto the E helices and loop L2. Thus, we focused on these elements, but
264
6.4.1 AI restriction of L2 dynamics.
Our previous single 200-ns simulation showed that the AI helix reduced the root mean square deviation
from the starting structure for all backbone atoms in loop L2 (residues 121-128) by ~ 1 Å and the RMSFs
by ~ 60% [6]. The more extensive analysis here suggests that the AI has specific effects on the dynamics
of L2 (modest) and the E helices (strong). The RMSF analysis we performed reports on the nanoscale
fluctuations of active site loops and residue side-chains. A 20 - 30% reduction in fluctuations, which we
observed over 20 × 1 s, could influence catalysis by limiting the range of conformational sampling, and
this may reduce the frequency of productive orientations for groups participating in catalysis [30].
However, given a kcat of 50 ms for fully active CCT, the dampening of nanosecond fluctuations may be
irrelevant.
On the other hand, the conversion of the L2 backbone at Lys-122 between and configurations
does occur on the nano-second timescale. The and Lys-122 configurations favour inward (into the
active site) and outward orientation, respectively. The loss of activity when the conserved Gly in loop L2
next door to Lys-122 was substituted with ala or pro supports a requirement for L2 backbone flexibility,
spurring the question as to whether the docking of the AI helix could impede transitions between the
and configurations of the backbone and in this way impede contact with substrates in the active site.
However, our analysis showed that although the configuration was highly correlated with the backbone
trap, Lys-122 contact with CTP was compatible with either configuration [Figure D.3]. Notably, the
configuration did not preclude an interaction of Lys-122 with CTP when the AI was present. The side
chain of Lys-122 can effectively contort to make contact with CTP. We cannot exclude a requirement for
the configuration (which is enabled by glycine at residue 123) for Lys-122 interaction with
phosphocholine or the transition state. Our simulations did not test this. Thus, the data do not provide
support for a lock-in of a particular L2 backbone conformation as a means for AI suppression of catalytic
265
function; however, a role for backbone flexibility of loop L2 to facilitate Lys-122 sequential access to
Another more obvious mechanism to reduce the frequency of productive conformations is to capture
alternative non-productive conformations. In CCT, the AI helix-turn forged frequent contacts with the
Lys-122 -amino group. Even with the active site occupied by CTP, the backbone carbonyls of the AI-
turn were effective at steering the catalytic lysine away from the substrate (the backbone trap increased
from 8-14% frequency to ~40% frequency). The reaction rate will scale inversely with the cumulative
time spent in non-productive alignments of a key catalytic residue. There are other enzymes that use
electrostatic steering to forge inhibitory contacts with an active site residue, but utilizing side-chain rather
than backbone carbonyls as the interloper. For example, Src kinases employ one or more basic residues in
the regulatory A loop to trap the glutamate in helix C that is a component of the KDE triad involved in
ATP binding. Activation by phosphorylation of certain tyrosines steers the arginines away from the
glutamate [5]. By analogy, in CCT, the catalytic residue Lys-122 is competitively pulled away from an
optimal orientation for substrate contact by the auto-inhibitory helix-turn, which presents a strongly
electronegative pull. Another example of electrostatic steering is found in PGE2 synthase, where an
oxidation step requiring Arg-126 is blockaded by an electrostatic interaction with Asp-49 [31]. In PGE2
synthase the cofactor glutathione (GSH) can overcome this inhibitory interaction, whereas in CCT,
effective desilencing requires larger domain movements (i.e. the dissociation of the AI helix-turn away
266
6.4.3 Silencing by constraining the E helices.
In addition to influencing the positioning of Lys-122, we suggest that the AI helices inhibit the CCT
reaction by preventing conformational change in the E helices. The simulations in the absence of the AI
helices with NOE-type restraints on helices EC show that the E helices are highly dynamic in a hinge
segment between residues ~210 and 215. The EN region is much less so. The hinge undergoes many
different contortions, enabling splaying apart of E helices to contact loop L2 at the base of the active
site. The contact with L2 can be maintained for hundreds of nanoseconds, and in one replicate, the
splayed conformation was acquired during the equilibration and persisted in both chains for the entire
production run (960 ns). The principle stabilizing force for this interaction is an aromatic cluster between
Phe-124 in L2 and the two tyrosines of the E hinge. This aromatic cluster was observed in 17 of 20
chains in the NOE simulations. In the other three chains the EC of that chain did not migrate toward L2,
but stayed linked to its partner EC. In addition, a hydrogen bond between the Tyr-213 hydroxyl in the
E hinge and the Phe-124 backbone carbonyl in L2 was observed for hundreds of ns in splayed
conformations. We note that Phe-124 of the aromatic cluster is not universally conserved. However, the
residues that can substitute for it are leucine (hydrophobic), arginine, and glutamine. The latter two may
replace a purely aromatic interaction with a π – cation/amine interaction [32]. It is intriguing that the
aromatic clusters that mediate the loop L2 – AI helix interaction may be replaced with an alternative
aromatic cluster between L2 and the E hinge upon dissociation of the AI helices.
The entire C-terminus of E may be intrinsically malleable in the absence of the AI helices.
There was no electron density for these residues in the crystal structure of CCT236, which contains all of
catalytic domain, the E, and the linker to domain M (but no AI helix). However, we were forced to put
some constraints on the C-terminus of the E helices, because without them the tail end entered into the
267
In vitro cross-linking also provided evidence that lipid-induced dissociation of the AI resulted in
increased distance between the two EC helices, consistent with the splaying observed in silico with the
AI helices removed. An alternative hypothesis, that lipid vesicles reduce the accessibility and/or reactivity
of the cross-linkers, is unlikely for two reasons: (i) the disassembly of the four-helix bundle should
increase the exposure of the reactive cysteines, leading to elevated cross-linking; and (ii) in separate
experiments, we found that the EC does not bury in the membrane in the CCT mem form. 2 Most
vesicles. The disulfide was located just C-terminal to the E hinge segment. This suggests that
reorganization and/or dissociation of the EC helices is required for activation of the membrane-bound
CCT, and that the four-helix bundle silences CCT by virtue of the constraining action on the two E
helices.
What is the potential impact of the E plasticity on catalysis? We know that membrane binding
displaces the AI helices, which would allow a large increase in conformational sampling of the E
helices. But perhaps only a few of those conformers are productive for speeding up catalysis. We
hypothesize that membrane binding of the M domain (and perhaps the short amphipathic linker segment
intervening between the E and domain M) promotes that productive conformation. As illustrated in the
captured images [Figure 6.5] and in the model shown in Figure 6.8, the bent helix E configurations
could enable close approach of the catalytic domain to the membrane surface. Whether this is important
for catalysis or for efficiency of PC synthesis remains to be probed in future studies. Delivery of product
(CDP-choline) at the membrane would facilitate its utilization in the last step of the CDP-choline
2
D.G. Knowles and R.B. Cornell, unpublished results
268
studies by George et al. [33, 34] showed that [3H]choline flux into PC in cultured glioma cells was not
reduced by excess CDP-choline delivered by cell permeabilization, providing evidence for metabolite
channeling within the CDP-choline pathway. Although an intriguing idea, CCT and the
phosphotransferase often localize to different membrane systems in cells [35]. We propose three
hypotheses for how the bent E with its close contacts with loop L2 could facilitate catalysis. (i) The bent
E could eliminate the backbone trap. In simulations without the AI helices, bent E helices enabled H-
bonding between the hydroxyl of Tyr-213 with the Phe-124 carbonyl, removing this carbonyl oxygen
from a trapping interaction with Lys-122. A hydrogen bond between these two atoms was observed in one
chain of the partially active CCT236 crystal structure [20]. Loss of H-bonding potential at Tyr-213 by
mutation to phenylalanine partially inhibited lipid activation, 3 implying a function for the Tyr-213
hydroxyl. (ii) Residues in the bent E helices may participate in one or more steps in the catalytic cycle.
Arg-223 at the C-terminus of the E made frequent interactions with the -phosphate of CTP that lasted
for up to 700 ns. Thus, upon membrane binding and acquisition of a bent E, Arg-223 may assist in the
binding of the substrate CTP. Alternatively, Arg-223, which bonds only to the -phosphate, may provoke
Lys-122 to interact with the -phosphate of CTP, phosphocholine, or the transition state. A third
possibility is that Arg-223 may assist in displacement of the product, diphosphate. Interestingly CCT-
R223S is one of several alleles in humans that suffer from an inherited disorder, SMD-CRD [15]. CCT-
R223S has a 16-fold lower kcat/Km for both substrates than the WT enzyme. 4 The contact frequency
between Ser-223 and CTP phosphates would be much lower than that for Arg-223. (iii) Residues from
the bent E may seal off the opening of the active site bounded by Loop L2 and the EN of both chains
3
J. Lee, S.G. Taneva, and R.B. Cornell, unpublished results
4
R.B. Cornell, unpublished results
269
[Figure 6.4-C]. This could assist catalysis by trapping the reaction intermediates and perhaps excluding
water. It is tempting to speculate that the bent helix conformation might be dominant in one step of the
catalytic cycle (e.g. in the lead-up to the chemical step) and that subsequently its reorganization to a more
Signaling helix pairs are emerging as agents of communication between domains of regulatory
proteins, such as receptor tyrosine kinases [36], two-component signaling systems [37], and cytokine
receptors [38-40]. The helix deformations that transduce inter-domain signals can occur within
microseconds after a trigger is applied in vitro [37], lending validity to our in silico results. We propose
that the conformational malleability of the E helix pair that bridges the membrane binding and catalytic
domains of CCT makes it an ideal element adapted by evolution for transducing signals from membrane
to active site.
270
Figure 6.8: New model for CCT activation.
In its soluble form, the two AI helices silence CCT by (i) capturing the catalytic lysine (Lys-122) with
backbone carbonyls and (ii) restraining helix αE dynamics. In response to increased anionic charge on the
membrane surface, the positively charged leash segment is attracted to the membrane, and its binding
facilitates dissociation of the AI helices. The αE helices gain conformational freedom, but the binding of
the linker segment to the membrane creates a platform to stabilize αE conformations that are productive
271
for catalysis. We propose that the bend at the αE hinge is featured in productive conformations. The linker
is represented as a simple trapezoid because its structure is unknown. The linker-membrane interaction
may be dynamic, enabling sampling of different αE (and linker) conformations during a catalytic cycle.
The bent αE helices and the folding of the linker brings the catalytic domain closer to the membrane
surface. Images are from NOE-restrained simulations. The disordered N and P regions are not displayed.
The crystal structure of the rat CCT dimer, CCT1-312(238-269), was used as the starting structure for
simulation (PDB code 4MVC) [6]. The N termini of each chain were acetylated, and the C termini were
amidated to eliminate charged ends. The crystal structure was solved in complex with CDP-choline. To
create a CTP-bound model, the crystal structure of GCT (PDB code 1COZ) [21] was first aligned to
4MVC using the superimpose module of Coot software [41]. Next, the CTP coordinates from the aligned
GCT were combined with the coordinates of the CCT protein from 4MVC. Hydrogens were added to the
CTP molecule using the program Avogadro [42]. The CTP acidic oxygens were deprotonated, giving it a
charge of -4 e. Hydrogens were added to the protein using the pdb2gmx module of the GROMACS
package. Each structure was solvated with TIP3P water molecules [43, 44] in a truncated dodecahedron-
shaped simulation box with a minimum distance of 1.2 nm around the protein. Na+ and Cl- ions were
added to neutralize the protein/ligand charge and then to obtain a 0.15 M ionic concentration.
MD simulations were run with the GROMACS version 4.6.7 package [45, 46]. The protein was
described using the all atom AMBER99SB-ILDN force field [47]. The parameters for CTP were taken
272
from Demir and Amaro [48] and were adapted for compatibility with GROMACS using the python script
ACPYPE [49].
The simulations fall into four different groups distinguished by the presence or absence of CTP,
and the presence or absence of the AI helix. In addition, we applied two methods for restraining the EC
helix, for a total of eight conditions, each with five replicates. In preliminary simulations without external
restraints, the C-terminal ends of the E helices were prone to unwinding and migration into the active
site pockets. To prevent this, we applied two different restraining methods [Figure D.9]. In the first, the
heavy backbone atoms of EC residues (216-223 inclusive) were restrained using position restraints with
a force constant of 1000 kJ mol-1 nm-2. This ensures these helices are locked in their crystal conformation
and pinned in space. In the second method, we applied NMR NOE-type distance restraints between all
pairs of heavy backbone atoms in the EC segment, using force constants of 1000 kJ mol-1 nm-2. This
maintained helicity, while allowing the helices freedom to move relative to each other. The side chain
All systems were energy minimized using the steepest descent algorithm, first with position
restraints applied on the non-hydrogen atoms of protein and CTP molecules, and then without. The
resulting structures were equilibrated by simulation for 40 ns at a temperature of 310 K using the v-
rescale thermostat [50] with a time constant of 1.0 ps and a pressure of 1 bar using the Berendsen pressure
coupling [51] with a p of 5.5 ps. This step was followed by 960 ns of production simulation at 310 K, in
which the pressure was maintained at 1 bar using the isotropic Parrinello-Rahman barostat [52, 53] and a
p of 5.5 ps. The van der Waals and the electrostatic interactions were both cut-off at 1.0 nm, whereas the
long-range electrostatic interactions were treated with the PME algorithm [54, 55]. The potential-shift-
Verlet modifier was used to shift both the van der Waals and coulomb potentials to zero at the cut-off. All
273
bonds were constrained using the LINCS algorithm [56] and a 2 fs time-step was used for the simulations.
Atomic coordinates were saved every 100 ps during the 40 ns equilibration and 960 ns production runs.
The coordinates of every 100 ps of each trajectory was fitted (rotated and translated) to the Cα
atoms of five stable beta strands on each monomer (residues 77-83, 106-113, 144-149, 165-169, and 191-
194), and the fitted trajectories were used for the analyses. Distances between select atoms, H-bonding
analysis, dihedrals at Lys-122, and RMSF analysis of select residues were carried out with GROMACS
tools. For the H-bond analysis, a value of 3 Å was used as the “donor-acceptor” distance cut-off. The
RMSF analysis used the coordinates at the end of the 40 ns equilibration as the reference for the change in
position of (i) heavy atoms of a specified residue side chain or (ii) The N, Cα, and carbonyl backbone
atoms of specified loops using the GROMACS module g_rmsf (with options -res and -nofit). Outputs of
the analyses were imported into Excel for averaging and statistical analyses. Unpaired 2-tailed t-tests
All CCT constructs were derived from Rattus norvegicus pcyt1a (α isoform; UniProtKB/Swiss-Prot
accession number P19836). All primer designs and codon substitutions were performed using
QuikChange site directed mutagenesis (Agilent). Schematics of all constructs are shown in Figure D.10.
The preparation of pET14b-CCT236 and pET14b-CCT312 was described in Lee et al. [6]. These were
used as the templates for engineering an Ala or Pro substitution at Gly-123 by site directed mutagenesis.
pET14b-CCT236 was also used for engineering a Trp replacement at Phe-124. All pET14b CCT
274
constructs contained a plasmid-derived His tag and a thrombin cleavage sequence at the N terminus
(MGSSH6SSGLVPRGSH).
C-terminally truncated proteolytic fragments were sometimes found as contaminants in the His-CCT312
preparations. To facilitate the removal of these fragments during purification, we prepared pET24a-
CCT312 with the His tag at the C-terminus (LEHHHHHHH). Fragments lacking the C-terminus would
not be retained on nickel-agarose. To prepare pET24a-CCT312-His, we amplified the open reading frame
(codons 1-312) and engineered NdeI and XhoI restriction sites at 5’ and 3’ ends, respectively, by standard
PCR methods using the template pAX142-His-CCT312, whose construction was described in Dennis et
al. [57]. The amplicon was inserted into NdeI/XhoI-cut pET24a. This construct was subsequently used to
create CCT312-His with two native tryptophans at Trp-151 and Trp-278 mutated to phenylalanine
(CCT312 (W151F/ W278F)). We then used the Trp-free construct to engineer CCT312-His F121W or
F124W.
The construction of CCT312 with five Cys to Ser mutations (pET24a-CCT312 (Cys-free))
utilized the template pAX142-HisCCT367-T207C, described in Huang et al. [22], which contained a
single cysteine at T207C in the context of a full-length CCT. The Cys-207 mutation was first converted
back to Thr by site-directed mutagenesis, and an internal KpnI/SacI fragment, which encompassed all five
Cys to Ser mutations was transferred into pET24a-CCT312. This construct was subsequently modified by
removal of the nuclear localization sequence (12RKRRK16) by site directed deletion. The NLS
modification was to eliminate CCT dimer-mediated vesicle cross-bridging and aggregation [58, 59]. The
resulting pET24a-CCT312 (Cys-free) NLS plasmid was then used to engineer a single Cys mutation at
Ala-217, and this construct was used for the cross-linking studies.
275
6.5.5 Expression and purification of CCT constructs.
All proteins were expressed in Escherichia coli (Rosetta-DE3) cells as His6 fusion proteins and were
purified using nickel-agarose affinity chromatography as previously described [6, 28, 58] with
modifications described below. The His tag was at the N terminus of all CCT236 constructs as well as
CCT312 (WT), CCT312 (G123A), and CCT312 (G123P). The tags were not cleaved after protein
purification. The His tag was at the C terminus of CCT312 (F121W), CCT312 (F124W), and CCT312
His-CCT236 constructs were purified from the 20,000 × g × 15 min supernatant fraction of the
cell lysates. They were eluted with 350 mM imidazole, 300 mM NaCl, 10 mM Tris, pH 8.0, and 2 mM
DTT and subjected to Q-Sepharose ion-exchange chromatography to remove the imidazole. The 312His
F121W and 312His F124W proteins were mostly in the 10,000 × g pellet and were purified after
solubilization of this pellet with 6 M guanidine hydrochloride at room temperature with occasional
vortexing over 30-60 min. The denatured proteins were refolded while bound on the nickel-agarose resin
using a 6 M to 0 M urea gradient in 50 mM NaH2PO4 (pH 8.0), 0.5 M NaCl, 25 mM imidazole and 1 mM
DTT at a flow rate of ~ 2 mL/min over 1 h. The renatured proteins were eluted and dialyzed against 200
The single Cys CCT312His (NLS) A217C protein used in cross-linking experiments was
purified from the 10,000 × g pellet as described above; however, the elution and dialysis buffers
contained TCEP as a reducing agent instead of DTT. TCEP is compatible with the bis-maleimide cross-
276
6.5.6 Preparation of lipid vesicles or micelles.
Lyso PC (18:1)/egg PG (4:1) micelles used for fluorescence spectroscopy were prepared as described in
Taneva et al. [10]. Small unilamellar vesicles, egg PC/egg PG (1/1), were prepared by sonication as
described [60], and were used for enzyme activity assays and cross-linking reactions.
Steady-state fluorescence anisotropy measurements were made using a Horiba Jobin Yvon FluoroLog 3
spectrometer equipped with polarizers. Intensities with excitation polarization either Vertical (V) or
Horizontal (H) and emission polarization either V or H (IVV, IVH, IHH, and IHV) were measured and the
anisotropy (r) was calculated by the software of the instrument. The temperature was controlled at 20 oC
with a thermo-regulated cell holder. The integration time was 1 s. The excitation wavelength was 295 nm,
the emission wavelength was 345 nm, and the excitation and emission bandwidths were 4 and 12 nm,
respectively. Anisotropy values are the average of 4-6 replicates for each sample. The protein
CCT activity was performed as described previously [61]. The standard reaction mixture contained 50 nM
mCi/mmol), and either 10 mM CTP (CCT312 constructs) or 20 mM CTP (CCT236 mutants). Incubations
277
6.5.9 Bis-maleimide cross-linking.
Immediately before use, the cross-linkers were dissolved in DMSO at a 25-50 mM concentration.
The single-Cys CCT312His (NLS; A217C) was purified and stored in 1 mM TCEP to prevent
oxidation of Cys residues. Chemical cross-linking was carried out at 37 ˚C for 10 min in a volume of 50
µl, containing ~2 µM CCT, 0.5 mM TCEP, and variable concentrations of cross-linkers and phospholipid
vesicles (egg PC/egg PG (1/1)). The reactions were initiated with the cross-linking agent and were
terminated with 5.5 mM DTT. Control reactions in parallel were pre-quenched with 5.5 mM DTT at time
zero. Aliquots were mixed with reducing SDS sample buffer, heated to 85 ˚C and evaluated using 11%
SDS-PAGE. Gels were stained with Coomassie Blue or proteins were electro blotted onto
poly(vinylidene difluoride) membranes (Immun-Blot PVDF, Bio-Rad) and probed with antibody directed
CCT312His (NLS; A217C) and Cys-free CCT312His (NLS) were purified from the 10,000 × g pellet
under denaturing conditions as described above. Low concentrations of the reducing agent DTT were
used throughout the protocol to prevent misfolding of the denatured proteins. The concentration of DTT
in the refolding, elution and dialysis buffer were 100, 50 and 10 µM, respectively. Gentle oxidation of the
single Cys CCT312His (NLS; A217C) was achieved by exposure to the atmospheric air during the
CCT activity.
278
6.6 Bibliography
[1]. Huse, M., & Kuriyan, J. (2002). The conformational plasticity of protein kinases. Cell, 109(3),
275-282.
[2]. Kazanietz, M. G., & Lemmon, M. A. (2011). Protein kinase C regulation: C1 meets C-tail.
[3]. Filippakopoulos, P., Müller, S., & Knapp, S. (2009). SH2 domains: Modulators of nonreceptor
[4]. Wu, W. I., Voegtli, W. C., Sturgis, H. L., Dizon, F. P., Vigers, G. P., & Brandhuber, B. J. (2010).
Crystal structure of human AKT1 with an allosteric inhibitor reveals a new mode of kinase
[5]. Xu, W., Doshi, A., Lei, M., Eck, M. J., & Harrison, S. C. (1999). Crystal structures of c-Src
[6]. Lee, J., Taneva, S. G., Holland, B. W., Tieleman, D. P., & Cornell, R. B. (2014). Structural basis
[7]. Dunne, S. J., Cornell, R. B., Johnson, J. E., Glover, N. R., & Tracey, A. S. (1996). Structure of
35(37), 11975-11984.
[8]. Johnson, J. E., Aebersold, R., & Cornell, R. B. (1997). An amphipathic α-helix is the principle
279
[9]. Johnson, J. E., Rao, N. M., Hui, S. W., & Cornell, R. B. (1998). Conformation and lipid binding
[10]. Taneva, S., Johnson, J. E., & Cornell, R. B. (2003). Lipid-induced conformational switch in the
[11]. Cornell, R. B. (2016). Membrane lipid compositional sensing by the inducible amphipathic helix
of CCT. Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids, 1861(8),
847-861.
[12]. Bigay, J., & Antonny, B. (2012). Curvature, lipid packing, and electrostatics of membrane
886-895.
[13]. Ding, Z., Taneva, S. G., Huang, H. K., Campbell, S. A., Semenec, L., Chen, N., & Cornell, R. B.
[14]. Yang, W., Boggs, K. P., & Jackowski, S. (1995). The association of lipid activators with the
increasing the affinity of the enzyme for CTP. Journal of Biological Chemistry, 270(41), 23951-
23957.
[15]. Hoover-Fong, J., Sobreira, N., Jurgens, J., Modaff, P., Blout, C., Moser, A., Kim, O.H., Cho, T.J.,
Cho, S.Y., Kim, S.J. & Jin, D. K. (2014). Mutations in PCYT1A, encoding a key regulator of
280
[16]. Testa, F., Filippelli, M., Brunetti-Pierri, R., Di Fruscio, G., Di Iorio, V., Pizzo, M., Torella, A.,
Barillari, M.R., Nigro, V., Brunetti-Pierri, N. & Simonelli, F. (2017). Mutations in the PCYT1A
gene are responsible for isolated forms of retinal dystrophy. European Journal of Human
[17]. Yamamoto, G.L., Baratela, W.A., Almeida, T.F., Lazar, M., Afonso, C.L., Oyamada, M.K.,
Suzuki, L., Oliveira, L.A., Ramos, E.S., Kim, C.A. & Passos-Bueno, M. R. (2014). Mutations in
PCYT1A cause spondylometaphyseal dysplasia with cone-rod dystrophy. The American Journal
[18]. Payne, F., Lim, K., Girousse, A., Brown, R.J., Kory, N., Robbins, A., Xue, Y., Sleigh, A.,
Cochran, E., Adams, C. & Borman, A. D. (2014). Mutations disrupting the Kennedy
phosphatidylcholine pathway in humans with congenital lipodystrophy and fatty liver disease.
[19]. Bork, P., Holm, L., Koonin, E. V., & Sander, C. (1995). The cytidylyltransferase superfamily:
Identification of the nucleotide‐binding site and fold prediction. Proteins: Structure, Function,
[20]. Lee, J., Johnson, J. E., Ding, Z., Paetzel, M., & Cornell, R. B. (2009). Crystal structure of a
284(48), 33535-33548.
[21]. Weber, C. H., Park, Y. S., Sanker, S., Kent, C., & Ludwig, M. L. (1999). A prototypical
281
[22]. Huang, H. K., Taneva, S. G., Lee, J., Silva, L. P., Schriemer, D. C., & Cornell, R. B. (2013). The
amphipathic helix flanking its active site. Journal of Molecular Biology, 425(9), 1546-1564.
regulation, and structure of an amphitropic enzyme required for membrane biogenesis. Progress
[24]. Helmink, B. A., Braker, J. D., Kent, C., & Friesen, J. A. (2003). Identification of lysine 122 and
[25]. Pattridge, K. A., Weber, C. H., Friesen, J. A., Sanker, S., Kent, C., & Ludwig, M. L. (2003).
glycerol and the role of lysine residues in catalysis. Journal of Biological Chemistry, 278(51),
51863-51871.
[27]. Emekli, U., Schneidman‐Duhovny, D., Wolfson, H. J., Nussinov, R., & Haliloglu, T. (2008).
[28]. Xie, M., Smith, J. L., Ding, Z., Zhang, D., & Cornell, R. B. (2004). Membrane binding modulates
[29]. Green, N. S., Reisler, E., & Houk, K. N. (2001). Quantitative evaluation of the lengths of
homobifunctional protein cross‐linking reagents used as molecular rulers. Protein Science, 10(7),
1293-1304.
282
[30]. Nashine, V. C., Hammes-Schiffer, S., & Benkovic, S. J. (2010). Coupled motions in enzyme
[31]. Brock, J.S., Hamberg, M., Balagunaseelan, N., Goodman, M., Morgenstern, R., Strandback, E.,
[32]. Burley, S. K., & Petsko, G. A. (1986). Amino‐aromatic interactions in proteins. FEBS Letters,
203(2), 139-143.
[33]. George, T. P., Morash, S. C., Cook, H. W., Byers, D. M., Palmer, F. S. C., & Spence, M. W.
intermediates. Biochimica et Biophysica Acta (BBA)-Lipids and Lipid Metabolism, 1004(3), 283-
291.
[34]. George, T. P., Cook, H. W., Byers, D. M., Palmer, F. B., & Spence, M. W. (1991). Channeling of
[35]. McMaster, C. R. (2018). From yeast to humans–roles of the Kennedy pathway for
[36]. Trenker, R., Call, M. J., & Call, M. E. (2016). Progress and prospects for structural studies of
39, 115-123.
[37]. Berntsson, O., Diensthuber, R.P., Panman, M.R., Björling, A., Gustavsson, E., Hoernke, M.,
Hughes, A.J., Henry, L., Niebling, S., Takala, H. & Ihalainen, J. A. (2017). Sequential
conformational transitions and α-helical supercoiling regulate a sensor histidine kinase. Nature
283
[38]. Lu, X., Gross, A. W., & Lodish, H. F. (2006). Active conformation of the erythropoietin receptor:
[40]. Matthews, E.E., Thévenin, D., Rogers, J.M., Gotow, L., Lira, P.D., Reiter, L.A., Brissette, W.H.
dimerization, rotation, and allosteric modulation. The FASEB Journal, 25(7), 2234-2244.
[41]. Emsley, P., Lohkamp, B., Scott, W. G., & Cowtan, K. (2010). Features and development of Coot.
[42]. Hanwell, M. D., Curtis, D. E., Lonie, D. C., Vandermeersch, T., Zurek, E., & Hutchison, G. R.
(2012). Avogadro: An advanced semantic chemical editor, visualization, and analysis platform.
[43]. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983).
Comparison of simple potential functions for simulating liquid water. The Journal of Chemical
[44]. Neria, E., Fischer, S., & Karplus, M. (1996). Simulation of activation free energies in molecular
[45]. Hess, B., Kutzner, C., Van Der Spoel, D., & Lindahl, E. (2008). GROMACS 4: Algorithms for
highly efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory
[46]. Pronk, S., Páll, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts, M.R., Smith, J.C.,
Kasson, P.M., Van Der Spoel, D. & Hess, B. (2013). GROMACS 4.5: A high-throughput and
highly parallel open source molecular simulation toolkit. Bioinformatics, 29(7), 845-854.
284
[47]. Lindorff‐Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., & Shaw, D.
E. (2010). Improved side‐chain torsion potentials for the Amber ff99SB protein force field.
[48]. Demir, O., & Amaro, R. E. (2012). Elements of nucleotide specificity in the Trypanosoma brucei
mitochondrial RNA editing enzyme RET2. Journal of Chemical Information and Modeling,
52(5), 1308-1318.
[49]. da Silva, A. W. S., & Vranken, W. F. (2012). ACPYPE-Antechamber python parser interface.
[50]. Bussi, G., Donadio, D., & Parrinello, M. (2007). Canonical sampling through velocity rescaling.
[51]. Berendsen, H. J., Postma, J. V., van Gunsteren, W. F., DiNola, A. R. H. J., & Haak, J. R. (1984).
Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics, 81(8),
3684-3690.
[52]. Nosé, S. (1984). A molecular dynamics method for simulations in the canonical ensemble.
[53]. Parrinello, M., & Rahman, A. (1981). Polymorphic transitions in single crystals: A new
[54]. Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N⋅log (N) method for
Ewald sums in large systems. The Journal of Chemical Physics, 98(12), 10089-10092.
[55]. Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., & Pedersen, L. G. (1995). A
smooth particle mesh Ewald method. The Journal of Chemical Physics, 103(19), 8577-8593.
[56]. Hess, B., Bekker, H., Berendsen, H. J., & Fraaije, J. G. (1997). LINCS: A linear constraint solver
285
[57]. Dennis, M. K., Taneva, S. G., & Cornell, R. B. (2011). The intrinsically disordered nuclear
localization signal and phosphorylation segments distinguish the membrane affinity of two
[58]. Taneva, S., Dennis, M. K., Ding, Z., Smith, J. L., & Cornell, R. B. (2008). Contribution of each
283(42), 28137-28148.
[59]. Taneva, S. G., Patty, P. J., Frisken, B. J., & Cornell, R. B. (2005). CTP:phosphocholine
44(26), 9382-9393.
Surface curvature, acyl chain length, and lipid-phase dependence for activation. Biochemistry,
30(24), 5881-5888.
[61]. Sohal, P. S., & Cornell, R. B. (1990). Sphingosine inhibits the activity of rat liver
11750.
286
Chapter Seven: Conclusions and Future Directions
7.1 Conclusions
Computer simulation can assist in a better understanding of the biomolecules, including lipid aggregates,
proteins, and nanoparticles [1]. A comprehensive review on how molecular simulation techniques,
delivery systems was provided in Chapter 3 [2]. In this thesis, three more examples for the successful
In Chapter 4, binary systems composed of POPC and KC2 were studied as a function of pH and
mixing ratio. The results strongly suggest that the interactions between neutral KC2 and POPC are highly
dependent on the pH level. Neutral KC2 were shown to segregated from POPC at high pH level.
However, the KC2H, corresponding to the low pH levels, stayed in the lipid-water interface and interact
with the POPC and solvent. This segregation was further found to be a function of mixing ratio. Systems
with less content of KC2 had larger proportion of KC2 stayed in the lipid-water interface. The effect of
KC2H on POPC was further shown to be similar to the effect of other cationic lipids on PC lipids
reported in literature [3]. Adding KC2H were shown to affect the ion distribution in vicinity of bilayer
surface. The POPC headgroup tilt angle, i.e. the angle between POPC P→N vector and the vector normal
to the bilayer surface were also shown to decrease upon adding KC2H to the system. Segregation of
neutral KC2 from POPC was proposed to be the underlying reason for the observed electron dense cores
in LNPs [4, 5], and limited drug release for LNPs containing KC2 [6].
In Chapter 5, DOPE and POPE HII systems were studied using MD simulations. Simulations,
using both semi-isotropic and anisotropic pressure couplings, could successfully predict the structural
parameters of HII systems as a function of hydration level and temperature. HII structure was shown to be
287
highly affected by the hydration level. The temperature and lipid type were two other factors affected the
HII structure. Dehydration resulted in the formation of longer lipid cylinders, with smaller R w, APL, and
SCD parameters. Increasing temperature was shown to decrease both the maximum hydration of HII lattice
and SCD parameters. Provided that a single measurement of dhex in the presence of excess water be
available, the maximum hydration of HII lattice could also be estimated using MD simulations.
experiments, and parameterization of new molecules using target data in HII phase was proposed.
In Chapter 6, MD simulation was used to study the structure and dynamics of CCT enzyme in
both presence and absence of two autoinhibitory (AI) helices [7]. The interactions between CCT and its
substrate (CTP) were also studied for each of these two conditions. The results suggested that AI helices
have a dual role in inhibiting CCT activity. AI helices were shown to affect the dynamics of loop L2 in
the active site. Both the dynamics of loop L2, and its interaction with CTP are believed to be important
for the catalytic activity of CCT. AI helices were also shown to repress the dynamics of two helices
(referred as αE helices). Upon the removal of AI helices, these αE helices were shown to bend towards
the active sites and interact with the substrates. This bending made the contacts between residues, e.g.
Arg-223, at the C-terminus of αE helices with CTP possible. Finally, a new model for the CCT activation
was proposed. In this model, the CCT structure with the bent αE helices was proposed as the
In addition to these specific scientific findings from each chapter, a more general conclusion can
also be made. Both MD simulation and experimental techniques have their own limitations which must be
considered when the data are interpreted and compared between two approaches. This will prevent any
288
There are several factors which determines the reliability of MD simulations [8]. The reliability
of an MD simulation and any computational technique is highly dependent on the assigned parameters to
the comprising molecules in the system. Parameters developed for any new molecule need to be well
tested and validated before being used. This validation should be done, preferably, by direct comparison
with available experimental data [9]. However, to make such a direct comparison, the systems in both
simulation and experiments should be equivalent as much as possible. If not, comparing the data between
Simulation time is another important factor to be considered [8, 10]. The appropriate time scale is
dependent on the questions asked and the properties of interest in the system. Short simulation times
could result in lack of enough sampling and, consequently, in inaccurate data and conclusions.
Regarding the experimental techniques, the data obtained from one technique might not be
enough to provide a correct/complete description for the molecular system under study. Thus,
characterizing a molecular system usually requires conducting several complementary experiments. The
study conducted on POPC/KC2 mixtures was such an example, where MD simulation, 2H NMR, SAXS,
and cryo-TEM were used to investigate the KC2 localization in binary mixtures. Investigation of CCT
enzyme activation and inhibition mechanism was another example, where both computational and
experimental techniques were combined in a complementary way to detect, test, and validate the
observations.
The studies and findings presented as well as the methods developed in this thesis could be used for
• KC2 is one of the most efficacious ICLs for gene silencing in liver. DLin-MC3-DMA (also known as
MC3) [Figure 7.1] is another ICL, which is known to be 10 times more potent than KC2 in silencing
289
a gene in hepatocytes in mice [11-13]. However, the underlying molecular interactions causing this
difference are not known. Simulations of MC3 and comparing the results with KC2 simulations might
assist in determining the structural factors responsible for this observed difference in efficacy, and
potentially assist in rational design of new ICLs with a higher efficacy. For instance, it would be
interesting to study interactions between MC3 and POPC to see if the observed segregation is
happening for MC3 or not. The force field parameters for MC3 and its protonated state are not
available and need to be developed. The protocol developed for parameterizing KC2 and KC2H could
Hydrogen atoms, except the protonation one, are not shown for clarity. Avogadro [14] and Chimera [15]
290
• LNPs containing ICLs also contain DSPC and cholesterol [Figure 1.4] [12]. Looking at interactions
between KC2 and KC2H with these molecules could be directly related to the nano-formulations. It
would be interesting to see if the KC2 segregation is also happening in presence of DSPC and
cholesterols, and if yes, how a change in DSPC/KC2/cholesterol mixing ratio will affect the KC2
solubility.
• This class of LNPs are the leading delivery systems for siRNA molecules [16, 17]. The siRNA
molecules are known to interact with KC2H inside the LNPs. Interactions between siRNAs and
KC2H in presence of water and ions could be useful to look at. As an example, effect of KC2H
concentration, siRNA sequence and length, and ion concentration and types are several factors to
study. Looking at these interactions could provide insight into the stability of the LNPs core structure
• Given the large size of LNPs (~ 100 nm in diameter), simulation of these systems in a full atomistic
resolution is not currently feasible. Therefore, coarse-grained simulations in which several atoms are
usually treated as a single interacting bead, should be used instead. MARTINI [18] is a popular
coarse-grained model which has been successfully used in many studies on biomolecules, including
siRNA-LNPs [4]. The validated all-atom parameters for KC2 and KC2H, and the atomistic
simulations conducted in this study can be further used for developing, validating, and improving
• The current siRNA-LNPs containing ICLs have reached the encapsulation efficacies of 100 % [19].
However, currently, only ca. 1-2 % of the internalized siRNAs are released to cytoplasm while others
are transferred to the lysosome and degraded [6]. Therefore, further optimization should be done on
drug release from endosomes. To do so, an understanding of the underlying mechanism and
interactions involved in the drug escape from endosome for this class of LNPs is required. According
to the proposed mechanism of drug release from endosome for these systems, endosomal membrane
291
destabilization is due to cationic-anionic lipids interactions between cationic lipids of siRNA-LNPs
and negatively charged lipids of endosomal membrane [20]. Forming ion-pairs with an effective
cone-shaped morphology was hypothesized to cause such a phase transition from lamellar to HII
phase, and consequently destabilizes the endosomal membrane required for drug release.
Therefore, study of KC2H interactions with negatively charged lipids in HII phase would be
interesting. Both early endosomes and late endosomes are known to contain negatively charged lipids,
e.g. phosphatidylserine (PS) and lysobisphosphatidic acid (LBPA), in their lipid composition [21].
The C36 FF parameters for LBPA lipids have been recently developed in our group [22]. Given the
validated parameters for KC2 and KC2H (Chapter 2 and 4), validated HII system setups (Chapter
5), and available parameters for LBPA, this study is possible now.
• Cholesterol is of special interest in lipid research, and is known to affect the structural properties of
lipids in both bilayers and HII phases [23]. It would be interesting to study the effect of cholesterol
concentration on structure of DSPS/KC2H or LBPA/KC2H systems in HII phase [Figure 7.2], and
see how cholesterol is distributed in HII phase. These simulations could potentially provide insight
into the role of cholesterol in HII formation and consequently in drug release.
292
Figure 7.2: HII phase corresponding to DSPS/KC2H/cholesterol (35/35/30).
Left) A single lipid tube composed of DSPS, KC2H and cholesterol, filled with water molecules. Red,
dark blue, and green spheres represent the nitrogen atom of DSPS, nitrogen atom of KC2H, and oxygen
atom in cholesterol molecules, respectively. Orange and light blue lines represent the DSPS and KC2H
tails, whereas the cholesterol body are colored in tan. Water represented as the cyan surface. Hydrogens
are not shown for clarity. Right) a small part of HII phase composed of such single water channel. Lipid
cylinders are arranged in a hexagonal symmetry. Blue box represents the triclinic simulation box. VMD
• Finally, the study on CCT enzyme, determined a new role for AI helices (inhibition of the αE
bending) and supported previously reported role (the backbone trap) [25]. These simulations also
resulted in new structures due to αE helices bending, as a result of in silico removal of AI helices.
These new structures were hypothesized to be corresponding to the active state of CCT upon its
binding and interaction with the PC-deficient lipid membranes [7]. These structures could be used as
the starting structure for future simulations in presence of lipid membranes. It would be interesting to
293
study how CCT enzyme and its residues interact with a PC-poor and PC-rich membranes. To do this
study, however, the structure for the linker region connecting the αE helices to the M domains should
be determined first.
• CCT enzyme with and without CDP-choline in its active sites has been previously studied using both
MD simulations and experimental techniques [25]. The focus of that study (200 ns simulations) was
on the effect of AI helices on loop L2 dynamics in presence and absence of CDP-choline. It would be
interesting to study this system again using the technique used in Chapter 6, i.e. using NOE
restraints, and compare the results with CCT-CTP system. This might assist in clarifying the role of
bent αEs in different stages of catalysis process. There were more hypothesis and mechanisms
regarding αE bending discussed in Chapter 6 which could be a subject for future experimental and
computational studies.
7.3 Bibliography
[1]. Schlick, T., Collepardo-Guevara, R., Halvorsen, L. A., Jung, S., & Xiao, X. (2011). Biomolecular
modeling and simulation: A field coming of age. Quarterly Reviews of Bophysics, 44(2), 191-
228.
[2]. Ramezanpour, M., Leung, S. S. W., Delgado-Magnero, K. H., Bashe, B. Y. M., Thewalt, J., &
1858(7), 1688-1709.
[3]. Gurtovenko, A. A., Patra, M., Karttunen, M., & Vattulainen, I. (2004). Cationic DMPC/DMTAP
[4]. Leung, A.K., Hafez, I.M., Baoukina, S., Belliveau, N.M., Zhigaltsev, I.V., Afshinmanesh, E.,
Tieleman, D.P., Hansen, C.L., Hope, M.J. & Cullis, P. R. (2012). Lipid nanoparticles containing
294
siRNA synthesized by microfluidic mixing exhibit an electron-dense nanostructured core. The
[5]. Kulkarni, J.A., Darjuan, M.M., Mercer, J.E., Chen, S., van der Meel, R., Thewalt, J.L., Tam,
Y.Y.C. & Cullis, P. R. (2018). On the formation and morphology of lipid nanoparticles
containing ionizable cationic lipids and siRNA. ACS nano, 12(5), 4787-4795.
[6]. Gilleron, J., Querbes, W., Zeigerer, A., Borodovsky, A., Marsico, G., Schubert, U., Manygoats,
K., Seifert, S., Andree, C., Stöter, M. & Epstein-Barash, H. (2013). Image-based analysis of lipid
[7]. Ramezanpour, M., Lee, J., Taneva, S. G., Tieleman, D. P., & Cornell, R. B. (2018). An auto-
constrains a pliable, domain-bridging helix pair. Journal of Biological Chemistry, 293(18), 7070-
7084.
[8]. Ingólfsson, H. I., Arnarez, C., Periole, X., & Marrink, S. J. (2016). Computational ‘microscopy’of
[9]. Poger, D., Caron, B., & Mark, A. E. (2016). Validating lipid force fields against experimental
[10]. Tieleman, D. P. (2010). Methods and parameters for membrane simulations. In Molecular
[11]. Jayaraman, M., Ansell, S.M., Mui, B.L., Tam, Y.K., Chen, J., Du, X., Butler, D., Eltepu, L.,
Matsuda, S., Narayanannair, J.K. & Rajeev, K. G. (2012). Maximizing the potency of siRNA
lipid nanoparticles for hepatic gene silencing in vivo. Angewandte Chemie International Edition,
51(34), 8529-8533.
295
[12]. Tam, Y., Chen, S., & Cullis, P. R. (2013). Advances in lipid nanoparticles for siRNA delivery.
[13]. Wan, C., Allen, T. M., & Cullis, P. R. (2014). Lipid nanoparticle delivery systems for siRNA-
[14]. Hanwell, M. D., Curtis, D. E., Lonie, D. C., Vandermeersch, T., Zurek, E., & Hutchison, G. R.
(2012). Avogadro: An advanced semantic chemical editor, visualization, and analysis platform.
[15]. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., &
Ferrin, T. E. (2004). UCSF Chimera—a visualization system for exploratory research and
[16]. Zimmermann, T.S., Lee, A.C., Akinc, A., Bramlage, B., Bumcrot, D., Fedoruk, M.N., Harborth,
J., Heyes, J.A., Jeffs, L.B., John, M., Judge, A. D. et al. (2006). RNAi-mediated gene silencing in
[17]. Arteta, M.Y., Kjellman, T., Bartesaghi, S., Wallin, S., Wu, X., Kvist, A.J., Dabkowska, A.,
Székely, N., Radulescu, A., Bergenholtz, J. & Lindfors, L. (2018). Successful reprogramming of
[18]. Marrink, S. J., Risselada, H. J., Yefimov, S., Tieleman, D. P., & De Vries, A. H. (2007). The
MARTINI force field: Coarse grained model for biomolecular simulations. The Journal of
[19]. Belliveau, N.M., Huft, J., Lin, P.J., Chen, S., Leung, A.K., Leaver, T.J., Wild, A.W., Lee, J.B.,
Taylor, R.J., Tam, Y.K. & Hansen, C. L. (2012). Microfluidic synthesis of highly potent limit-
size lipid nanoparticles for in vivo delivery of siRNA. Molecular Therapy-Nucleic Acids, 1, e37.
296
[20]. Hafez, I. M., Maurer, N., & Cullis, P. R. (2001). On the mechanism whereby cationic lipids
[21]. Kobayashi, T., Stang, E., Fang, K. S., de Moerloose, P., Parton, R. G., & Gruenberg, J. (1998). A
lipid associated with the antiphospholipid syndrome regulates endosome structure and function.
[22]. Kalu, Nnanya, Yoav Atsmon-Raz, Sanaz Momben Abolfath, Laura Lucas, Clare Kenney,
Stephen H. Leppla, D. Peter Tieleman, & Nestorovich, E. M. (2018). Effect of late endosomal
DOBMP lipid and traditional model lipids of electrophysiology on the anthrax toxin channel
[23]. Takahashi, H., Sinoda, K., & Hatta, I. (1996). Effects of cholesterol on the lamellar and the
[24]. Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: Visual molecular dynamics. Journal of
[25]. Lee, J., Taneva, S. G., Holland, B. W., Tieleman, D. P., & Cornell, R. B. (2014). Structural basis
297
References
Achar, S., & Puddephatt, R. J. (1994). Organoplatinum dendrimers formed by oxidative addition.
Adnan, A., Lam, R., Chen, H., Lee, J., Schaffer, D.J., Barnard, A.S., Schatz, G.C., Ho, D. & Liu, W. K.
(2011). Atomistic simulation and measurement of pH dependent cancer therapeutic interactions with
Afonin, K. A., Kasprzak, W. K., Bindewald, E., Kireeva, M., Viard, M., Kashlev, M., & Shapiro, B. A.
(2014). In silico design and enzymatic synthesis of functional RNA nanoparticles. Accounts of Chemical
Ajnai, G., Chiu, A., Kan, T., Cheng, C. C., Tsai, T. H., & Chang, J. (2014). Trends of gold nanoparticle-
based drug delivery system in cancer therapy. Journal of Experimental & Clinical Medicine, 6(6), 172-
178.
Albertazzi, L., Mickler, F.M., Pavan, G.M., Salomone, F., Bardi, G., Panniello, M., Amir, E., Kang, T.,
Killops, K.L., Bräuchle, C. & Amir, R. J. (2012). Enhanced bioactivity of internally functionalized
Alkilany, A. M., & Murphy, C. J. (2010). Toxicity and cellular uptake of gold nanoparticles: What we
Allen, T. M., & Cullis, P. R. (2004). Drug delivery systems: Entering the mainstream. Science,
303(5665), 1818-1822.
298
Allen, T. M., & Cullis, P. R. (2013). Liposomal drug delivery systems: From concept to clinical
Almutairi, A., Akers, W. J., Berezin, M. Y., Achilefu, S., & Fréchet, J. M. (2008). Monitoring the
Amado Torres, D., Garzoni, M., Subrahmanyam, A. V., Pavan, G. M., & Thayumanavan, S. (2014).
Anand, R., Ottani, S., Manoli, F., Manet, I., & Monti, S. (2012). A close-up on doxorubicin binding to γ-
cyclodextrin: An elucidating spectroscopic, photophysical and conformational study. RSC Advances, 2(6),
2346-2357.
Angelova, A., Angelov, B., Mutafchieva, R., Lesieur, S., & Couvreur, P. (2010). Self-assembled
multicompartment liquid crystalline lipid carriers for protein, peptide, and nucleic acid drug delivery.
Antila, H. S., Härkönen, M., & Sammalkorpi, M. (2015). Chemistry specificity of DNA–polycation
complex salt response: A simulation study of DNA, polylysine and polyethyleneimine. Physical
Aoun, B., Sharma, V. K., Pellegrini, E., Mitra, S., Johnson, M., & Mukhopadhyay, R. (2015). Structure
and dynamics of ionic micelles: MD simulation and neutron scattering study. The Journal of Physical
299
Arai, N., Yasuoka, K., & Zeng, X. C. (2013). A vesicle cell under collision with a Janus or homogeneous
Armstrong, J. K., Hempel, G., Koling, S., Chan, L. S., Fisher, T., Meiselman, H. J., & Garratty, G.
(2007). Antibody against poly (ethylene glycol) adversely affects PEG‐asparaginase therapy in acute
Arsawang, U., Saengsawang, O., Rungrotmongkol, T., Sornmee, P., Wittayanarakul, K., Remsungnen, T.,
& Hannongbua, S. (2011). How do carbon nanotubes serve as carriers for gemcitabine transport in a drug
Arteta, M.Y., Kjellman, T., Bartesaghi, S., Wallin, S., Wu, X., Kvist, A.J., Dabkowska, A., Székely, N.,
Radulescu, A., Bergenholtz, J. & Lindfors, L. (2018). Successful reprogramming of cellular protein
production through mRNA delivered by functionalized lipid nanoparticles. Proceedings of the National
Astruc, D., Boisselier, E., & Ornelas, C. (2010). Dendrimers designed for functions: From physical,
Avgoustakis, K., Beletsi, A., Panagi, Z., Klepetsanis, P., Karydas, A. G., & Ithakissios, D. S. (2002).
PLGA–mPEG nanoparticles of cisplatin: In vitro nanoparticle degradation, in vitro drug release and in
vivo drug residence in blood properties. Journal of Controlled Release, 79(1-3), 123-135.
Avila-Salas, F., Sandoval, C., Caballero, J., Guiñez-Molinos, S., Santos, L. S., Cachau, R. E., &
González-Nilo, F. D. (2012). Study of interaction energies between the PAMAM dendrimer and
300
nonsteroidal anti-inflammatory drug using a distributed computational strategy and experimental analysis
Badu, S. R., Melnik, R., Paliy, M., Prabhakar, S., Sebetci, A., & Shapiro, B. A. (2014). Modeling of RNA
nanotubes using molecular dynamics simulation. European Biophysics Journal, 43(10-11), 555-564.
Bagai, S., Sun, C., & Tang, T. (2012). Potential of mean force of polyethylenimine-mediated DNA
Bakry, R., Vallant, R. M., Najam-ul-Haq, M., Rainer, M., Szabo, Z., Huck, C. W., & Bonn, G. K. (2007).
Bandyopadhyay, S., Klein, M. L., Martyna, G. J., & Tarek, M. (1998). Molecular dynamics studies of the
hexagonal mesophase of sodium dodecylsulphate in aqueous solution. Molecular Physics, 95(2), 377-384.
Bao, G., Mitragotri, S., & Tong, S. (2013). Multifunctional nanoparticles for drug delivery and molecular
Baoukina, S., Monticelli, L., & Tieleman, D. P. (2013). Interaction of pristine and functionalized carbon
nanotubes with lipid membranes. The Journal of Physical Chemistry B, 117(40), 12113-12123.
Barauskas, J., Johnsson, M., & Tiberg, F. (2005). Self-assembled lipid superstructures: Beyond vesicles
Barnaby, S. N., Sita, T. L., Petrosko, S. H., Stegh, A. H., & Mirkin, C. A. (2015). Therapeutic
applications of spherical nucleic acids. In Nanotechnology-Based Precision Tools for the Detection and
301
Barnard, A., Posocco, P., Pricl, S., Calderon, M., Haag, R., Hwang, M.E., Shum, V.W., Pack, D.W. &
Smith, D. K. (2011). Degradable self-assembling dendrons for gene delivery: Experimental and
theoretical insights into the barriers to cellular uptake. Journal of the American Chemical Society,
133(50), 20288-20300.
Barnoud, J., Urbini, L., & Monticelli, L. (2015). C60 fullerene promotes lung monolayer collapse.
Barraza, L. F., Jiménez, V. A., & Alderete, J. B. (2015). Effect of PEGylation on the structure and drug
fluorouracil with native and PEGylated PAMAM‐G4. Macromolecular Chemistry and Physics, 216(16),
1689-1701.
Bartesaghi, A., Merk, A., Banerjee, S., Matthies, D., Wu, X., Milne, J. L., & Subramaniam, S. (2015). 2.2
348(6239), 1147-1151.
Bechinger, B., & Seelig, J. (1991). Interaction of electric dipoles with phospholipid head groups. A
deuterium and phosphorus-31 NMR study of phloretin and phloretin analogs in phosphatidylcholine
Beddoes, C. M., Case, C. P., & Briscoe, W. H. (2015). Understanding nanoparticle cellular entry: A
Bellini, R. G., Guimarães, A. P., Pacheco, M. A., Dias, D. M., Furtado, V. R., de Alencastro, R. B., &
Horta, B. A. (2015). Association of the anti-tuberculosis drug rifampicin with a PAMAM dendrimer.
302
Belliveau, N.M., Huft, J., Lin, P.J., Chen, S., Leung, A.K., Leaver, T.J., Wild, A.W., Lee, J.B., Taylor,
R.J., Tam, Y.K. & Hansen, C. L. (2012). Microfluidic synthesis of highly potent limit-size lipid
Bello, M.L., Junior, A.M., Vieira, B.A., Dias, L.R., de Sousa, V.P., Castro, H.C., Rodrigues, C.R. &
intercalated drugs arrangement into layered carrier material. PLoS ONE, 10(3), e0121110.
Berchane, N. S., Carson, K. H., Rice-Ficht, A. C., & Andrews, M. J. (2007). Effect of mean diameter and
polydispersity of PLG microspheres on drug release: Experiment and theory. International Journal of
Berendsen, H. J., Postma, J. V., van Gunsteren, W. F., DiNola, A. R. H. J., & Haak, J. R. (1984).
Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics, 81(8), 3684-
3690.
Berendsen, H. J., van der Spoel, D., & van Drunen, R. (1995). GROMACS: A message-passing parallel
Berger, O., Edholm, O., & Jähnig, F. (1997). Molecular dynamics simulations of a fluid bilayer of
Bernini, M. C., Fairen-Jimenez, D., Pasinetti, M., Ramirez-Pastor, A. J., & Snurr, R. Q. (2014). Screening
of bio-compatible metal–organic frameworks as potential drug carriers using Monte Carlo simulations.
303
Berntsson, O., Diensthuber, R.P., Panman, M.R., Björling, A., Gustavsson, E., Hoernke, M., Hughes,
A.J., Henry, L., Niebling, S., Takala, H. & Ihalainen, J. A. (2017). Sequential conformational transitions
and α-helical supercoiling regulate a sensor histidine kinase. Nature Communications, 8(1), 284.
Bhattacharya, R., Kanchi, S., Roobala, C., Lakshminarayanan, A., Seeck, O.H., Maiti, P.K., Ayappa,
K.G., Jayaraman, N. & Basu, J. K. (2014). A new microscopic insight into membrane penetration and
Bianco, A., Kostarelos, K., & Prato, M. (2005). Applications of carbon nanotubes in drug delivery.
Bianco, A., Kostarelos, K., Partidos, C. D., & Prato, M. (2005). Biomedical applications of functionalized
Bigay, J., & Antonny, B. (2012). Curvature, lipid packing, and electrostatics of membrane organelles:
Blanco, E., Shen, H., & Ferrari, M. (2015). Principles of nanoparticle design for overcoming biological
Bochicchio, D., Panizon, E., Monticelli, L., & Rossi, G. (2017). Interaction of hydrophobic polymers with
Böckmann, R. A., Hac, A., Heimburg, T., & Grubmüller, H. (2003). Effect of sodium chloride on a lipid
304
Bork, P., Holm, L., Koonin, E. V., & Sander, C. (1995). The cytidylyltransferase superfamily:
Identification of the nucleotide‐binding site and fold prediction. Proteins: Structure, Function, and
Borodin, O., Smith, G. D., & Jaffe, R. L. (2001). Ab initio quantum chemistry and molecular dynamics
22(6), 641-654.
Bouchemal, K. (2008). New challenges for pharmaceutical formulations and drug delivery systems
characterization using isothermal titration calorimetry. Drug Discovery Today, 13(21-22), 960-972.
Boyd, K. J., & May, E. R. (2018). BUMPy: A model-independent tool for constructing lipid bilayers of
varying curvature and composition. Journal of Chemical Theory and Computation, 14(12), 6642-6652.
Brancolini, G., Kokh, D. B., Calzolai, L., Wade, R. C., & Corni, S. (2012). Docking of ubiquitin to gold
Brar, S. K., & Verma, M. (2011). Measurement of nanoparticles by light-scattering techniques. TrAC
Briuglia, M. L., Rotella, C., McFarlane, A., & Lamprou, D. A. (2015). Influence of cholesterol on
liposome stability and on in vitro drug release. Drug Delivery and Translational Research, 5(3), 231-242.
Brock, J.S., Hamberg, M., Balagunaseelan, N., Goodman, M., Morgenstern, R., Strandback, E.,
essential for catalysis in microsomal prostaglandin E2 synthase. Proceedings of the National Academy of
305
Brooks, B.R., Brooks, C.L., Mackerell Jr, A.D., Nilsson, L., Petrella, R.J., Roux, B., Won, Y., Archontis,
G., Bartels, C., Boresch, S. & Caflisch, A. (2009). CHARMM: The biomolecular simulation program.
Bryant, G., Pope, J. M., & Wolfe, J. (1992). Low hydration phase properties of phospholipid mixtures.
Bunjes, H., & Unruh, T. (2007). Characterization of lipid nanoparticles by differential scanning
calorimetry, X-ray and neutron scattering. Advanced Drug Delivery Reviews, 59(6), 379-402.
Bunker, A. (2015). Molecular modeling as a tool to understand the role of poly (ethylene) glycol in drug
Burley, S. K., & Petsko, G. A. (1986). Amino‐aromatic interactions in proteins. FEBS Letters, 203(2),
139-143.
Bussi, G., Donadio, D., & Parrinello, M. (2007). Canonical sampling through velocity rescaling. The
Caffrey, M. (2003). Membrane protein crystallization. Journal of Structural Biology, 142(1), 108-132.
Cai, C., Wang, L., Lin, J., & Zhang, X. (2012). Morphology transformation of hybrid micelles self-
assembled from rod–coil block copolymer and nanoparticles. Langmuir, 28(9), 4515-4524.
Caminade, A. M., Laurent, R., & Majoral, J. P. (2005). Characterization of dendrimers. Advanced Drug
306
Carrasco-Sánchez, V., Vergara-Jaque, A., Zuñiga, M., Comer, J., John, A., Nachtigall, F.M., Valdes, O.,
Duran-Lara, E.F., Sandoval, C. & Santos, L. S. (2014). In situ and in silico evaluation of amine-and
73, 250-257.
Case, D.A., Cheatham, T.E., Darden, T., Gohlke, H., Luo, R., Merz, K.M., Onufriev, A., Simmerling, C.,
Wang, B. & Woods, R. J. (2005). The Amber biomolecular simulation programs. Journal of
Chaban, V. V., Savchenko, T. I., Kovalenko, S. M., & Prezhdo, O. V. (2010). Heat-driven release of a
drug molecule from carbon nanotubes: A molecular dynamics study. The Journal of Physical Chemistry
B, 114(42), 13481-13486.
Champion, J. A., & Mitragotri, S. (2006). Role of target geometry in phagocytosis. Proceedings of the
Chan, C. L., Majzoub, R. N., Shirazi, R. S., Ewert, K. K., Chen, Y. J., Liang, K. S., & Safinya, C. R.
(2012). Endosomal escape and transfection efficiency of PEGylated cationic liposome–DNA complexes
Chang, H. Y., Lin, Y. L., Sheng, Y. J., & Tsao, H. K. (2013). Structural characteristics and fusion
Chehel Amirani, M., & Tang, T. (2015). Binding of nucleobases with graphene and carbon nanotube: A
review of computational studies. Journal of Biomolecular Structure and Dynamics, 33(7), 1567-1597.
307
Chen, A. M., Zhang, M., Wei, D., Stueber, D., Taratula, O., Minko, T., & He, H. (2009). Co‐delivery of
doxorubicin and Bcl‐2 siRNA by mesoporous silica nanoparticles enhances the efficacy of chemotherapy
Chen, B. D., Yang, C. L., Yang, J. S., Wang, M. S., & Ma, X. G. (2013). Dynamic mechanism of HIV
replication inhibitor peptide encapsulated into carbon nanotubes. Current Applied Physics, 13(6), 1001-
1007.
Chen, H., & Ruckenstein, E. (2013). Formation and degradation of multicomponent multicore micelles:
Chen, L., Jiang, T., Cai, C., Wang, L., Lin, J., & Cao, X. (2014). Polypeptide-based “smart” micelles for
Chen, S., Tam, Y. Y. C., Lin, P. J., Sung, M. M., Tam, Y. K., & Cullis, P. R. (2016). Influence of particle
size on the in vivo potency of lipid nanoparticle formulations of siRNA. Journal of Controlled Release,
235, 236-244.
Chen, Y., Ma, P., & Gui, S. (2014). Cubic and hexagonal liquid crystals as drug delivery systems.
Chen, Z., & Rand, R. P. (1998). Comparative study of the effects of several n-alkanes on phospholipid
308
Chiu, C. C., Shinoda, W., DeVane, R. H., & Nielsen, S. O. (2012). Effects of spherical fullerene
Chiu, Y.L., Ho, Y.C., Chen, Y.M., Peng, S.F., Ke, C.J., Chen, K.J., Mi, F.L. & Sung, H. W. (2010). The
Contet, A., Pihan, E., Lavigne, M., Wengelnik, K., Maheshwari, S., Vial, H., Douguet, D. & Cerdan, R.
catalytic domains and is inhibited by a CDP-choline analog selected from a virtual screening. FEBS
curvature, acyl chain length, and lipid-phase dependence for activation. Biochemistry, 30(24), 5881-5888.
Cornell, R. B. (2016). Membrane lipid compositional sensing by the inducible amphipathic helix of CCT.
Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids, 1861(8), 847-861.
and structure of an amphitropic enzyme required for membrane biogenesis. Progress in Lipid Research,
59, 147-171.
Cox, P. (2012). Atomistic modelling of drug delivery systems. In Drug Design Strategies:
309
Cullis, P. R., & Kruijff, B. D. (1979). Lipid polymorphism and the functional roles of lipids in biological
Cullis, P. R., Hope, M. J., & Tilcock, C. P. (1986). Lipid polymorphism and the roles of lipids in
Cullis, P. R., Verkleij, A. J., & Ververgaert, P. H. (1978). Polymorphic phase behavior of cardiolipin as
detected by 31P NMR and freeze-fracture techniques. Effects of calcium, dibucaine and chlorpromazine.
Cutler, J. I., Auyeung, E., & Mirkin, C. A. (2012). Spherical nucleic acids. Journal of the American
da Rocha, E. L., Caramori, G. F., & Rambo, C. R. (2013). Nanoparticle translocation through a lipid
bilayer tuned by surface chemistry. Physical Chemistry Chemical Physics, 15(7), 2282-2290.
da Silva, A. W. S., & Vranken, W. F. (2012). ACPYPE-Antechamber python parser interface. BMC
Dai, X., Yue, Z., Eccleston, M. E., Swartling, J., Slater, N. K., & Kaminski, C. F. (2008). Fluorescence
intensity and lifetime imaging of free and micellar-encapsulated doxorubicin in living cells.
Dan, N. (2014). Nanostructured lipid carriers: Effect of solid phase fraction and distribution on the release
Dan, N. (2015). Drug release through liposome pores. Colloids and Surfaces B: Biointerfaces, 126, 80-86.
310
Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N⋅log (N) method for Ewald sums
Dass, C. R. (2008). Drug delivery in cancer using liposomes. In Drug Delivery Systems (pp. 177-182).
Humana Press.
Davis, J. H. (1979). Deuterium magnetic resonance study of the gel and liquid crystalline phases of
Deborah, M., Jawahar, A., Mathavan, T., Dhas, M. K., & Benial, A. M. F. (2015). Spectroscopic studies
Demir, O., & Amaro, R. E. (2012). Elements of nucleotide specificity in the Trypanosoma brucei
mitochondrial RNA editing enzyme RET2. Journal of Chemical Information and Modeling, 52(5), 1308-
1318.
Dennis, M. K., Taneva, S. G., & Cornell, R. B. (2011). The intrinsically disordered nuclear localization
signal and phosphorylation segments distinguish the membrane affinity of two cytidylyltransferase
Di Crescenzo, A., Aschi, M., & Fontana, A. (2012). Toward a better understanding of steric stabilization
when using block copolymers as stabilizers of single-walled carbon nanotubes (SWCNTs) aqueous
311
Di Marco, M., Sadun, C., Port, M., Guilbert, I., Couvreur, P., & Dubernet, C. (2007). Physicochemical
characterization of ultrasmall superparamagnetic iron oxide particles (USPIO) for biomedical application
Dickson, C. J., Madej, B. D., Skjevik, Å. A., Betz, R. M., Teigen, K., Gould, I. R., & Walker, R. C.
(2014). Lipid14: The Amber lipid force field. Journal of Chemical Theory and Computation, 10(2), 865-
879.
Di̇ nç, C. Ö., Ki̇ barer, G., & Güner, A. (2010). Solubility profiles of poly (ethylene glycol)/solvent
systems. II. Comparison of thermodynamic parameters from viscosity measurements. Journal of Applied
Ding, H. M., & Ma, Y. Q. (2012). Role of physicochemical properties of coating ligands in receptor-
Ding, H. M., & Ma, Y. Q. (2013). Controlling cellular uptake of nanoparticles with pH-sensitive
Ding, H. M., & Ma, Y. Q. (2014). Computer simulation of the role of protein corona in cellular delivery
Ding, H. M., & Ma, Y. Q. (2015). Theoretical and computational investigations of nanoparticle–
Ding, H. M., Tian, W. D., & Ma, Y. Q. (2012). Designing nanoparticle translocation through membranes
312
Ding, Z., Taneva, S. G., Huang, H. K., Campbell, S. A., Semenec, L., Chen, N., & Cornell, R. B. (2012).
cytidylyltransferase facilitates both silencing and activating functions. Journal of Biological Chemistry,
287(46), 38980-38991.
Doux, J. P., Hall, B. A., & Killian, J. A. (2012). How lipid headgroups sense the membrane environment:
Dowdy, S. F. (2017). Overcoming cellular barriers for RNA therapeutics. Nature Biotechnology, 35(3),
222-229
Duncan, R. (2003). The dawning era of polymer therapeutics. Nature Reviews Drug Discovery, 2(5), 347–
360.
Dunne, S. J., Cornell, R. B., Johnson, J. E., Glover, N. R., & Tracey, A. S. (1996). Structure of the
11984.
Durell, S. R., Brooks, B. R., & Ben-Naim, A. (1994). Solvent-induced forces between two hydrophilic
Durrant, J. D., & Amaro, R. E. (2014). LipidWrapper: An algorithm for generating large-scale membrane
313
Durzyńska, J., Przysiecka, Ł., Nawrot, R., Barylski, J., Nowicki, G., Warowicka, A., Musidlak, O. &
Goździcka-Józefiak, A. (2015). Viral and other cell-penetrating peptides as vectors of therapeutic agents
Dzieciuch, M., Rissanen, S., Szydłowska, N., Bunker, A., Kumorek, M., Jamróz, D., Vattulainen, I.,
Nowakowska, M., Rog, T. & Kepczynski, M. (2015). PEGylated liposomes as carriers of hydrophobic
Ellens, H., Siegel, D.P., Alford, D., Yeagle, P.L., Boni, L., Lis, L.J., Quinn, P.J. & Bentz, J. (1989).
Elzoghby, A. O., Samy, W. M., & Elgindy, N. A. (2012). Albumin-based nanoparticles as potential
controlled release drug delivery systems. Journal of Controlled Release, 157(2), 168-182.
Elzoghby, A. O., Samy, W. M., & Elgindy, N. A. (2012). Protein-based nanocarriers as promising drug
Emekli, U., Schneidman‐Duhovny, D., Wolfson, H. J., Nussinov, R., & Haliloglu, T. (2008). HingeProt:
Automated prediction of hinges in protein structures. Proteins: Structure, Function, and Bioinformatics,
70(4), 1219-1227.
Emsley, P., Lohkamp, B., Scott, W. G., & Cowtan, K. (2010). Features and development of Coot. Acta
Engelman, D. M. (2005). Membranes are more mosaic than fluid. Nature, 438(7068), 578-580.
314
Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., & Pedersen, L. G. (1995). A smooth
particle mesh Ewald method. The Journal of Chemical Physics, 103(19), 8577-8593.
Evans, D. F., & Wennerström, H. (1999). The colloidal domain: Where physics, chemistry, biology, and
Fatouros, D.G., Douroumis, D., Nikolakis, V., Ntais, S., Moschovi, A.M., Trivedi, V., Khima, B., Roldo,
M., Nazar, H. & Cox, P. A. (2011). In vitro and in silico investigations of drug delivery via zeolite BEA.
Fattal, E., Hillaireau, H., Mura, S., Nicolas, J., & Tsapis, N. (2012). Targeted delivery using
Feller, S. E., & Pastor, R. W. (1996). On simulating lipid bilayers with an applied surface tension:
Ficici, E., Andricioaei, I., & Howorka, S. (2015). Dendrimers in nanoscale confinement: The interplay
between conformational change and nanopore entrance. Nano Letters, 15(7), 4822-4828.
Filippakopoulos, P., Müller, S., & Knapp, S. (2009). SH2 domains: Modulators of nonreceptor tyrosine
Foloppe, N., & MacKerell, Jr, A. D. (2000). All‐atom empirical force field for nucleic acids: I. Parameter
optimization based on small molecule and condensed phase macromolecular target data. Journal of
315
Frenkel, D., & Smit, B. (2001). Understanding Molecular Simulation: From Algorithms to Applications
Fröhlich, E. (2012). The role of surface charge in cellular uptake and cytotoxicity of medical
Fu, I. W., Markegard, C. B., & Nguyen, H. D. (2014). Solvent effects on kinetic mechanisms of self-
assembly by peptide amphiphiles via molecular dynamics simulations. Langmuir, 31(1), 315-324.
Fu, I. W., Markegard, C. B., Chu, B. K., & Nguyen, H. D. (2013). The role of electrostatics and
amphiphiles via molecular dynamics simulations. Advanced Healthcare Materials, 2(10), 1388-1400.
Fusco, S., Huang, H.W., Peyer, K.E., Peters, C., Haberli, M., Ulbers, A., Spyrogianni, A., Pellicer, E.,
Sort, J., Pratsinis, S.E. & Nelson, B. J. (2015). Shape-switching microrobots for medical applications: The
influence of shape in drug delivery and locomotion. ACS Applied Materials & Interfaces, 7(12), 6803-
6811.
García-Fandiño, R., & Granja, J. R. (2013). Effect of organochloride guest molecules on the stability of
homo/hetero self-assembled α, γ-cyclic peptide structures: A computational study toward the control of
Gawrisch, K., Parsegian, V. A., Hajduk, D. A., Tate, M. W., Gruner, S. M., Fuller, N. L., & Rand, R. P.
316
George, T. P., Cook, H. W., Byers, D. M., Palmer, F. B., & Spence, M. W. (1991). Channeling of
George, T. P., Morash, S. C., Cook, H. W., Byers, D. M., Palmer, F. S. C., & Spence, M. W. (1989).
Ghosh, P., Han, G., De, M., Kim, C. K., & Rotello, V. M. (2008). Gold nanoparticles in delivery
Gilleron, J., Querbes, W., Zeigerer, A., Borodovsky, A., Marsico, G., Schubert, U., Manygoats, K.,
Seifert, S., Andree, C., Stöter, M. & Epstein-Barash, H. (2013). Image-based analysis of lipid
Gilleron, J., Querbes, W., Zeigerer, A., Borodovsky, A., Marsico, G., Schubert, U., Manygoats, K.,
Seifert, S., Andree, C., Stöter, M. & Epstein-Barash, H. (2013). Image-based analysis of lipid
Gillies, E. R., & Frechet, J. M. (2005). Dendrimers and dendritic polymers in drug delivery. Drug
Gin, D. L., Gu, W., Pindzola, B. A., & Zhou, W. J. (2001). Polymerized lyotropic liquid crystal
317
Giron, D. (2002). Applications of thermal analysis and coupled techniques in pharmaceutical industry.
González-Bulnes, P., Bobenchik, A. M., Augagneur, Y., Cerdan, R., Vial, H., Llebaria, A., & Mamoun,
C. B. (2011). PG12, a phospholipid analog with potent antimalarial activity inhibits Plasmodium
28940-28947
Gratton, S. E., Ropp, P. A., Pohlhaus, P. D., Luft, J. C., Madden, V. J., Napier, M. E., & DeSimone, J. M.
(2008). The effect of particle design on cellular internalization pathways. Proceedings of the National
Greaves, T. L., & Drummond, C. J. (2008). Ionic liquids as amphiphile self-assembly media. Chemical
Green, N. S., Reisler, E., & Houk, K. N. (2001). Quantitative evaluation of the lengths of
homobifunctional protein cross‐linking reagents used as molecular rulers. Protein Science, 10(7), 1293-
1304.
Gref, R., Domb, A., Quellec, P., Blunk, T., Müller, R. H., Verbavatz, J. M., & Langer, R. (1995). The
controlled intravenous delivery of drugs using PEG-coated sterically stabilized nanospheres. Advanced
Gruner, S. M., Cullis, P. R., Hope, M. J., & Tilcock, C. P. S. (1985). Lipid polymorphism: The molecular
basis of nonbilayer phases. Annual Review of Biophysics and Biophysical Chemistry, 14(1), 211-238.
318
Gruner, S. M., Rothschild, K. J., & Clark, N. A. (1982). X-ray diffraction and electron microscope study
of phase separation in rod outer segment photoreceptor membrane multilayers. Biophysical Journal,
39(3), 241-251.
Gruner, S.M., Tate, M.W., Kirk, G.L., So, P.T.C., Turner, D.C., Keane, D.T., Tilcock, C.P.S. & Cullis, P.
Guo, H., Qiu, X., & Zhou, J. (2013). Self-assembled core-shell and Janus microphase separated structures
of polymer blends in aqueous solution. The Journal of Chemical Physics, 139(8), 084907.
Guo, R., Mao, J., & Yan, L. T. (2013). Unique dynamical approach of fully wrapping dendrimer-like soft
Guo, X. D., Zhang, L. J., Wu, Z. M., & Qian, Y. (2010). Dissipative particle dynamics studies on
microstructure of pH-sensitive micelles for sustained drug delivery. Macromolecules, 43(18), 7839-7844.
Gupta, J., Nunes, C., Vyas, S., & Jonnalagadda, S. (2011). Prediction of solubility parameters and
Gurtovenko, A. A., Miettinen, M., Karttunen, M., & Vattulainen, I. (2005). Effect of monovalent salt on
cationic lipid membranes as revealed by molecular dynamics simulations. The Journal of Physical
Gurtovenko, A. A., Patra, M., Karttunen, M., & Vattulainen, I. (2004). Cationic DMPC/DMTAP lipid
319
Hafez, I. M., & Cullis, P. R. (2001). Roles of lipid polymorphism in intracellular delivery. Advanced
Hafez, I. M., Maurer, N., & Cullis, P. R. (2001). On the mechanism whereby cationic lipids promote
Hakem, I. F., Lal, J., & Bockstaller, M. R. (2004). Binding of monovalent ions to PEO in solution:
Hanwell, M. D., Curtis, D. E., Lonie, D. C., Vandermeersch, T., Zurek, E., & Hutchison, G. R. (2012).
Avogadro: An advanced semantic chemical editor, visualization, and analysis platform. Journal of
Hao, J., Cheng, Y., Ranatunga, R.U., Senevirathne, S., Biewer, M.C., Nielsen, S.O., Wang, Q. & Stefan,
M. C. (2013). A combined experimental and computational study of the substituent effect on micellar
4829-4838.
Harper, P. E., Mannock, D. A., Lewis, R. N., McElhaney, R. N., & Gruner, S. M. (2001). X-ray
He, J., Chipot, C., Shao, X., & Cai, W. (2013). Cyclodextrin-mediated recruitment and delivery of
320
He, X., Lin, M., Sha, B., Feng, S., Shi, X., Qu, Z., & Xu, F. (2015). Coarse-grained molecular dynamics
studies of the translocation mechanism of polyarginines across asymmetric membrane under tension.
He, X., Qu, Z., Xu, F., Lin, M., Wang, J., Shi, X., & Lu, T. (2014). Molecular analysis of interactions
between dendrimers and asymmetric membranes at different transport stages. Soft Matter, 10(1), 139-148.
Helmink, B. A., Braker, J. D., Kent, C., & Friesen, J. A. (2003). Identification of lysine 122 and arginine
42(17), 5043-5051.
Hess, B., Bekker, H., Berendsen, H. J., & Fraaije, J. G. (1997). LINCS: A linear constraint solver for
Hess, B., Kutzner, C., Van Der Spoel, D., & Lindahl, E. (2008). GROMACS 4: Algorithms for highly
efficient, load-balanced, and scalable molecular simulation. Journal of Chemical Theory and
Hirlekar, R., Jain, S., Patel, M., Garse, H., & Kadam, V. (2010). Hexosomes: A novel drug delivery
Hirota, S., de Ilarduya, C. T., Barron, L. G., & Szoka Jr, F. C. (1999). Simple mixing device to
Hobbs, S. K., Monsky, W. L., Yuan, F., Roberts, W. G., Griffith, L., Torchilin, V. P., & Jain, R. K.
(1998). Regulation of transport pathways in tumor vessels: Role of tumor type and microenvironment.
321
Hoover, W. G. (1985). Canonical dynamics: Equilibrium phase-space distributions. Physical Review A,
31(3), 1695-1697.
Hoover-Fong, J., Sobreira, N., Jurgens, J., Modaff, P., Blout, C., Moser, A., Kim, O.H., Cho, T.J., Cho,
S.Y., Kim, S.J. & Jin, D. K. (2014). Mutations in PCYT1A, encoding a key regulator of
Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., & Simmerling, C. (2006). Comparison of
multiple Amber force fields and development of improved protein backbone parameters. Proteins:
Hsueh, Y. W., Zuckermann, M., & Thewalt, J. (2005). Phase diagram determination for
Huang, H. K., Taneva, S. G., Lee, J., Silva, L. P., Schriemer, D. C., & Cornell, R. B. (2013). The
amphipathic helix flanking its active site. Journal of Molecular Biology, 425(9), 1546-1564.
Huang, L., & Roux, B. (2013). Automated force field parameterization for nonpolarizable and polarizable
atomic models based on ab initio target data. Journal of Chemical Theory and Computation, 9(8), 3543-
3556.
Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: Visual molecular dynamics. Journal of
322
Huse, M., & Kuriyan, J. (2002). The conformational plasticity of protein kinases. Cell, 109(3), 275-282.
Huynh, L., Neale, C., Pomès, R., & Allen, C. (2010). Systematic design of unimolecular star copolymer
Huynh, L., Neale, C., Pomès, R., & Allen, C. (2012). Computational approaches to the rational design of
nanoemulsions, polymeric micelles, and dendrimers for drug delivery. Nanomedicine: Nanotechnology,
Ingólfsson, H. I., Arnarez, C., Periole, X., & Marrink, S. J. (2016). Computational ‘microscopy’of cellular
Io, T., Fukami, T., Yamamoto, K., Suzuki, T., Xu, J., Tomono, K., & Ramamoorthy, A. (2009).
Islami, M., Mehrnejad, F., Doustdar, F., Alimohammadi, M., Khadem‐Maaref, M., Mir‐Derikvand, M., &
Taghdir, M. (2014). Study of orientation and penetration of LAH4 into lipid bilayer membranes: pH and
Israelachvili, J. N., Marčelja, S., & Horn, R. G. (1980). Physical principles of membrane organization.
Israelachvili, J. N., Mitchell, D. J., & Ninham, B. W. (1976). Theory of self-assembly of hydrocarbon
amphiphiles into micelles and bilayers. Journal of the Chemical Society, Faraday Transactions 2:
323
Izquierdo, M. (2005). Short interfering RNAs as a tool for cancer gene therapy. Cancer Gene Therapy,
12(3), 217-227.
Jabbari, E., Yang, X., Moeinzadeh, S., & He, X. (2013). Drug release kinetics, cell uptake, and tumor
Jain, T. K., Morales, M. A., Sahoo, S. K., Leslie-Pelecky, D. L., & Labhasetwar, V. (2005). Iron oxide
nanoparticles for sustained delivery of anticancer agents. Molecular Pharmaceutics, 2(3), 194-205.
Jain, V., Maingi, V., Maiti, P. K., & Bharatam, P. V. (2013). Molecular dynamics simulations of PPI
Jämbeck, J. P., & Lyubartsev, A. P. (2012). An extension and further validation of an all-atomistic force
field for biological membranes. Journal of Chemical Theory and Computation, 8(8), 2938-2948.
Jämbeck, J. P., & Lyubartsev, A. P. (2012). Derivation and systematic validation of a refined all-atom
force field for phosphatidylcholine lipids. The Journal of Physical Chemistry B, 116(10), 3164-3179.
Jämbeck, J. P., Eriksson, E. S., Laaksonen, A., Lyubartsev, A. P., & Eriksson, L. A. (2014). Molecular
dynamics studies of liposomes as carriers for photosensitizing drugs: Development, validation, and
simulations with a coarse-grained model. Journal of Chemical Theory and Computation, 10(1), 5-13.
Jang, S. H., Wientjes, M. G., Lu, D., & Au, J. L. S. (2003). Drug delivery and transport to solid tumors.
324
Janke, J. J., Bennett, W. D., & Tieleman, D. P. (2014). Oleic acid phase behavior from molecular
Jayaraman, M., Ansell, S.M., Mui, B.L., Tam, Y.K., Chen, J., Du, X., Butler, D., Eltepu, L., Matsuda, S.,
Narayanannair, J.K. & Rajeev, K. G. (2012). Maximizing the potency of siRNA lipid nanoparticles for
hepatic gene silencing in vivo. Angewandte Chemie International Edition, 51(34), 8529-8533.
Jeffs, L. B., Palmer, L. R., Ambegia, E. G., Giesbrecht, C., Ewanick, S., & MacLachlan, I. (2005). A
scalable, extrusion-free method for efficient liposomal encapsulation of plasmid DNA. Pharmaceutical
Jensen, L. B., Pavan, G. M., Kasimova, M. R., Rutherford, S., Danani, A., Nielsen, H. M., & Foged, C.
Jiang, Y., Zhang, D., Zhang, Y., Deng, Z., & Zhang, L. (2014). The adsorption-desorption transition of
double-stranded DNA interacting with an oppositely charged dendrimer induced by multivalent anions.
Jo, S., Cheng, X., Lee, J., Kim, S., Park, S.J., Patel, D.S., Beaven, A.H., Lee, K.I., Rui, H., Park, S. &
Lee, H. S. (2017). CHARMM‐GUI 10 years for biomolecular modeling and simulation. Journal of
Jo, S., Kim, T., Iyer, V. G., & Im, W. (2008). CHARMM‐GUI: A web‐based graphical user interface for
325
Johner, N., Harries, D., & Khelashvili, G. (2014). Curvature and lipid packing modulate the elastic
properties of lipid assemblies: Comparing HII and lamellar phases. The Journal of Physical Chemistry
Johnson, J. E., Aebersold, R., & Cornell, R. B. (1997). An amphipathic α-helix is the principle
Johnson, J. E., Rao, N. M., Hui, S. W., & Cornell, R. B. (1998). Conformation and lipid binding
Jones, S. P., Pavan, G. M., Danani, A., Pricl, S., & Smith, D. K. (2010). Quantifying the effect of surface
ligands on dendron–DNA interactions: Insights into multivalency through a combined experimental and
Jones, S.P., Gabrielson, N.P., Wong, C.H., Chow, H.F., Pack, D.W., Posocco, P., Fermeglia, M., Pricl, S.
& Smith, D. K. (2011). Hydrophobically modified dendrons: Developing structure− activity relationships
for DNA binding and gene transfection. Molecular Pharmaceutics, 8(2), 416-429.
Jorge, A. F., Dias, R. S., & Pais, A. A. (2012). Enhanced condensation and facilitated release of DNA
using mixed cationic agents: A combined experimental and Monte Carlo study. Biomacromolecules,
13(10), 3151-3161.
Jorge, A. F., Pereira, R. F., Nunes, S. C., Valente, A. J., Dias, R. S., & Pais, A. A. (2014). Interpreting the
326
Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983). Comparison of
simple potential functions for simulating liquid water. The Journal of Chemical Physics, 79(2), 926-935.
Jorgensen, W. L., Maxwell, D. S., & Tirado-Rives, J. (1996). Development and testing of the OPLS all-
atom force field on conformational energetics and properties of organic liquids. Journal of the American
Jouhet, J. (2013). Importance of the hexagonal lipid phase in biological membrane organization. Frontiers
Jusufi, A., DeVane, R. H., Shinoda, W., & Klein, M. L. (2011). Nanoscale carbon particles and the
Juul, S., Iacovelli, F., Falconi, M., Kragh, S.L., Christensen, B., Frøhlich, R., Franch, O., Kristoffersen,
E.L., Stougaard, M., Leong, K.W. & Ho, Y. P. (2013). Temperature-controlled encapsulation and release
of an active enzyme in the cavity of a self-assembled DNA nanocage. ACS nano, 7(11), 9724-9734.
Kaasgaard, T., & Drummond, C. J. (2006). Ordered 2-D and 3-D nanostructured amphiphile self-
assembly materials stable in excess solvent. Physical Chemistry Chemical Physics, 8(43), 4957-4975.
Kalu, Nnanya, Yoav Atsmon-Raz, Sanaz Momben Abolfath, Laura Lucas, Clare Kenney, Stephen H.
Leppla, D. Peter Tieleman, & Nestorovich, E. M. (2018). Effect of late endosomal DOBMP lipid and
traditional model lipids of electrophysiology on the anthrax toxin channel activity. Biochimica et
327
Kam, N. W. S., Liu, Z., & Dai, H. (2005). Functionalization of carbon nanotubes via cleavable disulfide
bonds for efficient intracellular delivery of siRNA and potent gene silencing. Journal of the American
Kang, M., Lam, D., Discher, D. E., & Loverde, S. M. (2015). Molecular modeling of block copolymer
Kann, B., Offerhaus, H. L., Windbergs, M., & Otto, C. (2015). Raman microscopy for cellular
investigations—From single cell imaging to drug carrier uptake visualization. Advanced Drug Delivery
Karatasos, K., Posocco, P., Laurini, E., & Pricl, S. (2012). Poly (amidoamine)‐based Dendrimer/siRNA
Karplus, M., & McCammon, J. A. (2002). Molecular dynamics simulations of biomolecules. Nature
Kataoka, K., Harada, A., & Nagasaki, Y. (2012). Block copolymer micelles for drug delivery: Design,
characterization and biological significance. Advanced Drug Delivery Reviews, 64, 37-48.
Kavyani, S., Amjad-Iranagh, S., & Modarress, H. (2014). Aqueous poly (amidoamine) dendrimer G3 and
G4 generations with several interior cores at pHs 5 and 7: A molecular dynamics simulation study. The
Kazanietz, M. G., & Lemmon, M. A. (2011). Protein kinase C regulation: C1 meets C-tail. Structure,
19(2), 144-146.
328
Kepczynski, M., Jamróz, D., Wytrwal, M., Bednar, J., Rzad, E., & Nowakowska, M. (2011). Interactions
of a hydrophobically modified polycation with zwitterionic lipid membranes. Langmuir, 28(1), 676-688.
Kim, C. S., Tonga, G. Y., Solfiell, D., & Rotello, V. M. (2013). Inorganic nanosystems for therapeutic
delivery: Status and prospects. Advanced Drug Delivery Reviews, 65(1), 93-99.
Kim, H., & Leal, C. (2015). Cuboplexes: Topologically active siRNA delivery. ACS nano, 9(10), 10214-
10226.
Kirtane, A. R., Siegel, R. A., & Panyam, J. (2015). A pharmacokinetic model for quantifying the effect of
vascular permeability on the choice of drug carrier: A framework for personalized nanomedicine. Journal
Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez, C.,
Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom additive
force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B, 114(23), 7830-
7843.
Klibanov, A. L., Maruyama, K., Torchilin, V. P., & Huang, L. (1990). Amphipathic polyethyleneglycols
effectively prolong the circulation time of liposomes. FEBS Letters, 268(1), 235-237.
Knecht, V., Mark, A. E., & Marrink, S. J. (2006). Phase behavior of a phospholipid/fatty acid/water
mixture studied in atomic detail. Journal of the American Chemical Society, 128(6), 2030-2034.
Knop, K., Hoogenboom, R., Fischer, D., & Schubert, U. S. (2010). Poly (ethylene glycol) in drug
delivery: Pros and cons as well as potential alternatives. Angewandte Chemie International Edition,
49(36), 6288-6308.
329
Kobayashi, T., Stang, E., Fang, K. S., de Moerloose, P., Parton, R. G., & Gruenberg, J. (1998). A lipid
associated with the antiphospholipid syndrome regulates endosome structure and function. Nature,
392(6672), 193-197.
Kohane, D. S. (2007). Microparticles and nanoparticles for drug delivery. Biotechnology and
Kolev, V., Ivanova, A., Madjarova, G., Aserin, A., & Garti, N. (2012). Molecular dynamics approach to
water structure of HII mesophase of monoolein. The Journal of Chemical Physics, 136(7), 074509.
Koltover, I., Salditt, T., Rädler, J. O., & Safinya, C. R. (1998). An inverted hexagonal phase of cationic
liposome-DNA complexes related to DNA release and delivery. Science, 281(5373), 78-81.
Kraszewski, S., Bianco, A., Tarek, M., & Ramseyer, C. (2012). Insertion of short amino-functionalized
single-walled carbon nanotubes into phospholipid bilayer occurs by passive diffusion. PLoS ONE, 7(7),
e40703.
Kraszewski, S., Picaud, F., Elhechmi, I., Gharbi, T., & Ramseyer, C. (2012). How long a functionalized
carbon nanotube can passively penetrate a lipid membrane?. Carbon, 50(14), 5301-5308.
Krishnan, Y., & Simmel, F. C. (2011). Nucleic acid based molecular devices. Angewandte Chemie
Kübart, S. A., & Keck, C. M. (2013). Laser diffractometry of nanoparticles: Frequent pitfalls &
overlooked opportunities. Journal of Pharmaceutical Technology and Drug Research, 2(1), 17.
330
Kulkarni, J. A., Tam, Y. Y. C., Chen, S., Tam, Y. K., Zaifman, J., Cullis, P. R., & Biswas, S. (2017).
Rapid synthesis of lipid nanoparticles containing hydrophobic inorganic nanoparticles. Nanoscale, 9(36),
13600-13609.
Kulkarni, J.A., Darjuan, M.M., Mercer, J.E., Chen, S., van der Meel, R., Thewalt, J.L., Tam, Y.Y.C. &
Cullis, P. R. (2018). On the formation and morphology of lipid nanoparticles containing ionizable
Kupiainen, M., Falck, E., Ollila, S., Niemelä, P., Gurtovenko, A.A., Hyvönen, M.T., Patra, M., Karttunen,
M. & Vattulainen, I. (2005). Free volume properties of sphingomyelin, DMPC, DPPC, and PLPC
Laasonen, K., & Klein, M. L. (1995). Molecular dynamics simulations of the structure and ion diffusion
in poly (ethylene oxide). Journal of the Chemical Society, Faraday Transactions, 91(16), 2633-2638.
Labiris, N. R., & Dolovich, M. B. (2003). Pulmonary drug delivery. Part I: Physiological factors affecting
588-599.
Lacerda, L., Bianco, A., Prato, M., & Kostarelos, K. (2006). Carbon nanotubes as nanomedicines: From
Lafleur, M., Bloom, M., Eikenberry, E. F., Gruner, S. M., Han, Y., & Cullis, P. R. (1996). Correlation
between lipid plane curvature and lipid chain order. Biophysical Journal, 70(6), 2747-2757.
Lafleur, M., Cullis, P. R., Fine, B., & Bloom, M. (1990). Comparison of the orientational order of lipid
331
Lafleur, M., Fine, B., Sternin, E., Cullis, P. R., & Bloom, M. (1989). Smoothed orientational order profile
Lai, L., & Barnard, A. S. (2015). Functionalized nanodiamonds for biological and medical applications.
Lai, L., & Barnard, A. S. (2015). Molecular and analytical modeling of nanodiamond for drug delivery
Lakowicz, J. R. (2006). Principles of fluorescence spectroscopy. Springer Science & Business Media.
Law, M., Jafari, M., & Chen, P. (2008). Physicochemical characterization of siRNA‐peptide complexes.
Lee, C. C., MacKay, J. A., Fréchet, J. M., & Szoka, F. C. (2005). Designing dendrimers for biological
Lee, H. (2013). Interparticle dispersion, membrane curvature, and penetration induced by single-walled
carbon nanotubes wrapped with lipids and PEGylated lipids. The Journal of Physical Chemistry B,
117(5), 1337-1344.
Lee, H. (2013). Membrane penetration and curvature induced by single-walled carbon nanotubes: The
effect of diameter, length, and concentration. Physical Chemistry Chemical Physics, 15(38), 16334-
16340.
332
Lee, H. (2013). Molecular dynamics studies of PEGylated single-walled carbon nanotubes: The effect of
PEG size and grafting density. The Journal of Physical Chemistry C, 117(49), 26334-26341.
Lee, H. (2014). Molecular modeling of PEGylated peptides, dendrimers, and single-walled carbon
Lee, H. (2015). Dispersion and bilayer interaction of single-walled carbon nanotubes modulated by
Lee, H., & Larson, R. G. (2011). Effects of PEGylation on the size and internal structure of dendrimers:
Self-penetration of long PEG chains into the dendrimer core. Macromolecules, 44(7), 2291-2298.
Lee, H., & Larson, R. G. (2011). Membrane pore formation induced by acetylated and polyethylene
5322.
Lee, H., & Pastor, R. W. (2011). Coarse-grained model for PEGylated lipids: Effect of PEGylation on the
size and shape of self-assembled structures. The Journal of Physical Chemistry B, 115(24), 7830-7837.
Lee, H., Choi, J. S., & Larson, R. G. (2011). Molecular dynamics studies of the size and internal structure
of the PAMAM dendrimer grafted with arginine and histidine. Macromolecules, 44(21), 8681-8686.
Lee, J., Cheng, X., Swails, J.M., Yeom, M.S., Eastman, P.K., Lemkul, J.A., Wei, S., Buckner, J., Jeong,
J.C., Qi, Y. & Jo, S. (2015). CHARMM-GUI input generator for NAMD, GROMACS, AMBER,
OpenMM, and CHARMM/OpenMM simulations using the CHARMM36 additive force field. Journal of
333
Lee, J., Johnson, J. E., Ding, Z., Paetzel, M., & Cornell, R. B. (2009). Crystal structure of a mammalian
CTP:phosphocholine cytidylyltransferase catalytic domain reveals novel active site residues within a
Lee, J., Taneva, S. G., Holland, B. W., Tieleman, D. P., & Cornell, R. B. (2014). Structural basis for
Lee, K. H., & Ytreberg, F. M. (2012). Effect of gold nanoparticle conjugation on peptide dynamics and
Lee, O. S., & Schatz, G. C. (2011). Computational simulations of the interaction of lipid membranes with
Lehtinen, J., Magarkar, A., Stepniewski, M., Hakola, S., Bergman, M., Róg, T., Yliperttula, M., Urtti, A.
& Bunker, A. (2012). Analysis of cause of failure of new targeting peptide in PEGylated liposome:
Molecular modeling as rational design tool for nanomedicine. European Journal of Pharmaceutical
Leung, A. K., Tam, Y. Y. C., Chen, S., Hafez, I. M., & Cullis, P. R. (2015). Microfluidic mixing: A
general method for encapsulating macromolecules in lipid nanoparticle systems. The Journal of Physical
Leung, A.K., Hafez, I.M., Baoukina, S., Belliveau, N.M., Zhigaltsev, I.V., Afshinmanesh, E., Tieleman,
D.P., Hansen, C.L., Hope, M.J. & Cullis, P. R. (2012). Lipid nanoparticles containing siRNA synthesized
334
by microfluidic mixing exhibit an electron-dense nanostructured core. The Journal of Physical Chemistry
C, 116(34), 18440-18450.
Li, J., Jiang, L., & Zhu, X. (2015). Computational studies of the binding mechanisms of fullerenes to
Li, J., Ouyang, Y., Kong, X., Zhu, J., Lu, D., & Liu, Z. (2015). A multi-scale molecular dynamics
Li, Y. (2015). Computer investigations of influences of molar fraction and acyl chain length of lipids on
Li, Y., & Hou, T. (2010). Computational simulation of drug delivery at molecular level. Current
Li, Y., Feng, D., Zhang, X., & Cao, D. (2015). Design strategy of cell-penetrating copolymers for high
Li, Y., Kröger, M., & Liu, W. K. (2014). Endocytosis of PEGylated nanoparticles accompanied by
structural and free energy changes of the grafted polyethylene glycol. Biomaterials, 35(30), 8467-8478.
Li, Y., Yue, T., Yang, K., & Zhang, X. (2012). Molecular modeling of the relationship between
Li, Y., Zhang, X., & Cao, D. (2014). A spontaneous penetration mechanism of patterned nanoparticles
335
Li, Y., Zhang, X., & Cao, D. (2015). Nanoparticle hardness controls the internalization pathway for drug
Li, Y.C., Rissanen, S., Stepniewski, M., Cramariuc, O., Róg, T., Mirza, S., Xhaard, H., Wytrwal, M.,
Kepczynski, M. & Bunker, A. (2012). Study of interaction between PEG carrier and three relevant drug
molecules: Piroxicam, paclitaxel, and hematoporphyrin. The Journal of Physical Chemistry B, 116(24),
7334-7341.
Li, Z. L., Ding, H. M., & Ma, Y. Q. (2012). Translocation of polyarginines and conjugated nanoparticles
Liao, Z. X., Peng, S. F., Ho, Y. C., Mi, F. L., Maiti, B., & Sung, H. W. (2012). Mechanistic study of
transfection of chitosan/DNA complexes coated by anionic poly (γ-glutamic acid). Biomaterials, 33(11),
3306-3315.
Lim, J., Pavan, G. M., Annunziata, O., & Simanek, E. E. (2012). Experimental and computational
evidence for an inversion in guest capacity in high-generation triazine dendrimer hosts. Journal of the
Lin, J., Zhang, H., Chen, Z., & Zheng, Y. (2010). Penetration of lipid membranes by gold nanoparticles:
Insights into cellular uptake, cytotoxicity, and their relationship. ACS nano, 4(9), 5421-5429.
Lin, X., Li, Y., & Gu, N. (2010). Nanoparticle's size effect on its translocation across a lipid bilayer: A
molecular dynamics simulation. Journal of Computational and Theoretical Nanoscience, 7(1), 269-276.
Lindahl, E., & Edholm, O. (2000). Mesoscopic undulations and thickness fluctuations in lipid bilayers
336
Lindorff‐Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., & Shaw, D. E.
(2010). Improved side‐chain torsion potentials for the Amber ff99SB protein force field. Proteins:
Liu, H., Chen, J., Shen, Q., Fu, W., & Wu, W. (2010). Molecular insights on the cyclic peptide nanotube-
Liu, L., Parameswaran, S., Sharma, A., Grayson, S. M., Ashbaugh, H. S., & Rick, S. W. (2014).
Molecular dynamics simulations of linear and cyclic amphiphilic polymers in aqueous and organic
Loverde, S. M. (2014). Computer simulation of polymer and biopolymer self-assembly for drug delivery.
Loverde, S. M., Klein, M. L., & Discher, D. E. (2012). Nanoparticle shape improves delivery: Rational
coarse grain molecular dynamics (rCG‐MD) of taxol in worm‐like PEG‐PCL micelles. Advanced
Lu, P. Y., & Woodle, M. C. (2008). Delivering small interfering RNA for novel therapeutics. In Drug
Lu, X., Gross, A. W., & Lodish, H. F. (2006). Active conformation of the erythropoietin receptor:
337
Luo, Q., Wang, Y., Yang, H., Liu, C., Ding, Y., Xu, H., Wang, Q., Liu, Y. & Xie, Y. (2014). Modeling
the interaction of interferon α-1b to bovine serum albumin as a drug delivery system. The Journal of
Luo, Z., & Jiang, J. (2012). pH-sensitive drug loading/releasing in amphiphilic copolymer PAE–PEG:
Integrating molecular dynamics and dissipative particle dynamics simulations. Journal of Controlled
Luzzati, V., & Husson, F. (1962). The structure of the liquid-crystalline phases of lipid-water systems.
Lyubartsev, A. P., & Rabinovich, A. L. (2016). Force field development for lipid membrane simulations.
Ma, Y. Q. (2012). pH-responsive dendrimers interacting with lipid membranes. Soft Matter, 8(9), 2627-
2632.
MacKerell Jr, A. D. (2004). Empirical force fields for biological macromolecules: Overview and issues.
Mackerell Jr, A. D., Feig, M., & Brooks III, C. L. (2004). Extending the treatment of backbone energetics
25(11), 1400-1415.
MacKerell Jr, A.D., Bashford, D., Bellott, M.L.D.R., Dunbrack Jr, R.L., Evanseck, J.D., Field, M.J.,
Fischer, S., Gao, J., Guo, H., Ha, S. & Joseph-McCarthy, D. (1998). All-atom empirical potential for
338
molecular modeling and dynamics studies of proteins. The Journal of Physical Chemistry B, 102(18),
3586-3616.
Magarkar, A., Karakas, E., Stepniewski, M., Róg, T., & Bunker, A. (2012). Molecular dynamics
simulation of PEGylated bilayer interacting with salt ions: A model of the liposome surface in the
Magarkar, A., Róg, T., & Bunker, A. (2014). Molecular dynamics simulation of PEGylated membranes
with cholesterol: Building toward the DOXIL formulation. The Journal of Physical Chemistry C,
118(28), 15541-15549.
Magarkar, A., Róg, T., & Bunker, A. (2014). Molecular dynamics simulation of inverse-phosphocholine
Maingi, V., Kumar, M. V. S., & Maiti, P. K. (2012). PAMAM dendrimer–drug interactions: Effect of pH
on the binding and release pattern. The Journal of Physical Chemistry B, 116(14), 4370-4376.
Makarucha, A. J., Todorova, N., & Yarovsky, I. (2011). Nanomaterials in biological environment: A
Malik, R., Genzer, J., & Hall, C. K. (2015). Proteinlike copolymers as encapsulating agents for small-
Marrink, S. J., & Mark, A. E. (2004). Molecular view of hexagonal phase formation in phospholipid
Marrink, S. J., & Tieleman, D. P. (2001). Molecular dynamics simulation of a lipid diamond cubic phase.
339
Marrink, S. J., & Tieleman, D. P. (2002). Molecular dynamics simulation of spontaneous membrane
Marrink, S. J., & Tieleman, D. P. (2013). Perspective on the Martini model. Chemical Society Reviews,
42(16), 6801-6822.
Marrink, S. J., De Vries, A. H., & Mark, A. E. (2004). Coarse grained model for semiquantitative lipid
Marrink, S. J., De Vries, A. H., & Tieleman, D. P. (2009). Lipids on the move: Simulations of membrane
pores, domains, stalks and curves. Biochimica et Biophysica Acta (BBA)-Biomembranes, 1788(1), 149-
168.
Marrink, S. J., Risselada, H. J., Yefimov, S., Tieleman, D. P., & De Vries, A. H. (2007). The MARTINI
force field: Coarse grained model for biomolecular simulations. The Journal of Physical Chemistry B,
111(27), 7812-7824.
Marrink, S. J., Sok, R. M., & Berendsen, H. J. C. (1996). Free volume properties of a simulated lipid
Marsh, D. (2011). Pivotal surfaces in inverse hexagonal and cubic phases of phospholipids and
Martincic, M., & Tobias, G. (2015). Filled carbon nanotubes in biomedical imaging and drug delivery.
340
Martínez, L., Andrade, R., Birgin, E. G., & Martínez, J. M. (2009). PACKMOL: A package for building
initial configurations for molecular dynamics simulations. Journal of Computational Chemistry, 30(13),
2157-2164.
Martinez-Veracoechea, F. J., & Frenkel, D. (2011). Designing super selectivity in multivalent nano-
Martinho, N., Florindo, H., Silva, L., Brocchini, S., Zloh, M., & Barata, T. (2014). Molecular modeling to
Martini, G., & Ciani, L. (2009). Electron spin resonance spectroscopy in drug delivery. Physical
Martin-Molina, A., Luque-Caballero, G., Faraudo, J., Quesada-Perez, M., & Maldonado-Valderrama, J.
(2014). Adsorption of DNA onto anionic lipid surfaces. Advances in Colloid and Interface Science, 206,
172-185.
Martins, A. F., Pereira, A. G., Fajardo, A. R., Rubira, A. F., & Muniz, E. C. (2011). Characterization of
Mashaghi, S., Jadidi, T., Koenderink, G., & Mashaghi, A. (2013). Lipid nanotechnology. International
341
Mathias, E. V., Liu, X., Franco, O., Khan, I., Ba, Y., & Kornfield, J. A. (2008). Model of drug-loaded
19
fluorocarbon-based micelles studied by electron-spin induced F relaxation NMR and molecular
Matthews, E.E., Thévenin, D., Rogers, J.M., Gotow, L., Lira, P.D., Reiter, L.A., Brissette, W.H. &
McLaughlin, C. K., Hamblin, G. D., & Sleiman, H. F. (2011). Supramolecular DNA assembly. Chemical
McMaster, C. R. (2018). From yeast to humans–roles of the Kennedy pathway for phosphatidylcholine
Meek, S. T., Greathouse, J. A., & Allendorf, M. D. (2011). Metal‐organic frameworks: A rapidly growing
Mehnert, W., & Mäder, K. (2012). Solid lipid nanoparticles: Production, characterization and
Mehnert, W., & Mäder, K. (2001). Solid lipid nanoparticles: Production, characterization and
342
Meli, V., Caltagirone, C., Falchi, A.M., Hyde, S.T., Lippolis, V., Monduzzi, M., Obiols-Rabasa, M.,
Rosa, A., Schmidt, J., Talmon, Y. & Murgia, S. (2015). Docetaxel-loaded fluorescent liquid-crystalline
Mendes, L. P., Delgado, J. M. F., Costa, A. D. A., Vieira, M. S., Benfica, P. L., Lima, E. M., &
Valadares, M. C. (2015). Biodegradable nanoparticles designed for drug delivery: The number of
Meneksedag-Erol, D., Sun, C., Tang, T., & Uludag, H. (2014). Molecular dynamics simulations of
polyplexes and lipoplexes employed in gene delivery. In Intracellular Delivery II (pp. 277-311). Springer,
Dordrecht.
Meneksedag-Erol, D., Tang, T., & Uludağ, H. (2014). Molecular modeling of polynucleotide complexes.
Meneksedag-Erol, D., Tang, T., & Uludağ, H. (2015). Probing the effect of miRNA on siRNA–PEI
Messina, P. V., Miguel Besada-Porto, J., & M Ruso, J. (2014). Self-assembly drugs: From micelles to
Metwally, A. A., & Hathout, R. M. (2015). Computer-assisted drug formulation design: Novel approach
Milano, G., Santangelo, G., Ragone, F., Cavallo, L., & Di Matteo, A. (2011). Gold nanoparticle/polymer
interfaces: All atom structures from molecular dynamics simulations. The Journal of Physical Chemistry
C, 115(31), 15154-15163.
343
Mills, T.T., Tristram-Nagle, S., Heberle, F.A., Morales, N.F., Zhao, J., Wu, J., Toombes, G.E., Nagle, J.F.
& Feigenson, G. W. (2008). Liquid-liquid domains in bilayers detected by wide angle X-ray scattering.
Milne, J.L., Borgnia, M.J., Bartesaghi, A., Tran, E.E., Earl, L.A., Schauder, D.M., Lengyel, J., Pierson, J.,
Patwardhan, A. & Subramaniam, S. (2013). Cryo‐electron microscopy–a primer for the non‐microscopist.
Molla, M. R., Rangadurai, P., Pavan, G. M., & Thayumanavan, S. (2015). Experimental and theoretical
Monticelli, L., & Tieleman, D. P. (2013). Force fields for classical molecular dynamics. In Biomolecular
Moquin, A., & Winnik, F. M. (2012). The use of field-flow fractionation for the analysis of drug and gene
delivery systems. In Field-Flow Fractionation in Biopolymer Analysis (pp. 187-206). Springer, Vienna.
Mu, Q., Hu, T., & Yu, J. (2013). Molecular insight into the steric shielding effect of PEG on the
conjugated staphylokinase: Biochemical characterization and molecular dynamics simulation. PLoS ONE,
8(7), e68559.
Mui, B.L., Tam, Y.K., Jayaraman, M., Ansell, S.M., Du, X., Tam, Y.Y.C., Lin, P.J., Chen, S.,
Narayanannair, J.K., Rajeev, K.G. & Manoharan, M. (2013). Influence of polyethylene glycol lipid
344
Müller, R. H., Runge, S. A., Ravelli, V., Thünemann, A. F., Mehnert, W., & Souto, E. B. (2008).
535-544.
Müller‐Plathe, F., & van Gunsteren, W. F. (1995). Computer simulation of a polymer electrolyte: Lithium
iodide in amorphous poly (ethylene oxide). The Journal of Chemical Physics, 103(11), 4745-4756.
Müllner, M., Dodds, S. J., Nguyen, T. H., Senyschyn, D., Porter, C. J., Boyd, B. J., & Caruso, F. (2015).
Size and rigidity of cylindrical polymer brushes dictate long circulating properties in vivo. ACS nano,
9(2), 1294-1304.
Murthy, V., Xu, Z. P., & Smith, S. C. (2015). Molecular modeling of layered double hydroxide
Murugan, K., Choonara, Y. E., Kumar, P., Bijukumar, D., du Toit, L. C., & Pillay, V. (2015). Parameters
Nagle, J. F., & Tristram-Nagle, S. (2000). Structure of lipid bilayers. Biochimica et Biophysica Acta
Namgung, R., Lee, Y.M., Kim, J., Jang, Y., Lee, B.H., Kim, I.S., Sokkar, P., Rhee, Y.M., Hoffman, A.S.
& Kim, W. J. (2014). Poly-cyclodextrin and poly-paclitaxel nano-assembly for anticancer therapy. Nature
Communications, 5, 3702.
345
Nandy, B., & Maiti, P. K. (2010). DNA compaction by a dendrimer. The Journal of Physical Chemistry
B, 115(2), 217-230.
Nandy, B., Maiti, P. K., & Bunker, A. (2012). Force biased molecular dynamics simulation study of
effect of dendrimer generation on interaction with DNA. Journal of Chemical Theory and Computation,
9(1), 722-729.
Nandy, B., Santosh, M., & Maiti, P. K. (2012). Interaction of nucleic acids with carbon nanotubes and
Nashine, V. C., Hammes-Schiffer, S., & Benkovic, S. J. (2010). Coupled motions in enzyme catalysis.
Navas, B.P., Lohner, K., Deutsch, G., Sevcsik, E., Riske, K.A., Dimova, R., Garidel, P. & Pabst, G.
(2005). Composition dependence of vesicle morphology and mixing properties in a bacterial model
Neria, E., Fischer, S., & Karplus, M. (1996). Simulation of activation free energies in molecular systems.
Nguan, H., Ahmadi, S., & Hashim, R. (2014). Molecular dynamics simulations of the lyotropic reverse
hexagonal (HII) of Guerbet branched-chain β-d-glucoside. Physical Chemistry Chemical Physics, 16(1),
324-334.
Nicolini, C., Kraineva, J., Khurana, M., Periasamy, N., Funari, S. S., & Winter, R. (2006). Temperature
and pressure effects on structural and conformational properties of POPC/SM/cholesterol model raft
346
mixtures—a FT-IR, SAXS, DSC, PPC and Laurdan fluorescence spectroscopy study. Biochimica et
Nie, S. Y., Lin, W. J., Yao, N., Guo, X. D., & Zhang, L. J. (2014). Drug release from pH-sensitive
polymeric micelles with different drug distributions: Insight from coarse-grained simulations. ACS
Nie, S. Y., Sun, Y., Lin, W. J., Wu, W. S., Guo, X. D., Qian, Y., & Zhang, L. J. (2013). Dissipative
particle dynamics studies of doxorubicin-loaded micelles assembled from four-arm star triblock polymers
Nisoh, N., Karttunen, M., Monticelli, L., & Wong-Ekkabut, J. (2015). Lipid monolayer disruption caused
Niven, R. W., Speer, M., & Schreier, H. (1991). Nebulization of liposomes. II. The effects of size and
Nosé, S. (1984). A molecular dynamics method for simulations in the canonical ensemble. Molecular
Nosé, S., & Klein, M. L. (1983). Constant pressure molecular dynamics for molecular systems. Molecular
Novina, C. D., & Sharp, P. A. (2004). The RNAi revolution. Nature, 430(6996), 161–164.
Nowakowska, M., Nawalany, K., Kępczyński, M., & Krawczyk, Z. (2009). Novel nanostructural hybride
materials for photodynamic theraphy. In Macromolecular Symposia (Vol. 279, No. 1, pp. 132-137).
347
Ouyang, D., Zhang, H., Herten, D. P., Parekh, H. S., & Smith, S. C. (2010). Structure, dynamics and
energetics of siRNA - Cationic vector complexation: A molecular dynamics study. The Journal of
Ouyang, D., Zhang, H., Parekh, H. S., & Smith, S. C. (2010). Structure and dynamics of multiple cationic
Ouyang, D., Zhang, H., Parekh, H. S., & Smith, S. C. (2011). The effect of pH on PAMAM dendrimer–
Pan, J., Heberle, F. A., Petruzielo, R. S., & Katsaras, J. (2013). Using small-angle neutron scattering to
detect nanoscopic lipid domains. Chemistry and Physics of Lipids, 170, 19-32.
Panczyk, T., Jagusiak, A., Pastorin, G., Ang, W. H., & Narkiewicz-Michalek, J. (2013). Molecular
dynamics study of cisplatin release from carbon nanotubes capped by magnetic nanoparticles. The
Panczyk, T., Warzocha, T. P., & Camp, P. J. (2010). A magnetically controlled molecular nanocontainer
as a drug delivery system: The effects of carbon nanotube and magnetic nanoparticle parameters from
Pandit, S. A., Bostick, D., & Berkowitz, M. L. (2003). Mixed bilayer containing
348
Pandit, S. A., Bostick, D., & Berkowitz, M. L. (2003). Molecular dynamics simulation of a
Pannuzzo, M., De Jong, D. H., Raudino, A., & Marrink, S. J. (2014). Simulation of polyethylene glycol
and calcium-mediated membrane fusion. The Journal of Chemical Physics, 140(12), 124905.
Parrinello, M., & Rahman, A. (1981). Polymorphic transitions in single crystals: A new molecular
Parveen, S., Misra, R., & Sahoo, S. K. (2012). Nanoparticles: A boon to drug delivery, therapeutics,
diagnostics and imaging. Nanomedicine: Nanotechnology, Biology and Medicine, 8(2), 147–166.
Pastor, R. W., Venable, R. M., & Feller, S. E. (2002). Lipid bilayers, NMR relaxation, and computer
Patel, S. K., Lavasanifar, A., & Choi, P. (2010). Molecular dynamics study of the encapsulation capability
of a PCL–PEO based block copolymer for hydrophobic drugs with different spatial distributions of
Pattni, B. S., Chupin, V. V., & Torchilin, V. P. (2015). New developments in liposomal drug delivery.
Pattridge, K. A., Weber, C. H., Friesen, J. A., Sanker, S., Kent, C., & Ludwig, M. L. (2003). Glycerol-3-
phosphate cytidylyltransferase: Structural changes induced by binding of CDP-glycerol and the role of
Pavan, G. M. (2014). Modeling the interaction between dendrimers and nucleic acids: A molecular
349
Pavan, G. M., & Danani, A. (2010). The influence of dendron's architecture on the “rigid” and “flexible”
behaviour in binding DNA—a modelling study. Physical Chemistry Chemical Physics, 12(42), 13914-
13917.
Pavan, G. M., Albertazzi, L., & Danani, A. (2010). Ability to adapt: Different generations of PAMAM
dendrimers show different behaviors in binding siRNA. The Journal of Physical Chemistry B, 114(8),
2667-2675.
Pavan, G. M., Mintzer, M. A., Simanek, E. E., Merkel, O. M., Kissel, T., & Danani, A. (2010).
Computational insights into the interactions between DNA and siRNA with “rigid” and “flexible” triazine
Pavan, G. M., Monteagudo, S., Guerra, J., Carrion, B., Ocana, V., Rodriguez-Lopez, J., Danani, A., C
Perez-Martinez, F. & Cena, V. (2012). Role of generation, architecture, pH and ionic strength on
successful siRNA delivery and transfection by hybrid PPV-PAMAM dendrimers. Current Medicinal
Pavan, G.M., Posocco, P., Tagliabue, A., Maly, M., Malek, A., Danani, A., Ragg, E., Catapano, C.V. &
Pricl, S. (2010). PAMAM dendrimers for siRNA delivery: Computational and experimental insights.
Payne, F., Lim, K., Girousse, A., Brown, R.J., Kory, N., Robbins, A., Xue, Y., Sleigh, A., Cochran, E.,
Adams, C. & Borman, A. D. (2014). Mutations disrupting the Kennedy phosphatidylcholine pathway in
humans with congenital lipodystrophy and fatty liver disease. Proceedings of the National Academy of
Sciences, 201408523.
350
Payne, F., Lim, K., Girousse, A., Brown, R.J., Kory, N., Robbins, A., Xue, Y., Sleigh, A., Cochran, E.,
Adams, C. & Borman, A. D. (2014). Mutations disrupting the Kennedy phosphatidylcholine pathway in
humans with congenital lipodystrophy and fatty liver disease. Proceedings of the National Academy of
Sciences, 201408523.
Pearson, R. M., Patra, N., Hsu, H. J., Uddin, S., Král, P., & Hong, S. (2012). Positively charged dendron
micelles display negligible cellular interactions. ACS Macro Letters, 2(1), 77-81.
Peng, L.X., Ivetac, A., Chaudhari, A.S., Van, S., Zhao, G., Yu, L., Howell, S.B., McCammon, J.A. &
Peng, S. F., Tseng, M. T., Ho, Y. C., Wei, M. C., Liao, Z. X., & Sung, H. W. (2011). Mechanisms of
cellular uptake and intracellular trafficking with chitosan/DNA/poly (γ-glutamic acid) complexes as a
Percec, V., Wilson, D.A., Leowanawat, P., Wilson, C.J., Hughes, A.D., Kaucher, M.S., Hammer, D.A.,
Levine, D.H., Kim, A.J., Bates, F.S. & Davis, K. P. (2010). Self-assembly of Janus dendrimers into
Pereira, P., Jorge, A. F., Martins, R., Pais, A. A., Sousa, F., & Figueiras, A. (2012). Characterization of
polyplexes involving small RNA. Journal of Colloid and Interface science, 387(1), 84-94.
Perttu, E. K., Kohli, A. G., & Szoka Jr, F. C. (2012). Inverse-phosphocholine lipids: A remix of a
351
Perutková, Š., Daniel, M., Dolinar, G., Rappolt, M., Kralj‐Iglič, V., & Iglič, A. (2009). Stability of the
inverted hexagonal phase. Advances in Planar Lipid Bilayers and Liposomes, 9, 237-278.
Petelska, A. D., & Figaszewski, Z. A. (2000). Effect of pH on the interfacial tension of lipid bilayer
Petrache, H. I., Dodd, S. W., & Brown, M. F. (2000). Area per lipid and acyl length distributions in fluid
Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., & Ferrin, T.
E. (2004). UCSF Chimera—a visualization system for exploratory research and analysis. Journal of
Phillips, J.C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, E., Chipot, C., Skeel, R.D., Kale,
L. & Schulten, K. (2005). Scalable molecular dynamics with NAMD. Journal of Computational
Pi, P., Qin, D., Lan, J.L., Cai, Z., Yuan, X., Xu, S.P., Zhang, L., Qian, Y. & Wen, X. (2015). Dissipative
particle dynamics simulation on the nanocomposite delivery system of quantum dots and poly (styrene-b-
ethylene oxide) copolymer. Industrial & Engineering Chemistry Research, 54(23), 6123-6134.
Piggot, T. J., Allison, J. R., Sessions, R. B., & Essex, J. W. (2017). On the calculation of acyl chain order
parameters from lipid simulations. Journal of Chemical Theory and Computation, 13(11), 5683-5696.
Plimpton, S. (1995). Fast parallel algorithms for short-range molecular dynamics. Journal of
352
Pluhackova, K., Kirsch, S. A., Han, J., Sun, L., Jiang, Z., Unruh, T., & Böckmann, R. A. (2016). A
critical comparison of biomembrane force fields: Structure and dynamics of model DMPC, POPC, and
Poger, D., Caron, B., & Mark, A. E. (2016). Validating lipid force fields against experimental data:
1556-1565.
Ponder, J. W., & Case, D. A. (2003). Force fields for protein simulations. In Advances in Protein
Posocco, P., Laurini, E., Dal Col, V., Marson, D., Karatasos, K., Fermeglia, M., & Pricl, S. (2012). Tell
me something I do not know. Multiscale molecular modeling of dendrimer/dendron organization and self-
Posocco, P., Pricl, S., Jones, S., Barnard, A., & Smith, D. K. (2010). Less is more–multiscale modelling
of self-assembling multivalency and its impact on DNA binding and gene delivery. Chemical Science,
1(3), 393-404.
Pourmousa, M., & Karttunen, M. (2013). Early stages of interactions of cell-penetrating peptide
penetratin with a DPPC bilayer. Chemistry and Physics of Lipids, 169, 85-94.
Pourmousa, M., Wong-ekkabut, J., Patra, M., & Karttunen, M. (2012). Molecular dynamic studies of
transportan interacting with a DPPC lipid bilayer. The Journal of Physical Chemistry B, 117(1), 230-241.
353
Pronk, S., Páll, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts, M.R., Smith, J.C., Kasson,
P.M., Van Der Spoel, D. & Hess, B. (2013). GROMACS 4.5: A high-throughput and highly parallel open
Pygall, S. R., Whetstone, J., Timmins, P., & Melia, C. D. (2007). Pharmaceutical applications of confocal
laser scanning microscopy: The physical characterization of pharmaceutical systems. Advanced Drug
Radlinska, E. Z., Gulik-Krzywicki, T., Lafuma, F., Langevin, D., Urbach, W., Williams, C. E., & Ober,
R. (1995). Polymer confinement in surfactant bilayers of a lyotropic lamellar phase. Physical Review
Ramezanpour, M., Lee, J., Taneva, S. G., Tieleman, D. P., & Cornell, R. B. (2018). An auto-inhibitory
helix in CTP:phosphocholine cytidylyltransferase hi-jacks the catalytic residue and constrains a pliable,
Ramezanpour, M., Leung, S. S. W., Delgado-Magnero, K. H., Bashe, B. Y. M., Thewalt, J., & Tieleman,
Ramos, B., El Mouedden, M., Claro, E., & Jackowski, S. (2002). Inhibition of CTP:phosphocholine
1068-1075.
Rand, R. P., & Fuller, N. L. (1994). Structural dimensions and their changes in a reentrant hexagonal-
354
Rand, R. P., Fuller, N. L., Gruner, S. M., & Parsegian, V. A. (1990). Membrane curvature, lipid
segregation, and structural transitions for phospholipids under dual-solvent stress. Biochemistry, 29(1),
76-87.
Rappolt, M., Hickel, A., Bringezu, F., & Lohner, K. (2003). Mechanism of the lamellar/inverse hexagonal
phase transition examined by high resolution X-ray diffraction. Biophysical Journal, 84(5), 3111-3122.
Razmimanesh, F., Amjad-Iranagh, S., & Modarress, H. (2015). Molecular dynamics simulation study of
chitosan and gemcitabine as a drug delivery system. Journal of Molecular Modeling, 21(7), 165.
Reinl, H., Brumm, T., & Bayerl, T. M. (1992). Changes of the physical properties of the liquid-ordered
phase with temperature in binary mixtures of DPPC with cholesterol: A 2H NMR, FT-IR, DSC, and
Rodríguez-Hidalgo, M.R., Soto-Figueroa, C., & Vicente, L. (2011). Mesoscopic simulation of the drug
release mechanism on the polymeric vehicle P(ST-DVB) in an acid environment. Soft Matter, 7(18),
8224-8230.
Rösler, A., Vandermeulen, G. W., & Klok, H. A. (2012). Advanced drug delivery devices via self-
assembly of amphiphilic block copolymers. Advanced Drug Delivery Reviews, 64, 270-279.
Rozmanov, D., Baoukina, S., & Tieleman, D. P. (2014). Density based visualization for molecular
Ruan, G., & Feng, S. S. (2003). Preparation and characterization of poly (lactic acid)–poly (ethylene
355
Saikia, N., Jha, A. N., & Deka, R. C. (2013). Dynamics of fullerene-mediated heat-driven release of drug
molecules from carbon nanotubes. The Journal of Physical Chemistry Letters, 4(23), 4126-4132.
Samanta, S., & Roccatano, D. (2013). Interaction of curcumin with PEO–PPO–PEO block copolymers: A
Sanchez‐Sanchez, A., Akbari, S., Moreno, A. J., Verso, F. L., Arbe, A., Colmenero, J., & Pomposo, J. A.
(2013). Design and preparation of single‐chain nanocarriers mimicking disordered proteins for combined
Santos, R., Ursu, O., Gaulton, A., Bento, A.P., Donadi, R.S., Bologa, C.G., Karlsson, A., Al-Lazikani, B.,
Hersey, A., Oprea, T.I. & Overington, J. P. (2017). A comprehensive map of molecular drug targets.
Santosh, M., Panigrahi, S., Bhattacharyya, D., Sood, A. K., & Maiti, P. K. (2012). Unzipping and binding
of small interfering RNA with single walled carbon nanotube: A platform for small interfering RNA
Sarukhanyan, E., Milano, G., & Roccatano, D. (2014). Coating mechanisms of single-walled carbon
nanotube by linear polyether surfactants: Insights from computer simulations. The Journal of Physical
Schlick, T., Collepardo-Guevara, R., Halvorsen, L. A., Jung, S., & Xiao, X. (2011). Biomolecular
modeling and simulation: A field coming of age. Quarterly Reviews of Biophysics, 44(2), 191-228.
356
Schnablegger, H., & Singh, Y. (2013). The SAXS guide. Anton Paar GmbH.
Scott, W.R., Hünenberger, P.H., Tironi, I.G., Mark, A.E., Billeter, S.R., Fennen, J., Torda, A.E., Huber,
T., Krüger, P. & van Gunsteren, W. F. (1999). The GROMOS biomolecular simulation program package.
Seddon, J. M. (1990). Structure of the inverted hexagonal (HII) phase, and non-lamellar phase transitions
headgroup to bilayer surface charge: Torsion angle constraints from dipolar and quadrupolar couplings in
Semple, S.C., Akinc, A., Chen, J., Sandhu, A.P., Mui, B.L., Cho, C.K., Sah, D.W., Stebbing, D., Crosley,
E.J., Yaworski, E. & Hafez, I. M. (2010). Rational design of cationic lipids for siRNA delivery. Nature
Seu, K. J., Cambrea, L. R., Everly, R. M., & Hovis, J. S. (2006). Influence of lipid chemistry on
membrane fluidity: Tail and headgroup interactions. Biophysical Journal, 91(10), 3727-3735.
(DOPE): Water system at subzero temperatures and at low water contents. Biochimica et Biophysica Acta
Shan, P., Shen, J.W., Xu, D.H., Shi, L.Y., Gao, J., Lan, Y.W., Wang, Q. & Wei, X. H. (2014). Molecular
dynamics study on the interaction between doxorubicin and hydrophobically modified chitosan
357
Sharma, A., & Sharma, U. S. (1997). Liposomes in drug delivery: Progress and limitations. International
Shcharbin, D., Shakhbazau, A., & Bryszewska, M. (2013). Poly (amidoamine) dendrimer complexes as a
platform for gene delivery. Expert Opinion on Drug Delivery, 10(12), 1687-1698.
Shen, J.W., Tang, T., Wei, X.H., Zheng, W., Sun, T.Y., Zhang, Z., Liang, L. & Wang, Q. (2015). On the
loading mechanism of ssDNA into carbon nanotubes. RSC Advances, 5(70), 56896-56903.
Sheng, Y., An, J., & Zhu, Y. (2015). Self-assembly of ABA triblock copolymers under soft confinement.
Shi, C., Guo, D., Xiao, K., Wang, X., Wang, L., & Luo, J. (2015). A drug-specific nanocarrier design for
Shi, C., Yuan, D., Nangia, S., Xu, G., Lam, K. S., & Luo, J. (2014). A structure–property relationship
Shinoda, W., DeVane, R., & Klein, M. L. (2012). Computer simulation studies of self-assembling
Shinoda, W., Discher, D. E., Klein, M. L., & Loverde, S. M. (2013). Probing the structure of PEGylated-
358
Shrestha, H., Bala, R., & Arora, S. (2014). Lipid-based drug delivery systems. Journal of Pharmaceutics,
2014.
Siegel, D. P., & Epand, R. M. (1997). The mechanism of lamellar-to-inverted hexagonal phase transitions
3089-3111.
Singh, R., & Lillard Jr, J. W. (2009). Nanoparticle-based targeted drug delivery. Experimental and
Skandani, A. A., Zeineldin, R., & Al-Haik, M. (2012). Effect of chirality and length on the penetrability
of single-walled carbon nanotubes into lipid bilayer cell membranes. Langmuir, 28(20), 7872-7879.
Sodt, A. J., & Pastor, R. W. (2013). Bending free energy from simulation: Correspondence of planar and
Sohal, P. S., & Cornell, R. B. (1990). Sphingosine inhibits the activity of rat liver CTP:phosphocholine
Soppimath, K. S., Aminabhavi, T. M., Kulkarni, A. R., & Rudzinski, W. E. (2001). Biodegradable
polymeric nanoparticles as drug delivery devices. Journal of Controlled Release, 70(1-2), 1-20.
Spanakis, M., Bouropoulos, N., Theodoropoulos, D., Sygellou, L., Ewart, S., Moschovi, A.M., Siokou,
A., Niopas, I., Kachrimanis, K., Nikolakis, V. & Cox, P. A. (2014). Controlled release of 5-fluorouracil
from microporous zeolites. Nanomedicine: Nanotechnology, Biology and Medicine, 10(1), 197-205.
359
Spoel, D.V.D., Lindahl, E., Hess, B., Buuren, A.R.V., Apol, E., Meulenhoff, P.J., Tieleman, D.P., Sijbers,
A.L.T.M., Feenstra, K.A., Drunen, R.V. & Berendsen, H. J. C. (2014). GROMACS User Manual
Version 4.6. 7.
Sridhar, A., Srikanth, B., Kumar, A., & Dasmahapatra, A. K. (2015). Coarse-grain molecular dynamics
study of fullerene transport across a cell membrane. The Journal of Chemical Physics, 143(2), 024907.
Srinivas, G., Mohan, R. V., & Kelkar, A. D. (2013). Polymer micelle assisted transport and delivery of
model hydrophilic components inside a biological lipid vesicle: A coarse-grain simulation study. The
Stepniewski, M., Pasenkiewicz-Gierula, M., Róg, T., Danne, R., Orlowski, A., Karttunen, M., Urtti, A.,
Yliperttula, M., Vuorimaa, E. & Bunker, A. (2011). Study of PEGylated lipid layers as a model for
PEGylated liposome surfaces: Molecular dynamics simulation and Langmuir monolayer studies.
Sternin, E., Bloom, M., & Mackay, A. L. (1983). De-Pake-ing of NMR spectra. Journal of Magnetic
Subashini, M., Devarajan, P. V., Sonavane, G. S., & Doble, M. (2011). Molecular dynamics simulation of
Sudhahar, C. G., Haney, R. M., Xue, Y., & Stahelin, R. V. (2008). Cellular membranes and lipid-binding
domains as attractive targets for drug development. Current Drug Targets, 9(8), 603-613.
Sun, B., & Chiu, D. T. (2005). Determination of the encapsulation efficiency of individual vesicles using
single-vesicle photolysis and confocal single-molecule detection. Analytical Chemistry, 77(9), 2770-2776.
360
Sun, C., & Tang, T. (2014). Study on the role of polyethylenimine as gene delivery carrier using
molecular dynamics simulations. Journal of Adhesion Science and Technology, 28(3-4), 399-416.
Sun, C., Lee, J. S., & Zhang, M. (2008). Magnetic nanoparticles in MR imaging and drug delivery.
Sun, C., Tang, T., & Uludaǧ, H. (2012). Molecular dynamics simulations for complexation of DNA with
2 kDa PEI reveal profound effect of PEI architecture on complexation. The Journal of Physical Chemistry
B, 116(8), 2405-2413.
Sun, C., Tang, T., & Uludağ, H. (2012). Probing the effects of lipid substitution on polycation mediated
Sun, C., Tang, T., & Uludag, H. (2013). A molecular dynamics simulation study on the effect of lipid
Sun, T., Guo, Q., Zhang, C., Hao, J., Xing, P., Su, J., Li, S., Hao, A. & Liu, G. (2012). Self-assembled
vesicles prepared from amphiphilic cyclodextrins as drug carriers. Langmuir, 28(23), 8625-8636.
Sun, X., Feng, Z., Hou, T., & Li, Y. (2015). Computational simulation of inorganic nanoparticle drug
Sun, X., Feng, Z., Zhang, L., Hou, T., & Li, Y. (2014). The selective interaction between silica
nanoparticles and enzymes from molecular dynamics simulations. PLoS ONE, 9(9), e107696.
Svenson, S., & Tomalia, D. A. (2005). Dendrimers in biomedical applications-reflections on the field.
361
Takahashi, H., Sinoda, K., & Hatta, I. (1996). Effects of cholesterol on the lamellar and the inverted
Tam, Y., Chen, S., & Cullis, P. R. (2013). Advances in lipid nanoparticles for siRNA delivery.
Tan, H., Wang, Z., Li, J., Pan, Z., Ding, M., & Fu, Q. (2013). An approach for the sphere-to-rod transition
Taneva, S., Dennis, M. K., Ding, Z., Smith, J. L., & Cornell, R. B. (2008). Contribution of each
membrane binding, and membrane cross-bridging. Journal of Biological Chemistry, 283(42), 28137-
28148.
Taneva, S., Johnson, J. E., & Cornell, R. B. (2003). Lipid-induced conformational switch in the
Taneva, S., Patty, P. J., Frisken, B. J., & Cornell, R. B. (2005). CTP:phosphocholine cytidylyltransferase
Taresco, V., Gontrani, L., Crisante, F., Francolini, I., Martinelli, A., D’Ilario, L., Bordi, F. & Piozzi, A.
(2015). Self-assembly of catecholic moiety-containing cationic random acrylic copolymers. The Journal
362
Tate, M. W., & Gruner, S. M. (1989). Temperature dependence of the structural dimensions of the
28(10), 4245-4253.
Team, I. (2004). Inkscape: A vector drawing tool. URL http://www. inkscape. org.
Teixeira, R.S., Cova, T.F., Silva, S.M., Oliveira, R., Araújo, M.J., Marques, E.F., Pais, A.A. & Veiga, F.
J. (2014). Lysine-based surfactants as chemical permeation enhancers for dermal delivery of local
Testa, F., Filippelli, M., Brunetti-Pierri, R., Di Fruscio, G., Di Iorio, V., Pizzo, M., Torella, A., Barillari,
M.R., Nigro, V., Brunetti-Pierri, N. & Simonelli, F. (2017). Mutations in the PCYT1A gene are
responsible for isolated forms of retinal dystrophy. European Journal of Human Genetics, 25(5), 651-
655.
Thakur, S. S., Parekh, H. S., Schwable, C. H., Gan, Y., & Ouyang, D. (2015). Solubilization of poorly
Thewalt, J. L., Tulloch, A. P., & Cushley, R. J. (1986). A deuterium NMR study of labelled n-alkanol
Thewalt, J. L., Wassall, S. R., Gorrissen, H., & Cushley, R. J. (1985). Deuterium NMR study of the effect
363
Thota, N., Luo, Z., Hu, Z., & Jiang, J. (2013). Self-assembly of amphiphilic peptide (AF) 6H5K15:
Coarse-grained molecular dynamics simulation. The Journal of Physical Chemistry B, 117(33), 9690-
9698.
Thurmond, R. L., Lindblom, G., & Brown, M. F. (1993). Curvature, order, and dynamics of lipid
Tian, F., Yue, T., Li, Y., & Zhang, X. (2014). Computer simulation studies on the interactions between
Tian, W. D., & Ma, Y. Q. (2012). Insights into the endosomal escape mechanism via investigation of
Tian, X., Sun, F., Zhou, X. R., Luo, S. Z., & Chen, L. (2015). Role of peptide self‐assembly in
Tieleman, D. P. (2010). Methods and parameters for membrane simulations. In Molecular Simulations
Tieleman, D. P., Marrink, S. J., & Berendsen, H. J. (1997). A computer perspective of membranes:
Molecular dynamics studies of lipid bilayer systems. Biochimica et Biophysica Acta (BBA)-Reviews on
Tironi, I. G., Sperb, R., Smith, P. E., & van Gunsteren, W. F. (1995). A generalized reaction field method
for molecular dynamics simulations. The Journal of Chemical Physics, 102(13), 5451-5459.
364
Todorova, N., Chiappini, C., Mager, M., Simona, B., Patel, I. I., Stevens, M. M., & Yarovsky, I. (2014).
Surface presentation of functional peptides in solution determines cell internalization efficiency of TAT
Trenker, R., Call, M. J., & Call, M. E. (2016). Progress and prospects for structural studies of
transmembrane interactions in single-spanning receptors. Current Opinion in Structural Biology, 39, 115-
123.
Tresset, G. (2009). The multiple faces of self-assembled lipidic systems. PMC Biophysics, 2(1), 3.
Tu, C. K., Chen, K., Tian, W. D., & Ma, Y. Q. (2013). Computational investigations of a peptide-
modified dendrimer interacting with lipid membranes. Macromolecular Rapid Communications, 34(15),
1237-1242.
Turner, D. C., & Gruner, S. M. (1992). X-ray diffraction reconstruction of the inverted hexagonal (HII)
Turovsky, T., Khalfin, R., Kababya, S., Schmidt, A., Barenholz, Y., & Danino, D. (2015). Celecoxib
encapsulation in β-casein micelles: Structure, interactions, and conformation. Langmuir, 31(26), 7183-
7192.
Utama, R. H., Dulle, M., Förster, S., Stenzel, M. H., & Zetterlund, P. B. (2015). SAXS analysis of shell
formation during nanocapsule synthesis via inverse miniemulsion periphery RAFT polymerization.
365
Vácha, R., Martinez-Veracoechea, F. J., & Frenkel, D. (2012). Intracellular release of endocytosed
Van Deenen, L. L. M., & De Gier, J. (1974). Lipids of the red cell membrane. The Red Blood Cell, 1,
147-211.
Van Gunsteren, W. F., Dolenc, J., & Mark, A. E. (2008). Molecular simulation as an aid to
Van Gunsteren, W.F., Bakowies, D., Baron, R., Chandrasekhar, I., Christen, M., Daura, X., Gee, P.,
Geerke, D.P., Glättli, A., Hünenberger, P.H. & Kastenholz, M. A. (2006). Biomolecular modeling: Goals,
Van Hoof, B., Markvoort, A. J., van Santen, R. A., & Hilbers, P. A. (2014). Molecular simulation of
protein encapsulation in vesicle formation. The Journal of Physical Chemistry B, 118(12), 3346-3354.
Van Lehn, R. C., & Alexander-Katz, A. (2011). Penetration of lipid bilayers by nanoparticles with
Van Lehn, R. C., & Alexander-Katz, A. (2013). Ligand-mediated short-range attraction drives
aqueous salt solution from atomistic molecular dynamics simulations. The Journal of Physical Chemistry
C, 117(39), 20104-20115.
Van Lehn, R. C., & Alexander-Katz, A. (2014). Free energy change for insertion of charged, monolayer-
366
Van Lehn, R.C., Ricci, M., Silva, P.H., Andreozzi, P., Reguera, J., Voïtchovsky, K., Stellacci, F. &
Alexander-Katz, A. (2014). Lipid tail protrusions mediate the insertion of nanoparticles into model cell
Van Meer, G., Voelker, D. R., & Feigenson, G. W. (2008). Membrane lipids: Where they are and how
Vanommeslaeghe, K., & MacKerell Jr, A. D. (2012). Automation of the CHARMM General Force Field
(CGenFF) I: Bond perception and atom typing. Journal of Chemical Information and Modeling, 52(12),
3144-3154.
Vanommeslaeghe, K., Hatcher, E., Acharya, C., Kundu, S., Zhong, S., Shim, J., Darian, E., Guvench, O.,
Lopes, P., Vorobyov, I. & Mackerell Jr, A. D. (2010). CHARMM general force field: A force field for
drug‐like molecules compatible with the CHARMM all‐atom additive biological force fields. Journal of
Vanommeslaeghe, K., Raman, E. P., & MacKerell Jr, A. D. (2012). Automation of the CHARMM
General Force Field (CGenFF) II: Assignment of bonded parameters and partial atomic charges. Journal
Velinova, M., Sengupta, D., Tadjer, A. V., & Marrink, S. J. (2011). Sphere-to-rod transitions of nonionic
surfactant micelles in aqueous solution modeled by molecular dynamics simulations. Langmuir, 27(23),
14071-14077.
Venable, R. M., Brown, F. L., & Pastor, R. W. (2015). Mechanical properties of lipid bilayers from
367
Venditto, V. J., & Szoka Jr, F. C. (2013). Cancer nanomedicines: So many papers and so few drugs!.
Veronese, F. M., & Pasut, G. (2005). PEGylation, successful approach to drug delivery. Drug Discovery
Viger, M. L., Sheng, W., McFearin, C. L., Berezin, M. Y., & Almutairi, A. (2013). Application of time-
resolved fluorescence for direct and continuous probing of release from polymeric delivery vehicles.
Vijayaraj, R., Van Damme, S., Bultinck, P., & Subramanian, V. (2013). Theoretical studies on the
Vitorino, C., Almeida, J., Gonçalves, L. M., Almeida, A. J., Sousa, J. J., & Pais, A. A. C. C. (2013). Co-
encapsulating nanostructured lipid carriers for transdermal application: From experimental design to the
Vukovic, L., Khatib, F. A., Drake, S. P., Madriaga, A., Brandenburg, K. S., Král, P., & Onyuksel, H.
(2011). Structure and dynamics of highly PEG-ylated sterically stabilized micelles in aqueous media.
Vuković, L., Madriaga, A., Kuzmis, A., Banerjee, A., Tang, A., Tao, K., Shah, N., Král, P. & Onyuksel,
15754.
368
Wan, C., Allen, T. M., & Cullis, P. R. (2014). Lipid nanoparticle delivery systems for siRNA-based
Wang, J., Wang, Y., & Liang, W. (2012). Delivery of drugs to cell membranes by encapsulation in PEG–
Wang, S., & Dormidontova, E. E. (2010). Nanoparticle design optimization for enhanced targeting:
Wang, S., & Dormidontova, E. E. (2012). Selectivity of ligand-receptor interactions between nanoparticle
Wang, T., Chipot, C., Shao, X., & Cai, W. (2010). Structural characterization of micelles formed of
Wang, X. Y., Zhang, L., Wei, X. H., & Wang, Q. (2013). Molecular dynamics of paclitaxel encapsulated
Wang, X., & Deserno, M. (2015). Determining the pivotal plane of fluid lipid membranes in simulations.
Wang, Y. L., Lu, Z. Y., & Laaksonen, A. (2012). Specific binding structures of dendrimers on lipid
369
Wang, Z., & Jiang, J. (2014). Dissipative particle dynamics simulation on paclitaxel loaded PEO–PPO–
PEO block copolymer micelles. Journal of Nanoscience and Nanotechnology, 14(3), 2644-2647.
Wang, Z., Wang, F., Su, C., & Zhang, Y. (2013). Computer simulation of polymer delivery system by
dissipative particle dynamics. Journal of Computational and Theoretical Nanoscience, 10(10), 2323-
2327.
Weber, C. H., Park, Y. S., Sanker, S., Kent, C., & Ludwig, M. L. (1999). A prototypical
7(9), 1113-1124.
Wen, X.F., Lan, J.L., Cai, Z.Q., Pi, P.H., Xu, S.P., Zhang, L.J., Qian, Y. & Wang, S. N. (2014).
White, S. H., King, G. I., & Cain, J. E. (1981). Location of hexane in lipid bilayers determined by neutron
Williams, P. S., Carpino, F., & Zborowski, M. (2009). Magnetic nanoparticle drug carriers and their study
Winter, N. D., Murphy, R. K., O’Halloran, T. V., & Schatz, G. C. (2011). Development and modeling of
370
Wong-Ekkabut, J., Baoukina, S., Triampo, W., Tang, I. M., Tieleman, D. P., & Monticelli, L. (2008).
Computer simulation study of fullerene translocation through lipid membranes. Nature Nanotechnology,
3(6), 363–368.
Wu, N., Wang, Q., & Arash, B. (2012). Ejection of DNA molecules from carbon nanotubes. Carbon,
50(13), 4945-4952.
Wu, W. I., Voegtli, W. C., Sturgis, H. L., Dizon, F. P., Vigers, G. P., & Brandhuber, B. J. (2010). Crystal
structure of human AKT1 with an allosteric inhibitor reveals a new mode of kinase inhibition. PLoS
Xiao, K., Li, Y., Luo, J., Lee, J.S., Xiao, W., Gonik, A.M., Agarwal, R.G. & Lam, K. S. (2011). The
effect of surface charge on in vivo biodistribution of PEG-oligocholic acid based micellar nanoparticles.
Xie, L. Q., Feng, J. W., Tian, W. D., & Ma, Y. Q. (2014). Generation-dependent gel-fluid phase transition
of membrane caused by a PAMAM dendrimer. Macromolecular Theory and Simulations, 23(8), 531-536.
Xie, M., Smith, J. L., Ding, Z., Zhang, D., & Cornell, R. B. (2004). Membrane binding modulates the
279(27), 28817-28825.
Xu, W., Doshi, A., Lei, M., Eck, M. J., & Harrison, S. C. (1999). Crystal structures of c-Src reveal
Yamamoto, G.L., Baratela, W.A., Almeida, T.F., Lazar, M., Afonso, C.L., Oyamada, M.K., Suzuki, L.,
Oliveira, L.A., Ramos, E.S., Kim, C.A. & Passos-Bueno, M. R. (2014). Mutations in PCYT1A cause
371
spondylometaphyseal dysplasia with cone-rod dystrophy. The American Journal of Human Genetics,
94(1), 113-119.
Yang, A. C., & Weng, C. I. (2010). Structural and dynamic properties of water near monolayer-protected
gold clusters with various alkanethiol tail groups. The Journal of Physical Chemistry C, 114(19), 8697-
8709.
Yang, L., & da Rocha, S. R. (2014). PEGylated, NH2-terminated PAMAM dendrimers: A microscopic
Yang, W., Boggs, K. P., & Jackowski, S. (1995). The association of lipid activators with the amphipathic
Yin, H., Kanasty, R. L., Eltoukhy, A. A., Vegas, A. J., Dorkin, J. R., & Anderson, D. G. (2014). Non-
viral vectors for gene-based therapy. Nature Reviews Genetics, 15(8), 541-555.
Yu, J., Becker, M. L., & Carri, G. A. (2011). The influence of amino acid sequence and functionality on
the binding process of peptides onto gold surfaces. Langmuir, 28(2), 1408-1417.
Yue, T., & Zhang, X. (2011). Molecular understanding of receptor-mediated membrane responses to
372
Zattoni, A., Roda, B., Borghi, F., Marassi, V., & Reschiglian, P. (2014). Flow field-flow fractionation for
the analysis of nanoparticles used in drug delivery. Journal of Pharmaceutical and Biomedical Analysis,
87, 53-61.
Zhan, B., Shi, K., Dong, Z., Lv, W., Zhao, S., Han, X., Wang, H. & Liu, H. (2015). Coarse-grained
2834-2844.
Zhang, L., Becton, M., & Wang, X. (2015). Designing nanoparticle translocation through cell membranes
by varying amphiphilic polymer coatings. The Journal of Physical Chemistry B, 119(9), 3786-3794.
Zhang, L., Wang, Z., Lu, Z., Shen, H., Huang, J., Zhao, Q., Liu, M., He, N. & Zhang, Z. (2013).
PEGylated reduced graphene oxide as a superior ssRNA delivery system. Journal of Materials Chemistry
B, 1(6), 749-755.
Zhang, S., Sun, H.J., Hughes, A.D., Moussodia, R.O., Bertin, A., Chen, Y., Pochan, D.J., Heiney, P.A.,
Klein, M.L. & Percec, V. (2014). Self-assembly of amphiphilic Janus dendrimers into uniform onion-like
dendrimersomes with predictable size and number of bilayers. Proceedings of the National Academy of
Sciences, 201402858.
Zhang, Y., Chan, H. F., & Leong, K. W. (2013). Advanced Materials and processing for drug delivery:
The past and the future. Advanced Drug Delivery Reviews, 65(1), 104-120.
Zhang, Y., Feller, S. E., Brooks, B. R., & Pastor, R. W. (1995). Computer simulation of liquid/liquid
interfaces. I. Theory and application to octane/water. The Journal of Chemical Physics, 103(23), 10252-
10266.
373
Zhao, M., Zhou, J., Su, C., Niu, L., Liang, D., & Li, B. (2015). Complexation behavior of oppositely
charged polyelectrolytes: Effect of charge distribution. The Journal of Chemical Physics, 142(20),
204902.
Zheng, L. S., Yang, Y. Q., Guo, X. D., Sun, Y., Qian, Y., & Zhang, L. J. (2011). Mesoscopic simulations
on the aggregation behavior of pH-responsive polymeric micelles for drug delivery. Journal of Colloid
Zheng, M., Pavan, G.M., Neeb, M., Schaper, A.K., Danani, A., Klebe, G., Merkel, O.M. & Kissel, T.
(2012). Targeting the blind spot of polycationic nanocarrier-based siRNA delivery. ACS nano, 6(11),
9447-9454.
Zhigaltsev, I. V., Belliveau, N., Hafez, I., Leung, A. K., Huft, J., Hansen, C., & Cullis, P. R. (2012).
Bottom-up design and synthesis of limit size lipid nanoparticle systems with aqueous and triglyceride
Zhigaltsev, I. V., Tam, Y. K., Leung, A. K., & Cullis, P. R. (2016). Production of limit size
nanoliposomal systems with potential utility as ultra-small drug delivery agents. Journal of Liposome
Zhuang, X., Dávila-Contreras, E. M., Beaven, A. H., Im, W., & Klauda, J. B. (2016). An extensive
simulation study of lipid bilayer properties with different head groups, acyl chain lengths, and chain
Zhuang, X., Makover, J. R., Im, W., & Klauda, J. B. (2014). A systematic molecular dynamics simulation
study of temperature dependent bilayer structural properties. Biochimica et Biophysica Acta (BBA)-
374
Ziebarth, J., & Wang, Y. (2010). Coarse-grained molecular dynamics simulations of DNA condensation
by block copolymer and formation of core−corona structures. The Journal of Physical Chemistry B,
114(19), 6225-6232.
Zimmermann, T.S., Lee, A.C., Akinc, A., Bramlage, B., Bumcrot, D., Fedoruk, M.N., Harborth, J.,
Heyes, J.A., Jeffs, L.B., John, M. & Judge, A. D. (2006). RNAi-mediated gene silencing in non-human
375
Appendix A: Parameters for the Parameterized Segment in the KC2 and KC2H
The final atom types, partial charges, and the bonded parameters for the parameterized segment
[Figure A.1] in KC2 and KC2H lipids are provided here as a reference. The index numbers in
these tables are the same with the atom index numbers in the .itp files for KC2 and KC2H lipids.
376
Figure A.1: Atom index numbers for the parameterized segment of KC2(H).
Atom index numbers are according to the KC2(H) .itp files. Atom 126 is the proton attached to the
nitrogen atom. This atom is therefore present in KC2H but not in KC2. All the hydrogens are shown with
grey lines representing the bonds, except the protonation hydrogen which is shown with a thick red line.
377
Table A-1: Parameters for the ring segment in KC2 and KC2H.
Type
Charge
Table A-2: Parameters for the amino segments in KC2 and KC2H.
Index
Atom
Partial
378
Table A-3: Parameters for the hydrogen atoms in KC2 and KC2H.
Atom 49, 50, 51, 52, 47, 48, 59, 60, 74, 75, 76, 78 83, 84, 85,
Index 61, 62, 63, 64 77, 79, 80, 81, 82, 89, 90 86, 87, 88
Type
Charge
Table A-4: Parameters for several bonds defined in KC2 and KC2H.
These parameters were added to the C36FF ffbonded.itp file, in the bonds section.
* b0 and kb represents the equilibrium bond length and the corresponding force constant in the harmonic
𝒏𝒎 𝒌𝑱 𝒎𝒐𝒍−𝟏 𝒏𝒎−𝟐
379
Table A-5: Parameters for several angles defined in KC2 and KC2H.
These parameters were added to the C36FF ffbonded.itp file, in the angles section.
* theta0 and ktheta are the equilibrium angle and the force constant in the harmonic term in the potential
function, respectively.
** ub0 and kub mean the equilibrium distance between i-k atoms in an i-j-k angle formed by three
connected atoms, and the force constant in the Urey-Bradley term in the potential function, respectively.
380
CG321 CG3C50 OG3C51 5 111.5 376.56 0 0
Table A-6: Parameters for several dihedrals defined in KC2 and KC2H.
These parameters were added to the C36FF ffbonded.itp file, in the proper dihedrals section.
* phi0 and kphi are the phase shift and the force constant in the sinusoidal expansion term in the potential
𝒅𝒆𝒈𝒓𝒆𝒆 𝒌𝑱 𝒎𝒐𝒍−𝟏
381
CG321 CTL2 CTL2 CTL2 9 180 0.594128 3
382
OG3C51 CG3C51 CG3C52 OG3C51 9 0 1.08784 3
* phi0 and kphi are the phase shift and the force constant in the sinusoidal expansion term in the potential
𝒅𝒆𝒈𝒓𝒆𝒆 𝒌𝑱 𝒎𝒐𝒍−𝟏
1 26 16 17 9 180 8.61063 1
1 26 16 17 9 180 0.98874 2
383
1 26 16 17 9 0 2.05603 3
1 26 16 17 9 180 0.36095 4
1 26 16 17 9 180 0.04944 6
1 26 16 18 9 180 8.61063 1
1 26 16 18 9 180 0.98874 2
1 26 16 18 9 0 2.05603 3
1 26 16 18 9 180 0.36095 4
1 26 16 18 9 180 0.04944 6
2 1 26 16 9 0 2.96923 1
2 1 26 16 9 0 0.52548 2
2 1 26 16 9 180 1.41779 3
2 1 26 16 9 0 0.71499 4
2 1 26 16 9 180 0.27977 6
7 15 16 17 9 180 8.61063 1
7 15 16 17 9 180 0.98874 2
7 15 16 17 9 0 2.05603 3
7 15 16 17 9 180 0.36095 4
7 15 16 17 9 180 0.04944 6
7 15 16 18 9 180 8.61063 1
384
7 15 16 18 9 180 0.98874 2
7 15 16 18 9 0 2.05603 3
7 15 16 18 9 180 0.36095 4
7 15 16 18 9 180 0.04944 6
8 7 15 16 9 0 2.96923 1
8 7 15 16 9 0 0.52548 2
8 7 15 16 9 180 1.41779 3
8 7 15 16 9 0 0.71499 4
8 7 15 16 9 180 0.27977 6
17 20 21 22 9 180 1.81701 1
17 20 21 22 9 0 0.95898 2
17 20 21 22 9 0 2.26375 3
17 20 21 22 9 180 0.63793 4
17 20 21 22 9 180 0.03559 6
20 21 22 23 9 180 0.90678 1
20 21 22 23 9 0 0.45645 2
20 21 22 23 9 0 0.23682 3
20 21 22 23 9 0 0.34473 4
20 21 22 23 9 180 0.27423 6
385
21 22 23 24 9 180 6.44817 1
21 22 23 24 9 0 3.26084 2
21 22 23 24 9 0 5.26565 3
21 22 23 24 9 180 0.40517 4
21 22 23 24 9 0 0.36597 6
21 22 23 25 9 180 6.44817 1
21 22 23 25 9 0 3.26084 2
21 22 23 25 9 0 5.26565 3
21 22 23 25 9 180 0.40517 4
21 22 23 25 9 0 0.36597 6
* phi0 and kphi are the phase shift and the force constant in the sinusoidal expansion term in the potential
𝒅𝒆𝒈𝒓𝒆𝒆 𝒌𝑱 𝒎𝒐𝒍−𝟏
1 26 16 17 9 180 8.61063 1
1 26 16 17 9 180 0.98874 2
386
1 26 16 17 9 0 2.05603 3
1 26 16 17 9 180 0.36095 4
1 26 16 17 9 180 0.04944 6
1 26 16 18 9 180 8.61063 1
1 26 16 18 9 180 0.98874 2
1 26 16 18 9 0 2.05603 3
1 26 16 18 9 180 0.36095 4
1 26 16 18 9 180 0.04944 6
2 1 26 16 9 0 2.96923 1
2 1 26 16 9 0 0.52548 2
2 1 26 16 9 180 1.41779 3
2 1 26 16 9 0 0.71499 4
2 1 26 16 9 180 0.27977 6
7 15 16 17 9 180 8.61063 1
7 15 16 17 9 180 0.98874 2
7 15 16 17 9 0 2.05603 3
7 15 16 17 9 180 0.36095 4
7 15 16 17 9 180 0.04944 6
7 15 16 18 9 180 8.61063 1
387
7 15 16 18 9 180 0.98874 2
7 15 16 18 9 0 2.05603 3
7 15 16 18 9 180 0.36095 4
7 15 16 18 9 180 0.04944 6
8 7 15 16 9 0 2.96923 1
8 7 15 16 9 0 0.52548 2
8 7 15 16 9 180 1.41779 3
8 7 15 16 9 0 0.71499 4
8 7 15 16 9 180 0.27977 6
17 20 21 22 9 180 1.81701 1
17 20 21 22 9 0 0.95898 2
17 20 21 22 9 0 2.26375 3
17 20 21 22 9 180 0.63793 4
17 20 21 22 9 180 0.03559 6
20 21 22 23 9 180 1.90831 1
20 21 22 23 9 180 0.50746 2
20 21 22 23 9 180 0.26074 3
20 21 22 23 9 0 0.50625 4
20 21 22 23 9 180 0.6132 6
388
21 22 23 24 9 180 0.18613 1
21 22 23 24 9 0 0.18274 2
21 22 23 24 9 0 2.52813 3
21 22 23 24 9 0 0.189 4
21 22 23 24 9 0 0.41145 6
21 22 23 25 9 180 0.18613 1
21 22 23 25 9 0 0.18274 2
21 22 23 25 9 0 2.52813 3
21 22 23 25 9 0 0.189 4
21 22 23 25 9 0 0.41145 6
389
Appendix B: Supplementary Information for Chapter Four
Although the parameters for POPC are available, optimized and well-tested in C36 FF [1], parameters for
KC2 and KC2H are not present and need to be developed. The step-by-step protocol for parameterization
of a novel molecule in C36 FF is documented in detail in several publications [2, 3]. In this study, several
computational tools, including CHARMM-GUI [4], GAAMP [5], and ParamChem [6, 7], were used to
develop reasonable models for KC2 and KC2H compatible with C36 FF.
The KC2(H) molecules were first divided into two main segments: lipid tails and headgroup. The
parameters for the tails were simply taken from DUPC downloaded from CHARMM-GUI [4], with atom
types taken from C36 FF. Next, the headgroup was further divided into two smaller segments (the ring
and the amino segments) and each segment was parameterized using the GAAMP server [5], with atom
types taken from CHARMM General force field (CGenFF) [3]. The parameters developed for the
segments were subsequently combined to give the parameter set for the whole lipid molecules. Where
needed, parameters were further modified in an iterative procedure until the desired level of agreement
between computational and experimental data was achieved (see below). A more detailed description of
the parameterization process, and the final parameter set are provided in section 2.4.1 of Chapter 2, and
Appendix A, respectively.
The amount of target data to validate parameters against are limited for the case of KC2 and KC2H. In
this study, the effect of adding KC2(H) on POPC palmitoyl chain order parameters in POPC/KC2(H)
390
The SCD profiles for the POPC palmitoyl chain extracted from 2H NMR spectra and calculated
from simulations of POPC/KC2(H) binary mixtures are shown in Figure B.1. In simulations, systems
with KC2 and KC2H correspond to the 2H NMR experiments conducted at pH 8.1 and 4.4, respectively.
The SCD parameters from simulations are in good agreement with the experimental data. The
overall trend in SCD profiles is similar between simulations and experiments. The acyl chain order
gradually decreases along the acyl chain towards the terminal methyl group. Also, in both simulations and
experiments, the pH seems to have no or little effect on acyl chain order parameter at 313 K. The only
difference was observed between pure POPC bilayers at two pH values. We note that the experimental
POPC palmitoyl chain order parameters are smaller at pH 4.4 than at pH 8.1. This is likely due to the
higher interfacial surface tension observed in POPC membranes at pH 4.4 [8]. In simulations, however,
the SCD parameters for palmitoyl chain are the same between two pH levels. That is mainly because the
effect of buffer and the higher surface tension were not considered in our simulations.
Finally, the SCD parameters calculated from simulations are slightly higher than the corresponding
experimental values. This is also true for POPC which the parameters are optimized and well tested in
C36 FF.
391
Figure B.1: Validation of developed models for KC2 and KC2H.
Smoothed deuterium order parameter (SCD) for the palmitoyl chain of POPC in POPC/KC2(H) binary
mixtures were measured using 2H NMR experiments (Left column). The same parameter was calculated
from simulations and smoothed (Right column). This comparison was done for two pH levels of 4.4 (a
392
B.2 KC2(H) Effects on POPC as a Function of Temperature and Mixing Ratio
The POPC-d31 order parameter (SCD) profiles for POPC/KC2(H) systems at different temperatures and
mixing ratios were measured using 2H NMR technique. The effect of KC2 and KC2H on POPC sn-1 acyl
chain SCD was shown to be a function of both temperature and mixing ratio [Figure B.2 and Figure B.3].
For both pH 4.4 and 8.1, and all the mixing ratios studied, higher temperature resulted in lower SCD
values as expected. For the pure POPC system, the POPC palmitoyl chain order parameters are
significantly smaller at pH 4.4 than pH 8.1. At higher temperatures (303 K and 313 K), pH has little to no
effect on chain order in POPC/KC2(H) systems. At 288 K and 293 K, KC2 has no effect on POPC chain
393
Figure B.2: SCD parameters for the palmitoyl chain (sn-1) of POPC at 288 K and 293 K.
Experimental SCD parameters (smoothed) measured using 2H NMR technique for binary POPC/KC2(H)
mixtures at (a) pH = 4.4 and T = 288 K, (b) pH = 8.1 and T = 288 K, (c) pH = 4.4 and T = 293 K, and (d)
Such an induced order in PC lipids due to adding cationic lipids to the system has been previously
reported [9]. Including cationic lipids (DMTAP) in a zwitterionic (DMPC) membrane was reported to
induce acyl chain order in DMPC in a concentration-dependent way, up to a 0.50 mole fraction of
DMTAP [9]. Increasing the DMTAP mole fraction above 0.50, however, started to decrease the induced
394
order. To explain this observation, the authors proposed a scenario as follows. Given that both DMTAP
and DMPC have the same acyl chains, a small amount of DMTAP tends to reorient the nearby DMPC
headgroups tilt angle to smaller angles. This will result in a more packed structures with a reduced area
per lipid and increased acyl chain order. Upon increase in the DMTAP concentration beyond a limit, the
electrostatic repulsions between the positively charged lipids dominate which result in the membrane
395
Figure B.3: SCD parameters for the palmitoyl chain (sn-1) of POPC at 303 K and 313 K.
Experimental SCD parameters (smoothed) measured using 2H NMR technique for binary POPC/KC2(H)
mixtures at (a) pH = 4.4 and T = 303 K, (b) pH = 8.1 and T = 303 K, (c) pH = 4.4 and T = 313 K, and (d)
The same scenario might also be valid here and could explain the induced chain order at pH 4.4
[Figure B.2]. However, since the acyl chains in KC2H and POPC are not the same, the behavior of
POPC/KC2H systems are rather more complex to explain. This complex pattern (i.e. temperature-
concentration dependency) of induced order might be a collective effect of several factors, including the
396
interfacial surface tension [8], which is likely relaxed due to KC2H, the observed decrease in POPC P→N
vector tilt angle (see section B.6) [9], lipid-ion interactions (see section B.5) [10], and lipid-lipid
interactions [9, 11]. Other factors might also contribute to such a complex behavior. The so-called free
volume in the center of the bilayer [12] could be one such factor. In a study by Kupianien et al. [13], it
was shown that a reduction in the free volume of membrane result in a tighter molecular packing and
If neutral KC2 are segregated and confined in the POPC bilayer center, systems with a higher
concentration of KC2 are expected to result in systems with larger repeat spacings (d-spacing) at higher
pH values. According to this expectation and the observations from simulations, small angle X-ray
scattering experiments were carried out on POPC multilamellar vesicles containing 0, 20 or 30 mol %
KC2 at pH 8.1, 8.5 or 10 [Figure B.4]. The major finding is that added KC2 causes a dramatic reduction
in structure correlations in the sample, as exemplified by the broad scattering peak [Figure B.4]. A
second result is that the bilayer repeat spacing increases from 65 angstroms to 68 angstroms upon the
addition of 20 mol % KC2 to POPC at pH 8.1. Increasing the pH to 8.5 results in a reduction in lamellar
repeat spacing to 67.5 angstroms, and increasing it to 10 brings the lamellar repeat spacing back to 65
angstroms. These changes in lamellar repeat spacing can be explained by the presence of a small fraction
of positively charged KC2H at pH 8.1 and 8.5 and are consistent with the observations of B. Pozo Navas
et al. [14] on POPC/POPG mixtures. KC2 is silent in the SAXS patterns, and is likely separating into oily
droplets associated with the POPC lamellae and thereby disrupting their stacking regularity.
397
Figure B.4: SAXS curves for POPC/KC2 multilamellar vesicles at several basic pH values.
SAXS curves for POPC multilamellar vesicles containing 0, 20 or 30 mol % KC2 at the indicated pH.
Disregard of the KC2H content in the mixtures, simulations suggest that all KC2H stay in the lipid-water
interface [Figure 4.1]. Figure B.5 shows the symmetrized electron densities for the POPC phosphorus
atoms in POPC/KC2H systems at different mixing ratios, pH = 4.4 and temperature 313 K. For the pure
398
POPC, the inter-leaflet distance is approximately 3.72 nm. Adding KC2H had little to no effect on POPC
P-P distance.
The electron density of the phosphorus atoms of POPC were calculated from simulations of binary
POPC/KC2H mixtures at pH = 4.4 and T = 313 K. The X axis shows the relative distance from the
bilayer center. Each double-headed arrow represents the peak-to-peak distance for a mixing ratio,
corresponding to the P-P inter-leaflet distance. The P-P distance for the pure POPC system at 313 K is
399
B.5 KC2H Repels Na Ions from and Attracts Cl Ions to the Lipid-Water Interface
Given the positive charge of amino group in KC2H, the electrostatic interactions with both negatively and
positively charged groups (e.g. ions and POPC headgroups) in the system are expected. To study how
KC2H affect the structural properties of the system, we further looked at the ion distribution and POPC
Electron densities for the Cl and Na ions in POPC/KC2H systems simulated at acidic pH and
temperature 313 K are shown in Figure B.6. Adding KC2H to the system redistributes the ions. For the
pure POPC system, the Cl ions density is zero till ca. 1.8 nm from the bilayer center [Figure B.6-a]. It
then increases smoothly and reaches its maximum at ca. 2.9 nm. Finally, it decreases slightly and reaches
a plateau corresponding to a uniform distribution of ions in the aqueous phase. Adding KC2H to the
POPC system populates the hydration layer with Cl ions. The Cl ion density in the hydration layer was
directly correlated with the KC2H concentration in the bilayer. Furthermore, in the presence of KC2H, the
peak corresponding to the maximum Cl ion density were shifted towards the bilayer center, compared to
400
Figure B.6: Effect of KC2H on ion distribution.
Symmetrized electron densities of a) Cl and b) Na ions for systems with different mixing ratios simulated
at pH = 4.4 and T = 313 K. Assuming the bilayers are centered at zero, only the densities for the positive
positions are shown. The X axis shows the direction normal to the bilayer surface.
Na ions [Figure B.6-b] are distributed deeper in the hydration layer, compared to the Cl ions. The
Na density peaks around 1.8 nm. The Na density near the bilayer surface is slightly less than Na density
in aqueous phase. Adding KC2H to POPC repels the Na ions from the bilayer.
The Cl and Na ions electron densities in pure POPC system [Figure B.6] agree with previous
studies on ion distributions in lipid bilayers [15, 16]. Furthermore, Na ions were found to be deeper in the
hydration layer compared to Cl ions, as expected. Figure B.6 suggests that adding KC2H affects the ion
distribution along the bilayer normal. Upon adding KC2H to the system, the Cl ions are attracted towards
the lipid-water interface, while the Na ions are repelled from the interface. This behavior is also consistent
with previous reports on membranes containing cationic lipids [9, 15]. Presence of positively charged
KC2H in the membrane increases the effective positive charge of the membrane surface. This
401
consequently attracts the negatively charged Cl ions and deplete the lipid-water interface from positively
charged Na ions.
The tilt angle for the POPC headgroup (i.e. the angle between the POPC P→N vector and outward normal
to the leaflet) were calculated for each system at 313 K and are shown in Figure B.7. For the pure POPC
simulations, this angle is approximately 70 degrees [Figure B.7], in a close agreement with previous
studies [17-19]. Adding KC2H to the POPC bilayers is shown to affect the distribution of POPC
headgroups. More KC2H in the system resulted in smaller tilt angles so that the POPC choline group
moves towards the aqueous phase. For instance, 30 mol% KC2H reduced this angle from ~ 70 to ~ 43
degrees. KC2, on the other hand, did not have any effect on the average POPC headgroup angle
distribution.
402
Figure B.7: Effect of KC2 and KC2H on POPC P→N vector angle distribution.
POPC headgroup angle distribution for each simulated system at T = 313 K are shown. The X axis shows
the angle between the POPC P→N vector with the outward direction normal to the leaflet. Adding KC2H
to the system reduces the POPC headgroup tilt angle at the acidic conditions but leaves it intact at the
Similar to the change in the ion distributions, the POPC headgroup will also interact with cationic
lipids in the system. These electrostatic interactions will straighten the POPC headgroup and move the
choline group to the aqueous phase [9, 20]. Consistent with what was observed in a study by Gurtovenko
403
et al. [9] on zwitterionic-cationic lipid mixtures, including positively charged lipids in the PC bilayer
shifts the PC headgroup tilt angle to the smaller angles, i.e. make it more perpendicular to the bilayer
surface. The observed changes in the POPC headgroup tilt angle in this study might partially explain why
B.7 Bibliography
[1]. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J., Mondragon-Ramirez,
C., Vorobyov, I., MacKerell Jr, A.D. & Pastor, R. W. (2010). Update of the CHARMM all-atom
additive force field for lipids: Validation on six lipid types. The Journal of Physical Chemistry B,
114(23), 7830-7843.
[2]. Foloppe, N., & MacKerell, Jr, A. D. (2000). All‐atom empirical force field for nucleic acids: I.
Parameter optimization based on small molecule and condensed phase macromolecular target
[3]. Vanommeslaeghe, K., Hatcher, E., Acharya, C., Kundu, S., Zhong, S., Shim, J., Darian, E.,
Guvench, O., Lopes, P., Vorobyov, I. & Mackerell Jr, A. D. (2010). CHARMM general force
field: A force field for drug‐like molecules compatible with the CHARMM all‐atom additive
[4]. Jo, S., Kim, T., Iyer, V. G., & Im, W. (2008). CHARMM‐GUI: A web‐based graphical user
[5]. Huang, L., & Roux, B. (2013). Automated force field parameterization for nonpolarizable and
polarizable atomic models based on ab initio target data. Journal of Chemical Theory and
404
[6]. Vanommeslaeghe, K., & MacKerell Jr, A. D. (2012). Automation of the CHARMM General
Force Field (CGenFF) I: Bond perception and atom typing. Journal of Chemical Information and
[7]. Vanommeslaeghe, K., Raman, E. P., & MacKerell Jr, A. D. (2012). Automation of the
CHARMM General Force Field (CGenFF) II: Assignment of bonded parameters and partial
[8]. Petelska, A. D., & Figaszewski, Z. A. (2000). Effect of pH on the interfacial tension of lipid
[9]. Gurtovenko, A. A., Patra, M., Karttunen, M., & Vattulainen, I. (2004). Cationic DMPC/DMTAP
[10]. Pandit, S. A., Bostick, D., & Berkowitz, M. L. (2003). Molecular dynamics simulation of a
[11]. Pandit, S. A., Bostick, D., & Berkowitz, M. L. (2003). Mixed bilayer containing
[12]. Marrink, S. J., Sok, R. M., & Berendsen, H. J. C. (1996). Free volume properties of a simulated
[13]. Kupiainen, M., Falck, E., Ollila, S., Niemelä, P., Gurtovenko, A.A., Hyvönen, M.T., Patra, M.,
Karttunen, M. & Vattulainen, I. (2005). Free volume properties of sphingomyelin, DMPC, DPPC,
and PLPC bilayers. Journal of Computational and Theoretical Nanoscience, 2(3), 401-413.
[14]. Navas, B.P., Lohner, K., Deutsch, G., Sevcsik, E., Riske, K.A., Dimova, R., Garidel, P. & Pabst,
405
[15]. Gurtovenko, A. A., Miettinen, M., Karttunen, M., & Vattulainen, I. (2005). Effect of monovalent
salt on cationic lipid membranes as revealed by molecular dynamics simulations. The Journal of
[16]. Böckmann, R. A., Hac, A., Heimburg, T., & Grubmüller, H. (2003). Effect of sodium chloride on
[17]. Pluhackova, K., Kirsch, S. A., Han, J., Sun, L., Jiang, Z., Unruh, T., & Böckmann, R. A. (2016).
A critical comparison of biomembrane force fields: Structure and dynamics of model DMPC,
POPC, and POPE bilayers. The Journal of Physical Chemistry B, 120(16), 3888-3903.
[18]. Doux, J. P., Hall, B. A., & Killian, J. A. (2012). How lipid headgroups sense the membrane
phosphatidylcholine headgroup to bilayer surface charge: Torsion angle constraints from dipolar
[20]. Bechinger, B., & Seelig, J. (1991). Interaction of electric dipoles with phospholipid head groups.
[21]. Kulkarni, J.A., Darjuan, M.M., Mercer, J.E., Chen, S., van der Meel, R., Thewalt, J.L., Tam,
Y.Y.C. & Cullis, P. R. (2018). On the formation and morphology of lipid nanoparticles
containing ionizable cationic lipids and siRNA. ACS nano, 12(5), 4787-4795.
406
Appendix C: Supplementary Information for Chapter Five
Anisotropic pressure coupling has been used in many simulations on HII phase [1-3]. In those studies, the
pressure in each direction was treated independently, and both the simulation box size and shape were
allowed to change. This way, system can undergo deformations and form new phases, if energetically
favorable. Assuming the lipid cylinders in HII phase are along the Z axis, HII systems are symmetric in
the XY-plane. Therefore, simulations of HII phase could also be conducted using the semi-isotropic
pressure coupling, where the pressure in Z direction and in XY-plane are treated independently. From a
computational point of view, it is important to evaluate the extent which each pressure coupling could
affect the structural properties of HII systems. All the results presented in Chapter 5 were based on the
simulations conducted with an applied anisotropic pressure coupling, although the shape of simulation
box kept fixed. A few systems in this study were also simulated using semi-isotropic coupling [Table
Figure C.1 shows the dhex for DOPE HII systems calculated from simulations at 303 K using two
different pressure coupling algorithms. The dhex values calculated from two approaches agree with each
other. One exception was the HII system with 26 nw, which the predicted dhex from anisotropic simulation
is higher than the interpolated value (according to the fitted line) from semi-isotropic simulations.
The molecular systems were further inspected visually. For the hydration levels less or near the
maximum hydration (i.e. up to 20 nw), systems simulated using both coupling methods were similar in
their water core shapes at each hydration level. However, using semi-isotropic and anisotropic couplings
caused differences in HII structures at higher hydration levels. For instance, the systems with 26 nw and
407
30 nw simulated using anisotropic coupling were stretched in the Y direction, whereas the system with
To conclude, both semi-isotropic and anisotropic could be used for hydration levels less or near
the maximum hydration. For the higher hydration levels, choosing the appropriate coupling method
408
Figure C.1: Lattice distance as a function of hydration for DOPE HII systems (semi-isotropic
Lattice distances for DOPE HII systems were calculated from simulations conducted at T = 303 K using
either semi-isotropic or anisotropic pressure couplings and results are compared. The red line is the linear
fit to the data from semi-isotropic simulations. The orange box points to several systems simulated using
anisotropic pressure coupling (black circles) which, based on visualization inspection, the water core
shape of these systems tended to deviate or deviated from their cylindrical/hexagonal shape. Error bars
represent the standard deviation of means, and their size are comparable to the symbols.
409
C.2 Bibliography
[1]. Marrink, S. J., & Mark, A. E. (2004). Molecular view of hexagonal phase formation in
[2]. Marrink, S. J., De Vries, A. H., & Mark, A. E. (2004). Coarse grained model for semiquantitative
[3]. Knecht, V., Mark, A. E., & Marrink, S. J. (2006). Phase behavior of a phospholipid/fatty
acid/water mixture studied in atomic detail. Journal of the American Chemical Society, 128(6),
2030-2034.
410
Appendix D: Supplementary Information for Chapter Six
Figure D.1: Dynamics of active site loops and side chains based on the 20 simulations with
411
RMSF of heavy atoms of the indicated residues, excluding backbone, were computed as described in the
Methods. The data are means average deviation of 10 replicates (5 CCT dimer simulations × 2 chains
each) for each condition indicated in the matrix at the top of the figure. The asterisks above each bar
indicate that the value is significantly different (*, P < 0.05; **, P < 0.0005) with reference to the first set
(white bar). The results resemble those for the analyses on 20 simulations using NOE restraints [Figure
6.3] with two exceptions. In the NOE-restrained analysis there was no significant effect of the AI on
R196 or K122 RMSF. Whereas in the PR analyses R196 was significantly restrained by the AI (RMSF
decreased 27%); and K122 showed a ~25% higher RMSF in the absence of AI and CTP. An explanation
for the lower RMSF for K122 in the NOE restrained simulations (Figure 6.3; RMSF = 2.3 0.6 Å) is that
the Ec helices bend to make contact with L2 and this restrains the L2 backbone and K122. This bending
412
Figure D.2: Dynamics of Loop L2 in vitro.
A) Location of F-121 and F-124, sites in L2 for tryptophan substitution. Image is from a simulation with
AI helices, and PR applied to αEC. The residues highlighted in ball and stick form a hydrophobic pocket
for F124. B) Effect of lipid vesicles on the fluorescence anisotropy of tryptophans in loop L2. Purified
CCTs were incubated with or without lipid (Lyso PC/PG (4/1) micelles) for 5 min prior to acquisition of
steady state fluorescence anisotropy for the indicated Trp residue. The data are averages ± S.D. of 5
413
Supplemental text related to Figure D.3.
The data showing that non-glycine substitutions at G123 in L2 inhibited CCT activity spurred an
investigation of the impact of the L2 backbone configuration on contact preferences for the catalytic
residue, K122. We examined the L2 backbone configurations of all 40 simulations (20 with positional
restraints on the EC helix and 20 with NOE restraints) and found facile switching between the and
backbone configurations, which strictly correlated with a switch in the , angles at G123 from glycine-
restricted values to -helical values. We mined the simulations to generate correlation maps between the
dihedrals at K122 and H-bonding partners for K122. We found that the AI helix promoted the
configuration at K122, but only when CTP was absent. With CTP present there was an even distribution
The correlation analysis revealed that the interaction of K122 with CTP was compatible with
either the or configuration at K122 in the absence of the AI; but the AI helix hampered bonds to CTP
when K122 was in the configuration [Figure D.3-B]. The interaction with the backbone trapping atoms
was strongly correlated with the configuration [Figure D.3, C and D]. In PR simulations with the AI
present and where the backbone trap interaction was dominant, extending for >750 ns, the configuration
was highly preferred and was also maintained for >750 ns. This occurred in 8 of 10 chains. In the NOE
simulations with AI present 6 of 10 chains showed continuous backbone traps for >500 ns, and the L2
configuration was β in five and in one of the six. In simulations lacking the AI helix we also found that
the backbone trap correlated strongly with the configuration at K122 [Figure D.3-C].
These data suggest that the configuration at K122 does not preclude an interaction with CTP,
but that the configuration is strongly preferred for trapping by the backbone atoms. This is a reasonable
414
outcome because in the configuration the lysine is directed away from the interior of the active site
We examined individual time plots to further probe whether the configuration prevents K122
interaction with CTP when the AI is present. When CTP and AI were both present, interaction of K122
with the AI was strongly correlated with the configuration, while contact with CTP was only weakly
correlated with the configuration [Figure D.3, B-D]. In 7 of 19 chains with the AI present there were
instances of prolonged contacts with CTP (> 100 ns) while the L2 backbone configuration was ; 12
while the L2 backbone was . Figure D.3-E shows some representative time-plots comparing H-bonding
to CTP and backbone configuration. Even when the AI was present and L2 was in the configuration,
These findings suggest that the conformational switching of the L2 backbone between and
structure does not dictate strict silencing and activating positions for K122; the presence of CTP in the
active site can over-ride the influence of the AI on the L2 backbone configuration. In other words, the
mechanism whereby the AI helix silences catalysis is not primarily by freezing the L2 loop at K122 into a
conformation that prevents it from engaging substrate. An alternative hypothesis, that remains untested, is
that this backbone flexibility simply facilitates K122 sequential access to different partners during a
catalytic cycle, including phosphocholine, which was not present in any of the simulations. The
configuration may be required for K122 to extend all the way to the phosphocholine phosphate, which is
415
Figure D.3: Correlations between K122 backbone dihedrals and H-bonding partners.
A–D) The data from 5 NOE and 5 PR-restrained simulations for each condition described in the matrix at
the top of this figure was mined for an α or βconfiguration at the K122 backbone and corresponding H-
bonding partner for K122. Alpha was defined as Ψ angle < 60o, and β was defined as Ψ angle > 60o. An
H-bond to CTP, F124, or F293 was defined by a distance < 4 Å to any oxygen. A) Frequency of α versus
β configurations for K122. B-D) Frequency of K122 H-bonds to CTP, F124, or F293 while sampling
either the α or β configuration. The data for the NOE set and PR set were averaged to generate the error
bars. E) Time correlations between H-bonds and backbone dihedrals. Upper panels: The Ψ angle for
K122 in the presence of both the AI and CTP are plotted versus simulation time. Representative results
from four separate chains are shown. Lower panels: Corresponding distance between K122 ε-amino and
any CTP phosphate oxygen. The red dashed lines indicate H-bonding distance (2.5 -3.5 Å).
416
Figure D.4: Stability of 4-helix bundle during 1000 ns simulation time.
Snapshot of one replicated simulation at 40 ns (left) and 1000 ns (middle), where both AI helices remain
part of the 4-helix bundle for the duration. Right image is a snapshot of the endpoint of a separate
simulation where the N-terminal portion of AI-2 (yellow) unravels, yet the αE helices remain associated.
The hydrophobic cluster that stabilizes packing of the αE helices at the cross-over is highlighted in ball
and stick (I206, I209, V210). The side chains of the aromatic/hydrophobic residues involved in AI – L2
interactions are highlighted in stick representation (F124, F289, M292, F293). Color coding of chains is
as in previous images.
417
Figure D.5: Alpha-E hinge dynamics enables contacts with the active site.
A) R223 in helix αEC contacts CTP. R223 of chain A αE contacts the -phosphate of CTP in the active
site of chain B, without eliminating the K122 - phosphate contact. The aromatic cluster is captured in
the opposite chain, composed of Y213 and Y216 in the αE and F124 in L2. B) RMSF of the αE hinge.
The residues of the hinge are 211-215, inclusive. Data are means ± Ave Dev of 5 NOE simulations for
each condition (10 chains). P values for the effect of the AI are indicated with ** for P < 0.005, and # for
P ≤ 0.001.
418
Figure D.6: Alpha-E configuration 3.
The αEC helices have dissociated and one is in close contact with L2, the other is not. With few
exceptions this is usually not long-lived and transitions to configuration 2, 4 or 5 within 20 ns.
419
Figure D.7: Lipid vesicles reduce inter-chain cross-linking between αEc.
A separate reaction of CCT312 (A217C) with BMOE was imaged by immunoblot using an antibody
against the catalytic domain, and the signal was quantified using Image Quant 5.2. The data are means
range of 2 determinations.
420
Figure D.8: CCT(A217C) binding to lipid vesicles via domain M is not impaired by oxidative
disulfide formation.
Oxidized CCT-312(A217C) or its Cys-free parent was incubated with (■) or without (◊) 10 mM DTT for
containing egg PC/egg PG/Dansyl-PE (47.5/50/2.5). The increase in fluorescence of Dansyl-PE, due to
FRET from Trp-278 in domain M, was measured with a Varian Eclipse spectrofluorimeter. As a negative
control, CCT-236 with no M domain and a single Trp (W151) in the catalytic domain was tested in the
presence of 10 mM DTT (○, right panel). The excitation wavelength was 280 nm and the emission was
scanned between 294 and 550 nm. FRET denotes the fluorescence increase at 520 nm, calculated as the
fractional increase relative to the Dansyl-PE-containing vesicles in the absence of CCT, where
fluorescence of vesicles alone = 0. Data are means range of 2 determinations; lines drawn to guide the
eye.
421
Figure D.9: PR vs NOE restraints.
The heavy backbone atoms in the αEC of both monomers (residues 216 to 223 inclusive, colored blue and
red) were treated either with PR or NOE restraints. PR restraints pin these atoms in space (with respect to
the coordinate system origin), which maintains the helicity of the segments and fixes the inter-helical
distances for the duration of the simulation. The rest of the enzyme can fluctuate freely. The NOE
restraints can be imagined as a group of virtual bonds between the heavy backbone atoms in each αE C
segment that maintain the helicity of each independently from the other αEC segment. Consequently, the
422
Figure D.10: Detailed description of CCT mutant constructs.
The length of the CCT mutant constructs and the locations of the His-tag are displayed. The green box
(residues 40-223) indicates the region corresponding to the resolved segment in the X-ray structure
(PDB:4MVC). The positions of the mutated residues are indicated by red bars. For the single tryptophan
and single cysteine mutants the sites of the native Trps and Cys that were mutated to Phe or Ser are also
indicated. The deleted NLS sequence (residues 12-16) served to prevent vesicle aggregation.
423
Appendix E: Copyright Permissions
As the first author of this manuscript, no permission was required for reusing it in this thesis.
424
Permission from the journal of Biochimica et Biophysica Acta (BBA) - Biomembranes:
Kulkarni, J. A., Darjuan, M. M., Mercer, J. E., Chen, S., van der Meel, R., Thewalt, J. L., Tam Y. Y. C. &
Cullis, P. R. (2018). On the Formation and Morphology of Lipid Nanoparticles Containing Ionizable
425
426
Permission for Figure 1.4-left:
Tam, Y. Y. C., Chen, S., & Cullis, P. R. (2013). Advances in lipid nanoparticles for siRNA
Copyright permissions are owned by the authors and this figure was reproduced (adapted) with
permission from Dr. Yuen Yi C. Tam, Dr. Sam Chen, and Dr. Pieter R. Cullis.
427
Permission from co-authors/collaborators to include our study as the Chapter 3 of this thesis:
428
Permission from co-authors/collaborators to include our study as the Chapter 4 of this thesis:
429
Permission from co-authors/collaborators to include our study as the Chapter 5 of this thesis:
430
Permission from co-authors/collaborators to include our study as the Chapter 6 of this thesis:
431