Atomic coordination reflects peptide immunogenicity

Georgios S. Antipas1, 2*, Anastasios E. Germenis3, 2

1

Division of Materials Technology, National Technical University of Athens, Greece,

Department of Molecular Medicine, Hellenic Pasteur Institute, Greece, 3School of

Medicine, University of Thessaly, Greece
Submitted to Journal:
Frontiers in Molecular Biosciences
Specialty Section:
Mathematics of Biomolecules
ISSN:
2296-889X
Article type:
Original Research Article

s
i
v
o
r
P
Received on:
06 Oct 2015
Accepted on:
19 Dec 2015

Provisional PDF published on:

19 Dec 2015

l
a
n
io

Frontiers website link:

www.frontiersin.org

Citation:
Antipas GS and Germenis AE(2015) Atomic coordination reflects peptide immunogenicity. Front. Mol.
Biosci. 2:77. doi:10.3389/fmolb.2015.00077
Copyright statement:
2015 Antipas and Germenis. This is an open-access article distributed under the terms of the
Creative Commons Attribution License (CC BY). The use, distribution and reproduction in other
forums is permitted, provided the original author(s) or licensor are credited and that the original
publication in this journal is cited, in accordance with accepted academic practice. No use,
distribution or reproduction is permitted which does not comply with these terms.

This Provisional PDF corresponds to the article as it appeared upon acceptance, after peer-review. Fully formatted PDF
and full text (HTML) versions will be made available soon.

Frontiers in Molecular Biosciences | www.frontiersin.org

Atomic coordination reflects peptide immunogenicity

G.S.E. Antipas1,3 and A.E. Germenis2,3
1

Division of Materials Technology, National Technical University of Athens, Zografou Campus,

Athens 15780, Greece, Tel. +30 210 7722037, Email: [email protected]
2
Department of Immunology & Histocompatibility, School of Medicine, University of Thessaly,
Biopolis, Larissa 41110, Greece, Tel. +30 2410 685728, Email: [email protected]
3
Department of Molecular Medicine, Hellenic Pasteur Institute, 127 Vasilissis Sofias Avenue, Athens
11521, Greece

Abstract
We demonstrated that the immunological identity of variant peptides may be accurately
predicted on the basis of atomic coordination of both unprotonated and protonated tertiary
structures, provided that the structure of the native peptide (index) is known. The metric
which was discovered to account for this discrimination is the coordination difference
between the variant and the index; we also showed that increasing coordination difference in
respect to the index was correlated to a correspondingly weakening immunological outcome
of the variant. Additionally, we established that this metric quickly seizes to operate beyond
the peptide scale, e.g. over a coordination shell inclusive of atoms up to a distance of 7
away from the peptide or over the entire pMHC-TCR complex. Analysis of molecular orbital
interactions over a range of formal charges further revealed that the N-terminus of the
agonists was always able to sustain a stable ammonium (NH3+) group which was consistently
absent in antagonists. We deem that the presence of NH3+ constitutes a secondary observable
with a biological consequence, signifying a change in T cell activation. While our analysis of
protonated structures relied on the quantum chemical relaxation of the H species, the results
were consistent over a wide range of peptide charge and spin polarization conditions.

o
r
P

o
i
s
i
v

l
a
n

Keywords
pMHC-TCR interaction; atomic pair correlation; short range order; cumulative coordination;
functional avidity; structure-function relationship

Introduction
The original work on the characterization of the Class I Tax antigen along with three,
artificially synthesized, of its variants which were declared as biologically diverse while
stereochemically similar1 set the stage for the quantification of the control exerted by
peptide tertiary structure on the synapse formed by a peptide-Major Histocompatibility
Complex (pMHC) from a T cell receptor (TCR). However, the passage of the next decades
would reveal that the causal condition operating on the pMHC-TCR structure-function
relationship would remain alarmingly elusive. This is largely due to the appreciable span of
length scales between three and four orders of magnitude which separates the molecular
level of the immune response from any plausible fundamental mechanism operating on the
atomic or electronic level. Over the same period, a substantial amount of research was
devoted to the phenomenology involved in the immune synapse2,3, with particular emphasis

on peptide anchorage inside the MHC4 as well as on the issue of docking of the TCR with the
pMHC complex and its subsequent thermodynamic stability5,6, both of which were discussed
within the premise of binding energetics. However, predicting a biological outcome on the
basis of p-MHC binding energy or pMHC-TCR free energy has never been proven as
consistently possible.
Previously we suggested that biological function may not be predicted based on
energetics7,8, as all bonded (and indeed, certain non-bonded) interactions occur within the first
coordination shell of short range order; in the same work we argued that, due to the
uniformity of bond lengths (which are directly correlated to bond energies) across the entirety
of protein tertiary structure, binding energetics will tend to be degenerate. Here, we will
exemplify this claim for the case of the Tax antigen9 - a transcriptional regulatory protein of
the human T-cell leukemia virus posing an attractive target for anti-cancer vaccine
development10 due to its critical role in HTLV-1-associated leukemogenesis11-13 - and three of
its artificially synthesized variants1 by additionally showing that coordination carries the
signature of peptide immunological identity. The direct relationship between coordination
and peptide function will, furthermore, be shown to constitute a physical observable of
biological function in both, unprotonated and protonated peptide tertiary structure.

Materials and methods

l
a
n

Peptides
The antigens used were the cognate HTLV-1 Tax peptide (LLFGYPVYV, PDB entry 1AO7)
(index peptide), the weak agonist (or null peptide) V7R (LLFGYPRYV, PDB entry 1QSE), the
weak antagonist Y8A (LLFGYPVAV, Protein data bank PDB entry 1QSF) and the
antagonist P6A (LLFGYAVYV, PDB entry 1QRN). All complexes were presented by HLAA201 and were bound to the human A6TCR1; furthermore, all complexes have been
functionally characterized by cell assays as well as by kinetic and thermodynamic
measurements1. In alignment to previous work of ours7,8, the current analysis has been
performed in the gas phase. The assumption of absence of water molecules from the cleft is
reasonable on the basis of a reported entropic advantage14.

o
r
P

o
i
s
i
v

Peptide protonation and the induction of polarization

Protein structures resolved by XRD are unprotonated as their H species is not detectable by X
rays due to the low form factor of the hydrogen atom15. In principle, in-silico peptide
protonation requires calculation of the interaction energies between pairs of charged residues
and the average protonation of each residue must be determined from the electrostatic
energies. For a characteristic Monte Carlo solution of the problem of residue protonation, the
reader is referred to early work by Beroza et al16. Here, peptide backbone and side chain
bonds on PDB structures were saturated by hydrogen atoms, followed by ab initio relaxation
of the H species. During H relaxation, the C, N and O species were kept immobile in their invitro crystallized positions. We also took into account the possibility of charge and spin on
peptide structure by calculating the electronic structure for a range of allowed spin
multiplicities up to the quintet state. A complete set of charge and spin polarization conditions
is given in our precursor work7,8. All peptides were assumed to experience a neutral pH;
furthermore, all protonated structures were initially zwitterionic and two different sets of ab
initio calculations were carried out. In the first set, all peptides were fully protonated
zwitterions. In the second set, the hydroxyl functional groups on the hydrophilic phenol rings
were deprotonated.

Ab initio calculations
All-electron, spin unrestricted DFT calculations were performed with the Amsterdam density
functional (ADF) program17,18. Electron exchange and correlation was initially addressed by
the BLYP19,20 functional within the generalized gradient approximation (GGA). Singleelectron wavefunctions were expanded using the TZ2P uncontracted Slater-type orbital (STO)
basis set, (a triple- basis set with two sets of polarization functions) for all atoms. Nonaufbau occupations were discarded.
Calculation of pair correlation functions
Coordination refers to atomic structure and expresses the tendency of atoms to be surrounded
by other atoms (the coordination number indicates the number of atoms surrounding a
reference atom). The calculation of atomic coordination is based on the initial formulation of
a histogram of interatomic distances, via calculation of the distances between all atom pairs
and the assignment of these distances to bins of a predefined size (e.g. 0.1 ; choice of the
most appropriate bin size is a matter of experimentation7,8,21-23 but does not affect the results if
chosen to be sufficiently small). By convention, the ith bin is assigned all interatomic
distances, R, within the range r R < r+r, where r = i*r and r is the bin size. For
example, the 10th bin is assigned all interatomic distances 1.0 R < 1.1 , the 11th bin
includes all distances for which 1.1 R < 1.2 , etc. The number of distances in each bin
then represents the value of the histogram in that bin. If all atom species are considered the
histogram is representative of the total coordination, whereas if calculations are restricted to
specific atom species the histogram represents a partial. In the current study, both the total
and all of the partial histograms were calculated for the peptide structures. Numerical
manipulation of the histogram yields two important pair correlation functions which lead up
to the atomic coordination number: the Pair Distribution Function (PDF), also symbolised as
g(r), and the Radial Distribution Function (RDF), symbolised as R(r).
The PDF is a statistical representation of interatomic distances24. The PDF was
calculated by initially constructing the histogram of interatomic distances in respect to the real
space coordinate, r. Calculation of the histogram involved the partition of space into bins,
with a bin size equal to 0.1 . The PDF is expressed as
N N
1
g (r )
(1)
(r rij )
2Nr 2 0 j 1 i j

o
r
P

o
i
s
i
v

l
a
n

where N is the number of peptide atoms, is the Dirac delta function and 0 is the number
density N/V, where V is the volume of the simulation box containing the peptide and rij is the
distance between the ith and jth atoms. The RDF was then calculated as
R(r) = 4r20g(r)

(2)

and integrated to estimate atomic coordination, n, of any atom within a spherical shell defined
by interatomic distances r1 and r2 as follows
r2

r2

n rr12 R(r)dr 40 g(r)r 2dr

r1

r1

In the latter expression, the cumulative coordination for each peptide up to any value of
interatomic distance r2 may also be computed by setting r1 equal to zero. Similarly, the
running difference between any pair of such cumulative coordination integrals may be
calculated either as total coordination (i.e. disregarding atom species) or as partial

(3)

coordination (i.e. for selected atom species pairs). All calculations of PDF, RDF and
coordination were performed with the PRDF software 25,26.

Results
Unprotonated peptide structures
Unprotonated peptide structures appeared similar, as judged by the distances between alpha
carbon atoms (C) on the N- and C-terminus residues (see Fig.1.), all distances being of the
order of 22 . Further comparison of the C distances revealed that antagonist, P6A, was
marginally more confined than the agonist, Tax. The increased confinement of the P6A
peptide was also reflected on its elevated density (equal to 0.309146 g/cm3, number density:
0.014444 atoms/3) as compared to that of Tax (density: 0.302837 g/cm3, number density:
0.014175 atoms/3). Also compared to Tax, the increased density of the weak agonist, V7R,
(equal to 0.317370 g/cm3, number density: 0.014818 atoms/3), and the low density of the
weak antagonist, Y8A, (density: 0.262836 g/cm3, number density: 0.012275: atoms/3)
indicated respective over- and under-coordination, the latter probably reflecting more
fundamental differences of backbone/sidechain conjugation27.
The total and partial PDF of unprotonated peptide structures are shown in Fig.2a to 2d.
The peak positions in every PDF curve underpinned our previously discussed theme of bond
length uniformity within the first coordination shell7,8, the latter extending up to
approximately 1.6 as indicated by the total PDF datasets. More precisely, the first
coordination shell primarily comprised C-C and C-N bonded interactions, which were
respectively 1.5 and 1.3 in length, regardless of the peptide. In fact, the consistency in the
lengths of the C-C and C-N partials was characteristic of the entire pMHC-TCR complexes,
as portrayed in Fig.2e to 2h. Short range order features were visible up to 4 , altogether
faded beyond approximately 5 to 7 and this was common to the structures of both the
isolated peptide and the entire pMHC-TCR complex. Within the realm of short range order,
the average coordination, number density and density of atomic clusters centered on each of
the C, N and O species were also studied, with the aim of highlighting features which would
be prominently indicative of agonist-antagonist discrimination. Two types of structures were
considered: isolated peptides, the results for which are shown in Fig.3a to 3d, and the peptide
inclusive of its cleft environment to within 7 from each peptide atom, shown in Fig.3e to
3h. The results indicated that average cluster coordination, density and number density were
always higher on the variants in comparison to Tax; therefore peptide biological function may
not be readily discerned on the basis of the first coordination shell. Perhaps surprisingly, the
cleft environment to within 7 from peptide atoms was also found to be devoid a trend
signifying a clear structure-function relationship.
Previously we established that cumulative coordination difference of the protonated
variant peptides in respect to the index yields a direct correlation to immunological
identity7,8,21,22 and, more precisely, that cumulative under-coordination is characteristic of the
antagonist; at this point we inquire if this might additionally stand true for unprotonated
structures and the results are shown in Fig.4, for the cases of the isolated peptides, the cleft
environment up to 7 from each peptide and the entire pMHC-TCR complex. Both the total
PDF (Fig.4a) and the C-C partial (Fig.4b) were found to be correlated to peptide
immunological identity (i.e. antagonist peptides were under-coordinated in respect to Tax), in
agreement with our previous results of protonated structures; in contrast, he rest of the PDF
partials (Fig.4c to 4f) were not relevant. Furthermore, the total PDF was not found to reflect
function either on the scale of the immediate environment of the peptide see Fig. 4g and 4h)
or in the premise of the entire pMHC-TCR complex (see Fig.4i and 4j). Therefore, we

o
r
P

o
i
s
i
v

l
a
n

envisage that the physical observable of pMHC-TCR functional avidity is closely associated
with the peptide scale.
Peptide protonation
Protonated tertiary structures were subjected to various deprotonation combinations of their
hydrophilic side chain hydroxyl groups8 (see the groups marked by cyan arrows in Fig.5a), in
order to cater for the possibility of excessive alkalinity experienced by the side chain in the
TCR micro environment. The compounded effect of the different deprotonation/charge/spin
combinations as reflected on the binding energy of the peptides is shown in Fig.5b to 5e. On
the basis of total binding energy (Fig.5b), the models of the weakly interacting variants (Y8A
and V7R) were observed to lie in the lowest and highest energy positions, respectively.
Unfortunately, between the two energy extremes, there was pronounced overlap among the
binding energies of the agonist and the antagonist. Moreover, the energy constituents
(electrostatic, Pauli and orbital, shown in Fig. 5c to 5e respectively) did not reveal a solid
trend in regard to peptide functionality and, in fact, confirmed that the binding energies of the
agonist and antagonist models were degenerate; due to these factors peptide binding energy
was rendered unsuitable as a criterion of functional avidity. We, thus, proceeded to examine
the pair distribution function of peptide protonated structures.
The only meaningful differences between the PDF curves of the unprotonated and
protonated tertiary structures comprised partials of the H species and these are shown in Fig.6,
for the case of spin unpolarized peptides. Principal contributions to the total PDF were from
the C-H partial, which also exhibited the most varied behaviour in respect to peaks beyond the
first coordination shell; more specifically, V7R (Fig.6b) was the only structure showing a
peak at 1.9 while P6A (Fig.6d) exhibited a peak at 2.8 instead of 2.7 shown by the rest of
the structures, a sign of lower second-shell coordination (longer interatomic distances). Aside
from the value cast on the total PDF (and its C-C partial) of unprotonated peptide structures,
the protonated structures of the Tax and its variant peptides have already been shown to
reflect pMHC-TCR functional avidity7,8 through the total PDF (also see Fig.7a). Our current
results indicate that primarily the C-H (Fig.7b) and secondarily the H-H partials were also
correlated to functional avidity, while the rest of the H partials were not (see Fig.7c to 7e).

o
r
P

o
i
s
i
v

l
a
n

Contribution of residue interactions towards coordination differences

As already indicated by the total PDFs of both unprotonated and protonated peptide structures
(see Fig.4a and Fig.7a), coordination deviations in respect to Tax were correlated to peptide
immunological identity at interatomic distances beyond 5 and up to the limit of short range
order at approximately 7 . Within this range we proceeded to calculate the coordination,
R(r)dr, between backbone and side chain atoms for each of the three partials which were most
actively implicated in peptide functionality, i.e. C-C, C-H and H-H. The results indicated that
a very specific subset of residue interactions was always responsible for coordination
differences and we, thus, present in Fig.8 the case of non-polarized peptides without loss of
generality. As may be deduced by comparison of Fig.8a, 8d and 8g, by far the highest
contributors to peptide coordination deviation from Tax were the side chain atoms of the
substituted residues, i.e. residues 6, 7 and 8 for P6A, V7R and Y8A respectively. P6A undercoordination was underpinned by C-C interactions contributed solely by residue 6 side chain
atoms (6S) at a substantial 81%; the corresponding C-C percentages for Y8A and V7R were
77% and 64%, respectively. Of equal importance were coordination deviation contributions
by side chain C-H partials on substituted residues. These were 85%, 73% and 69% for P6A,
Y8A and V7R, respectively. In some contrast, the H-H partial contributions were not as
important, since the Y8A 8S contribution (see Fig.8i) was only 39% towards the peptides
total under-coordination in respect to Tax (these contributions for P6A and V7R were high, at
79% and 69% respectively).
5

Quantum descriptors
It is reasonable to expect that there may be potential for relationships to be drawn between the
topology of the charge density and coordination. Accordingly, the Laplacian of the electron
density for non-polarized peptide structures, presented in Fig.9, revealed a rather counterintuitive similarity among agonists and antagonists; more specifically, it appeared that all
peptide backbones and triggering side chains were loci of high electron kinetic energy (charge
depletion) while hydrophobic side chains remained wrapped within regions of high potential
energy (charge concentration). This uniformity of motifs across peptides did not allow any
consistent conclusions to be drawn in regard to peptide functionality; hence a spectrum of
formal charge and spin polarization combinations were introduced on the tertiary peptide
structures, with the aim of detecting structural and/or electronic motifs consistent with
immunological function. This analysis also considered the possibility of deprotonation of the
terminal (hydroxyl) groups residing on the hydrophilic side chains.
The introduction of spin polarization produced a very consistent difference in the
structures of the agonist vs. that of the agonist peptides: all agonists maintained a zwitterionic
state while antagonists did not and this behaviour was observed across all the charge/spin
combinations studied. A select such case is presented in Fig.10, in which we compare
peptides of similar charges and spin polarizations. Regardless of their functionality, all
peptides redistributed electron density such that there was an outflow of charge from
backbone atoms and an inflow of net charge into side chain (mostly O species) atoms, as
shown by the Voronoi deformation density (VDD) topology. However, spin density topology
(shown as iso-density surfaces in Fig.10) was decidedly different among peptides of different
functionalities and was accompanied by formation of an ammonium (NH3+) group particular
only to the Tax and V7R agonists. Although spin density was delocalized over all triggering
side chains, it was manifested as a lobe encapsulating the N terminus H atoms of the Tax
peptide (see Fig.10a), while it always engulfed the H atom abstracted from the N terminus of
the antagonist structures (see Fig.10 c and d). However, spin density was not delocalized over
the N terminus of weak antagonist V7R. N terminus atomic orbital (AO) (sum of a1g and t1u)
contributions towards formation of molecular orbitals are shown in Fig.11. Again, our outlook
for the analysis of AO contributions across the four peptides was qualitative; the data raised
two features of interest. The first was that both antagonist peptides (Fig.11c and 11d)
comprised valence molecular orbitals (MO) primarily made up by majority (spin up) H and
minority (spin down) contributions towards formation of the highest-lying frontier orbitals
(i.e. Highest Occupied Molecular Orbital HOMO, as well as HOMO-1). The second, and
perhaps most prominent, feature was that the native peptide, Tax, contributed its N terminus
H density almost exclusively towards formation of the structures low-lying Lowest Occupied
Molecular Orbital (LUMO) see H contribution at 0.0048 a.u. in Fig11a; this was in direct
contrast to the three other peptides compared and was consistent across all spin polarized
models examined in the case of Tax.

o
r
P

o
i
s
i
v

l
a
n

Discussion
PDF fluctuations gradually tailed off with increasing values of the real space coordinate, r. In
practice, no appreciable structural features existed beyond 5 and were certainly absent
beyond 7 (see Fig.2 for unprotonated and Fig.6 for protonated structure PDFs). Also, from
the RDF expression (see eq.2) it may be seen that the term r2 will naturally assign greater
weight onto larger interatomic spacings. Broadly speaking, 5 is the scale of single residues,
hence the RDF suggested that any interatomic interactions that are to be instrumental for
peptide coordination are to be found in the range of 5 to 7 . In the case of the peptide
environment inclusive of structure within a radius of 7 from peptide atoms (see Fig.3e
through to 3h), coordination differences among peptides ranged up to 5% (between the Tax
6

and P6A) while in the case of the entire complexes (Fig.3a through to 3d) these differences
were negligible. Accordingly, we are inclined to propose that pMHC-TCR functional avidity
is not reflected on scales beyond that of the peptide. Moreover, the energetics involved in
coordination did not present a singularly meaningful trend even on the scale of single
peptides, as seen by the overlapping binding energies shown in Fig.5b. The extensive
similarity among partials over scales ranging from the peptide up to the pMHC-TCR
complex, shown in Fig.2, should be considered as supportive of the theme of first
coordination shell energy degeneracy, raised by us previously8.
One of the most important findings of the current work was the correlation between
the total PDF of unprotonated tertiary structures and pMHC-TCR functional avidity. This
relation, which was expressed by peptide cumulative coordination differences from the index,
is depicted in Fig.4. Coordination based on the total PDF was underpinned primarily by the
C-C partial (Fig.4b) with contributions from the N-N partial (Fig.4e), the latter being
particularly relevant to the weak agonist, V7R. Cumulative coordination based on the total
PDF of unprotonated structures (the utility of the latter has been raised in our previous
work7,8) was in full agreement with coordination based on the protonated models (Fig.7a).
Interestingly, coordination differences within the short range limit of the peptides (i.e. up to a
radial distance of 7 ) were in agreement with functional avidity measurements (see Fig.4h)
but they seized to be of relevance on the scale of the entire pMHC-TCR complex (Fig.4i and
4j). As already mentioned, along with C-C, both of the C-H and H-H partials (Fig.7b and 7e
respectively) were correlated to peptide functionality and this correlation was controlled by
the conformation of side chains on the substituted residues. It is notable that, in the case of the
agonists, P6A and V7R, the side chains of the substituted residues were hydrophobic (buried
in the MHC). Since coordination deviations arise mainly from hydrophobic side chains,
sufficient rigidity of hydrophobic portions8,14,28 would favor the possibility that peptide
agonistic potential is inborn to the pMHC complex upon its conformation and its presentation
on the surface of the antigen presenting cell and, hence, independent of the TCR.

o
r
P

o
i
s
i
v

l
a
n

Conclusions
The main points established to determine the link between selective agonist coordination and
peptide immunological identity are:
1. The structural expression of agonism occurs over interatomic distances within the range of
5-7 , i.e. beyond the typical length of the residue; we therefore nominate this length as
the interaction limit below which peptide immunological identity may not be defined. Any
bonded interaction occurring over interatomic distances smaller than this critical scale will
be reflective of coordination shell degeneracy, a term we introduced to describe the
uncorrelated behaviour between peptide immunological identity and structure.
2. It is unclear to which degree peptide tertiary structure is shaped by structural adaptations
made by both the pMHC and the TCR, during formation of the immune synapse.
However, as most of the peptide tertiary structure is hydrophobic, the correlation of
immunological identity to interatomic distances beyond a critical 5-7 suggests that the
structural expression of immunological identity (i.e. selective agonist coordination or the
absence of it) must be in place already upon presentation of the immunocompetent peptide
by the MHC, on the provision that the MHC cleft does not undergo structural adaptations
larger than 5-7 during antigen recognition by the TCR.
3. Weak agonist and weak antagonist peptides are, respectively, markedly over- and undercoordinated in comparison to the native peptide. Hence the coordination requirement for
an antagonist peptide is, indeed, very close to that of the agonist, albeit the antagonist is

most categorically under-coordinated. This selective agonist coordination was found to be

reflected upon the C-C, C-H and H-H PDF partials.
4. Despite other structural similarities, the electronic structure of an agonist vs. that of an
antagonist peptide is profoundly different. We determined that selective agonist
coordination could be explained on the basis of specific charge and spin polarization
combinations acting on peptide electronic structure in the gas phase. In the premise of
polarization, antagonists were always found to maintain a stable ammonium group on
their N-termini, which was altogether inaccessible to antagonists. Moreover, the density of
states revealed that the index (Tax) contributed the bulk of the electron density associated
with the H species of its ammonium group towards the peptides LUMO.
Author contributions
G.A. designed and performed the analysis and wrote the main body of the paper. Both authors
discussed the results and commented on the manuscript at all stages.
Competing financial interests
The authors declare no competing financial interests.

l
a
n

Acknowledgments
The authors would like to thank Professor M. Manousakis, Director General, Hellenic Pasteur
Institute, Athens, Greece for his support.

o
i
s
i
v

References
1
Ding, Y.-H., Baker, B. M., Garboczi, D. N., Biddison, W. E. & Wiley, D. C. Four A6TCR/Peptide/HLA-A2 Structures that Generate Very Different T Cell Signals Are
Nearly Identical. Immunity 11, 45-56 (1999).
2
van der Merwe, P. A. The TCR triggering puzzle. Immunity 14, 665-668,
doi:10.1016/S1074-7613(01)00155-8 (2001).
3
van Der Merwe, P. A. & Davis, S. J. Vol. 21 659-684 (2003).
4
Agudelo, W. A. & Patarroyo, M. E. Quantum Chemical Analysis of MHC-Peptide
Interactions for Vaccine Design. Mini-Reviews in Medicinal Chemistry 10, 746-758,
doi:10.2174/138955710791572488 (2010).
5
Wan, S., Coveney, P. V. & Flower, D. R. Molecular Basis of Peptide Recognition by
the TCR: Affinity Differences Calculated Using Large Scale Computing. Journal of
Immunology 175, 1715-1723 (2005).
6
Wan, S., Flower, D. R. & Coveney, P. V. Toward an atomistic understanding of the
immune synapse: large-scale molecular dynamics simulation of a membraneembedded TCR-pMHC-CD4 complex. Molecular immunology 45, 1221-1230,
doi:10.1016/j.molimm.2007.09.022 (2008).
7
Antipas, G. S. E. & Germenis, A. E. The quantum chemical causality of pMHC-TCR
biological avidity: peptide atomic coordination data and the electronic state of agonist
N termini. Data In Brief 3, 180-184, doi:10.1016/j.dib.2015.02.021 (2015).
8
Antipas, G. S. E. & Germenis, A. E. Quantum chemical calculations predict biological
function: The case of T cell receptor interaction with a peptide/MHC class I. Frontiers
in Chemistry 3, doi:10.3389/fchem.2015.00009 (2015).
9
Tanaka, Y. et al. Single-cell analysis of T-cell receptor repertoire of HTLV-1 Taxspecific cytotoxic T cells in allogeneic transplant recipients with adult T-cell
leukemia/lymphoma. Cancer research 70, 6181-6192, doi:10.1158/0008-5472.CAN10-0678 (2010).

o
r
P

10

11

12

13

14

15
16

17
18
19
20

21

22

23
24

25
26
27

28

Sundaram, R. et al. A novel multivalent human CTL peptide construct elicits robust
cellular immune responses in HLA-A*0201 transgenic mice: implications for HTLV-1
vaccine design. Vaccine 21, 2767-2781 (2003).
Kannagi, M. et al. Predominant recognition of human T cell leukemia virus type I
(HTLV-I) pX gene products by human CD8+ cytotoxic T cells directed against
HTLV-I-infected cells. International immunology 3, 761-767 (1991).
Pique, C. et al. Evidence for the chronic in vivo production of human T cell leukemia
virus type I Rof and Tof proteins from cytotoxic T lymphocytes directed against viral
peptides. The Journal of experimental medicine 191, 567-572 (2000).
Elovaara, I. et al. High human T cell lymphotropic virus type 1 (HTLV-1)-specific
precursor cytotoxic T lymphocyte frequencies in patients with HTLV-1-associated
neurological disease. The Journal of experimental medicine 177, 1567-1573 (1993).
Schamel, W. W. & Reth, M. The TCR binding site does move. Proceedings of the
National Academy of Sciences of the United States of America 104, 16398-16399,
doi:10.1073/pnas.0708462104 (2007).
Stojilovic, N. Why Cant We See Hydrogen in X-ray Photoelectron Spectroscopy?
Journal of Chemical Education 89, 1331-1332, doi:10.1021/ed300057j (2012).
Beroza, P., Fredkin, D., Okamura, M. & Feher, G. Protonation of interacting residues
in a protein by a Monte Carlo method: application to lysozyme and the photosynthetic
reaction center of Rhodobacter sphaeroides. Proceedings of the National Academy of
Sciences 88, 5804-5808 (1991).
Amsterdam Density Functional (ADF) program, Theoretical Chemistry,
(http://www.scm.com/ADF/), Vrije Universiteit, Amsterdam (2014).
te Velde, G. et al. Chemistry with ADF. Journal of Computational Chemistry 22, 931967, doi:10.1002/jcc.1056 (2001).
Becke, A. D. Density-functional exchange-energy approximation with correct
asymptotic behavior. Physical Review A 38, 3098-3100 (1988).
Lee, C., Yang, W. & Parr, R. G. Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density. Physical Review B 37, 785-789,
doi:10.1103/PhysRevB.37.785 (1988).
Antipas, G. S. E. & Germenis, A. E. The coordination of unprotonated peptide tertiary
structure as a metric of pMHCTCR functional avidity. Data In Brief 5, 342-347,
doi:10.1016/j.dib.2015.09.009 (2015).
Antipas, G. S. E. & Germenis, A. E. Human Cytomegalovirus variant peptides adapt
by decreasing their total coordination upon binding to a T cell receptor. Data In Brief
4, 492-499, doi:10.1016/j.dib.2015.07.019 (2015).
Antipas, G. S. E. & Germenis, A. E. Physical observables and quantum causality of
peptide immunological identity. (submitted) (2015).
Antipas, G. S. E. et al. A containerless study of short-range order in high-temperature
FeSiAlCaMgCrCuNi oxide systems. Journal of Molecular Structure 1019,
151-158, doi:10.1016/j.molstruc.2012.03.056 (2012).
PRDF: software for the calculation of atomic pair correlation and short-range
ordering, (http://users.ntua.gr/gantipas/prdf/ ), NTUA, Athens (2014).
Antipas, G. S. E. PRDF: navigating the amorphous short-range order. Annals of
Materials Research, article ref. 201511121346 (2015).
Li, Y., Yang, Y., He, P. & Yang, Q. QM/MM study of epitope peptides binding to
HLA-A*0201: the roles of anchor residues and water. Chemical biology & drug
design 74, 611-618, doi:10.1111/j.1747-0285.2009.00896.x (2009).
Gakamsky, D. M., Luescher, I. F. & Pecht, I. T cell receptor-ligand interactions: a
conformational preequilibrium or an induced fit. Proceedings of the National

o
r
P

o
i
s
i
v

l
a
n

Academy of Sciences of the United States of America 101, 9063-9066,

doi:10.1073/pnas.0402840101 (2004).

o
r
P

o
i
s
i
v

10

l
a
n

Figure captions
Figure 1. (a) to (d): unprotonated structures of the Tax, V7R, Y8A and P6A, respectively. The
TCR alpha and beta chains (not shown) are located underneath the peptides while the MHC
alpha chain (also not shown) is located over the peptide and encapsulates it. In all peptides,
hydroxyl groups attached to the phenyl side chain of residue 5 point towards the alpha chain
of the TCR. In every structure the distance in between the alpha carbon (C) atoms of the
N- and C-terminus residues are also shown. Atom colour notation is C grey, N blue and O
red and is followed throughout. Peptide mutations in respect to Tax are highlighted in
purple.
Figure 2. Unprotonated structure total and partial PDF curves (symbolized as g(r) and gA-B,
respectively where A and B are any of the C, N or O species). From a) to d): single
peptides Tax, V7R, Y8A and P6A, respectively. From e) to h): entire pMHC-TCR complexes
of the Tax, V7R, Y8A and P6A peptides, respectively. All interatomic distances have been
rounded up to the first decimal digit. Each of the partial curves has been normalized by the
average number density of the peptide.

l
a
n

Figure 3. Short range order statistics (total coordination number, CN, number density and
density, up to interatomic distances of 4 ) for unprotonated pMHC-TCR complexes of a)
Tax, b) V7R, c) Y8A and d) P6A. Also, short range order statistics for unprotonated complex
environments including atoms up to a distance of 7 from peptide atoms: e) Tax, f) V7R, g)
Y8A and h) P6A. Statistics are shown for the C, N and O species as differently coloured bars.
Weighted averages for each atom are shown in parentheses and are defined as (xiNi)/Ni,
where xi is the ith PDF bin and Ni is the histogram value.

o
r
P

o
i
s
i
v

Figure 4. Cumulative coordination differences of unprotonated structures in respect to Tax.

From a) to f) single peptides, g) to h) peptides and a surrounding shell inclusive of atoms up
to 7 from each of the peptide atoms and i) to j) the entire pMHC-TCR complexes.

Figure 5. a) The protonated Tax structure in which H atoms are shown in white colour. We
deemed that each of the protonated peptides could potentially give rise to two stereochemical
variants: one in which the hydroxyl groups attached to phenol side chains would be fully
protonated and one in which the hydroxyls would be deprotonated; these stereochemical
variants are referred to as p and u respectively throughout the text. b) Total binding
energies of a range of protonated peptides, sorted in ascending energy order. The
decomposition of binding energy into its constituent electrostatic, Pauli repulsion and orbital
interaction parts is shown in c) to e), respectively.
Figure 6. PDF diagrams of the H species partials for protonated, spin unpolarized peptide
models: a) Tax, b) V7R, c) Y8A and d) P6A.
Figure 7. Cumulative coordination differences of the H species partials for protonated
peptide structures.
Figure 8. Coordination, R(r)dr, per type of interaction (vertical axes) for the C-C, C-H and HH partials of non-polarized peptide structures. All graphs include contributions by Tax for the
sake of comparison: (a) to (c) P6A, (d) to (f) Y8A and (g) to (i) V7R. All data refer to
interatomic distances within the range 5-7 . Each interaction represents a group of
interatomic distances and is symbolized by the number of the residue hosting each atom and a
designation denoting a backbone, B, or side chain, S, atom position. For example,
11

interaction 2B6S denotes the coordination sum which arises from all interatomic distances
between backbone atoms (B) on residue 2 and side chain atoms (S) on residue 6.
Figure 9. Laplacian of the electron density of non-polarized peptide structures. All triggering
side chains are fully protonated. From a to d: Tax z0 s0, V7R z1 s0, Y8A z0 s0 and P6A z0
s0. Charge concentration (high potential energy) and charge depletion (high kinetic energy)
are shown as red and blue surfaces, respectively. All iso-density surfaces are drawn at 0.2 a.u.
Figure 10. Voronoi deformation density (VDD) and spin density of selected spin polarized
peptides. From a to d: Tax (charge -2, polarization 2), V7R, Y8A and P6A (charge -2,
polarization 2). All peptides had a formal charge of -2 and a spin polarization of 2 except V7R
for which these quantities were -1 and 1, respectively. On all structures the terminal hydroxyl
groups of the triggering side chains were deprotonated. Every inset depicts a magnification of
the peptide N terminus, which shows atom Voronoi charges and bond lengths. All spin
density surfaces are drawn at 0.002 a.u. The VDD bar values are also in a.u.
Figure 11. MO contributions (fractions) by N terminus atomic orbital (AO) populations. From
a to d: Tax (charge (z) -2, spin polarization (s) 2), V7R (z-1, s1), Y8A (z-1, s2) and P6A (z-2,
s2) respectively. All triggering side chains were deprotonated. Each bar corresponds to the
sum of a1g and t1u contributions either by the N atom or by the group of the three H atoms on
the N terminus. Majority (spin up) and minority (spin down) populations are shown as
positive and negative AO fractions, respectively.

o
r
P

o
i
s
i
v

12

l
a
n

o
r
P

o
i
s
i
v
Figure 1.

13

l
a
n

o
r
P

o
i
s
i
v

Figure 2.
14

l
a
n

o
r
P

o
i
s
i
v

Figure 3.
15

l
a
n

o
r
P

o
i
s
i
v

Figure 4.
16

l
a
n

o
r
P

o
i
s
i
v

Figure 5.
17

l
a
n

o
r
P

o
i
s
i
v

Figure 6.
18

l
a
n

o
r
P

o
i
s
i
v

Figure 7.
19

l
a
n

o
r
P

o
i
s
i
v

Figure 8.

20

l
a
n

o
r
P

o
i
s
i
v

Figure 9.

21

l
a
n

o
r
P

o
i
s
i
v

Figure 10.
22

l
a
n

o
r
P

o
i
s
i
v

Figure 11.
23

l
a
n