Article

Cite This: ACS Nano XXXX, XXX, XXX−XXX www.acsnano.org

Microphysiological Engineering of Self-

Assembled and Perfusable Microvascular Beds
for the Production of Vascularized Three-
Dimensional Human Microtissues
Jungwook Paek,†,○ Sunghee E. Park,†,○ Qiaozhi Lu,† Kyu-Tae Park,† Minseon Cho,† Jeong Min Oh,†
Keon Woo Kwon,† Yoon-suk Yi,† Joseph W. Song,† Hailey I. Edelstein,‡ Jeff Ishibashi,§,∥ Wenli Yang,‡
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Jacob W. Myerson,⊥ Raisa Y. Kiseleva,⊥ Pavel Aprelev,⊥ Elizabeth D. Hood,⊥ Dwight Stambolian,#
Patrick Seale,§,∥ Vladimir R. Muzykantov,⊥ and Dongeun Huh*,†,‡,∇
†
Department of Bioengineering, ‡Institute for Regenerative Medicine, §Institute for Diabetes, Obesity and Metabolism (IDOM),
Downloaded via BUFFALO STATE on July 19, 2019 at 12:25:44 (UTC).

∥
Department of Cell and Developmental Biology, ⊥Department of Systems Pharmacology and Translational Therapeutics and
Center for Translational Targeted Therapeutics and Nanomedicine, and #Department of Ophthalmology, Perelman School of
Medicine, and ∇NSF Science and Technology Center for Engineering Mechanobiology, University of Pennsylvania, Philadelphia,
Pennsylvania 19104, United States
*
S Supporting Information

ABSTRACT: The vasculature is an essential component of

the circulatory system that plays a vital role in the
development, homeostasis, and disease of various organs
in the human body. The ability to emulate the architecture
and transport function of blood vessels in the integrated
context of their associated organs represents an important
requirement for studying a wide range of physiological
processes. Traditional in vitro models of the vasculature,
however, largely fail to offer such capabilities. Here we
combine microfluidic three-dimensional (3D) cell culture
with the principle of vasculogenic self-assembly to engineer
perfusable 3D microvascular beds in vitro. Our system is
created in a micropatterned hydrogel construct housed in
an elastomeric microdevice that enables coculture of primary human vascular endothelial cells and fibroblasts to achieve
de novo formation, anastomosis, and controlled perfusion of 3D vascular networks. An open-top chamber design adopted
in this hybrid platform also makes it possible to integrate the microengineered 3D vasculature with other cell types to
recapitulate organ-specific cellular heterogeneity and structural organization of vascularized human tissues. Using these
capabilities, we developed stem cell-derived microphysiological models of vascularized human adipose tissue and the
blood−retinal barrier. Our approach was also leveraged to construct a 3D organotypic model of vascularized human lung
adenocarcinoma as a high-content drug screening platform to simulate intravascular delivery, tumor-killing effects, and
vascular toxicity of a clinical chemotherapeutic agent. Furthermore, we demonstrated the potential of our platform for
applications in nanomedicine by creating microengineered models of vascular inflammation to evaluate a nanoengineered
drug delivery system based on active targeting liposomal nanocarriers. These results represent a significant improvement
in our ability to model the complexity of native human tissues and may provide a basis for developing predictive
preclinical models for biopharmaceutical applications.
KEYWORDS: organ-on-a-chip, vasculature, 3D culture, nanomedicine, cancer, adipose, retina

V ascularity is an essential anatomical feature of virtually

every organ in the human body. By providing a
conduit for hemodynamic flow, blood vessels con-
stitute the main component of the circulatory system
indispensable to the development and homeostasis of various
Received: January 25, 2019
Accepted: June 13, 2019
organ systems.1 In particular, the vasculature is responsible for Published: June 13, 2019

© XXXX American Chemical Society A DOI: 10.1021/acsnano.9b00686

ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 1. A microengineered in vitro 3D culture platform to produce self-assembled and perfusable microvascular beds. a. Our
microengineered cell culture device is used to engineer in vivo-like 3D networks of blood vessels. b. The device consists of a cell culture
chamber with a top opening and two parallel microchannels used for controlled vascular perfusion. c. This multilayered device is fabricated
by sealing microfabricated channels against a blank PDMS slab and bonding a PDMS ring used as a medium reservoir. d. To form the
vasculature, template needles are inserted into the microchannels (Step 1), and this step is followed by the injection of endothelial cells and
fibroblasts suspended in an ECM hydrogel solution into the cell culture chamber (Step 2). Once gelation is complete, the needles are pulled
out to generate hollow microchannels (Step 3), which are then seeded with vascular endothelial cells to create endothelialized and externally
accessible compartments on both sides of the cell-laden scaffold (Step 4). e. During cell culture, the endothelial cells in the hydrogel undergo
vasculogenesis and self-assemble into interconnected endothelial tubes that together form a 3D multicellular structure reminiscent of
microvascular plexus in vivo. The vessels in the hydrogel also anastomose with the endothelium in the side channels to form a perfusable
network. f. The open top of the cell culture chamber can be used for culturing another cell type, such as epithelial cells, directly on the
surface of the vessel-containing ECM hydrogel scaffold to mimic the structural organization of vascularized 3D tissues in a more realistic
manner.

mediating the exchange of gases, nutrients, metabolites, to represent the higher-level architecture of the in vivo
immune components, hormones, and other biomolecules vasculature characterized by complex three-dimensional (3D)
between the blood and perfused tissues during diverse networks of interconnected and perfusable endothelial tubes.14
physiological processes.2−5 Studies have also shown that This limitation is particularly problematic in modeling small
vascular complications due to abnormal changes in the blood vessels in the microcirculatory system that play an
architecture, biological phenotype, and hemodynamic environ- essential role in the regulation of tissue perfusion.15,16
ment of blood vessels are associated with a wide variety of Conventional approaches, most of which are based on
pathological conditions including diabetes,6 obesity,7 hyper- monoculture of endothelial cells, also present major challenges
tension,8 age-related macular degeneration,9 and cancer.10 to reconstituting the structural and functional integration of
Therefore, the ability to model and examine the structure and the vasculature with other cell types in an organ-specific
function of the vasculature is of critical importance in manner.17 To address these important drawbacks of traditional
investigating the physiology of vascularized tissues and organs vascular models, advanced techniques are needed to more
in the human body. faithfully recapitulate the native complexity of the vasculature
Over the past few decades, the significant need for such in an integrated physiological context.
capabilities has stimulated considerable research efforts to Here we demonstrate the feasibility of engineering an in vitro
develop physiologically relevant in vitro models of the platform to transform primary culture of human endothelial
vasculature. Many of these early studies focused on cells into microphysiological models of blood vessels. Our
incorporating perfusable cell culture chambers to simulate approach builds upon advanced in vitro techniques derived
the dynamic flow environment of blood vessels.11,12 By from recent progress in the development of microengineered
recapitulating flow-generated physiological mechanical cues perfusable blood vessels18−27 and hydrogel-based micro-
not captured in static culture, these model systems have proven physiological systems14,28,29 to combine microfluidic 3D cell
instrumental in advancing our fundamental understanding of culture with the principle of vasculogenesis, a critical
vascular endothelial cells.11,12 With the increasing attention to developmental process by which new blood vessels are formed
the multiscale nature of physiological and disease processes in de novo.30 Specifically, we used a template-assisted fabrication
the vascular system,13 there is now a rapidly growing need to technique to build a hybrid microfluidic platform that contains
expand the scope of research beyond the cellular level to perfusable microchannels in an extracellular matrix (ECM)
investigate integrated vascular structure and function at the hydrogel scaffold housed in a multilayered elastomeric device.
tissue and organ levels. Despite significant advances, however, This system was used for primary culture of human vascular
existing in vitro vascular models are greatly limited in their endothelial cells and fibroblasts to emulate the process of
ability to meet this important need. Two-dimensional (2D) vasculogenesis and to induce spontaneous formation of
monolayers of endothelial cells often used in these systems fail perfusable 3D vascular networks distributed throughout the
B DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 2. Microfluidic tissue engineering of perfusable 3D microvessels. a. Production of blood vessels in our model begins with the insertion
of thin needles (shown as dark objects) guided by the side microchannels. Scale bar, 400 μm. b. With the needles in place, the culture
chamber is filled with a fibrinogen solution mixed with thrombin. The 1-μm microbeads fluorescently labeled green were added to the matrix
solution for visualization purposes. Scale bar, 400 μm. c. The side channels created by removing the needles from the solidified fibrin
scaffold are injected with 1-μm red fluorescent beads. Scale bar, 200 μm. d, e. Over time, coculture of RFP-HUVECs (red) and human lung
fibroblasts (not shown) in the hydrogel leads to the formation of network structures. Scale bars, (d) 250 μm and (e) 200 μm. f.
Immunofluorescence staining of CD31 expressed by the endothelial tubes in the self-assembled vascular network. Scale bar, 100 μm. g. 3D
confocal rendering of 3D vessels distributed throughout a 400 μm-thick hydrogel construct. The inset shows a cross-section of a vessel with
an open lumen (white dotted line). Scale bar, 200 μm. h. The physical interaction between the self-assembled vessels (red) and the
endothelial lining of the side channels (green) is visible at the hydrogel−channel interface (white dotted line). Scale bar, 100 μm. After
anastomosis, the engineered vessels become perfusable as visualized by the flow of (i) 70 kDa FITC−dextran (scale bar, 200 μm) and (j) 1-
μm fluorescent beads. The blue and yellow lines in j show the trajectories of two fluorescent particles flowing through the vasculature (see
Supplementary Movie 1).

scaffold. Importantly, we implemented an open-top design in (Figure 1d, step 1). Subsequently, an ECM hydrogel precursor
this cell culture system to enable integration of the micro- solution mixed with a suspension of primary human vascular
engineered vasculature with other types of specialized tissue. endothelial cells and fibroblasts is injected into the cell culture
We demonstrate this capability by presenting stem cell-based chamber and enzymatically solidified to form a cell-laden
microphysiological models of vascularized human adipose hydrogel scaffold (Figure 1d, step 2). After gelation, the
tissue and the interface between the retina and the choroidal needles are gently removed from the microchannels to create
blood vessels. Moreover, we show that tumor spheroids can be hollow circular channels (Figure 1d, step 3), which are then
incorporated into our engineered microvascular beds to seeded with vascular endothelial cells to form an endothelium
construct an in vitro model of vascularized adenocarcinoma on the luminal surface (Figure 1d, step 4).
in the human lung. We demonstrate the feasibility of using this During culture, the endothelial cells dispersed throughout
disease model as a high-content drug screening platform by the ECM scaffold spread, proliferate, interconnect, and self-
simulating the efficacy and vascular toxicity of a chemo- assemble into a network of microvessels with open lumens,
therapeutic drug in current clinical use for advanced lung approximating vasculogenesis during embryonic development
cancer. Finally, we present microengineered models of vascular and adult vascular growth in vivo (Figure 1e). The fibroblasts
inflammation for in vitro evaluation of nanoengineered in the hydrogel play an essential role in this process by
liposomes designed for targeted intravascular drug delivery. producing soluble factors and matrix proteins that have been
shown to promote lumen formation and the stability of
MODEL DESIGN AND CONSTRUCTION microvascular structures.31 Another critical step in the
Our microphysiological model is constructed in an elastomeric construction of our model is to establish physical connection
cell culture device made out of poly(dimethylsiloxane) between the vascular network formed in the hydrogel scaffold
(PDMS) that consists of multiple compartments necessary and the flow channels on the sides. This is accomplished by
for in vitro tissue engineering and controlled perfusion of 3D vascular anastomosis that occurs spontaneously via fusion of
microvascular networks (Figure 1a). The microengineered the microvascular tubes in the hydrogel scaffold with the
device consists of a cell culture chamber which is open to an endothelial lining of the side microchannels (Figure 1e).32
upper medium reservoir and flanked by two parallel micro- Completion of this process provides a means to interface the
channels (Figure 1b). To make this device, a PDMS layer microengineered vasculature with the external world and to
containing microfabricated channel features with a large generate pressure gradient across the hydrogel scaffold to
opening is bonded to a rectangular PDMS ring and a blank perfuse the vascular network in a controlled manner.
slab of PDMS (Figure 1c). In the first step of establishing cell Our platform also makes it possible to incorporate other
culture in this device, a long needle is inserted into each types of cells into the perfusable microvascular beds to model
microchannel through the inlet access ports oriented sideways various kinds of vascularized human tissues. For example, the
C DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

opening of the cell culture chamber can be used to grow and eventually led to the formation of cellular networks that
epithelial cells directly on the surface of the hydrogel scaffold resembled the primitive vascular plexus generated by vasculo-
and to form a multilayered tissue unit that resembles the genesis in vivo (Day 6 in Figure 2e). Immunostaining analysis
mucosa and its underlying vascularized stroma or other also showed intricate networks of interconnected endothelial
epithelial-endothelial interfaces responsible for key physio- tubes with an average diameter of 24 ± 7.05 μm (mean ± SD)
logical processes such as barrier function (Figure 1f). It is also that exhibited robust expression of cluster of differentiation 31
possible to embed cells with specialized functions in the ECM (CD31) (Figure 2f).
scaffold along with endothelial cells and fibroblasts to mimic It should be noted that our engineered microvessels were
vascularized parenchymal tissues. In this case, the opening of distributed throughout the hydrogel construct with a thickness
the chamber provides a means to increase media supply and of 400 μm (Figure 2g), which is significantly thicker than ECM
facilitate diffusive transport of nutrients, oxygen, and soluble scaffolds typically used in other microfluidic vascular
factors into the hydrogel to support cell growth and models.18−27 We speculate that the ability to vascularize
differentiation prior to the formation of perfusable vasculature. thick tissues in our system is enabled mainly by the open-top
Importantly, these approaches can be implemented in an design of the device that greatly increases the area of contact
organ-specific manner by using cells derived from the same between the hydrogel scaffold and cell culture medium to
organ. In any case, the open-top design of our system provides improve the supply of nutrients and oxygen deep into the
easy access to the cultured tissues and facilitates the process of engineered 3D tissue constructs. This capability makes our
engineering the multicellular complexity and microenviron- platform attractive for recapitulating three-dimensionality of
ment of these models. the native vasculature.
Perfusion of the Microengineered Vascular Net-
RESULTS/DISCUSSION works. Next, we investigated the perfusability of our
Engineering of 3D Vascular Networks in Perfusable engineered vasculature. Since the microvessels were situated
ECM Hydrogel Scaffolds. Our study first focused on within the hydrogel scaffold, their intravascular space was not
demonstrating the method of using needle templates to readily accessible from the external compartment, making it
generate perfusable hydrogel cell culture scaffolds. When the challenging to achieve vascular perfusion in our model. To
needles were inserted into the access ports, the two parallel address this issue, we cultured HUVECs in the side
microchannels physically guided them to slide along the entire microchannels and allowed them to connect with the 3D
length of our device without noticeable deflection (Figure 2a). vessels in the hydrogel through the process of vascular
The fibrinogen solution injected in the subsequent step anastomosis.32,33 In these experiments, HUVECs expressing
advanced along the cell culture compartment without moving green fluorescent protein (GFP) were seeded into the channels
the inserted needles and filled the entire chamber within a 1 day after hydrogel loading to visualize anastomosis. After
minute (Figure 2b). During this process, we did not observe seeding, GFP-HUVECs were continuously perfused with
any spillage of the solution through the open top of the medium to form confluent monolayers within the first 2 days
chamber presumably because the meniscus of the solution was of culture, after which they began to migrate into the gel. The
pinned at the edge of the opening due to surface tension. endothelial sprouts formed by this directional migration then
Following gelation for 15 min, the needle templates were appeared to establish connections with the nascent vessels
removed from the fibrin scaffold to generate flow channels on assembled in the scaffold and integrate into the vascular
both sides of the hydrogel construct in the culture chamber network (Figure 2h). The same pattern of anastomosis was
(Figure 2c). When we introduced a cell culture medium into observed along the entire length of the interface between the
the side microchannels and the medium reservoir to hydrate cell culture chamber and the endothelialized side channels.
the device, the micropatterned scaffold remained stable We then asked whether the anastomosed endothelium in the
without showing significant structural changes. side channels provided fluidic access to the lumens of the
Our microengineered system containing this fibrin scaffold engineered vessels in the hydrogel. To examine this question,
was then used as a coculture platform to produce 3D vascular we injected a fluorescently labeled solution containing 70 kDa
networks. To demonstrate the proof-of-principle of this FITC-dextran into one of the endothelialized microchannels
approach, we created a model system using red fluorescent and generated a pressure gradient across the vascularized
protein (RFP)-expressing human umbilical vein endothelial hydrogel. Under this condition, the solution was observed to
cells (HUVECs) and human lung fibroblasts as representative enter the microvessels and flow in the direction of applied
cell populations. After hydrogel loading, the cells were pressure gradient, eventually reaching the microchannel on the
uniformly dispersed throughout the gel and began to spread other side of the cell culture chamber (Figure 2i). During flow,
within 24 h of seeding. During culture, the embedded cells the dextran solution remained in the intravascular space
were observed to undergo active proliferation as illustrated by a without leaking into the surrounding scaffold, illustrating
continuous increase in the intensity of RFP in the cell culture barrier integrity of the engineered vessels. We also conducted
chamber (Figure 2d). The increasing cell number, however, similar experiments using 1 μm fluorescent particles to show a
had negligible effects on the structural integrity of the scaffold, continuous flow of the microbeads through the vascular
and the cell-laden hydrogel construct remained stable and network (Figure 2j; Supplementary Movie 1). These results
firmly attached to the surface of the chamber for the entire clearly demonstrate the perfusability of the anastomosed 3D
duration of culture (>10 days). Importantly, microfluorimetric vasculature in our model. Given the small size of the
imaging of the cells revealed the development of network engineered vessels, the data also suggest the possibility of
structures in the fibrin hydrogel (Figure 2e). As early as on day using our system to simulate microvascular perfusion in a
2, HUVECs appeared significantly elongated and began to physiologically relevant 3D environment.
organize into chord-like structures (Day 2 in Figure 2e). These Microphysiological Model of Vascularized Human
morphological changes became more pronounced over time Adipose Tissue. In the next phase of our study, we set out to
D DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 3. Construction of vascularized human adipose tissue in vitro. a, b. The highly vascularized human white adipose tissue is recreated in
our model by coculturing hASCs and hAMECs in a fibrin hydrogel scaffold to induce adipogenesis and vasculogenesis simultaneously. c.
Over a period of 2 weeks, hAMECs organize themselves into a 3D vascular network visualized by immunostaining of CD31. Scale bar, 100
μm. d. Concurrently, hASCs differentiate into adipocytes recognized by their intracellular lipid droplets visible in phase contrast
micrographs taken on day 40. Scale bar, 100 μm (scale bar for magnified micrograph, 50 μm). e. Adipogenic differentiation of hASCs is
further evidenced by their expression of perilipin-1 after 40 days of culture. Scale bar, 100 μm. f, g. As culture progresses, the number of lipid
droplet-containing cells increases. Scale bar, 100 μm. h. The size of lipid droplets in the differentiated cells also increase with time. i. A
micrograph of vascularized adipose tissue construct after 40-day culture in our device. Scale bar, 100 μm. j. Adipogenic differentiation occurs
more rapidly in the coculture of hAMECs and hASCs as compared with the hASC monoculture model. k. ELISA analysis of conditioned
media collected from our engineered adipose tissue reveals increased leptin production due to the vasculature. The results in j and k indicate
the significant contribution of endothelial cells to adipogenic differentiation of hASCs. Data are presented as mean ± SEM. **P < 0.01 and
***P < 0.001 (n = 3).

leverage the demonstrated capabilities of our platform to adipogenesis and vasculogenesis simultaneously (Figure 3b).
develop specialized in vitro models that combined the We included collagen in the hydrogel due to its promotive
engineered vasculature with other types of cells to emulate effects on adipogenesis39−41 and also engineered the scaffold
vascularized human tissues. To this end, we chose adipose by adjusting ECM concentrations and ratios to achieve the
tissue as a model system and explored the feasibility of using level of stiffness that has been shown to support adipogenic
our microdevice to mimic human subcutaneous white adipose differentiation of stem cells (Figure S2).42−44 The 3D
tissue (WAT) (Figure 3a). As the major site of energy storage coculture configuration was based on previous reports that
in the body, WAT is mostly composed of adipocytes that store adipose-derived stem cells have the capacity to produce soluble
and release lipids in accordance with systemic energy levels factors to trigger and support the formation of the vasculature
and also secrete a variety of hormones to regulate in adipose tissue.45 Consistent with these results, hAMECs
metabolism.34 Conventionally, the study of WAT has relied grown in our model underwent robust vasculogenesis in the
predominantly on the monolayer culture of adipocytes. This presence of hASCs and formed a 3D network of
traditional approach, however, has been shown to prevent interconnected endothelial tubes throughout the hydrogel
adipocytes from assuming their physiological 3D morphology scaffold (Figure 3c). Interestingly, the generated vessels
and to induce the expression of fibrotic and inflammatory appeared to be smaller yet more organized and denser than
phenotypes.35 The conventional models are also greatly limited those created by HUVECs (Figure S3), illustrating the
in their ability to recapitulate structural and functional coupling microvascular origin of hAMECs. In comparison to HUVECs,
of adipocytes with other cell types that play a critical role in the the hAMECs were also assembled into 3D vessels more slowly
physiological function of WAT. In particular, WAT is highly but the resultant vasculature remained stable and maintained
vascularized and relies on vascular transport of nutrients, its network structure during prolonged culture (over 40 days).
oxygen, cells, and various soluble factors to maintain tissue Considering that similar differences occur in traditional 3D
homeostasis and to carry out its specialized functions.36 culture models (data not shown), these results are not specific
Studies have also shown that the development and repair of to the microfluidic culture environment, but they highlight the
WAT occur in coordination with the vasculature.37,38 Despite significant effect of endothelial phenotype on vascular
the increasing recognition of the vasculature as a critical development.
component of WAT, modeling this essential feature using In concurrence with this process of vasculogenesis, the
traditional in vitro techniques remains a significant challenge. hASCs embedded in the same scaffold began to differentiate
To suggest an alternative in vitro strategy to tackle this into adipocytes. Phase contrast imaging of the construct
problem, we focused on using the microengineered vascular showed clusters of cells throughout the gel that contained
platform to model the vascularized architecture of human large, spherical organelles (Figure 3d). These intracellular
WAT. In this study, we embedded human adipose-derived structures were stained positive with perilipin-1, which is a
stem cells (hASCs) and primary human adipose microvascular specific marker of lipid droplets,46,47 indicating the differ-
endothelial cells (hAMECs) isolated from WAT in a hydrogel entiation of hASCs into adipocytes (Figure 3e). The number
scaffold composed of type I collagen and fibrin and cocultured of these cells and the size of their lipid droplets increased with
them for extended periods (over 40 days) to induce time, and the differentiated adipocytes were clearly visible
E DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 4. Microengineered model of the blood−retinal barrier. a. The blood−retinal barrier is formed by a monolayer of RPEs and the
choroidal capillary bed in the outer layers of the retina in the human eye. b. This multilayered tissue−tissue interface is modeled in our
device by culturing human iPSC-derived RPEs on the surface of a hydrogel construct containing a network of blood vessels. c. CD31 staining
of microvessels constructed by primary human retinal microvascular endothelial in coculture with choroidal fibroblasts embedded in a fibrin
scaffold. The image was taken on day 14. Scale bar, 50 μm. d. The iPSC-derived RPEs proliferate and form a confluent monolayer that
expresses well-defined intercellular tight junctions (ZO-1) after 14 days of culture. Scale bar, 50 μm. e−g. Production of basement
membrane proteins by RPEs is visualized by immunostaining of laminin (red) accumulated on the basolateral side of the epithelium.
Analysis of immunofluorescence after 14 days of culture shows substantially increased laminin deposition in coculture of RPEs and vessels.
Scale bar, 50 μm. h. Prolonged culture (14 days) in our model also leads to the pigmentation of RPEs as evident from the formation of dark
granules in the RPEs shown by bright-field microscopy. Scale bar, 20 μm. i. The pigmented RPEs show robust expression of melanosomes in
the cytoplasmic compartment. Scale bar, 25 μm. j. The number of melanosome-expressing RPEs increases significantly due to the presence
of choroidal vessels in the hydrogel scaffold. k, l. The beneficial effect of the engineered blood vessels is further demonstrated by the
increased expression of RPE-specific marker (RPE65) in the coculture (14 days) model. Scale bar, 50 μm. Data are presented as mean ±
SEM. ***P < 0.001 (n = 3).

throughout the entire scaffold after 30 days of culture (Figure conditioned media collected from our devices yielded
3f−h). Of note is that when the hASCs were grown in a closed measurable amounts of leptin in both models. The level of
cell culture chamber using the same media and scaffold, the leptin in the vascularized construct, however, was found to be
number of differentiated adipocytes was greatly reduced for the significantly higher than that measured in the monoculture
same duration of culture (14 days) (Figure S4). We suspect device. For instance, coculture of hASCs with hAMECs for 15
that this difference is due to the increased supply of nutrients days resulted in a more than 2-fold increase in the
and adipogenic factors to the hASCs through the opening of concentration of leptin as compared to a monoculture of
the culture chamber. As demonstrated in Figure 3i, 40-day hASCs (Figure 3k). Our results also showed increasing leptin
culture in our system produced highly vascularized 3D tissue concentrations over time in both cases, but this trend occurred
constructs interspersed with a large number of adipocytes. in a more accelerated manner in the vascularized construct
During prolonged culture over 40 days, the lipid droplets (Figure 3k). While further characterization is necessary, these
continued to increase in size and number, forming densely data demonstrate that the tissue-specific vasculature engi-
packed tissue constructs (Figure S5). neered in our system allows for more efficient and rapid
Importantly, these results also led us to ask whether our differentiation of hASCs into adipocytes, providing strong
engineered microvasculature had any significant effects on justification for inclusion of vascular components in modeling
hASC differentiation in our model. Blood vessels in adipose adipose tissue in vitro. These types of vascularized models may
tissue have been suggested as the key component of an provide a research platform to simulate and mechanistically
adipogenic niche that provides supporting signals for adipocyte investigate various physiological and pathophysiological
development in vivo.38,45,48 Inspired by these studies, we processes in human adipose tissue.
compared adipocyte differentiation and maturation in our Microengineering of the Blood−Retinal Barrier. The
vascularized model to those in a monoculture of hASCs open-top design of the cell culture chamber in our system
established in the same microdevice. First, we examined the provides a means not only to increase medium supply to the
difference in the rate of adipogenic differentiation by cells cultured in the hydrogel scaffold but also to form planar
comparing the numbers of cells with lipid droplets over time. tissue layers directly on the surface of the engineered vascular
As shown in Figure 3j, the differentiation of hASCs into constructs to mimic the interface between blood vessels and
adipocytes occurred regardless of the vasculature, but the parenchymal tissues in an organ-specific manner. To
vascularized models exhibited a substantially higher rate of demonstrate this capability, we developed a microphysiological
differentiation, yielding nearly twice as many lipid droplet- system designed to reconstruct the barrier between the retinal
containing cells as a monoculture of hASCs over 28 days. We pigment epithelium and the microvasculature in the outer
also measured the production of leptin, which is one of the layers of the retina in the human eye. This tissue−tissue
most important adipokines secreted by mature adipocytes to interface, termed the outer blood−retinal barrier (oBRB), is
regulate energy balance and insulin sensitivity.49,50 Our formed by tight junctions between retinal pigment epithelial
enzyme-linked immunosorbent assay (ELISA) analysis of cells (RPEs) anchored to a specialized basement membrane
F DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 5. Vascularized solid tumor-on-a-chip for testing the efficacy and vascular toxicity of chemotherapy. a. To mimic malignant solid
tumors in the lung, multicellular cancer spheroids are cocultured with endothelial cells and lung fibroblasts in the hydrogel compartment of
our device to generate vascularized tumor constructs. b. During culture, the cancer spheroids composed of A549 cells (green) and RFP-
HUVECs (red) grow in size and undergo structural integration with the surrounding vasculature formed by HUVECs embedded in the same
hydrogel. Scale bar, 50 μm. c. A micrograph of a vascularized tumor spheroid perfused with fluorescent microbeads (blue). The outline of
the spheroid is depicted with a white dotted line. Scale bar, 50 μm. d. For drug testing, the vascularized tumor construct is perfused with a
clinical dose of paclitaxel. Drug treatment for 2 days leads to significant reduction in the viability of cancer cells, demonstrating the tumor
killing effect of paclitaxel. Scale bar, 20 μm. e. In a given spheroid, the majority of dead cells are found in the inner two-thirds of the spheroid
(area A). f. Our model also allows for detection and analysis of drug-induced vascular toxicities. Paclitaxel causes the loss and disintegration
of the vasculature, resulting in significantly reduced vascular density, average vessel diameter, total vessel length, and junction number. Scale
bar, 50 μm. Vascular perfusion with paclitaxel for 2 days also induces (g) endothelial apoptosis and (h) oxidative stress. Scale bars, 50 μm. i.
The deleterious potential of paclitaxel to elicit vascular inflammation is shown by endothelial expression of ICAM-1. Scale bar, 50 μm. j. The
activated endothelial lining of the paclitaxel-perfused vessels permits the recruitment of circulating immune cells as demonstrated by
endothelial adhesion of T cells (green) injected into the engineered vasculature. Scale bar, 50 μm. Data are presented as mean ± SEM. **P <
0.01 and ***P < 0.001 (n = 3).

known as Bruch’s membrane that separates the neural retina hydrogel underwent vasculogenic self-assembly over a period
from the capillary bed of the choroid (Figure 4a).51,52 The of 7 days to develop a dense meshwork of perfusable
oBRB plays an essential role in maintaining the homeostasis of microvessels that were roughly 10−25 μm in diameter (Figure
the retina by regulating fluid and molecular transport between 4c). In parallel, the iPSC-derived RPEs seeded into the
the outer retina and the blood in choroidal vessels.53 opening of the culture chamber adhered to the surface of the
Compromised integrity of the oBRB has been implicated in hydrogel scaffold and proliferated continuously until they
the development of retinal diseases such as age-related macular formed a densely packed monolayer of cuboidal cells. After 14
degeneration.9,54,55 days of culture, the structural integrity of the epithelial barrier
To reconstitute this specialized barrier, we combined our was demonstrated by a well-defined network of intercellular
vascular engineering platform with human inducible pluri- tight junctions (Figure 4d), which is a critical feature of the
potent stem cells (iPSCs) as a renewable source of RPEs. retinal pigment epithelium in vivo necessary for physiological
Specifically, we used human RPEs derived from long-term barrier function of the oBRB.59,60 Our characterization also
(140 days) culture and differentiation of iPSCs to form a revealed the production and deposition of basement
confluent epithelial monolayer on the exposed surface of the membrane protein, laminin, on the basolateral side of the
underlying hydrogel construct (Figure 4b). The lower epithelial monolayer (Figure 4e). The laminin deposits
compartment of the device was used to mimic the micro- accumulated underneath the epithelium and formed a thin
vascular bed of the choroid by culturing primary human retinal sheet reminiscent of Bruch’s membrane in vivo. Interestingly,
microvascular endothelial cells and choroidal fibroblasts in an this process of matrix synthesis and remodeling occurred in a
ECM hydrogel made out of fibrin mixed with type I collagen, vasculature-dependent manner. When the RPEs were cultured
which has been shown to promote the structural and functional alone without the engineered vessels in the hydrogel
differentiation of RPEs.56−58 As was the case with HUVECs compartment, the extent of laminin deposition was reduced
and hAMECs, the retinal endothelial cells cultured in the by 50% as compared to that measured in the coculture system
G DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

at the same point (Figures 4f,g). This result illustrates types of studies has been the limited capacity of conventional
significant vascular contributions to basement membrane in vitro techniques to model the vasculature in the integrated
formation in our model. Although the relevance of this finding context of its associated tissues undergoing pathological
to the in vivo context needs to be validated, our observation processes. Motivated by this problem, we explored whether
raises the possibility that the choroid may have the capacity to our microvascular engineering approach could be leveraged to
influence matrix production by RPEs and play an important mimic diseased human tissues that reside in vascularized 3D
role in the maintenance and remodeling of Bruch’s membrane environments.
and the oBRB. For this investigation, we used lung cancer as a model
Another finding of interest was the development of dark disease and developed a microphysiological system that
granules in the cytoplasmic compartment of RPEs that were combined the self-assembled perfusable microvasculature
clearly visible in bright-field imaging of the epithelium (Figure with 3D tumor spheroids (Figure 5a). As a first step to
4h) and also showed strong immunofluorescence for construct this model, we established a mixed coculture of
melanosomes (Figure 4i). These results illustrate pigmentation human lung adenocarcinoma cells (A549) and endothelial cells
of the cultured RPEs, which is an important characteristic of (HUVECs) in low-attachment wells to form multicellular
their in vivo counterparts that allows the oBRB to absorb spheroids reminiscent of solid tumors in the lung. The
scattered light for clear vision and to scavenge reactive radical composite cancer spheroids were then injected into our device
species to protect the retina from oxidative damage.53,61 and embedded in an ECM hydrogel scaffold that contained
Importantly, the expression of this key RPE phenotype was endothelial cells and lung fibroblasts. During culture, the
enhanced by the vasculature in the choroidal compartment, as spheroids became progressively larger, presumably due to the
evidenced by a more than 2.5-fold increase in immunohis- proliferation of cancer cells, while maintaining their spherical
tochemical analysis of melanosome-containing RPEs in the shape and structural integrity without significant cellular
oBRB model in comparison to a monoculture of RPEs (Figure outgrowth (Figure 5b). Simultaneously, the endothelial cells
4j). When the RPEs in monoculture were treated with in the hydrogel underwent proliferation and self-assembly into
endothelial conditioned media, the level of melanosome a 3D vascular network throughout the scaffold over a period of
expression did not show significant changes compared to the 7 days. During this process, most of the tumor spheroids were
untreated monoculture group and remained substantially lower fully enveloped by the vasculature and appeared to be
than that measured in the coculture model (Figure S6), integrated with their surrounding microvessels (Figure 5b).
suggesting the importance of the short-range interactions Perfusability of the vascularized tumors and their local
between the RPEs and the vasculature for RPE pigmentation. microenvironment was demonstrated by the flow of
Many of the pigmented RPEs were also found to express high fluorescent microbeads through the vasculature in and around
levels of RPE-specific markers such as RPE65 (Figure 4k). the cancer spheroids (Figure 5c).
Similar to laminin deposition and pigmentation, the expression Given the increasing attention to the potential of micro-
of RPE65 was significantly upregulated by coculture of RPEs physiological systems for drug testing, we then investigated the
with the underlying microvasculature (Figure 4l), highlighting feasibility of using our engineered tumor model as a screening
the importance of vascular components for the induction of platform to evaluate the effects of anticancer drugs. This study
physiological phenotype in RPEs. was conducted using paclitaxel as a model drug that represents
Taken together, these results demonstrate the suitability of cytotoxic chemotherapeutic agents in current clinical use for
the iPSC-derived RPE cells in coculture with retina-specific the treatment of solid tumors in various types of cancer
endothelial cells for the purpose of modeling the architecture including lung cancer.67,68 To simulate intravascular drug
and cellular phenotype of the oBRB in our microphysiological delivery to solid tumors during chemotherapy, we injected
system. The compartmentalized design of our device made it clinical concentrations of paclitaxel into the endothelialized
possible to replicate the relative spatial arrangement of the side channel and generated continuous flow through the
retinal pigment epithelium and the underlying choroidal engineered vasculature (Figure 5d). Vascular perfusion of our
vascular bed in a more realistic manner than is possible with model under this condition for 2 days led to arrested growth
conventional models based on Transwell inserts.62,63 As shown and significant damage of the lung tumor constructs, resulting
here, our system may provide a robust platform to create in approximately 20% reduction in the viability of cells in the
coculture models with capabilities to mimic the integrated cancer spheroids as compared to the vehicle control group
organization of vascularized parenchymal tissues in various (Figure 5d). It was noted that the drug-induced cytotoxicity
organs. showed spatial heterogeneity with the majority of dead cells
Microphysiological Model of Vascularized Solid found near the center of the spheroids (Figure 5e). This result
Tumor for High-Content Drug Screening. The vasculature is in contrast to the localized tumor-killing effect of paclitaxel at
plays an important role in a wide variety of diseases. As the outer surface of spheroids typically observed in traditional
exemplified by various types of cancer and inflammatory in vitro models based on static culture of avascular cancer
disorders, blood vessels are central to delivery and exchange of spheroids (Figure S7). While further investigation is needed,
cells and soluble factors that mediate complex biological this difference may be attributed to the vascularized
processes underlying the development of pathological architecture of our engineered tumor tissue that facilitated
conditions.64 The vasculature can also contribute to disease drug transport into the inner regions of the spheroids.
progression and exacerbation by undergoing significant Another important goal of our study was to explore the use
structural and functional changes to support growth, of the vascularized tumor model for the study of paclitaxel-
maintenance, and expansion of diseased tissues.65,66 Despite induced vascular toxicities. This investigation was motivated by
extensive in vivo evidence, however, our ability to model and two lines of evidence suggesting the clinical significance and
investigate the pathophysiological significance of the vascula- implications of endothelial toxicity due to chemotherapy. First,
ture in vitro remains rudimentary. A key challenge in these studies have discovered that classic cytotoxic drugs designed to
H DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

Figure 6. Vascular inflammation models for evaluation of nanoengineered drug carriers. a, b. A schematic diagram of azide-functionalized
fluorescent DPPC-cholesterol liposomes. The monoclonal antibody against ICAM-1 coupled to their surface allows them to recognize and
bind to ICAM-1 on the activated endothelial surface during inflammatory responses. c. Intravascular flow of TNF for 6 h results in
endothelial expression of ICAM-1. Scale bar, 100 μm. d, e. The anti-ICAM-1 nanocarriers (NCs) adhere to the TNF-stimulated endothelial
surface. Scale bars, (d) 100 μm and (e) 25 μm. f. The non-TNF-treated blood vessels show minimal liposome binding. Scale bar, 100 μm. g,
h. The IgG-coupled liposomes do not bind to the blood vessels. Scale bar, 100 μm. i. Local inflammation is simulated by injecting LPS-
soaked microbeads directly into the vascularized hydrogel through the opening of the culture chamber. j, k. The localized source of
inflammation results in graded activation of endothelial ICAM-1. The expression level is inversely proportional to the distance from the LPS
beads. Scale bar, 100 μm. l, m. Binding of anti-ICAM-1 NCs perfused through the vasculature corresponds to the spatial pattern of ICAM-1
activation, further demonstrating the active targeting capability of the liposomal NCs. Scale bar, 100 μm.

target cancer cells can also exert detrimental effects on the lining in a large fraction of drug-treated vessels. Quantification
basic function of endothelial cells.69 As a result, several of vascular architecture also revealed significant reduction in
cytotoxic cancer drugs, including paclitaxel, are now consid- the average vessel diameter, total vessel length, and number of
ered as potential therapeutic agents to disrupt the vasculature vessel junctions over the same period (Figure 5f). The
in the tumor microenvironment and to inhibit cancer detrimental effect of paclitaxel was further evidenced by the
angiogenesis.70,71 Systematic and accurate preclinical evalua- regression of the interconnected tubular network to disor-
tion of this added therapeutic value, however, has been ganized endothelial chord-like structures, as well as significant
challenged by the limited capacity of traditional in vitro cancer reduction (Figure 5f). The drug-induced changes in vascular
models to reproduce the vascularized architecture of solid architecture were accompanied by the activation of apoptotic
tumors. Second, clinical evidence has shown that vascular pathways in endothelial cells, as demonstrated by their robust
toxicities due to anticancer drugs may be used as a predictive expression of caspase-3/7 throughout the perfused vasculature
marker to identify asymptomatic patients at high risk of (Figure 5g). These results are consistent with previous
developing cardiovascular complications during the course of findings72,73 and can be explained by the well-documented
chemotherapy.70 These studies highlight the potential clinical activity of paclitaxel to disrupt the microtubule network, which
benefit of evaluating chemotherapy-induced endothelial has been shown to cause detachment and apoptosis of
dysfunction, but this line of investigation is still in need of endothelial cells.70,71
advanced in vitro techniques to model and interrogate toxic While endothelial cell death is central to chemotherapy-
responses of the tumor vasculature. induced vascular toxicities, cytotoxic cancer drugs have the
Therefore, we set out to test the utility of our engineered capacity to elicit various other types of endothelial dysfunction
cancer model for analysis of adverse vascular responses to that contribute to adverse vascular responses.74,75 For example,
chemotherapy. To capture the complexity of such responses, paclitaxel can promote the production of reactive oxygen
our study relied on microfluorimetric techniques to measure species (ROS) in endothelial cells by increasing the activity of
the effect of paclitaxel on a range of phenotypic end points reduced nicotinamide adenine dinucleotide phosphate
indicative of vascular structure and biochemical function. As (NADPH), which can be injurious to the vasculature.76,77
shown in Figure 5f, the engineered vasculature perfused with a Consistent with these findings, paclitaxel treatment in our
clinical dose of paclitaxel underwent marked structural changes model led to a 2.5-fold increase in the level of ROS, as
illustrated by the disruption of its characteristic network compared to the control group (Figure 5h). Given the
architecture. Within 2 days of perfusion, vascular density established role of ROS as key mediators of inflammation,78 we
decreased significantly due to a significant loss of endothelial also examined whether paclitaxel triggered inflammatory
I DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

responses in the engineered vasculature. This study focused on for specific interaction with activated endothelial cells in the
endothelial expression of intercellular adhesion molecule 1 vasculature of inflamed tissues (Figure 6b).93
(ICAM-1) as a representative readout of vascular inflamma- To test these engineered liposomes, we established a model
tion. In contrast to the vehicle control that yielded no of inflamed blood vessels by perfusing the self-assembled
measurable signals in immunohistochemical analysis of ICAM- vasculature in our device with tumor necrosis factor (TNF)-α
1, positive staining was clearly visible in a significant fraction of at 10 ng/mL for 6 h. The cytokine treatment did not induce
vessels perfused with paclitaxel (Figure 5i). When human T significant changes in vascular architecture but was effective for
lymphocytes were injected into the drug-treated vascular eliciting inflammatory responses, as evidenced by a more than
network, a large number of cells attached to the endothelial 6-fold increase in endothelial expression of ICAM-1 compared
lining and remained adherent, further demonstrating drug- to the untreated group (Figure 6c). When the liposomes were
induced activation of endothelial cells (Figure 5j). Under the injected into the TNF-treated vascular network, we observed
same flow conditions, however, the control devices showed abundant binding to the blood vessels during vascular
negligible endothelial recruitment of lymphocytes. perfusion (Figure 6d, Figure S9). Confocal microscopy
Taken together, these data indicate that paclitaxel can revealed that the liposomes were bound to the endothelial
induce endothelial oxidative stress and inflammation in our lining of the vasculature and remained adherent after multiple
microengineered device, which have been suggested as the key steps of washing (Figure 6e). Under the same flow conditions,
potential mechanisms of cytotoxic drug-induced vascular however, the control blood vessels without TNF stimulation
toxicities.76,77 It remains to be examined, however, how the exhibited nearly negligible vascular retention of the liposomes
toxic responses of the tumor-specific vasculature are different (Figure 6f,h; Figure S9). Importantly, when the ICAM-1
from those of normal blood vessels, which is an important antibody was substituted with nonspecific immunoglobulin G
consideration for the development of more effective anticancer (IgG), the vessels perfused with liposome-containing media
drugs with reduced side effects.79 Nevertheless, the results showed no detectable fluorescence regardless of TNF
presented here show the possibility of using this model to treatment (Figure 6g). These results indicate that the
develop a preclinical platform for high-content phenotypic functionalized liposomal nanocarriers have the ability to
analysis of vascular responses to chemotherapy. recognize and specifically bind to their target (ICAM-1) on
Evaluation of Nanoengineered Drug Carriers in the activated vascular endothelium.
Vascular Inflammation Models. Advances in nanotechnol- While this study serves to verify the target-specificity of the
ogy have led to the development of various colloidal particles engineered liposomes, our model based on global stimulation
of the vasculature with TNF fails to mimic the reality of how
of submicrometer size that can encapsulate and carry
the pathophysiological environment of inflamed tissues is
therapeutic payloads for nanomedicine applications.80,81
created in vivo. In many cases, inflammation in the body is
Research has demonstrated that these nanocarriers provide a
triggered in a localized manner, which gives rise to biochemical
promising approach to increase the solubility and stability of
gradients of soluble pro-inflammatory mediators due to their
drug compounds while also improving their pharmacokinetics
diffusion from the local source of inflammation into the
and biodistribution.82−85 Moreover, the physicochemical
surrounding environment.97,98 To capture this important
properties of nanocarriers can readily be engineered for aspect of the in vivo environment and model local
controlled release and specific delivery of therapeutic agents, inflammation, we generated lipopolysaccharide (LPS)-laden
offering a means to minimize adverse off-target effects.86−88 microbeads and injected them into the vascularized hydrogel
For the development of these specialized nanomaterials, through the open top of the culture chamber (Figure 6i). Our
traditional cell culture has proven instrumental in testing and analysis of the vasculature after 4 h of incubation clearly
optimizing the design of engineered drug delivery vehicles.89,90 showed spatially graded endothelial activation where the level
Conventional in vitro models, however, have limited capacity of ICAM-1 expression was the highest in the vicinity of the
to accurately recapitulate transport and biological interactions LPS-microbeads and decreased gradually with increasing
of nanocarriers in the physiological 3D environment of distance from the source of inflammation (Figure 6j,n). This
vascularized and perfused human tissues. As a result, current is presumably because the engineered vessels were subjected to
methods of assessing the efficacy and safety of therapeutic concentration gradients of LPS in the hydrogel scaffold and
nanoparticles rely predominantly on animal studies. Therefore, therefore became activated to varying degrees in a dose-
we asked if our microengineering platform could be used to dependent manner. Importantly, vascular perfusion of this
construct advanced in vitro models for preclinical screening of model with anti-ICAM-1 liposomes resulted in corresponding
nanoengineered drug carriers. zonation of their endothelial adhesion. As illustrated in Figure
In this study, we selected liposomes as a model delivery 6l, the vast majority of blood vessels in the region adjacent to
system due to their well-established utility and potential for the injected endotoxin beads exhibited strong fluorescence due
applications in nanomedicine.91−93 Based on the central role of to extensive binding of green fluorescent liposomes. The
inflammation in the development and progression of a wide intensity of nanocarrier-generated fluorescence, however, was
range of diseases,94,95 the focus of our study was to examine reduced significantly in other zones farther away from the
the performance of liposomal drug carriers engineered to target microbeads (Figure 6m).
the pathophysiological niche of inflamed tissues. For this Taken together, these data provide in vitro evidence that
investigation, we first generated azide-functionalized fluores- supports the potential of our engineered nanocarriers for
cent DPPC-cholesterol liposomes with an average hydro- target-specific drug delivery. It should be noted that our study
dynamic diameter of 135 nm using thin film hydration was focused on a limited set of end points and did not include
techniques (Figure S8).96 The surface of the liposomes was parenchymal tissues that play an important role in inflamma-
then functionalized with a monoclonal antibody against ICAM- tory disease processes.99 Despite these limitations, however,
1 (Figure 6a) to confer active targeting capabilities necessary our results suggest the possibility of leveraging the advanced
J DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

capabilities to mimic vascularized and perfused 3D human as blood vessels in the body are fully capable of supporting the
tissues in microphysiological systems for preclinical evaluation development and maintenance of various specialized tissues in
of nanoengineered materials for therapeutic applications. a self-sufficient manner. Modeling this in vivo-like functional
capacity in our engineered vasculature should be considered as
CONCLUSIONS an important goal for future investigations. Previous work on
As exemplified by considerable recent advances in the vascularized in vitro 3D tissue constructs may provide valuable
development of organs-on-a-chip and microphysiological insights and guidance for this effort.14,108 Studies should also
systems,100 major efforts are being made toward developing explore the possibility of leveraging our techniques to
in vitro technologies with advanced capabilities to emulate the reconstruct different-sized blood vessels, which may then be
integrated structure and function of human physiological combined to build a multiscale vascular model capable of
systems. Our work represents an important contribution to this mimicking the structural and functional hierarchy of the
emerging trend in that it introduces a potentially powerful vascular system. As demonstrated in our study using cancer
approach to reverse engineer the vasculature, which is an spheroids as a model system, our platform makes it possible to
essential element of the human body. By simulating the natural integrate 3D tissue constructs with perfusable vessels.
process of de novo blood vessel formation during development, Extending this approach to in vitro culture of organoids and
our method makes it possible to assemble 3D networks of tissue explants may provide a potential solution to address the
interconnected and perfusable endothelial tubes that resemble major challenges associated with controlling and manipulating
microvascular beds in vivo. The microengineered cell culture their local microenvironment in 3D culture systems.109 Finally,
system developed in this study also provides a modular efforts should be made to address the ultimate question of
platform to recapitulate the structural and functional whether our advanced in vitro systems are more physiologically
association of blood vessels with other types of specialized relevant and predictive of human responses than laboratory
cells and thus to create more realistic in vitro models of animals.
vascularized 3D tissues. For applications in drug development, Our work represents a significant advance in our ability to
our results suggest that these vascularized models are reverse engineer the essential unit of the vascular system and
advantageous for simulating drug delivery in the native tissue its structural and functional integration with other types of
microenvironment and may offer capabilities for high-content specialized tissues. We believe that the 3D culture platform and
screening of drug-induced vascular dysfunction, as well as for biologically inspired tissue engineering strategy developed in
the development of advanced therapeutic approaches. this study will contribute to advancing the state-of-the-art for in
While this study demonstrates the feasibility and potential of vitro modeling of complex physiological systems for the study
our microvascular engineering approach, it should be noted of human health and disease.
that it still leaves significant room for further investigations to
improve the fidelity of the microengineered vasculature. METHODS/EXPERIMENTAL DETAILS
Cellular heterogeneity is one of the important features of Device Fabrication. We used standard soft lithography
blood vessels and their perivascular environment that has yet techniques to fabricate a medium reservoir and a poly-
to be fully recapitulated in our model. In particular, it is (dimethylsiloxane) (PDMS) slab patterned with recessed features of
possible to incorporate pericytes that are intimately associated microchannels and a cell culture chamber as shown in Figure 1b. The
with the endothelial lining of the microvasculature and serve as cross-sectional dimensions of the cell culture chamber and the side
a key regulator of vascular structure and function in a variety of microchannels were 1600 μm (width) × 400 μm (height) and 400
physiological processes.101,102 Similarly, immune cells found in μm (width) × 400 μm (height), respectively (Figure S1). The size of
the perivascular region, such as macrophages103,104 and mast the opening in the cell culture chamber was 1 mm (width) × 3 mm
(length). The medium reservoir was produced by molding PDMS
cells,105 are another important resident cell population that into a square ring with the dimensions of 12 mm (width) × 12 mm
may contribute to reconstituting physiological vascular (length) × 4 mm (height). For fabrication of these components,
responses to external stimuli in our model. Although our PDMS (Sylgard 184, Dow Corning) base was mixed thoroughly with
study has shown perfusability of the microengineered vessels, a curing agent at a weight ratio of 10:1 (base:curing agent) and
their functional characteristics need to be investigated and poured onto 3D-printed masters (ProtoLab). After degassing, PDMS
validated in a more rigorous manner. As demonstrated in the was fully cured in an oven maintained at 65 °C. The hardened PDMS
vascular toxicity study, for example, our data indicate transport was then removed from the molds, and inlet and outlet ports were
of soluble intravascular content into the extravascular space, made sideways in the channel slab to gain fluidic access to the
but more quantitative and systematic analysis of vascular microchannels and cell culture chamber. For device assembly, the
PDMS slab containing channel features was stamped against a thin
permeability is required to improve our ability to engineer layer of uncured PDMS prepared by spin-coating at 1500 rpm for 5
blood vessels that are more representative of their in vivo min and sealed against a thin PDMS slab that was used as a bottom
counterparts. Given the extensive evidence showing the layer of the device. The same technique was used to bond the
importance of hemodynamics in the vascular system,106,107 medium reservoir to the upper surface of the channel slab. The
further research efforts may be necessary to engineer the assembled device was baked at 65 °C to fully cure the PDMS adhesive
characteristics of vascular perfusion (e.g., shear stress, layer.
pulsatility) with the goal of recapitulating the physiologically Cell Culture. To demonstrate the proof-of-principle for engineer-
relevant hemodynamic microenvironment. ing 3D microvascular beds in our device presented in Figures 1 and 2,
Our results have demonstrated that the open-top design we used primary human umbilical vein endothelial cells (HUVECs)
and primary normal human lung fibroblasts (NHLFs). HUVECs and
makes it possible to increase the supply of nutrients and NHLFs were cultured in 25 cm 2 flasks according to the
oxygen and to meet complex media requirements during manufacturer’s protocols using endothelial cell growth medium
coculture of multiple cell types necessary for generating (EGM)-2 (CC-3162, Lonza) and fibroblast growth medium
complex 3D tissue models. On the other hand, this enabling (FGM)-2 media (CC-3132, Lonza), respectively. Cells between
design feature may also reflect the limitation of our approach passage 3 and 5 were used for device culture.

K DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

For modeling the adipose tissue (Figure 3), we used multipotent fluid reservoir containing dye/bead solution was inserted into the inlet
human adipose-derived stem cells (hASCs) (7510, ScienCell) and of one of the side microchannels while keeping its outlet blocked.
primary human adipose microvascular endothelial cells (hAMECs) Next, the inlet of the other side microchannel was sealed, but the
(7200, ScienCell). The adipose and vascular cells were cultured and outlet of this channel was left open. This configuration created a
maintained with hASC (PM-1, Zenbio) and hAMEC (1001, gradient of hydrostatic pressures across the hydrogel scaffold in a
ScienCell) growth media 25 cm2 flasks following the protocols diagonal direction, providing driving force for the flow of fluorescent
provided by the manufacturers. For best results, cells were used for tracers through the vessels. Vascular perfusion was monitored and
microfluidic culture before they reach passage 6. visualized using an epi-fluorescence microscope (Axio Observer D1,
A microengineered model of the outer blood−retinal barrier Zeiss).
(oBRB) shown in Figure 4was constructed using primary human Construction of the Adipose Tissue Model. To form
retinal endothelial cells (ACBRI 181, Cell Systems), primary human vascularized adipose tissue in our microdevice, we first resuspended
ocular choroid fibroblasts (#6620, ScienCell), and induced trypsinized hASCs and hAMECs in their growth media and mixed
pluripotent stem cell (iPSC)-derived human retinal pigment epithelial them with collagen type I (final concentration: 2 mg/mL; 354236, BD
cells (RPEs). The endothelium growth medium (4Z0-500, Cell bioscience), fibrinogen (final concentration: 10 mg/mL, Sigma) and
Systems) and fibroblast specific medium (#2301, ScienCell) were thrombin (final concentration: 1 U/mL, Sigma). The final
used to maintain the retinal endothelial cells and choroidal fibroblasts, concentrations of hASCs and hAMECs were 2.5 × 106 cells/mL
respectively. For culture of iPSC-derived RPEs, the cells were cultured and 5 × 106 cells/mL, respectively. The cell-containing hydrogel
with specialized RPE-THT medium composed of 118 mL of DMEM/ solution was then injected into the cell culture chamber of our
F12 (D6421, Sigma), 118 mL of alpha-MEM (M8042, Sigma), 2.5 microdevice, which was kept in a regular cell culture incubator for
mL of sodium pyruvate (11360-070, ThermoFisher), 2.5 mL of gelation. Once the template needles were removed, a 1:1 mixture of
MEM-NEAA (11140-050, ThermoFisher), 1.25 mL of N1 supple- EGM-2 (22011, PromoCell) and subcutaneous basal medium (BM-1,
ment (N6530, Sigma), 1.25 mL of Glutamax (35050-061, Thermo- ZenBio) was introduced into the top medium reservoir and the side
Fisher), 62.5 mg of Taurine (T0625, Sigma), 5 μg of hydrocortisone microchannels. To induce adipogenesis, we used BM-1 supplemented
(H0396, Sigma), and 0.00325 μg of triiodothrionine (T5516, Sigma). with 20 nM of insulin (Actrapid, Novo Nodisk), 1 μM of
Before use, cells were cultured in complete growth media in 75 cm2 dexamethason (D2915, Sigma), and 0.5 mM of 3-isobutyl-1-
flasks according to the manufacture’s protocol. methylxanthine (IBMX, I7018, Sigma). This media condition was
Formation of Cell-Laden ECM Hydrogel Scaffolds in the maintained for 7 days to induce the differentiation of hASCs into
Microdevice. Prior to cell culture, the fully assembled device was adipocytes. Without the adipogenic factors in the media, hASC
first sterilized by exposing it to ultraviolet (UV) light (Electro-lite differentiation did not occur during this period. After 7 days of
ELC-500) for at least 30 min. Subsequently, the cell culture chamber culture, the supplements were replaced with 20 nM of insulin
was filled with a buffer solution containing 0.2 mg/mL of (Actrapid, Novo Nodisk) and 1 μM of dexamethason (D2915,
sulfosuccinimidyl 6-(4′-azido-2′-nitrophenylamino)hexanoate (Sulfo- Sigma). The cells were cultured with this maintenance medium for
SANPAH, 13414, Covachem) and irradiated with 365 nm UV light another 40 days. During this period, the media in the top reservoir
for 10 min to create surface coating for enhanced adhesion of ECM and the microchannels were replenished every other day.
hydrogel to PDMS surfaces of the cell culture chamber. Detection and Quantification of Leptin. To analyze leptin
After rinsing with DPBS (356008, Corning) twice, removable secretion from the microengineered adipose tissue, the medium in the
templates were inserted into the device through the side micro- top reservoir was collected on days 7, 15, and 40 of culture. We used a
channels (Figure 1d). We used a syringe needle with an outer human leptin ELISA kit (ab179884, abcam) to measure the
diameter of 400 μm (Jensen Global) as a template. Once the needles concentration of leptin in the samples. In the first step of analysis,
were positioned in place, 20 μL of a cell suspension solution 50 μL of conditioned medium was added to each well, followed by the
containing fibrinogen (10 mg/mL; F8630, Sigma), thrombin (1 U/ injection of another 50 μL of antibody solution. The plate was then
mL; T7513, Sigma), aprotinin (0.15 U/mL; A1153, Sigma), HUVECs sealed and incubated on an orbital shaker set at 400 rpm for 1 h at
(2.5 × 106 cells/mL), and NHLFs (5 × 106 cells/mL) was gently room temperature (RT). Subsequently, each well was washed 3 times
injected into the cell culture chamber. The device was then left in a with 350 μL of wash buffer provided by the manufacturer of the
cell culture incubator at 37 °C and 5% CO2 for 30 min. Upon ELISA kit and received 100 μL of tetramethylbenzidine (TMB)
completion of the gelation step, the needles were removed from the solution. After 10 min of incubation in the dark, 100 μL of stop
microchannels, and EGM-2 medium was added to the medium solution was added to each well. Finally, we used a multimode plate
reservoir and the side microchannels. reader to measure the optical density of samples in each well (M200,
Microfluidic Cell Culture. Following the formation of cell-laden Tecan).
fibrin hydrogel construct, the side microchannels were incubated with Modeling of the Blood−Retinal Barrier. Retinal cells between
a fibronectin solution (25 μg/mL in PBS; 356008, Corning) for 3 h at passage 3 and 5 were used to create a model of the oBRB. First, we
37 °C to create an ECM coating on the channel surface. Afterward, prepared 20 μL of a cell suspension solution containing type I
the channels were washed once with EGM-2, and 10 μL of HUVEC collagen (final concentration: 2 mg/mL; 354236, BD bioscience),
suspension (1 × 107 cells/mL) was introduced into both channels. fibrinogen (final concentration: 10 mg/mL; F8630, Sigma), thrombin
The seeded cells were then allowed to attach to the channel surface (final concentration: 1 U/mL; Sigma), aprotinin (final concentration:
over a period of 3 h. Once the cells established firm adhesion, external 0.15 U/mL; Sigma), primary human retinal microvascular endothelial
medium reservoirs were inserted into the channel inlets, and the cells (2.5 × 106 cells/mL), and primary human choroidal fibroblasts
outlet of each channel was connected to a syringe pump (Chemyx) (5 × 106 cells/mL). This mixture solution was injected into the cell
operating on a withdrawal mode to generate continuous flow of culture chamber and gelled in a cell culture incubator at 37 °C and 5%
culture media at a volumetric flow rate of 70 μL/h. This perfusion CO2 for 30 min. Subsequently, the needles were removed from the
culture condition was maintained to allow the endothelial cells to microchannels, and EGM-2 medium was added to the top medium
form confluent monolayers on the surface of the hollow circular reservoir and the side microchannels. Following one-day incubation,
channels in the hydrogel scaffold created by the template needles and retinal endothelial cells were seeded into the side channels and
to induce anastomosis between the endothelium lining and the self- cultured under flow conditions to induce vascular anastomosis as
assembled vasculature in the hydrogel. described above. After 24 h, iPSC-derived human RPEs were plated
Testing of Vascular Perfusability. To investigate the perfus- and grown on the open top surface of the cell-laden ECM scaffold.
ability of the engineered microvasculature, we used 75 kDa FITC- Culture media were replaced every day.
dextran (50 μg/mL in PBS; 46945, Sigma) and fluorescently labeled Preparation of Lung Cancer Spheroids. Tumor spheroids were
1-μm microbeads (FluoSpheres; F-8815, ThermoFisher) as flow generated by using a A549 human lung adenocarcinoma cell line (Cell
tracers. In the first step to generate flow through the vasculature, a Biolabs, Inc.). A549 cells were cultured in ultralow attachment 96-well

L DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

plates (7007, Corning) with DMEM/F12 medium (D6421, Sigma) constant stream of nitrogen gas until visibly dried and then lyophilized
supplemented with 10% fetal bovine serum (FBS, 1082147, for 1 to 2 h to remove any residual solvent. Dried lipid films were
ThermoFisher) and 1% penicillin/streptomycin (P/S). For each hydrated with buffer (sterile Dulbecco’s phosphate buffered saline
well, 100 μL of cell suspension containing 1000 A549 cells were (hereafter PBS), then went through three freeze−thaw cycles using
added, and the cells were allowed to settle and form a loose aggregate. liquid N2/50 °C water bath, and then extruded 10 cycles through 200
After overnight culture, 100 μL of medium was pipetted into each nm polycarbonate filters (Avanti Polar Lipid). Dynamic light
well. A549 spheroids generated over a period of 3 days were harvested scattering (DLS) measurements of hydrodynamic particle size,
for use in our microdevices. distribution, and PDI were taken at each step of formulation from
Production and Paclitaxel Treatment of the Vascularized extrusion and subsequent modifications at 1:125 dilution in PBS using
Tumor Model. To create a vascularized lung tumor construct in our a Zetasizer Nano ZSP (Malvern Panalytical, Malvern UK). Liposome
device, A549 spheroids were mixed with HUVECs and NHLFs in a concentration following extrusion was assessed with Nanosight
hydrogel solution containing fibrinogen (final concentration: 10 mg/ nanoparticle tracking analysis (Malvern).
mL; F8630, Sigma), thrombin (final concentration: 1 U/mL; Sigma), Strained Alkyne Modification of Antibodies for Conjuga-
and aprotinin (final concentration: 0.15 U/mL; Sigma). This solution tion to Liposomes. IgG and anti-ICAM monoclonal-Antibody
was introduced into the device to form a fibrin construct in the (mAb) were modified with dibenzylcyclooctyne-PEG4-NHS ester
culture chamber. The cells were cultured with EGM-2 for 7 days, (Jena Bioscience; Thuringia, Germany). The proteins, buffered in
during which the vascular network was formed and integrated with PBS and adjusted to pH 8.3 with 1 M NaHCO3 buffer, were reacted
A549 spheroids embedded in the hydrogel. For paclitaxel treatment, for 1 h at RT at a ratio of 1:7 mAb:NHS ester PEG4DBCO. Post-
the engineered vessels in the tumor construct were perfused with reaction, the mixture was buffer-exchanged with an amicon 10k
EGM-2 medium containing 2.5 μM paclitaxel for 48 h using the MWCO centrifugal filter (MilliporeSigma, Burlington MA) to remove
vascular perfusion technique described above. unreacted NHS ester PEG4DBCO. The extent of modification was
Viability Assay of A549 Spheroids. Upon completion of assessed via optical absorbance at 309 nm (corresponding to DBCO
paclitaxel treatment, a mixture of calcein AM and ethidium absorbance maximum, extinction coefficient 12000 M−1 cm−1),
homodimer-1 (2 μM and 4 μM in culture medium, respectively) compared to absorbance at 280 nm (corresponding to IgG
was introduced into the device and incubated at 37 °C for 20 min. absorbance, extinction coefficient 204000 M−1 cm−1).
Subsequently, the device was washed with DPBS three times and Ligand Conjugation to Liposomes. DBCO-modified antibodies
examined using an inverted epi-fluorescence microscope (Axio (200 per liposome) were incubated with azide-functionalized
Observer D1, Zeiss). For quantitative analysis, we estimated the liposomes overnight at RT. After incubation, the liposome−antibody
fraction of live and dead cells in each spheroid by measuring the mixture was characterized with dynamic light scattering (DLS),
average intensity of fluorescence generated by calcein AM and purified using size exclusion chromatography, concentrated to the
ethidium homodimer-1, respectively. In each device, five spheroids original volume against centrifugal filters (Amicon), and again
were used for our analysis. characterized with DLS.
Measurement of ROS Production and Caspase-3/7 Activa- Production of the Local Vascular Inflammation Model. LPS
tion. Following paclitaxel treatment, the CellROXgreen (5 μM in microbeads were produced by soaking gelatin microbeads (PCHMP-
DPBS; C10444, ThermoFisher) and CellEvent Caspase-3/7 Green (5 GB, Thies Technology) in 10 μg/mL LPS solution for 24 h. When
μM in DPBS; C10423, ThermoFisher) were used to measure vessel formation in the fibrin hydrogel (final concentration: 10 mg/
oxidative stress and apoptosis of endothelial cells, respectively. The mL; F8630, Sigma) was complete on day 7, the LPS microbeads were
dye solutions were injected into the vasculature and incubated at 37 mixed with type I collagen (final concentration: 2 mg/mL, BD
°C for 30 min. After three washing steps using DPBS, the vessels were bioscience) and injected into a preformed small cavity within the
imaged using an inverted epi-fluorescence microscope (Axio Observer fibrin scaffold. Upon gelation of the collagen−bead mixture (20 min
D1, Zeiss). To generate quantitative data, we averaged fluorescence at 37 °C), the device was incubated with EGM-2 for 4 h at 37 °C.
intensity measured from three regions of interest in each device. Liposome Binding Assay. Following the preparation of the local
T Lymphocytes Adhesion Assay. Human T lymphocytes were inflammation model, the anti-ICAM-1 liposomes were injected into
obtained from the Human Immunology Core at the University of the vessels through one of the side microchannels and allowed to flow
Pennsylvania and cultured in RPMI-1640 (11875-093, Thermo- through the vasculature. After 2 h of perfusion in a cell culture
Fisher) with 10% FBS, 1% P/S, and 1 μg/mL of Phytohemagglutinin- incubator, the device was washed with DPBS three times and
M (PHA-M; 11082132001, Sigma) for 3 days. Subsequently, we inspected to measure liposome binding in the vascular network.
changed the medium to RPMI-1640 with 10% FBS, 1% P/S, and 20 Immunostaining. For immunostaining, cells in our microfluidic
ng/mL of interleukin-2 (IL-2; 200-02, PeproTech) and cultured the devices were fixed in 4% paraformaldehyde for 15 min at RT, washed
cells for another three days. To test endothelial adhesion of T with DPBS, and permeabilized with 0.25% Triton X-100 for 10 min.
lymphocytes in our model, we labeled the cells with a fluorescent dye Subsequently, a blocking step was performed using 3% bovine serum
(CellTracker Green, ThermoFisher) and resuspended them in EGM- albumin (BSA) overnight at 4 °C. The cells were then incubated
2 (Lonza) at the final concentration of 3 × 106 cells/mL. The cell overnight at 4 °C with primary antibodies. For imaging of the self-
suspension was then injected into the vessels through one of the side assembled vessels, we used mouse monoclonal anti-CD31 (ab24590,
microchannels and allowed to flow through the vasculature for 1 h in 1:100, Abcam) and rabbit polyclonal antivimentin (ab92547, 1:200,
a cell culture incubator. At the completion of perfusion, the device Abcam) primary antibodies. For analysis of lipid droplets in the
was washed with DPBS three times and examined to analyze the adipose tissue model, the cells were incubated with rabbit polyclonal
number of adhered T cells in the vascular network. antiperilipin-1 antibody (ab3526, 1:100, Abcam). Visualization of
Liposome Preparation. Azide functionalized liposomes were RPEs in the oBRB model was achieved by using rabbit polyclonal
prepared by thin film hydration techniques similar to those previously antizonula occludens-1 (anti-ZO-1, ab7878, 1:100, Abcam), mouse
described.96 Briefly, lipids DPPC (1,2-dipalmitoyl-sn-glycero-3- monoclonal anti-RPE 65 (ab13826, 1:1000, Abcam), rabbit
phosphocholine), cholesterol, Top Fluor PC (1-palmitoyl-2-(dipyrro- monoclonal anti-Melanoma (ab137078, 1:400, Abcam), and mouse
metheneboron difluoride)undecanoyl-sn-glycero-3-phosphocholine), monoclonal F-actin (ab205, 1:100, Abcam) antibodies. Rabbit
and azide PEG2000DSPE (1,2-distearoyl-sn-glycero-3-phosphoethanol- polyclonal antilaminin antibody (ab11575, 1:200, Abcam) was used
amine-N-[azido(polyethylene glycol)-2000]) (All phospholipids to stain the basement membrane between RPEs and the choroid.
purchased from Avanti Polar Lipids, Alabaster, AL) in chloroform After incubation with primary antibody, the cells were washed
were combined in a borosilicate glass tube at a total lipid twice with PBS and incubated for 1 h at RT with secondary antibody
concentration of 20 mM. The lipid film was composed of 57.5 mol (A-11037, ThermoFisher; A-32723, ThermoFisher; ab150080
%:40 mol %:2 mol %:0.5 mol % DPPC/cholesterol/azide Abcam; ab150117, Abcam). We also used Hoechst (33342,
PEG2000DSPE/Top Fluor PC. Lipid solutions were subjected to a ThermoFisher) for nuclear staining. Fluorescence images of the

M DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

cells were captured by an inverted epi-fluorescence microscope National Science Foundation (CMMI:15-48571), the Paul G.
equipped with a confocal laser scanning module (LSM 800; Carl Allen Family Foundation, the Alternatives Research and
Zeiss, Germany). The obtained images were processed using the ZEN Development Foundation, and the University of Pennsylvania.
software (Zeiss, Germany).
SEM Imaging of Hydrogel Scaffolds. Scanning electron
D.H. is a recipient of the NIH Director’s New Innovator
microscopy (SEM) was conducted at the CDB Microscopy Core Award and the Cancer Research Institute Technology Impact
(Perelman School of Medicine, University of Pennsylvania). The Award.
hydrogel scaffolds were washed three times with 50 mM Na-
cacodylate buffer (pH 7.3) and fixed for 2 h with 2% glutaraldehyde in REFERENCES
50 mM Na-cacodylate buffer. Dehydration steps were then performed
in 30%, 50%, 75%, 95%, and 100% ethanol for 1.5 h. Subsequently, (1) Fishman, A. P. Endothelium: A Distributed Organ of Diverse
the scaffolds were incubated for 20 min in 50% HMDS in ethanol Capabilities. Ann. N. Y. Acad. Sci. 1982, 401, 1−8.
followed by three changes of 100% HMDS (Sigma-Aldrich Co.) and (2) Durand, M. J.; Gutterman, D. D. Diversity in Mechanisms of
overnight air-drying as described previously.110 Finally, the samples Endothelium-Dependent Vasodilation in Health and Disease. Micro-
were mounted on stubs and sputter coated with gold palladium. The circulation 2013, 20, 239−247.
specimens were observed and photographed using a Quanta 250 FEG (3) Cines, D. B.; Pollak, E. S.; Buck, C. A.; Loscalzo, J.; Zimmerman,
scanning electron microscope (FEI, Hillsboro, OR, USA) at 10 kV G. A.; McEver, R. P.; Pober, J. S.; Wick, T. M.; Konkle, B. A.;
accelerating voltage. Schwartz, B. S.; Barnathan, E. S.; McCrae, K. R.; Hug, B. A.; Schmidt,
AFM Measurement for Hydrogel Stiffness. Atomic force A. M.; Stern, D. M. Endothelial Cells in Physiology and in the
microscopy (AFM, MFP-3D-BIO, Asylum) was used to characterize Pathophysiology of Vascular Disorders. Blood 1998, 91, 3527−3561.
the mechanical property of hydrated hydrogel scaffolds used in the (4) Pugsley, M. K.; Tabrizchi, R. The Vascular System. An Overview
adipose and retina models shown in Figures 3 and 4. A Au coated of Structure and Function. J. Pharmacol. Toxicol. Methods 2000, 44,
cantilever (SCONT tip, NANOSENSORS) with a spring constant of 333−40.
14.58 pN/nm and a pyramid indenter was used to obtain force− (5) Rubin, L. L.; Staddon, J. M. The Cell Biology of the Blood-Brain
indentation curves. Young’s modulus was calculated from the force Barrier. Annu. Rev. Neurosci. 1999, 22, 11−28.
indentation data using the AtomicJ software. (6) Clarkson, P.; Celermajer, D. S.; Donald, A. E.; Sampson, M.;
Statistical Analysis. Statistical significance of the obtained data Sorensen, K. E.; Adams, M.; Yue, D. K.; Betteridge, D. J.; Deanfield, J.
was evaluated by a two-tailed t test. Data generated by three E. Impaired Vascular Reactivity in Insulin-Dependent Diabetes
independent experiments were presented as mean ± SEM. Mellitus Is Related to Disease Duration and Low Density Lipoprotein
Cholesterol Levels. J. Am. Coll. Cardiol. 1996, 28, 573−579.
ASSOCIATED CONTENT (7) Ziccardi, P.; Nappo, F.; Giugliano, G.; Esposito, K.; Marfella, R.;
*
S Supporting Information Cioffi, M.; D’Andrea, F.; Molinari, A. M.; Giugliano, D. Reduction of
The Supporting Information is available free of charge on the Inflammatory Cytokine Concentrations and Improvement of
ACS Publications website at DOI: 10.1021/acsnano.9b00686. Endothelial Functions in Obese Women after Weight Loss over
One Year. Circulation 2002, 105, 804−809.
Additional details on the dimensions of the device, the (8) Panza, J. A.; Quyyumi, A. A.; Callahan, T. S.; Epstein, S. E. Effect
AFM measurement and SEM imaging of hydrogel of Antihypertensive Treatment on Endothelium-Dependent Vascular
scaffolds, the morphological features of blood vessels, Relaxation in Patients with Essential Hypertension. J. Am. Coll.
the differentiation of hASCs, the engineered adipose Cardiol. 1993, 21, 1145−1151.
tissue, the contribution of choroidal vasculature on RPE (9) Ferrara, N. Vascular Endothelial Growth Factor and Age-Related
maturation, the drug-induced tumor toxicity, the Macular Degeneration: from Basic Science to Therapy. Nat. Med.
characterization of nanocarriers, and the nanocarrier 2010, 16, 1107−1111.
binding (PDF) (10) Pezzella, F. Cancer and Blood Vessels: A Complex Relation-
ship. J. Can. Sci. Res. 2016, DOI: 10.4172/2576-1447.1000S1-002.
Movie showing continuous flow of 1 μm fluorescent (11) Chrobak, K. M.; Potter, D. R.; Tien, J. Formation of Perfused,
particles observed through the vascular network (AVI) Functional Microvascular Tubes. Microvasc. Res. 2006, 71, 185−196.
(12) Booth, R.; Noh, S.; Kim, H. A Multiple-Channel, Multiple-
AUTHOR INFORMATION Assay Platform for Characterization of Full-Range Shear Stress Effects
Corresponding Author on Vascular Endothelial Cells. Lab Chip 2014, 14, 1880−1890.
*(D.H.) E-mail: [email protected]. (13) Zahedmanesh, H.; Lally, C. Multiscale Modeling in Vascular
Disease and Tissue Engineering. In Multiscale Computer Modeling in
ORCID Biomechanics and Biomedical Engineering; Gefen, A., Ed.; Springer
Jungwook Paek: 0000-0003-4624-1996 Berlin Heidelberg: Berlin, Heidelberg, 2013; pp 241−258.
Sunghee E. Park: 0000-0002-2909-6507 (14) Miller, J. S.; Stevens, K. R.; Yang, M. T.; Baker, B. M.; Nguyen,
Elizabeth D. Hood: 0000-0002-6829-8339 D. H.; Cohen, D. M.; Toro, E.; Chen, A. A.; Galie, P. A.; Yu, X.;
Chaturvedi, R.; Bhatia, S. N.; Chen, C. S. Rapid Casting of Patterned
Author Contributions Vascular Networks for Perfusable Engineered Three-Dimensional
○
(J.P. and S.E.P.) These authors contributed equally to this Tissues. Nat. Mater. 2012, 11, 768−774.
work. (15) Scharrer, E. The Functional Significance of the Capillary Bed in
Notes the Brain of the Opossum. Anat. Rec. 1939, 75, 319−340.
The authors declare no competing financial interest. (16) Maton, A. Human Biology and Health; Prentice Hall:
Englewood Cliffs, NJ, 1993.
ACKNOWLEDGMENTS (17) Hasan, A.; Paul, A.; Vrana, N. E.; Zhao, X.; Memic, A.; Hwang,
Y. S.; Dokmeci, M. R.; Khademhosseini, A. Microfluidic Techniques
We thank G. Al, M. Farrell, J. Seo, G.S. Worthen, and A. for Development of 3D Vascularized Tissue. Biomaterials 2014, 35,
Georgescu for their input and technical assistance. This work 7308−7325.
was supported by the National Institutes of Health (NIH) (18) Ehsan, S. M.; Welch-Reardon, K. M.; Waterman, M. L.;
(1DP2HL127720-01, 1UG3TR002198-01, 1UC4DK104196- Hughes, C. C.; George, S. C. A Three-Dimensional In Vitro Model of
01, U24DK076169), the Cancer Research Institute, the Tumor Cell Intravasation. Integr. Biol. 2014, 6, 603−610.

N DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

(19) Wang, X.; Phan, D. T.; Sobrino, A.; George, S. C.; Hughes, C. Desorption/Ionization-Time of Flight Mass Spectrometry Reveals
C.; Lee, A. P. Engineering Anastomosis Between Living Capillary an Active Metabolism of the Extracellular Matrix of 3T3-L1
Networks and Endothelial Cell-Lined Microfluidic Channels. Lab Adipocytes. Proteomics 2004, 4, 3855−3863.
Chip 2016, 16, 282−290. (40) Mariman, E. C.; Wang, P. Adipocyte Extracellular Matrix
(20) Kim, S.; Lee, H.; Chung, M.; Jeon, N. L. Engineering of Composition, Dynamics and Role in Obesity. Cell. Mol. Life Sci. 2010,
Functional, Perfusable 3D Microvascular Networks on a Chip. Lab 67, 1277−1292.
Chip 2013, 13, 1489−1500. (41) Chun, T. H.; Hotary, K. B.; Sabeh, F.; Saltiel, A. R.; Allen, E.
(21) Nashimoto, Y.; Hayashi, T.; Kunita, I.; Nakamasu, A.; D.; Weiss, S. J. A Pericellular Collagenase Directs the 3-Dimensional
Torisawa, Y. S.; Nakayama, M.; Takigawa-Imamura, H.; Kotera, H.; Development of White Adipose Tissue. Cell 2006, 125, 577−591.
Nishiyama, K.; Miura, T.; Yokokawa, R. Integrating Perfusable (42) Guilak, F.; Cohen, D. M.; Estes, B. T.; Gimble, J. M.; Liedtke,
Vascular Networks with a Three-Dimensional Tissue in a Microfluidic W.; Chen, C. S. Control of Stem Cell Fate by Physical Interactions
Device. Integr. Biol. 2017, 9, 506−518. with the Extracellular Matrix. Cell Stem Cell. 2009, 5, 17−26.
(22) Oh, S.; Ryu, H.; Tahk, D.; Ko, J.; Chung, Y.; Lee, H. K.; Lee, T. (43) Smith, L. R.; Cho, S.; Discher, D. E. Stem Cell Differentiation
R.; Jeon, N. L. Open-Top” Microfluidic Device for In Vitro Three- Is Regulated by Extracellular Matrix Mechanics. Physiology 2018, 33,
Dimensional Capillary Beds. Lab Chip 2017, 17, 3405−3414. 16−25.
(23) Yeon, J. H.; Ryu, H. R.; Chung, M.; Hu, Q. P.; Jeon, N. L. In (44) Huebsch, N.; Arany, P. R.; Mao, A. S.; Shvartsman, D.; Ali, O.
Vitro Formation and Characterization of a Perfusable Three- A.; Bencherif, S. A.; Rivera-Feliciano, J.; Mooney, D. J. Harnessing
Dimensional Tubular Capillary Network in Microfluidic Devices. Traction-Mediated Manipulation of the Cell/Matrix Interface to
Lab Chip 2012, 12, 2815−2822. Control Stem-Cell Fate. Nat. Mater. 2010, 9, 518−526.
(24) Chen, M. B.; Whisler, J. A.; Fröse, J.; Yu, C.; Shin, Y.; Kamm, R. (45) Freiman, A.; Shandalov, Y.; Rozenfeld, D.; Shor, E.; Segal, S.;
D. On-Chip Human Microvasculature Assay for Visualization and Ben-David, D.; Meretzki, S.; Egozi, D.; Levenberg, S. Adipose-Derived
Quantitation of Tumor Cell Extravasation Dynamics. Nat. Protoc. Endothelial and Mesenchymal Stem Cells Enhance Vascular Network
2017, 12, 865−880. Formation on Three-Dimensional Constructs In Vitro. Stem Cell Res.
(25) Whisler, J. A.; Chen, M. B.; Kamm, R. D. Control of Perfusable Ther. 2016, 7, 5.
Microvascular Network Morphology Using a Multiculture Micro- (46) Sztalryd, C.; Brasaemle, D. L. The Perilipin Family of Lipid
fluidic System. Tissue Eng., Part C 2014, 20, 543−552. Droplet Proteins: Gatekeepers of Intracellular Lipolysis. Biochim.
(26) Chen, Y. C.; Lin, R. Z.; Qi, H.; Yang, Y.; Bae, H.; Melero- Biophys. Acta, Mol. Cell Biol. Lipids. 2017, 1862, 1221−1232.
Martin, J. M.; Khademhosseini, A. Functional Human Vascular (47) Itabe, H.; Yamaguchi, T.; Nimura, S.; Sasabe, N. Perilipins: A
Network Generated in Photocrosslinkable Gelatin Methacrylate Diversity of Intracellular Lipid Droplet Proteins. Lipids Health Dis.
Hydrogels. Adv. Funct. Mater. 2012, 22, 2027−2039. 2017, 16, 83.
(27) Bajaj, P.; Schweller, R. M.; Khademhosseini, A.; West, J. L.; (48) Cao, Y. Adipose Tissue Angiogenesis as a Therapeutic Target
Bashir, R. 3D Biofabrication Strategies for Tissue Engineering and for Obesity and Metabolic Diseases. Nat. Rev. Drug Discovery 2010, 9,
Regenerative Medicine. Annu. Rev. Biomed. Eng. 2014, 16, 247−276.
107−115.
(28) Linville, R. M.; Boland, N. F.; Covarrubias, G.; Price, G. M.;
(49) Harris, R. B. Direct and Indirect Effects of Leptin on Adipocyte
Tien, J. Physical and Chemical Signals That Promote Vascularization
Metabolism. Biochim. Biophys. Acta, Mol. Basis Dis. 2014, 1842, 414−
of Capillary-Scale Channels. Cell. Mol. Bioeng. 2016, 9, 73−84.
423.
(29) Zhang, W.; Wray, L. S.; Rnjak-Kovacina, J.; Xu, L.; Zou, D.;
(50) Stern, J. H.; Rutkowski, J. M.; Scherer, P. E. Adiponectin,
Wang, S.; Zhang, M.; Dong, J.; Li, G.; Kaplan, D. L.; Jiang, X.
Leptin, and Fatty Acids in the Maintenance of Metabolic Homeostasis
Vascularization of Hollow Channel-Modified Porous Silk Scaffolds
through Adipose Tissue Crosstalk. Cell Metab. 2016, 23, 770−784.
with Endothelial Cells for Tissue Regeneration. Biomaterials 2015, 56,
(51) Cunha-Vaz, J.; Bernardes, R.; Lobo, C. Blood-Retinal Barrier.
68−77.
(30) Yancopoulos, G. D.; Davis, S.; Gale, N. W.; Rudge, J. S.; Eur. J. Ophthalmol. 2011, 21, 3−9.
(52) Campbell, M.; Humphries, P. The Blood-Retina Barrier: Tight
Wiegand, S. J.; Holash, J. Vascular-Specific Growth Factors and Blood
Vessel Formation. Nature 2000, 407, 242−248. Junctions and Barrier Modulation. Adv. Exp. Med. Biol. 2013, 763,
(31) Morin, K. T.; Tranquillo, R. T. In Vitro Models of Angiogenesis 70−84.
and Vasculogenesis in Fibrin Gel. Exp. Cell Res. 2013, 319, 2409− (53) Bhutto, I.; Lutty, G. Understanding Age-Related Macular
2417. Degeneration (AMD): Relationships between the Photoreceptor/
(32) Diaz-Santana, A.; Shan, M.; Stroock, A. D. Endothelial Cell Retinal Pigment Epithelium/Bruch’s Membrane/Choriocapillaris
Dynamics during Anastomosis In Vitro. Integr. Biol. 2015, 7, 454−466. Complex. Mol. Aspects Med. 2012, 33, 295−317.
(33) Song, J. W.; Bazou, D.; Munn, L. L. Anastomosis of Endothelial (54) Jager, R. D.; Mieler, W. F.; Miller, J. W. Age-Related Macular
Sprouts Forms New Vessels in aTissue Analogue of Angiogenesis. Degeneration. N. Engl. J. Med. 2008, 358, 2606−2617.
Integr. Biol. 2012, 4, 857−862. (55) Ambati, J.; Fowler, B. J. Mechanisms of Age-Related Macular
(34) Rosen, E. D.; Spiegelman, B. M. Adipocytes as Regulators of Degeneration. Neuron 2012, 75, 26−39.
Energy Balance and Glucose Homeostasis. Nature. 2006, 444, 847− (56) Kamao, H.; Mandai, M.; Okamoto, S.; Sakai, N.; Suga, A.;
853. Sugita, S.; Kiryu, J.; Takahashi, M. Characterization of Human
(35) Tilg, H.; Moschen, A. R. Adipocytokines: Mediators Linking Induced Pluripotent Stem Cell-Derived Retinal Pigment Epithelium
Adipose Tissue, Inflammation and Immunity. Nat. Rev. Immunol. Cell Sheets Aiming for Clinical Application. Stem Cell Rep. 2014, 2,
2006, 6, 772−783. 205−218.
(36) Gray, S. L.; Vidal-Puig, A. J. Adipose Tissue Expandability in (57) Nugroho, R. W. N.; Harjumäki, R.; Zhang, X.; Lou, Y. R.;
the Maintenance of Metabolic Homeostasis. Nutr. Rev. 2007, 65, S7− Yliperttula, M.; Valle-Delgado, J. J.; Ö sterberg, M. Quantifying the
S12. Interactions between Biomimetic Biomaterials - Collagen I, Collagen
(37) Chen, C. W.; Corselli, M.; Peault, B.; Huard, J. Human Blood- IV, Laminin 521 and Cellulose Nanofibrils - by Colloidal Probe
Vessel-Derived Stem Cells for Tissue Repair and Regeneration. J. Microscopy. Colloids Surf., B 2019, 173, 571−580.
Biomed. Biotechnol. 2012, 2012, 597439. (58) Sorkio, A. E.; Vuorimaa-Laukkanen, E. P.; Hakola, H. M.;
(38) Butler, J. M.; Kobayashi, H.; Rafii, S. Instructive Role of the Liang, H.; Ujula, T. A.; Valle-Delgado, J. J.; Ö sterberg, M.; Yliperttula,
Vascular Niche in Promoting Tumour Growth and Tissue Repair by M. L.; Skottman, H. Biomimetic Collagen I and IV Double Layer
Angiocrine Factors. Nat. Rev. Cancer 2010, 10, 138−146. Langmuir-Schaefer Films as Microenvironment for Human Pluri-
(39) Bouwman, F.; Renes, J.; Mariman, E. A Combination of Protein potent Stem Cell Derived Retinal Pigment Epithelial Cells.
Profiling and Isotopomer Analysis Using Matrix-Assisted Laser Biomaterials 2015, 51, 257−269.

O DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

(59) Rizzolo, L. J. Development and Role of Tight Junctions in the (80) De Jong, W. H.; Borm, P. J. Drug Delivery and Nanoparticles:
Retinal Pigment Epithelium. Int. Rev. Cytol. 2007, 258, 195−234. Applications and Hazards. Int. J. Nanomed. 2008, 3, 133−149.
(60) Rizzolo, L. J.; Peng, S.; Luo, Y.; Xiao, W. Integration of Tight (81) Duncan, R. The Dawning Era of Polymer Therapeutics. Nat.
Junctions and Claudins with the Barrier Functions of the Retinal Rev. Drug Discovery 2003, 2, 347−360.
Pigment Epithelium. Prog. Retinal Eye Res. 2011, 30, 296−323. (82) Senior, J.; Gregoriadis, G. Is Half-Life of Circulating Liposomes
(61) Boulton, M. E. Studying Melanin and Lipofuscin in RPE Cell Determined by Changes in Their Permeability? FEBS Lett. 1982, 145,
Culture Models. Exp. Eye Res. 2014, 126, 61−67. 109−114.
(62) Geisen, P.; McColm, J. R.; Hartnett, M. E. Choroidal (83) Seki, J.; Sonoke, S.; Saheki, A.; Fukui, H.; Sasaki, H.; Mayumi,
Endothelial Cells Transmigrate across the Retinal Pigment Epithelium T. A Nanometer Lipid Emulsion, Lipid Nano-Sphere (LNS), as a
but Do Not Proliferate in Response to Soluble Vascular Endothelial Parenteral Drug Carrier for Passive Drug Targeting. Int. J. Pharm.
Growth Factor. Exp. Eye Res. 2006, 82, 608−619. 2004, 273, 75−83.
(63) Song, H. B.; Jun, H. O.; Kim, J. H.; Lee, Y. H.; Choi, M. H.; (84) Wissing, S. A.; Kayser, O.; Müller, R. H. Solid Lipid
Kim, J. H. Disruption of Outer Blood-Retinal Barrier by Toxoplasma Nanoparticles for Parental Drug Delivery. Adv. Drug Delivery Rev.
Gondii-Infected Monocytes Is Mediated by Paracrinely Activated 2004, 56, 1257−1272.
FAK Signaling. PLoS One 2017, 12, No. e0175159. (85) Panyam, J.; Zhou, W. Z.; Prabha, S.; Sahoo, S. K.; Labhasetwar,
(64) Khazaei, M.; Moien-Afshari, F.; Laher, I. Vascular Endothelial V. Rapid Endo-Lysosomal Escape of Poly(DL-Lactide-Co-Glycolide)
Function in Health and Diseases. Pathophysiology 2008, 15, 49−67. Nanoparticles: Implications for Drug and Gene Delivery. FASEB J.
(65) Goel, S.; Duda, D. G.; Xu, L.; Munn, L. L.; Boucher, Y.; 2002, 16, 1217−1226.
Fukumura, D.; Jain, R. K. Normalization of the Vasculature for (86) Muzykantov, V.; Muro, S. Targeting Delivery of Drugs in the
Treatment of Cancer and Other Diseases. Physiol. Rev. 2011, 91, Vascular System. Int. J. Transp. Phenom. 2011, 12, 41−49.
1071−1121. (87) Koren, E.; Torchilin, V. P. Drug Carriers for Vascular Drug
(66) Roudsari, L. C.; West, J. L. Studying the Influence of Delivery. IUBMB Life 2011, 63, 586−595.
Angiogenesis in In Vitro Cancer Model Systems. Adv. Drug Delivery (88) Ferrari, M. Cancer Nanotechnology: Opportunities and
Rev. 2016, 97, 250−259. Challenges. Nat. Rev. Cancer 2005, 5, 161−171.
(67) Rowinsky, E. K. Clinical Pharmacology of Taxol. J. Natl. Cancer (89) Farokhzad, O. C.; Khademhosseini, A.; Jon, S.; Hermmann, A.;
Inst. Monogr. 1993, 15, 25−37. Cheng, J.; Chin, C.; Kiselyuk, A.; Teply, B.; Eng, G.; Langer, R.
(68) Ramalingam, S.; Belani, C. P. Paclitaxel for Non-Small Cell Microfluidic System for Studying the Interaction of Nanoparticles and
Lung Cancer. Expert Opin. Pharmacother. 2004, 5, 1771−1780. Microparticles with Cells. Anal. Chem. 2005, 77, 5453−5459.
(69) Svilaas, T.; Lefrandt, J. D.; Gietema, J. A.; Kamphuisen, P. W. (90) Ahn, J.; Ko, J.; Lee, S.; Yu, J.; Kim, Y.; Jeon, N. L. Microfluidics
Long-Term Arterial Complications of Chemotherapy in Patients with in Nanoparticle Drug Delivery; from Synthesis to Pre-Clinical
Cancer. Thromb. Res. 2016, 140, S109−S118. Screening. Adv. Drug Delivery Rev. 2018, 128, 29−53.
(70) Soultati, A.; Mountzios, G.; Avgerinou, C.; Papaxoinis, G.; (91) Mainardes, R. M.; Silva, L. P. Drug Delivery Systems: Past,
Pectasides, D.; Dimopoulos, M.; Papadimitriou, C. Endothelial Present, and Future. Curr. Drug Targets 2004, 5, 449−455.
Vascular Toxicity from Chemotherapeutic Agents: Preclinical (92) Moghimi, S. M.; Szebeni, J. Stealth Liposomes and Long
Evidence and Clinical Implications. Cancer Treat. Rev. 2012, 38, Circulating Nanoparticles: Critical Issues in Pharmacokinetics,
473−484. Opsonization and Protein-Binding Properties. Prog. Lipid Res. 2003,
(71) Cameron, A. C.; Touyz, R. M.; Lang, N. N. Vascular 42, 463−478.
Complications of Cancer Chemotherapy. Can. J. Cardiol. 2016, 32, (93) Hood, E. D.; Greineder, C. F.; Shuvaeva, T.; Walsh, L.; Villa, C.
852−862. H.; Muzykantov, V. R. Vascular Targeting of Radiolabeled Liposomes
(72) Semb, K. A.; Aamdal, S.; Oian, P. Capillary Protein Leak with Bio-Orthogonally Conjugated Ligands: Single Chain Fragments
Syndrome Appears to Explain Fluid Retention in Cancer Patients Provide Higher Specificity than Antibodies. Bioconjugate Chem. 2018,
Who Receive Docetaxel Treatment. J. Clin. Oncol. 1998, 16, 3426− 29, 3626−3637.
3432. (94) Medzhitov, R. Inflammation 2010: New Adventures of an Old
(73) Vassilakopoulou, M.; Mountzios, G.; Papamechael, C.; Flame. Cell 2010, 140, 771−776.
Protogerou, A.D.; Aznaouridis, K.; Katsichti, P.; Venetsanou, K.; (95) Rajendran, P.; Rengarajan, T.; Thangavel, J.; Nishigaki, Y.;
Dimopoulos, M.-A.; Ikonomidis, I.; Papadimitriou, C.A. Paclitaxel Sakthisekaran, D.; Sethi, G.; Nishigaki, I. The Vascular Endothelium
Chemotherapy and Vascular Toxicity as Assessed by Flow-Mediated and Human Diseases. Int. J. Biol. Sci. 2013, 9, 1057−1069.
and Nitrate-Mediated Vasodilatation. Vasc. Pharmacol. 2010, 53, (96) Hood, E. D.; Greineder, C. F.; Dodia, C.; Han, J.; Mesaros, C.;
115−121. Shuvaev, V. V.; Blair, I. A.; Fisher, A. B.; Muzykantov, V. R.
(74) Clements, M. K.; Jones, C. B.; Cumming, M.; Daoud, S. S. Antioxidant Protection by PECAM-Targeted Delivery of a Novel
Antiangiogenic Potential of Camptothecin and Topotecan. Cancer NADPH-Oxidase Inhibitor to the Endothelium In Vitro and In Vivo. J.
Chemother. Pharmacol. 1999, 44, 411−416. Controlled Release 2012, 163, 161−169.
(75) Vacca, A.; Ribatti, D.; Iurlaro, M.; Merchionne, F.; Nico, B.; (97) Vodovotz, Y.; Csete, M.; Bartels, J.; Chang, S.; An, G.
Ria, R.; Dammacco, F. Docetaxel Versus Paclitaxel for Antiangio- Translational Systems Biology of Inflammation. PLoS Comput. Biol.
genesis. J. Hematother. Stem Cell Res. 2002, 11, 103−118. 2008, 4, No. e1000014.
(76) Serizawa, K.; Yogo, K.; Aizawa, K.; Tashiro, Y.; Takahari, Y.; (98) Vodovotz, Y. Deciphering the Complexity of Acute
Sekine, K.; Suzuki, T.; Ishizuka, N.; Ishida, H. Paclitaxel-Induced Inflammation Using Mathematical Models. Immunol. Res. 2006, 36,
Endothelial Dysfunction in Living Rats Is Prevented by Nicorandil via 237−245.
Reduction of Oxidative Stress. J. Pharmacol. Sci. 2012, 119, 349−358. (99) Kemp, W.; Burns, D. K.; Brown, T. G. Inflammation and
(77) Alexandre, J.; Hu, Y.; Lu, W.; Pelicano, H.; Huang, P. Novel Repair. In Pathology: The Big Picture; McGraw-Hill Education: 2007;
Action of Paclitaxel Against Cancer Cells: Bystander Effect Mediated Chapter 2, pp 13−22.
by Reactive Oxygen Species. Cancer Res. 2007, 67, 3512−3517. (100) Esch, E. W.; Bahinski, A.; Huh, D. Organs-On-Chips at the
(78) Mittal, M.; Siddiqui, M. R.; Tran, K.; Reddy, S. P.; Malik, A. B. Frontiers of Drug Discovery. Nat. Rev. Drug Discovery 2015, 14, 248−
Reactive Oxygen Species in Inflammation and Tissue Injury. Antioxid. 260.
Redox Signaling 2014, 20, 1126−1167. (101) Bergers, G.; Song, S. The Role of Pericytes in Blood-Vessel
(79) Michailidou, M.; Brown, H. K.; Lefley, D. V.; Evans, A.; Cross, Formation and Maintenance. Neuro-Oncology. 2005, 7, 452−464.
S. S.; Coleman, R. E.; Brown, N. J.; Holen, I. Microvascular (102) Koike, N.; Fukumura, D.; Gralla, O.; Au, P.; Schechner, J. S.;
Endothelial Cell Responses In Vitro and In Vivo: Modulation by Jain, R. K. Tissue Engineering: Creation of Long-Lasting Blood
Zoledronic Acid and Paclitaxel? J. Vasc. Res. 2010, 47, 481−493. Vessels. Nature 2004, 428, 138−139.

P DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX
ACS Nano Article

(103) Moore, E. M.; West, J. L. Harnessing Macrophages for

Vascularization in Tissue Engineering. Ann. Biomed. Eng. 2019, 47,
354−365.
(104) Lewis, C. E.; Harney, A. S.; Pollard, J. W. The Multifaceted
Role of Perivascular Macrophages in Tumors. Cancer Cell 2016, 30,
18−25.
(105) Chillo, O.; Kleinert, E. C.; Lautz, T.; Lasch, M.; Pagel, J. I.;
Heun, Y.; Troidl, K.; Fischer, S.; Caballero-Martinez, A.; Mauer, A.;
Kurz, A. R. M.; Assmann, G.; Rehberg, M.; Kanse, S. M.; Nieswandt,
B.; Walzog, B.; Reichel, C. A.; Mannell, H.; Preissner, K. T.; Deindl,
E. Perivascular Mast Cells Govern Shear Stress-Induced Arterio-
genesis by Orchestrating Leukocyte Function. Cell Rep. 2016, 16,
2197−2207.
(106) Nerem, R. M. Hemodynamics and the Vascular Endothelium.
J. Biomech. Eng. 1993, 115, 510−514.
(107) Secomb, T. W. Hemodynamics. Compr. Physiol. 2016, 6, 975−
1003.
(108) Choi, N. W.; Cabodi, M.; Held, B.; Gleghorn, J. P.; Bonassar,
L. J.; Stroock, A. D. Microfluidic Scaffolds for Tissue Engineering.
Nat. Mater. 2007, 6, 908−915.
(109) Dutta, D.; Heo, I.; Clevers, H. Disease Modeling in Stem Cell-
Derived 3D Organoid Systems. Trends Mol. Med. 2017, 23, 393−410.
(110) Braet, F.; De Zanger, R.; Wisse, E. Drying Cells for SEM,
AFM and TEM by Hexamethyldisilazane: A Study on Hepatic
Endothelial Cells. J. Microsc. 1997, 186, 84−87.

Q DOI: 10.1021/acsnano.9b00686
ACS Nano XXXX, XXX, XXX−XXX