A 3D-printed blood-brain barrier model with tunable topology and cell-matrix interactions

Recent developments in digital light processing (DLP) can advance the structural and biochemical complexity of perfusable in vitro models of the blood–brain barrier. Here, we describe a strategy to functionalize complex, DLP-printed vascular models with multiple peptide motifs in a single hydrogel. Different peptides can be clicked into the walls of distinct topologies, or the peptide motifs lining channel walls can differ from those in the bulk of the hydrogel. The flexibility of this approach is used to both characterize the effects of various bioactive domains on endothelial coverage and tight junction formation, in addition to facilitating astrocyte attachment in the hydrogel surrounding the endothelialized vessel to mimic endothelial–astrocyte interaction. Peptides derived from proteins mediating cell-extracellular matrix (e.g. RGD and IKVAV) and cell–cell (e.g. HAVDI) adhesions are used to mediate endothelial cell attachment and coverage. HAVDI and IKVAV-lined channels exhibit significantly greater endothelialization and increased zonula-occluden-1 (ZO-1) localization to cell–cell junctions of endothelial cells, indicative of tight junction formation. RGD is then used in the bulk hydrogel to create an endothelial–astrocyte co-culture model of the blood–brain barrier that overcomes the limitations of previous platforms incapable of complex topology or tunable bioactive domains. This approach yields an adjustable, biofabricated platform to interrogate the effects of cell-matrix interaction on blood–brain barrier mechanobiology.


Introduction
Advances in 3D bioprinting increase the complexity and branching of predetermined vascular topologies within hydrogels.Specifically, the development of digital light processing (DLP) facilitates the patterning of interpenetrating networks of any predetermined geometry within photocrosslinkable hydrogels [1].This fabrication technique has mostly been limited to self-assembling monomers [2] that cannot incorporate covalently bound biochemical cues, since the polymer backbone binds to itself.However, thiol-ene based reactions in poly-ethylene glycol (PEG) norbornene have recently been incorporated with DLP workflows to yield an additional tunable parameter through a crosslinker, which in turn frees up additional binding domains [3].This multi-arm polymeric Michael Addition type scheme [4] provides the ability to utilize secondary crosslinking of thiolated binding motifs derived from cell-cell and cell-extracellular matrix (ECM) adhesions into the lining of channels after the geometry has been printed.This strategy allows for tuning of cell-matrix interactions of endothelial cells in complex vascular networks.
However, coupling DLP with thiol-ene reaction schemes also facilitates binding of peptides within the bulk of the hydrogel.Near UV photochemical Norrish type I-α cleavage drives selective thiolene peptide functionalization to the polymer backbone during the printing process.These materials include both synthetic, PEG-based formulations as well as naturally derived biomaterials like hyaluronan.Different reaction systems for selective peptide modification includes Huisgen cycloaddition via PEG tetra-azide [5] in addition to thiol-ene mediated polymerization via norbornene, given its high number of n-domains which create faster reaction kinetics and subsequently lower exposure times for live cell printing.Selective and spatial control of peptide binding using this approach has been demonstrated by many groups in both 2D and 3D topologies [4].These techniques use click chemistry to selectively bind peptides to a desired region within the hydrogel framework [6,7].The ability to bind peptides both to the lining of 3D-printed vessels and within the bulk of the hydrogel enables the development of new coculture vascular models.
One biological system that can be modeled using 3D-printed hydrogels with spatial segregation of bioactive domains is the blood-brain barrier, given the importance of endothelial-astrocyte interaction.Astrocytes extend endfeet to the basal side of endothelial cells and this structure is unique to vasculature in the central nervous system.Although our previous in vitro studies have shown that shear stress is capable of tight junction formation in the absence of glial cells [8], astrocyte interaction has been shown to facilitate the endothelial response to inflammatory stimuli and to alter the expression of tissue plasminogen activator in cultured blood-brain barrier endothelial cells [9].In vivo, astrocyte interaction with the endothelium has been found to mediate a wide range of processes including transport associated with the glymphatic system [10] Therefore, the ability of an in vitro model to incorporate endothelial-astrocyte interaction is crucial to accurately mimic this aspect of the blood-brain barrier [11].
The primary benefit of engineered microenvironments is the ability to specify available binding motifs that dictate cell function and more closely mimic in vivo conditions.For instance, the composition and structure of the extracellular matrix has been demonstrated to affect endothelial cell function and flow mechanosensing by dictating integrin binding and subsequent signaling transduction [12].The endothelium binds to a laminin-rich basement membrane that exhibits a host of peptide motifs known to interact with integrins on the basal surface of the cell.Previous studies have demonstrated that laminin-associated motifs including arginineglycine-aspartic acid (RGD) and isoleucine-lysinevaline-alanine-valine (IKVAV) can create engineered microenvironments that mimic native cell-matrix interaction and support endothelial growth and tubulogenesis [13].RGD in particular has been used to facilitate endothelial attachment in both 2D [14,15] and 3D [16] systems.The RGD motif binds to a wide array of integrin isoforms, perhaps the most well-characterized being αVβ3 [17], that are known to mediate the endothelial response to shear stress [18].The peptide motif has also been shown to exert protective effects on glial cells [19] making it a candidate to enable astrocyte attachment within the bulk of 3D-printed hydrogels.
In this study, we describe a process to print vascular topologies in PEG-Norbornene hydrogels using DLP, followed by the spatial patterning of peptide motifs via a secondary cross-linking process both in the walls of the channels and within the bulk of the hydrogel.This approach provides a platform to interrogate the effect of cell-matrix interactions on endothelial coverage and lumen formation, and to implement endothelial-astrocyte interaction in 3Dprinted topologies.

Functionalizing cell binding motifs
Radicalized driven thiol-ene reaction schemes have been utilized previously for functionalizing biomaterials [6].Several previously used thiolated peptide cell binding motifs were functionalized to the walls of the printed geometry.Fluorescent labeled peptides were first synthesized via peptide synthesizer.Red-fluorescent protein-labeled peptide (GCDDDK-RFP) and green fluorescent protein labeled peptide (RADA16-GFP) were both synthesized to demonstrate selective and controlled protein functionalization within different complex geometries.Lyophilized RFP and GFP were both brought to room temperature and diluted to 10 mg ml −1 in 1X PBS.The bifurcation geometry print included RFP peptide in the bulk solution at 1:1000, while 300 µl of GFP was flowed into the channel at 10 µg ml −1 and hit with 10 mW cm −2 for 180 s.Process was repeated in the same manner for channel functionalization of the 'Rowan University' (RU) geometry.Similarly, HAVDI(HAVDIGGGC), derived from N-Cadherein and RGD(GCGYGRGDSPG), derived from multiple ECM proteins, were synthesized and purchased from Genscript.IKVAV(GCGGGIKVAVG), a laminin derived motif, was fabricated using a peptide synthesizer.About 0.5 wt% Irgacure I2959 was diluted to 0.05 wt% in PBS and thiolated peptide was added to a final concentration of 5 mM.About 50 µl of peptide solution was injected into the channel and 10 mW cm −2 was radiated onto surface of gel for 60 s.This process was repeated after inverting the gel to assure full coverage of the peptide onto the surface of the channel.Gels were then washed out 2X for 30 min in PBS to remove any unbound peptide.

Endothelial cell seeding
All endothelial cell experiments were performed with P23-25 hCMEC/D3 gifted from Dr Robert Nagele's lab at the Rowan School of Osteopathic Medicine.The passage number represents the passages since original immortalization in the laboratory of Prof. Pierre Olivier Couraud.P23 hCMEC/D3 endothelial cells (with and without transfection with RFP-LifeAct) were cultured until 90% confluent, trypsinized, and suspended at a concentration of 50 M ml −1 for hydrogel cell seeding.Hydrogels were seeded with 22 µl of 12.5 M ml −1 and placed face up at the bottom of six well plate with 1 mL of EGM-2 incubated at 37 • C/5% CO 2 .The cell seeding technique included 90 • rotations every 30 min with additional seeding every hour for 4 h [20].Following initial seeding, 8.8 µl of 12.5 M ml −1 HCMEC/D3 were injected into the bifurcation geometry for vessel coating.Hydrogels were kept in static culture for at least four days in EGM-2.

Mechanical testing
PEG-Norbornene was bioprinted at the same polymer density used to generate cell-seeded hydrogels.Briefly, gels were printed as discs with dimensions of 13 × 2 mm and washed out in 1X DPBS for 2 h at room temperature on a rocker.Gels were then stored at 4 • C overnight, prior to incubation in 37 • C/5% CO 2 for three days to mimic the conditions used for cell-seeded hydrogels.Gels were then moved onto a Discovery HR-3 rheometer fitted with a 20 mm flat parallel plate.Axial compression was set to a rate of 250 µm s −1 to 50% of the original gel height to measure force-displacement curves, which were used to calculate elastic modulus by measuring the slope of the stress-strain curve at approximately 10% strain.A camera was connected to the instrument to measure Poisson's ratio by taking images to measure both the transverse and axial displacement during compression.

Permeability testing
To assess the integrity of the barrier, permeability testing was required.Four different conditions were examined based upon peptide and mono or dual culture modality within the printed vasculature.HAVDI and RGD were chosen as peptides of interest decorated inside the lumen wall as previously described with either P24-P25 LifeAct hCMEC/D3 or both live cell printed P6-7 normal human astrocytes (NHAs) and the former.Printed vasculature was cultured for five days then transferred to a perfusion chamber and ported.Following, the device was loaded onto an epifluorescent microscope fitted with an environmental chamber set to 37 • C, 5% CO 2 , and 95% humidity.Vessels were perfused with 4 kDa dextran FITC at a flow rate of 5 µl min −1 using a syringe pump for 10 min.The flow rate was chosen to assure fully developed flow and to maintain consistency with previous work.Images were taken at 30 s intervals and the diffusion coefficients were established using the following equation: where P is the permeability coefficient, dI/dt is the rate of change in fluorescence intensity outside the 3D vessel, r is the vessel radius, and I 0 is the fluorescence intensity inside the 3D vessel.

Immunocytochemistry
Following static and/or flow culture within the vessels, hydrogels were washed 3X in PBS and placed in 4% paraformaldehyde for 20 min.Gels were washed in PBS and a primary monoclonal antibody for zonula-occluden-1 (ZO-1) (Cell Signaling Technologies) was added at 1:250 in PBS and incubated at 4 • C for 48 h.Gels were then washed for 10 min 3X in PBS on the rocker.Far red-labeled ZO-1 secondary antibody (Cell Signaling Technologies) was diluted 1:250 in PBS along with Hoesch at 2 ug ml −1 .Solution was added to the gels and incubated for 1 h at 37 • C. Gels were then washed and imaged on a Nikon Eclipse Ti Confocal microscope.

Quantification of cell coverage and ZO-1 localization
ImageJ was used for all image processing and analysis.For cell coverage assays, three unique gels were evaluated per condition.A z-projection was performed to create images for each gel.Each image was normalized to the same threshold intensity, set to the same ROI area, and made into a binary image prior to calculation of percent area coverage.Similarly in the ZO-1 localization images, the three ROIs evaluated were from three unique gels per condition.Which were analyzed to provide an average and standard deviation.ZO-1 intensity was quantified in both static and perfused culture at different locations within the branched network.The image intensity along approximately 70 µm lines across the image plane were plotted to quantify ZO-1 localization to cell-cell junctions.Root mean squares of this data were calculated to compare ZO-1 localization between experimental groups.

Perfusion of endothelial-astrocyte co-culture models
Vessels cultured in static conditions for four days were perfused using an Arduino-controlled peristaltic pump.Gels were ported inside a perfusion chamber with 20-gauge PTFE catheter tips connected to Luer locks with Tygon tubing (McMaster Carr).Vessels were connected in parallel through a dampener with EGM-2 for 24 h at a flow rate of approximately 5 ml min −1 , corresponding to a peak 4.4 Pa shear stress in the branches and a minimum 0.17 Pa in the region directly downstream of the bifurcation.

Confocal microscopy
Imaging was performed on a Nikon Eclipse Ti Confocal Microscope to assess cell coverage and morphology in 3D over time within printed microvessels.
Images were taken on days 1, 3, and 5 following cell seeding.

Statistical analysis
A two-factor analysis of variance (ANOVA) with posthoc Tukey's test was used to test for significance in the two variables of time and peptide coating in the cell coverage assay.In order to test for significance in the ZO-1 localization experiments, a one-factor ANOVA with post-hoc Tukey's test was utilized for each peptide condition tested.Subsequently, a Shapiro Wilk test was used to test for normality in the data set in figure 4. The test yielded a p-value of 0.4156, indicating that the data was normal.A two-factor ANOVA with post-hoc Tukey's test was used to test for significant differences in permeability between peptide conditions with and without the presence of astrocytes in the bulk.

Working principle of DLP-printed PEG-Nor hydrogels
Stepwise chain-growth polymerization of PEG-Nor (PEG-norbornene) was used to print hydrogels with a DLP-based LumenX+ bioprinter (figure 1(A)).Subsequent photo-patterning of R-peptides using an Irgacure 2959 photoinitiator under UV light (320-390 nm) yielded topological binding cues.Two different bioprinting workflows were demonstrated as a proof-of-concept for functionalizing multiple peptide motifs within a single topology.First, thiolated rhodamine-peptide was functionalized to the bulk hydrogel network during the print, followed by perfusing a thiolated GFP-labeled peptide into a bifurcation network topology and clicking peptides into the walls of the vessel (figure 1(B)).This protocol was the basis of the 3D-printed blood-brain barrier model, since it demonstrated the ability to click distinct bioactive domains into the bulk of the hydrogel and the channel walls.Therefore, a peptide motif intended to facilitate astrocyte attachment could be different from a motif in the lining of the channel designed for endothelial attachment.The second workflow involved printing two separate, interpenetrating topologies and perfusing them with two peptides to demonstrate a different approach to patterning multiple peptide motifs within a single hydrogel.

Spatial patterning of fluorescent motifs in 3D PEG-Nor networks
Using the first workflow described in figure 1(B), GFP-labeled peptides were flowed into and clicked into the lining of a branched topology within the hydrogel, which was printed within rhodaminelabeled peptides functionalized into the bulk (figure 2(A)).A cross section of the channel shows an evenly distributed, uniform coating of GFP-labeled peptide with an equal distribution of rhodaminelabeled peptide in the bulk (figure 2(B)).A profile plot along the x-axis of the gel overlays red and green peptide intensities across the plotted area (figure 2(C)).
The result of the second post-processing workflow shown in figure 1(B), lining two distinct, interpenetrating channels within the same bulk matrix with two different peptides, is shown in figure 2(D).The 'RU' geometry, for 'Rowan University' , includes multiple planes in the Z-axis demonstrating the ability to apply this technique within complex 3D vascular models.

Effect of peptide motif on endothelialization in DLP-printed channel
To study endothelial attachment in channels lined with different peptides, cylindrical voids (650 µm diameter) were generated within the PEG-Nor hydrogels.Cell coverage in channels lined with three different peptides (IKVAV, HAVDI, and RGD) in addition to a negative control (no peptide clicked to the surface) were measured over time.Cell coverage did not increase in the control condition, although spherical aggregates formed within the peptide-free cylindrical channels (figure 3(A)).IKVAV, a binding motif of the basement membrane protein laminin, exhibited no noticeable difference in vessel coverage throughout the five days static culture.Rather, the cells adhered immediately to the peptide and began to spread (figure 3(C)).Cells in the channel lined with HAVDI, the active sequence of the cell-cell junction protein N-cadherin, initially appeared to favor cell-cell junctions and binding to each other.However, over the course of the experiment cells began to spread and cover the walls (figure 3(B)).Finally, cells in channels lined with RGD, a common motif in multiple ECM proteins, exhibited immediate spreading on PEG-norbornene and progressed linearly overtime (figure 3(D)).Interestingly, in each of the three conditions, endothelial cells began to develop monolayer coverage even in the absence of shear stress that has previously been shown to encourage endothelialization in collagen-based hydrogels [20].Quantification of vessel coverage indicated that both the HAVDI and IKVAV-lined channels exhibited significantly higher coverage than the control condition in static culture at day 5 (figure 3(E)).

Assessment of tight junction integrity within DLP-printed vessels
A crucial indicator of a robust and resilient barrier formation is localization of tight junction-associated proteins to the cell-cell junctions of the endothelium.Zonula occluden-1 (ZO-1), a tight junction scaffolding protein that is indicative of barrier formation [21] in hCMEC/D3 endothelial cells, was imaged in cells cultured in the peptide-lined channels in static conditions for five days.To quantify the images, intensity plots of the ZO-1 staining were measured and

Implementing endothelial-astrocyte interaction in a 3D-printed, branched topology
The primary benefit of DLP over other methods to create channels within hydrogels is the ability to print branched, physiologically relevant topologies that mimic the complex fluid flow regimes found in vivo.Therefore, the topology showed in figure 2(A) was chosen to verify the ability to 3Dprint an endothelial-astrocyte co-culture model.The mechanical properties of the DLP-printed hydrogels can also be modified by changing the light intensity and duration of printing to mimic central nervous system tissue.Supplemental figure 1 indicates that the printing parameters used to fabricate the hydrogels resulted in a mean elastic modulus of approximately 4 kPa (supplemental figure 1), which is on the same order of magnitude of previous measurements of brain tissue [22,23].However, an important distinction should be noted between viscoelastic tissue and elastic PEG-Nor hydrogels.First, full coverage of the branched topology with only endothelial cells was demonstrated in figure 5(A), prior to seeding the hydrogel with astrocytes.The RGD peptide motif was used to demonstrate that full channel coverage was possible even with a peptide motif that did not result in the highest ZO-1 localization to cell-cell junctions.
To assure that the printing process could incorporate live cells without reducing viability, a Live/Dead assay was performed with astrocytes in the PEG-Nor formulation containing RGD peptide.Figure 5(B) shows representative epifluorescent images of the live/dead assay for each of the three bioink formulations that were tested and subsequent viability of 83% for bioink 1. Bioink 1 formulation showed the highest viability due to lower near UV exposure times despite being in the presence of a higher concentration of cytotoxic LAP.Finally, to demonstrate successful printing of endothelial-astrocyte co-culture, bioink 1 was lined with the peptide that induced the highest ZO-1 localization to the cell-cell junctions, HAVDI, and seeded with endothelial cells.The channels were cultured for four days to allow full coverage of the channel prior to application of flow for 24 h.The channels were probed with anti-ZO-1 and anti-GFAP to show astrocytes in the bulk with endothelial cells lining the branched geometry at several locations in the channel (figure 5(C)).Additionally, a stitched confocal scan was completed demonstrating the spatial control of spread astrocytes in the bulk with barrier formation of an endothelium in the lumen (figure 5(D)).

Measuring the effect of astrocyte co-culture on barrier function
Permeability experiments were conducted to determine whether the presence of astrocytes affected the formation of tight junctions by the endothelium.Straight, cylindrical channels lined with either HAVDI or RGD peptide were seeded with endothelial cells and cultured in static conditions for five days.These hydrogels were fabricated with and without astrocytes to interrogate the effect of co-culturing the cells on barrier formation.Representative FITC dextran imaging is shown in figures 6(A) and (B). Figure 6(C) provides quantification of dextran permeability assays for channels with HAVDI or RGD motifs lining the surface.Channels lined with HAVDI exhibited significantly lower permeability coefficients, which is consistent with the ZO-1 staining in figure 4. The presence of astrocytes had no significant effect on permeability in HAVDIlined channels.However, astrocytes did significantly improve the barrier in RGD-lined channels, suggesting that the presence of the glial cells affected endothelial cells that were unable to form robust tight junctions.

Discussion
These results establish the efficacy and broad applicability of combining DLP with thiol-ene based photoinks to tune cell-ECM and cell-cell interactions within perfusable network topologies.The workflows described here demonstrate the ability to covalently bind one peptide in the bulk of the hydrogel and a different peptide to the lining of the channel wall, as well as to click different peptides into the lining of separate, interpenetrating networks.The former is used to create a model of the blood-brain barrier with increased functionality compared to previous approaches: both in terms of its branched, networklike topology and also its flexibility in tuning bioactive domains used to facilitate both endothelial and astrocyte attachment within the model.The relatively fast reaction kinetics of the PEG-Norbornene ameliorates the cell death associated with live cell printing, as evidenced by the viability of the astrocytes after the printing, and the three-day incubation of the co-culture prior to fixation.Overall, this approach provides the tools to create a new generation of 3D blood-brain barrier models to interrogate the effects of mechanical and biochemical stimuli on the integrity of the barrier.
As demonstrated in the results, the endothelial cells are significantly affected by the bioactive domains clicked into the lining of the channels.The IKVAV and HAVDI motifs provide the most favorable microenvironment for both vessel coverage and development of barrier function in hCMEC/D3 endothelial cells.The efficacy of IKVAV could be due to the prevalence of laminin in the basement membrane, which therefore provided a more native cell-ECM interaction for the endothelial cells.Several studies have demonstrated the benefit of recreating elements of the basement membrane to increase endothelial attachment in vitro [24,25].In contrast, HAVDI is a sequence from the extracellular domain of N-cadherin, which mediates cell-cell junctions [26][27][28].Therefore, clicking HAVDI into the channel wall simulated cell-cell adhesion on the basal side of the endothelial cells.Unlike the IKVAV condition, the cells in the HAVDI-lined channels took several days to cover the vessel wall.Yet, ZO-1 staining, and permeability testing indicated that the HAVDI motif resulted in the highest barrier integrity.However, the molecular mechanisms underlying this dynamic response as well as the development of endothelial barrier function in peptide-lined vessels remains unresolved.Previous studies have found that N-cadherin junctions increase RhoA activation and decrease Rac1 activation [29], which counters a recent finding that shear stress stabilizes tight junctions by decreasing RhoA activation [25].Future work can interrogate how specific peptide sequences affect both cytoskeletal-mediated and transcriptomic changes in endothelial cells.
The primary advantage of printing the PEG-Nor hydrogels with DLP is its ability to fabricate perfusable, multivascular topologies that mimic in vivo vasculature.In contrast to two photon-based approaches to spatially pattern photocrosslinkable ligands within three-dimensional hydrogels [14], DLP produces complex vascular topologies that support physiological flow rates.However, one of the current limitations of this technology is its limited resolution: the LumenX+ printer used in these studies has a minimum printing resolution of 50 µm, which substantially limits the ability to print small caliber vessel and capillary-scale vasculature.This feature is especially important for blood-brain barrier models, since the neurovascular unit is most relevant at the capillary scale.Nonetheless, the resolution used in this study is limited by the printer, not by the photoink and workflow used to generate the 3D-printed blood-brain barrier model.As printing technology improves its resolution, the same approach described here can be used to facilitate incorporation of bioactive domains to control endothelial cell and astrocyte attachment within 3D-printed topologies.
Despite the limitation in its resolution, the model described here can be used as a new platform to study the mechanobiology of the blood-brain barrier.Heterogeneous distributions of shear stress, even disturbed flow, can be applied within single network topologies to improve our understanding of the effects of complex fluid flow on the integrity of the barrier.Combining DLP with computational fluid dynamics provides a platform to correlate shear stress magnitude and gradients with changes in endothelial transcriptomics and proteomics.Moreover, the effect of peptide binding on the response of the endothelium to shear stress can also be interrogated in this platform.In addition to incorporating new peptide motifs not used in this study, equimolar ratios of multiple peptides can be covalently bound to the wall to better mimic the heterogeneous peptide sequences available to cells in the native basement membrane.This approach can be used to determine how changes in the extracellular matrix affects shear stress mechanotransduction.Finally, the advance described here is not limited to modeling of the blood-brain barrier, since there are multiple biological systems that feature prominent interaction between vascular and stromal cell types that can be modeled using this same workflow.

Figure 1 .
Figure 1.(A) Reaction schematic of Norrish Type 1α cleavage step growth photopolymerization of 20 kDa 8-arm PEG-norbornene via PEG-Dithiol crosslinker, followed by synthesis of photopatterning thiolated peptide motifs via Irgacure 2959.(B) Schematic of two distinct 3D bioprinting workflows for material functionalization of various peptide motifs, yielding spatially controlled covalently bound peptides in the bulk as well as lining the inside of the channels.

Figure 2 .
Figure 2. (A) 4X confocal stitched image of rhodamine peptide (GCDDDK-RFP) printed inside the bulk PEG-norbornene hydrogel and green fluorescent peptide (RADA16-GFP) functionalized to the walls of the bifurcation geometry (Scale Bar = 1 mm).(B) Cross section of the inlet displaying full coverage of RADA16-GFP covalently bound to the walls in 3D (Scale Bar = 100 µm).(C) Fluorescent peptide intensity plot over the cross section in (B) for both green and red fluorescent peptide (GFP, RFP).(D) Proof-of-concept for multiple and selective peptide functionalization in complex, interpenetrating topologies (Scale Bar = 1 mm).

Figure 3 .
Figure 3.Time course images of P23 transfected LifeAct HCMEC/D3 seeded cells at 50 M mL −1 captured on days 1, 3, and 5 in straight 650 µm channels with covalently bound peptide motifs.(A) Control, no peptide functionalized to the wall of the channel.(B) HAVDI, derived from N-Cadherin, (C) IKVAV, derived from laminin, and (D) RGD, derived from multiple ECM proteins, adhered to the walls of the vessel.(E) Percent cell coverage quantification within the vessel across five days for each peptide condition ( * = p < 0.05).Scale Bars = 100 µm.n = 3 per condition.

Figure 5 .
Figure 5. (A) Stitched bifurcation vessel with P23 LifeAct HCMEC/D3 cells lining the walls of the channel with RGD peptide (Scale Bar = 1 mm) and (A,i.)cross-section of endothelium in the branch (Scale Bar = 50 µm).(B) Live/Dead assay of printed astrocytes with different formulations (Bioinks 1 (B,i.), 2 (B,ii.), and 3(B,iii.))and viability after five days in culture, n = 3 per condition.(C, C,i) Composite images showing co-culture of endothelial cells and astrocytes in different locations of the 3D-printed vessel model (Scale Bars = 100 µm).(D) Stitched z-stack scan (Scale Bar = 1 mm) of entire bifurcation showing astrocytes in the bulk and endothelium coverage with ZO-1 localization.

Figure 6 .
Figure 6.Representative images of 4 kDa dextran-FITC perfused through vessels decorated with both (A) HAVDI and (B) RGD both without NHA.(C) Quantitative measurements of permeability coefficients with and without NHA in the bulk of the hydrogel.(Scale Bars = 100 µm).n = 3 per condition.

Table 1 .
Printing parameters for three bioink formulations.PEG-Dithiol concentration and laser power are constant across all formulations.
• C. The gel was then washed in medium on a rocker at 37 • C for 2 h, replacing medium every 30 min.A live/dead assay was performed on day 0 and day 5 of NHA culture inside of the 3D printed bifurcation model.About 2 µm of calcein AM and 4 µm of ethidium homodimer was added to 10 mL of PBS and the gels were incubated at 37 • C/5% CO 2 for 1 h prior to imaging.