Microvasculature‐on‐a‐Post Chip That Recapitulates Prothrombotic Vascular Geometries and 3D Flow Disturbance

Stenosis, characterized by partial vessel narrowing, alters blood hemodynamics and can lead to unpredictable thrombosis. Existing models struggle to accurately represent the complex vascular geometries and hemodynamics involved in such conditions. To address this challenge, a microvasculature‐on‐a‐post chip is developed to mimic partially stenotic vascular geometries and thrombogenicity, featuring isolated 3D micropost structures with variable sizes that recreate disturbed flow profiles. To emulate the diseased vessel wall, the post microfluidics are vascularized with a confluent layer of endothelial cells. Subsequently, human blood is perfused through the endothelialized post microfluidics, observing the temporal and spatial thrombotic response governed by Virchow's triad, including vessel wall injury, hemodynamic disturbance, and hypercoagulability. The innovative model offers valuable insights into stenosis‐induced thrombosis and endothelial behavior, paving the way for improved assessment of thrombotic risks associated with stenotic vessels. This advanced microfluidic platform also offers new approaches for evaluation of prothrombotic phenotypes and cardiovascular risk assessment in the future.


Introduction
Partial stenosis is a complex condition characterized by the intermediate narrowing of a blood vessel, which can cause pathological hemodynamics such as elevated shear, [1] turbulence, [2] and vorticity, [3] leading to unpredictable thrombosis.Stenosed vessels can result from atherosclerotic plaque which alters blood hemodynamics and causes endothelial damage, creating a thromboinflammatory microenvironment.This exacerbates the risk of plaque rupture and the subsequent emboli formation that can lodge in smaller vessels, leading to reduced blood flow and ischemia in the heart or brain. [4]However, the relationship between partial stenotic vascular anatomies and thrombosis in the context of flow disturbance remains elusive due to the lack of reliable methodologies and customizable model systems.
Clinically, thrombi retrieved from patients with acute ischemic stroke are dominated by red blood cell (RBC)-and are plateletrich.However, thrombi from patients with venous thromboembolism are dominated by RBCs and fibrin with little platelet composition. [5]The significance of the heterogenous characteristics of thrombus is not widely studied from the mechanobiological perspective.Therefore, it is critical to study the effect of hemodynamics on thrombosis development in partially stenosed vessels commonly caused by atherosclerosis.Nevertheless, exsiting in vivo models do not fully recapitulate the biomechanics and pathology of the unpredictable clot dynamics of diseased

F. Passam Central Clinical School Faculty Medicine and Health
The University of Sydney Camperdown, NSW 2006, Australia vasculature, [6] making it challenging to study the mechanism.Although, microfluidic models have been used to study the effect of disturbed flow on endothelial cell response, neutrophil adhesion, or platelet aggregation, [7] they lack the integration of 3D disturbed flow and functional endothlium to fully capture the vascular mechanobiology.
To address this challenge, we developed a 3D microvasculature-on-a-post chip device that emulates stenosis or pre-existing thrombi caused by atherosclerosis in the vasculature system.These devices incur flow disturbance and shear gradients, promoting the risk of occlusive blood clot formation.Incorporating 3D posts of varying sizes within our microchannel enables the simulation of different extents of intermediate stenosis with an array of shear gradients in blood vessels (Figure 1).Through the perfusion of human blood, we can visualize and quantify the platelet and fibrin deposition induced by disturbed fluid flow, inflamed endothelium, and blood hypercoagulability -Virchow's Triad [8] (Figure 1).Our endothelialized microfluidic post channelprovides valuable insights into the assessment of thinflammation-induced thrombosis upon endothelial dysfunction, the impact of hydrodynamic changes on the behavior of the endothelial cells, and the development and progression of thrombosis.These insights can help pave the way for the development of new and more effective antithrombotic diagnosis and therapeutic strategies.

Development of the Endothelialized Microfluidic Post Device
To replicate the vessel stenosis, we first fabricated the post microchannel using hybrid dual-layer soft lithography as previously described. [9]The microfluidic channel was designed with a width of 200 μm and a height of 70 μm, emulating the microvascular dimensions.Within the channel, eight identical cylindrical 3D microposts were housed at intervals (straight region) of 1000 μm (Figure 2A,i,ii).This arragement generated identical hydrodynamics around each post without interfering with one another (Figure S1A-C, Supporting Information).To examine the impact of flow disturbance, the diameter of the post geometries can be customized to either D = 80 or 40 μm (Figure 2A,iii), while maintaining a fixed height of H = 20 μm.
To incorporate the physiological function of the vessel wall, we then cultured human umbilical endothelial cells (HUVECs) to endothelialize the post microfluidics.Notably, we overcame the technical challenge to produce a confluent endothelial monolayer not only on the straight region but also on the 3D microposts with tight cell-cell junction (see the Experimental Section and Figure S1D in the Supporting Information).Specifically, reproducible and complete endothelialization were achieved throughout the entire microchannel (Figure 2B), including the channel side, top (z = 70 μm) and bottom (z = 0 μm), as well as the apex (z = 50 μm) and side of the posts (Figure 2B).The F-actin expression further confirmed that the endothelial cells grossly achieved confluency and showed physiologically relevant morphologies (Figure 2C).Of note, larger posts (D = 80 μm) can accommodate ≈4-5 endothelial cells on their apexes (Figure 2C,i), while there were only ≈1-2 endothelial cells attached on the apex of smaller post of D = 40 μm (Figure 2C,ii).

Impact of Post Induced Flow Disturbance on Endothelial Morphology
Under static condition, both the bodies and nuclei of endothelial cells in the straight region in the middle of the consecutive posts and on the post displayed more circular shape (Figure 3A).To evaluate how hydrodynamic forces caused by the microposts could alter the endothelial shape, orientation and polarization, we introduced continuous flow for 24 h using a peristaltic pump (Figure 3).Interestingly, when exposed to  0 = 600 s −1 bulk shear rate, the cell body of endothelium in the straight region displayed elongated along the direction of flow (Figure 3B) with the cell aspect ratio (the ratio of the long axis to short axis) increased from 2 to 3 (Figure 3E), while nucleus aspect ratio remained similar at ≈1.5 (Figure 3F).However, on the large (D = 80 μm) post, flow disturbance only elongated nuclei with increased aspect ratio from 1.56 to 1.67 (Figure 3B, yellow arrows; Figure 3F) but not the main body (Figure 3E) of the endothelial cells.Together, fluidic shear stress regulated endothelial cell morphology, and the nucleus could sense the strength and direction of the elevated shear flow induced by the stenosed post geometry (cf. Figure 5E,F) . [10]ditionally, we analyzed the orientation of the endothelium in response to shear flow, defined by the angle between the flow direction and the longest axis of an endothelial cell. [11]Under static condition, the endothelial cells in both straight, large and small post regions orientated and polarized randomly, and the addition of flow culture transformed the endothelial alignment parallel to the flow direction (Figure 3A,B; Figure S2A,B, Supporting Information).Compared to the static condition, the proportion of parallelly aligned endothelial cells in the straight region and on the large and small post increased significantly to 94.5%, 66.7%, and 72.4% respectively (Figure 3A,B,G).We further ex-amined the endothelial cell polarization which depicts the angle between the shear flow vector and the Golgi-nucleus vector [12] that forms by connecting the nucleus and the center of the Golgi body (Figure 3C,D) (see the Experimental Section).We divided the endothelial polarity into three categories: parallel polarization (0°-45°and 315°-360°around the flow axis), vertical polarization (45°-135°and 225°-315°around the flow axis) and antiparallel polarization (135°-225°around the flow axis) (Figure S3, Supporting Information).The results revealed that endothelial cells in static culture mostly polarized vertically to the flow, with Golgi bodies positioned on the top and bottom of the nucleus relative to the flow direction (Figure 3C,H; Figure S2C, Supporting Information).When exposed to shear flow, the Golgi bodies were mostly front-parallel to the nucleus (Figure 3D,H, Parallel 51%; Figure S2D, Supporting Information), while the proportion of vertical polarization in the straight region decreased from 58% to 24% (Figure 3H).However, when on the post, endothelial polarization was disrupted by the flow disturbance, with parallelly polarized endothelium on large post increased from 15% to 26% while that on small post decreased from 35% to 31% (Figure 3H; Figure S2D, Supporting Information).

Engineered Microvasculature Responds in a Physiological Manner to Proinflammatory Vascular Injury
Vessel wall injury in response to proinflammatory environments results in a disruptive hemodynamic state, leading to thrombosis. [13]In order to mimic the thrombo-inflammatory state, we treated the microvasculature-on-a-post chip with phorbol-12-myristate-13-acetate (PMA) (Figure 4).Subsequently, we compared the expression of inflammatory markers, cytoskeletal morphology and the junctional integrity of the endothelial cells situated on both the post and straight regions.
Upon PMA treatment, the endothelial cells on the post exhibited a more circular morphology with a cellular aspect ratio of ≈1.31, while the untreated cells were more elongated and had a higher aspect ratio at around 2.25 (Figure 4A-C).It was also observed that PMA treatment disorganizes the endothelial actin cytoskeleton predominantly with short and disrupted actin stress fibers and actin rich nodules in singular (podosomes, square) and clustered form (rosettes, circle), as opposed to the untreated endothelium with a well-organized and aligned actin stress fibers (Figure 4B).Additionally, the PMA-stimulated endothelial cells on the post had shorter F-actin fibers (13.73 μm) and occupied a smaller percentage of the set cellular area (25.54%) compared to the control (length = 25.33 μm; area = 47.61%)(Figure 4E,F).Similar trends of the cellular aspect ratio, F-actin length and Factin occupied area were observed in the endothelium of the straight region (Figure 4C,E,F).
Regarding the endothelial barrier characteristics, we further found that PMA-treated endothelium exhibited a significant decrease in PECAM-1 expression compared to the untreated cells on both the post and the straight regions (Figure 4A,D).Inflamed endothelial cells underwent active shedding of the PECAM-1, leading to the loss of cell-cell junction and downregulation of PECAM-1 expression. [14]Notably, we also observed a thromboinflammatory endothelial profile in response to PMA treatment (Figure 4G).The endothelium in the straight region had a 3.4 and 3.5 folds higher fluorescent intensity of ICAM-1 and Eselectin respectively after 1.5 h PMA treatment (Figure 4G,H).This trend in inflammatory behavior was mirrored in post cells compared to nontreated conditions (Figure 4G,H).Remarkably, cells situated on both small and large posts exhibited an average of 3 folds higher ICAM-1 (14.92 a.u.), and fivefolds higher E-selectin expression (15.95 a.u.) compared to those situated on the straight channel without PMA stimulation (Figure 4H).This further resulted in even higher ICAM-1 (24.64 a.u.) and E-selectin (25.45 a.u.) signal densities for the PMA stimulated endothelial cells on the post (Figure 4H).These results established the microvasculature-on-a-post chip as an in vitro humanized model for inflammation induced thrombosis.

Endothelial Injury and Flow Disturbance Trigger Microvascular Thrombosis
To recapitulate the thrombotic response, citrated whole blood was recalcified and perfused through the microvasculature-on-a-post chip at a bulk shear rate  0 = 600 s −1 for 5 min (Figure 2A,i,ii; Figure 5; Figure S4A, Supporting Information).Confocal microscopy was used to capture the platelet aggregation (white) and fibrin formation (magenta) at the first post in real time (Figure 5A).In untreated and healthy endothelium, there was little platelet and fibrin deposition during the blood perfusion (Figure 5A,B, 1st row; 5C and D).However, upon PMA stimulation, the inflamed endothelium in both large and small posts caused platelets to adhere and form small aggregates, followed by fibrin formation and large thrombosis predominantly at the downstream of the post after 2 min of blood perfusion (Figure 5A,B, 2nd row; Figure 5C,D).The initially deposited fibrin served as the core of thrombi, which continously grew outward forming an outer shell region (Figure 5A,B; Videos S1 and S2, Supporting Information).These findings are consistent with the previous observations demonstrating formation of a tear droplike tapering structure of the clot under flow that comprises characteristics of a nonpermeable core and flexible shell region of a growing thrombus. [15]Moreover, both large and small PMA treated endothelialized posts showed a continuous increase in fibrin fluorescence intensities (Figure 5A,B,D); while the fibrin surface area showed an increase-then-decrease pattern after about 2.5 min of blood perfusion (Figure 5C), which might be due to the contraction of activated platelets within the fibrin clot.Strikingly, the PMA treated small post led to bigger clot with higher fibrin area than the large post.
We hypothesized that the thrombotic response was influenced as a result of not only the endothelial injury but also the hemodynamic disturbances induced by the stenotic geometry.This is due to the fact that platelets and fibrin predominantly deposited around the post but not the straight regions.To correlate the hemodynamics with thrombosis, we numerically reconstructed the geometries and surface topography from the 3D confocal images of the microvasculature-on-a-post chip. [16]Then we conducted computational fluid dynamic (CFD) simulation to map the disturbed blood flow profile (see the Experimental Section and Figure 5E,F).The blood flow transverses the post and produces a polynomial distribution along the post surface, with maxima of shear rates at the apical edges of both the small (Figure 5E,  max = 1746 s −1 ) and large (Figure 5F,  max = 1926 s −1 ) post, resulting in the activation and exacerbation of thrombotic process in those regions.We then compared the experimental outcome of the thrombotic dynamics on posts of different sizes that incurred altered flow disturbances.Notably, the small post resulted in larger fibrin clot than the large post (Figure 5A,B, 2nd row; Figure 5C,D), probably due to the 14.3% higher shear rate induced by the small post (Figure 5F,  ave = 857 s −1 ) than the large post (Figure 5E,  ave = 750 s −1 ).

Thrombotic Responses in the Distinct Hemodynamic Zones of Microvasculature-on-a-Post Chip
To analyze the spatial distribution of the thrombotic response, we further defined three different hemodynamic zones (Figure 6A): 1) Zone 1-the pre-stenotic region at the upstream to the post; 2) Zone 2-the stenotic region around the post; 3) Zone 3-the post-stenotic region at the downstream to the post.In untreated endothelium, both small and large posts incurred a negligible platelet and fibrin deposition across the three zones (Figure 6B-E; Figure S4B, Supporting Information).In the inflamed condition, fibrin deposition was significantly increased on both large and small posts, predominantly located in the Zone 3 compared to the Zones 2 and 1 (Figure 6B,C,D; Figure S4B, Supporting Information).However, significant increase in platelet volume was only observed predominantly in Zone 3 of the inflamed large but not small post (Figure 6B,C,E).We then quantified the percentage of fibrin and platelet volume in each zone relative to the total volume.Notably, after PMA treatment in large post, the fibrin volume increased to 66% and 34% in Zones 3 and 2, respectively (Figure 6F), while the platelet volume increased significantly to 74% in Zone 3 but slightly decreased slightly to 15% in Zone 2 (Figure 6G).The inflamed small post also exhibited an increased fibrin (69%) and platelet (69%) percentage in the Zone 3, but a similar fibrin (31%) and increased platelet (23%) percentages in the Zone 2 (Figure 6F,G).This could be attributed to the low shear pocket located at the downstream region of the post (Figure 5E), facilitating fibrin and platelet accumulation in Zone 3.
In terms of the fibrin/platelet ratio of the thrombus calculated based on the mean fluorescence intensities per unit area (μm 2 ), an equal and negligible proportion of fibrin and platelets were observed in each zone of healthy endothelium (Figure 6B,C,H).However, in PMA-treated endothelium, larger thrombi were accompanied with ≈33-fold and 43-fold higher accumulation of fibrin contents than platelets in the Zones 2 and 3 of the small post respectively (Figure 6B,C,H).Moreover, lower proportion of fibrin/platelet ratio in Zones 2-3 can be found on the inflamed large post rather than the small post, indicating that platelets were more prone to deposit in Zones 2-3 of large post than small post (Figure 6H).We also observed that in the untreated condition, the peak shear at Zone 2 was accommodated by relatively denser platelets than Zones 1 and 3 (Figure S4C,i, Supporting Information).A similar trend was observed in fibrin depositiondramatically denser fibrin contents occupied Zones 2 and 3 in both large and small posts region in PMA treated chips (Figure S4C,ii, Supporting Information).
Next, we further analyzed the 3D distribution of the thrombus formed in the microvasculature-on-a-post chip.We divided the post region into three planes along the z-axis: Plane a) the top region of 50 < z < 70 μm; Plane b) the middle region of 25 < z < 50 μm; Plane c) the bottom region of 0 < z < 25 μm (Figure 7A).Without PMA treatment on the endothelium, little platelets and fibrin were found in the Planes (a)-(c) of both large and small posts (Figure 7A).However, after PMA treatment, fibrin was predominantly formed in both Planes (b) and (c) (Figure 7A,B), while platelet was found to accumulate significantly in Plane (c) than Planes (a) and (b) (Figure 7A,C) for both large and small posts.Moreover, in PMA-treated chips, fibrin-rich clots were predominantly found in Plane (b), with relatively fewer fibrin and platelet aggregates at the Planes (a) and (c) for both large and small posts (Figure 7D).It can be interred that the Plane (b) with higher shear rate gradient serves as a core and the initial site of thrombus growth.In this regard, the clot formation in our microvasculature-on-a-post chip model was not only zone but also plane dependent; in other words, the 3D flow disturbance coupled with endothelial injury contributed to the heterogeneity of the thrombotic response.

Discussion
In this study, we developed a microvasculature-on-a-post chip with a stenosed vessel geometry to simulate protrusion caused by atherosclerotic plaque.Notably, we employed a dual-layer lithography technique to create an isolated 3D post structure with variable sizes, which mimics atherosclerotic stenosis and recapitulates a 3D flow profile in a microfluidic channel. [9]We then functionalized the chip by forming a confluent monolayer of endothelium to simulate the vessel wall in direct contact with blood circulation.Our investigation focused on the impact of continuous flow disturbance on endothelial morphology and the endothelium's induced inflammatory response.We observed disrupted inflammatory marker expression, cytoskeletal morphology and junctional integrity in endothelial cells located on both the post and straight regions.Lastly, we perfused recalcified human blood and visualized the thrombotic response regulated by the Virchow's triad, which includes vessel wall injury, hemodynamic disturbance and hypercoagulability.Our observations revealed blood clot formation over time and a heterogeneous composition of platelet and fibrin contents within different zones and planes of the thrombus formed around the post structure, which is associated with the size of the stenosed post structure and the altered shear rate profile across the 3D microenvironment of stenosis.
Our model features microfluidic channels with diameters of ≈100-200 μm, effectively simulating arterioles and venules. [9,17]his allows for a more accurate assessment of blood flow dynamics, shear rate effects, and thrombogenic potential. [18]Shear rate, defined as the velocity gradient in a fluid, plays a crucial role in blood flow dynamics and has significant implications for vascular health. [19]17b] This can induce very high shear rates on the post.19c,21] High shear rates on the post play a vital role in platelet activation and aggregation, which are critical for hemostasis and thrombosis. [22]Future research aims to explore various blood flow conditions while recognizing the importance of establishing flow pattern relevance between large-scale vessels and microscale microfluidics for validation purposes. [18]ur system used PMA to trigger endothelial inflammation with decreased PECAM-1 intensity, increased ICAM-1 and  E-selectin intensity and disorganized F-actin cytoskeleton. [23]Injured endothelial cells by PMA have downregulated PECAM-1 expression and can undergo active shedding of the extracellular domain of the PECAM-1 molecule; this shedding could contribute to the loss of cell-cell adhesion and its expression in junction. [14]Moreover, ICAM-1 and E-selectin are major activatory endothelial surface glycoproteins that bind to CD15/CD18 receptors, facilitating neutrophil adhesion and subsequent thromboinflammation. [9,24] Our results were consistent with previous work underpinning increased endothelial integrin expression and actin remodeling in response to inflammation. [23]he phenotypic changes in the characteristics of inflamed endothelium serve as a prothrombotic factor that triggers the coagulation cascade in thrombosis. [25]ur results showed that without endothelial injury, healthy intact endothelium alone led to a few micro-thrombi formations in a straight channel and post areas but did not induce large, stabilized fibrin clots.Some thrombi formed at the site of vessel lesions can be washed away and travel to the major organs like the heart, lungs, and brain.Elevated shear on the inflamed 3D post aggravates platelet activation, adhesion and aggregation followed by polymerization of fibrinogen monomer into a fibrin network that can stabilize the blood clot by providing mechanical stability and fibrinolysis resistance.We observed considerable heterogeneous composition of platelet and fibrin contents within different zones and planes of the thrombus formed around the post structure, which is associated with the altered shear rate profile across the zones and planes.Accumulating studies suggest that fibrin content varies in thrombi, highlighting an essential role for fibrin in thrombus composition that can influence the fate of the thrombus, such as stability, embolization, and breakage.In in vivo animal models, disseminated emboli contained less fibrin/platelet contents than stable thrombi.This is because fibrin plays a vital role in clot stability and anchors clots at the site of lesion, thus preventing them from migrating to distant blood vessels. [26]However, a higher proportion of fibrin-rich thrombus can be responsible for the occlusion of a large cerebral vessel. [27]nsights into thrombus formation and constituents regulated by 3D blood flow disturbances and vessel injury may serve as novel predictors of thrombosis and could guide interventional and therapeutic strategies.Leveraging the advantages of our microvasculature-on-a-post chip model, the correlation between 3D flow disturbance, platelet, and fibrin characteristics (platelet aggregate density and distribution, fibrin length, thickness, porosity, branching and stiffness) and thrombus stability are future intriguing future areas of study to predict the risk of thrombus complications as well as to improve the efficiency of antithrombotic intervention.More importantly, with high efficiency and small amount of blood sample, our microfluidic chip can be used to screen the proper choice of anticoagulant or antiplatelet drugs in patients with stenotic vessels.

Experimental Section
Microfluidic Chip Fabrication-Dual-Layer Photolithography: To emulate the geometries of vessel stenosis, a multilayer microfluidic channel (width y 0 = 200 μm and height z 0 = 70 μm) was designed and fabricated with an array of circular posts (eight posts being 1,000 μm apart from each other in the x-direction) in predefined diameter (D = 80 and 40 μm) and height (H = 20 μm).
After being designed in AutoCAD software (version 2015; AutoDesk, San Rafael, CA, USA), the microchannel mold was fabricated using superimposed layers in a combination of dry and wet photoresists. [28]Firstly, a 6-inch silicon wafer was cleaned and baked at 200 °C for 10 min, followed by 30 s of hexamethyldisilazane vapor priming at 120 °C.A 50 μm constant height dry photoresist film (DJ micro laminate SUEX) was then laminated on the silicon wafer at 65 °C to create the channel using a direct lithography writer (Heidelberg MLA-100, 4500 mJ cm −2 ).After the first write, the wafer was rinsed with DI water and dried with dinitrogen.A 20 μm layer of liquid photoresist SU-8 (MicroChem Inc., Westborough, MA, USA) was further spin-coated on top of the patterned 50 μm layer at 4100 rpm for 45 s.After baking for five min at 90 °C, the second lithography is conducted using the same direct writer at 360 mJ cm 2 .When the lithography was finished, the wafer was taken for five min post baking at 90 °C, followed by PGMEA rinsing until the excess photoresist was completely removed.To prevent permanent PDMS adhesion, the double-patterned wafers were treated with silane in a vacuum for 2 h and became the master molds.
Microfluidic Chip Fabrication-Microfluidic Post PDMS Casting Microfluidic Fabrication: To fabricate the microchannel, PDMS (Sylgard 184 by Dow Corning) was mixed at a 10:1 ratio (w/w) of elastomer to curing agent and then poured onto the wafer master molds and cured at 60-80 °C for 4 h.The cured PDMS blocks were then removed from the wafer, and the inlet reservoir and outlet were punched by the 6 and 1 mm biopsy punchers (World Precision Instruments) respectively.Blood cells and agonist solution were introduced via the inlet reservoir (volume 60-200 μL), while the outlet was connected to a PHD ULTRA syringe pump (Harvard Apparatus) to withdraw the fluid from the channel.Before assembling the microfluidic chip, dust from the bottom of the PDMS chip was removed by scotch tape.To assemble a functional microchannel, a thin glass coverslip (#1, 22 × 40 mm; Thermo Scientific) was bonded to the PDMS post chip using a plasma cleaner (Harrick Plasma) at the bottom portion of the device. [24]ndothelization of Microfluidic Chips: Human umbilical endothelial cells (HUVECs) were obtained from Thermo Fisher Scientific and cultured with EGM-2 medium (EGM-2 BulletKit, Lonza).Once reached 80%-90% confluency, HUVECs (passages 3-7) were washed with phosphatebuffered saline (PBS, ThermoFisher) and detached by trypsin/EDTA solution (ThermoFisher).After centrifuge, HUVECs were resuspended in EGM-2 medium at a seeding density of ≈5 × 10 6 cells mL −1 .Prior to HUVECs seeding, the microfluidic chip was sterilized with 80% ethanol for 20 min and washed thrice with PBS.Then the entire channels were coated with 100 μg mL −1 human plasma fibronectin (Thermo Fisher), incubating in a 4 °C fridge overnight.The channels were then rinsed with PBS twice, and then 10 μL of prepared HUVECs suspension was injected into each microchannel prefilled with culture medium.The microfluidic chip was immediately flipped upside down to allow HUVECs attachment to the post apex for 20 min.Then the chip was flipped again to allow HUVECs to attach to the bottom surface for 20 min.After that, the EGM-2 medium was added to the reservoirs to culture HU-VECs statically overnight, and the endothelialized microfluidic chip was completed.
In the study of the effect of endothelial cell inflammation on platelet and fibrin deposition, the endothelialized post chip was treated with PMA (50 ng mL −1 in serum-free medium EBM-2), a potent diacylglycerol (DAG) mimicking phorbol ester with protein kinase C (PKC) mediated angiogenic properties, [23,29] for 1.5 h at static conditions at 37 °C.Then the post chip was washed with complete medium EGM-2 twice.Then the post chip was ready for further examination.
In order to observe the effects of shear forces on the endothelial phenotype, the endothelialized microfluidic post chip was connected with the peristaltic pump (Harvard, P70).Then, the microfluidic chip was exposed to EGM-2 medium for 24 h continuously at a volume flow rate of 8.17 μL min −1 for a bulk shear rate of 600 s −1 without damping flow.
Blood Collection: All procedures involving blood collection from healthy donors were approved by the University of Sydney Human Research Ethics Committee (HREC, project 2014/244).All human donor blood samples were obtained with written informed consent.Blood was slowly drawn from the vein of a healthy volunteer.For whole blood perfusion studies, blood was anticoagulated in 3.2% sodium citrate vacutainers (Becton Dickinson).
After staining for 1 h, the whole blood was recalcified with CaCl2 immediately before use to achieve the desired concentration of 4mM CaCl 2 .Blood was introduced at the channel inlet of the PDMS block, which was cut into a 200 μL reservoir.The flow was induced with a 2 mL syringe connected to the outlet, withdrawn by a PHD ULTRA pump (Harvard Apparatus).The flow rate was set to generate a bulk (wall) shear rate of 600 s −1 .Platelet aggregation, fibrin deposition, and inflammatory biomarker expression on endothelial cells were monitored with a combined threecolor confocal ×30 silicon oil objective.Fluorescent images were captured using FLUOVIEW software, version 2.6 (Olympus).Fluorescently labeled platelet aggregates were quantitatively analyzed on a frame-by-frame basis offline using Imaris (Bitplane AG, Oxford Instruments).To examine endothelial and thrombosis biomarkers, the microvasculature-on-a-post chip was fixed with 4% paraformaldehyde after blood perfusion, and then thoroughly washed with PBS.Subsequently, the post chip was imaged using an Olympus FV3000RS confocal microscope.
Immunostaining and Confocal Microscopy: The microvasculature-on-apost chip was fixed with 4% paraformaldehyde, then thoroughly washed with PBS and permeabilized with 1% Triton X-100 (Roche) at room temperature for 1 h.Then the chip was blocked with 5% bovine serum albumin (BSA, Sigma-Aldrich).Subsequently, the chip was incubated with anti-VE-Cadherin-Alexa488 antibody (Invitrogen; 1:100 dilution) for endothelial cells for 1 h at 37 °C.After washing with PBS, the chip was stained with Hoechst 33 342 (Abcam; 1:1000 dilution) or Phalloidin (Abcam; 1:4000 dilution) for nuclei and F-actin for 20 min at room temperature, respectively.After washing, the microfluidic device was imaged using an Olympus FV3000RS laser scanning confocal microscope.
To quantify CD31 expression, live HUVECs in the chips treated with/without 50 ng mL −1 PMA were incubated with mouse antihuman CD31-Alexa488 antibody (Abcam; 1:200 dilution) in complete EGM-2 medium, for 30 min in the incubator.The chips were then washed with EGM-2 and imaged immediately.To investigate inflammatory marker expression, HUVECs were fixed with 4% paraformaldehyde for 10 min and blocked with 5% BSA for 1 h after treatment as mentioned above.The chips were first incubated with primary mouse anti-CD62E antibody (Invitrogen; 1:200 dilution) overnight at 4 °C, and then incubated with secondary donkey antimouse Alexa Fluor 555 antibody (Invitrogen; 1:500 dilution) for 1 h at room temperature the following day.They were then stained with anti-ICAM-1-Alexa488 antibody (Stock In-House; 1:200 dilution) in 2% BSA for 2 h at room temperature.
Z-projection immunofluorescence images were obtained using max intensity projection on ImageJ software.Signal overlaps between channels were then removed using ImageJ's LUMos Spectral Unmixer Plugin (https: //imagej.net/plugins/lumos-spectral-unmixing)with 1 fluorophore, 50 replicates, and 100 iterations.Individual channels were then converted to an 8-bit binary scale and measured for their mean gray value using a randomized Region of Interest (ROI) ImageJ macro for each image in both untreated and PMA-treated conditions.
Endothelial Cell Morphology Quantification-Aspect Ratio Analysis: To quantify the influence of the PMA drug treatment on the endothelium, the aspect ratio of endothelial cell bodies and the circularity of the nuclei were calculated.The aspect ratio was defined as the ratio of the long axis to the short axis automatically calculated by Imaris.Manual editing was optional to have a better detection efficiency (correct detection of cell nuclei, shapes, and area, etc.) for the autodetection of cells performed by Imaris.
Endothelial Cell Morphology Quantification-Orientation and Polarization Analysis: The endothelial cell orientation was quantified by calculating the angle between the flow direction and the "cell vector" determined by the long axis of each endothelial cell.The orientation of the endothelial cells was quantified to vary from 0°to 180°.To plot the bar graph in Figure 3G, the endothelial orientation was categorized into aligned (0°-45°and 135°-180°) and unaligned (45°-135°) cells with the flow direction.The proportion of cells aligned and unaligned was calculated accordingly.
To quantify the polarization of the endothelial cells, the "polarization angle" was defined as the angle between the vector of flow direction and the vector formed by the nucleus and Golgi body.The polarization of the endothelial cells was quantified to vary from 0°to 360°.Similarly, to plot the bar graph presented in Figure 3H, the endothelial cells were categorized into three types: parallel (135°-225°), vertical (45°-135°, and 225°-315°), and antiparallel (0°-45°and 315°-360°) to the flow direction.The angle calculation for orientation and polarization was automatically generated from detection of Imaris and a handmade script in Excel (Microsoft Office) which was validated by ImageJ.The proportion of cells polarized parallelly, vertically, and antiparallelly to the flow direction was calculated accordingly.
Computational Fluid Dynamic Analysis: In order to map the flow field and consequently calculate the shear rate, the geometries for endothelial cells coated microchannels were reconstructed from the 3D confocal images using Imaris.Then, the reconstructed geometries were imported as STL.file in ANSYS SPACECLAIM software, and further processing has been performed.A Finite Volume Method was applied to solve the governing equations of fluid flow utilizing ANSYS Fluent version 2020 R1.It is assumed that the fluid is Newtonian, incompressible, with constant properties and the flow is laminar (Re < 1) and steady state.Following these assumptions, the governing equations of fluid flow can be summarized as ∇.V = 0 ( 1 ) where V, , P, and  denote the velocity vector, the density of the fluid, the pressure, and the dynamic viscosity, respectively.The physical properties of blood at 37 ○ C were considered in the simulations; density was set to 1060 kg m −3 and the dynamic viscosity was set to 0.00345 Pa.sThe boundary conditions were considered as zero-gauge pressure at the inlet, and 5.88 μL min −1 outflow was applied to the outlet of the channel to mimic the experiments.The no-slip boundary conditions were applied to the walls of the microchannel.A second order upwind discretization scheme was applied to the continuity and momentum equations, and a second order method was applied to pressure.The Coupled scheme was incorporated for pressure-velocity coupling.Finally, the convergence criteria were considered as 10 −6 for the continuity and momentum equations.
In order to ensure that the results are independent of the implemented grids, a set of grid studies has been performed for both geometries with similar grid sizes.The geometry of the microchannel with the 40 μm posts has been discretized to 200 7601, 298 5366, and 403 5209 computational cells and the microchannel with the 80 μm posts were discretized to 200 6419, 298 0963, and 407 6052 computational cells.The average shear rate on the first post of the channel has been considered for the quantitative analysis.The average shear rate is calculated on the post by equation below, where A is the surface area of the post The quantity of the average shear rate on the first post for the microchannel with 40 μm post using the three grids are 852.79,856.94, and 861.20 s −1 .This parameter is also calculated for the microchannel with 80 μm post and the values are 748.05,749.93, and 751.12 s −1 .This critical parameter varies less than 0.5% in the consecutive simulation verifying the results are independent of the selected grids.
Thrombus Quantification: To construct 3D rendered images of thrombus, spot, and surface function in Imaris were used to detect platelets and fibrin in a thrombus architecture respectively.To calculate the platelet and fibrin volume, the 3D confocal images were cut into three zones and planes according to the definition in Figures 6 and 7. Then the absolute volume of platelets and fibrin in each zone and plane was calculated by Imaris software using the surface function.To quantify platelet number, spheres with spot sizes around 2 μm were counted in Imaris.To quantify the fluorescent intensity of fibrin, Imaris was used to export the fluorescent intensity in arbitrary unit.
Statistical Analysis: All graphical data are presented by GraphPad Prism 9.0.Statistical differences between each group were tested by unpaired, two-tailed Student's t-test and a one-way ANOVA test.A p-value below 0.05 was accepted as significant.

Figure 1 .
Figure 1.Microvasculature-on-a-post chip concept and summary of experimental procedures: 1) identify pathological vascular anatomies implicated in atherosclerosis; 2) fabricate the microfluidic post device to model the stenotic vascular geometries; 3) perform endothelization and perfuse whole in the post microfluidic devices; 4) visualize thrombotic pathogenesis in the endothelialized microfluidic post device.

Figure 2 .
Figure 2. Bioengineered microvasculature-on-a-post chip.A) i) Schematic diagram of the design of the post microfluidics.Image not to scale.ii) A representative image of the actual post microfluidic device bonded to a glass slide.Image not to scale.iii) Scanning electron microscope (SEM) image of the post structure with diameter of D = 80 μm.Scale bar = 40 μm.B) Confocal images of the endothelialized post microfluidics with 5 consecutive posts (D = 80 μm, H = 20 μm) from the side (xz), top plane (z = 70 μm), post plane (z = 50 μm), and bottom plane (z = 0 μm) view.Scale bar = 200 μm.C).Zoom in view of the post structure with diameter of D = 80 μm, H = 20 μm i) and D = 40 μm, H = 20 μm ii).Scale bar = 20 μm.

Figure 3 .
Figure 3. Fluid flow induced endothelial morphology change.A,B) Confocal images of the endothelium (indicated by green: ve-cadherin and blue: nucleus) in a straight channel (1st column), at post apex with D = 80 μm (2nd column) and D = 40 μm (3rd column) after 24 h of static (A;  0 = 0 s −1 ) and flow (B;  0 = 600 s −1 ) culture.Scale bar = 20 μm.C,D) Confocal images of the endothelium (indicated by white: Golgi body and blue: nucleus) in a straight channel (1st column), at post apex with D = 80 μm (2nd column) and D = 40 μm (3rd column) after 24 h of static at  0 = 0 s −1 C) and flow culture at  0 = 600 s −1 D).Scale bar = 20 μm.E) The aspect ratio of endothelial cell bodies, defined by the ratio of long axis to short axis of cell body.The mean value was calculated using descriptive statistics in GraphPad Prism.F) The aspect ratio of endothelial cell nucleus, defined by the ratio of long axis to short axis of nucleus.The mean value was calculated using descriptive statistics in GraphPad Prism.G) The proportion of endothelial cells aligned (in between 45°around the flow axis) or not aligned with the flow direction).H) Quantification of endothelial cell polarization relative to the flow direction (Parallel: in between 0°and 45°and 315°and 360°; Vertical: in between 45°and 135°and 225°and 315°; Antiparallel: in between 135°and 225°).*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001, assessed by a one-way ANOVA and Tukey's multiple comparison test.Data represents n ≥ 6 chips and n ≥ 20 cells.

Figure 5 .
Figure 5. Endothelial injury and hemodynamic disturbance caused thrombotic response.A,B) Time-lapse confocal microscopy images taken during fibrin clot formation with blood perfusion at 600 s −1 and respective 3D rendered images of the thrombus after 5 min blood perfusion in D = 80 μm post chip A) and D = 40 μm post post microfluidics B) with or without PMA treatment.Scale bar = 50 μm.C,D) The surface area D) and the fluorescence intensities of fibrin in a growing thrombus over a period of 5 min blood perfusion.E-F) CFD simulation of endothelialized post chip channel constructed from confocal 3D images were performed at a bulk wall shear rate of 600 s −1 to model shear rate distribution of endothelialized post microfluidics with different geometries: D = 80 μm E) and D = 40 μm F).Data represents mean ± s.e.m. of n ≥ 2 independent chips in duplicate.

Figure 6 .
Figure 6.Platelet and fibrin distribution across different zones.A) Schematic representation of thrombi formed in endothelialized post microfluidics respectively after blood perfusion with a shear rate of 600 s −1 .Zones 1-3 are defined as the following: Zone 1 (pre-stenosis), Zone 2 (stenosis), and Zone 3 (post-stenosis).B,C) Confocal z-projection 2D images of platelets (row 1, white), fibrin (row 2, magenta) and merged images with endothelium (row 3, green) on the endothelialized post microfluidics in the absence (column 1) and presence (column 2) of PMA treatment in D = 80 μm B) and D = 40 μm C) post chips.Scale bar = 50 μm.D,E) The absolute volume of fibrin D) and platelet E) in the three Zones 1-3.F,G) Percentage distribution of platelet volume F) and fibrin volume G) in a thrombus formed on the chips.H) The ratio of fibrin to platelet on post microfluidics calculated by dividing the fibrin volume by the platelet volume across Zones 1-3.Data represents mean ± s.e.m. of n ≥ 2 independent chips in duplicate.

Figure 7 .
Figure 7. Platelet and fibrin distribution along different planes.A) Z-projection confocal images of Plane a (roof of channel; z = 50-70 μm), Plane b (apex of post; z = 25-50 μm) and Plane c (cover slip; z = 0 -25 μm) on post chips with diameter of D = 80 μm and D = 40 μm in the presence and absence of PMA treatment.Scale bar: 50 μm.B,C) The absolute volume of fibrin D) and platelet E) in the three Planes (a)-(c).D) The ratio of fibrin to platelet on post chips calculated by dividing the fibrin volume by the platelet volume across the Planes (a)-(c).Data represents mean ± s.e.m. of n ≥ 2 independent chips in duplicate.