Peristaltic pumps adapted for laminar flow experiments enhance in vitro modeling of vascular cell behavior

Endothelial cells (ECs) are the primary cellular constituent of blood vessels that are in direct contact with hemodynamic forces over their lifetime. Throughout the body, vessels experience different blood flow patterns and rates that alter vascular architecture and cellular behavior. Because of the complexities of studying blood flow in an intact organism, particularly during development, the field has increasingly relied on in vitro modeling of blood flow as a powerful technique for studying hemodynamic-dependent signaling mechanisms in ECs. While commercial flow systems that recirculate fluids exist, many commercially available pumps are peristaltic and best model pulsatile flow conditions. However, there are many important situations in which ECs experience laminar flow conditions in vivo, such as along long straight stretches of the vasculature. To understand EC function under these contexts, it is important to be able to reproducibly model laminar flow conditions in vitro. Here, we outline a method to reliably adapt commercially available peristaltic pumps to study laminar flow conditions. Our proof-of-concept study focuses on 2D models but could be further adapted to 3D environments to better model in vivo scenarios, such as organ development. Our studies make significant inroads into solving technical challenges associated with flow modeling and allow us to conduct functional studies toward understanding the mechanistic role of shear forces on vascular architecture, cellular behavior, and remodeling in diverse physiological contexts.

Endothelial cells (ECs) are the primary cellular constituent of blood vessels that are in direct contact with hemodynamic forces over their lifetime. Throughout the body, vessels experience different blood flow patterns and rates that alter vascular architecture and cellular behavior. Because of the complexities of studying blood flow in an intact organism, particularly during development, the field has increasingly relied on in vitro modeling of blood flow as a powerful technique for studying hemodynamic-dependent signaling mechanisms in ECs. While commercial flow systems that recirculate fluids exist, many commercially available pumps are peristaltic and best model pulsatile flow conditions. However, there are many important situations in which ECs experience laminar flow conditions in vivo, such as along long straight stretches of the vasculature. To understand EC function under these contexts, it is important to be able to reproducibly model laminar flow conditions in vitro. Here, we outline a method to reliably adapt commercially available peristaltic pumps to study laminar flow conditions. Our proof-of-concept study focuses on 2D models but could be further adapted to 3D environments to better model in vivo scenarios, such as organ development. Our studies make significant inroads into solving technical challenges associated with flow modeling and allow us to conduct functional studies toward understanding the mechanistic role of shear forces on vascular architecture, cellular behavior, and remodeling in diverse physiological contexts.
Peristaltic pumps have often been utilized in system designs because of their simplicity, their relatively low cost, and their ability to keep aseptic conditions as the pumped liquid is fully contained and does not directly contact any mechanical parts of the system. However, the caveat of peristaltic pumps is their inherent pulsation, because of intermittent contact of the pump rollers with the tubing that carries the liquid. This creates a temporal change in flow velocity profiles differing from steady flow velocities in laminar flow (32), altering the forces experienced by the surrounding environment-in this case, the cells. In similar systems that require pumps to transport fluids, such as fuel injectors, the implementation of pulsation dampeners, devices that reduce the pulsation in fluids by compensating the discontinuity of the flow with an applied force of equal magnitude, is necessary to protect the integrity of the ducts that carry the liquid. There are many ways to generate dampeners, but the most common methods offset force by liquid displacement, pressure regulation valves, or diaphragms/elastomers (33,34).
While significant research has been done to build reliable platforms capable of generating physiologic levels of laminar versus oscillatory or pulsatile shear forces, systems remain inaccessible to the broad vascular biology research community for a number of reasons. For many, the price of a full commercial system capable of altering flow patterns, rates, or type prevents use of this modeling technique. Microfluidic systems and raised reservoir low-angle inlet platforms, while fantastic at modeling steady-state laminar flow, remain complicated to fabricate or set up (35,36). Peristaltic pumps are cost-effective alternatives but, as described, are best suited to mimic pulsatory/oscillatory flow conditions (33). To account for this, we have adapted and optimized a commercially available multihead peristaltic pump system that has its own controller software (37) to reproducibly perform steady-state laminar flow or pulsatile flow experiments (Fig. 1A). Here, we will outline the steps needed to create pulse dampeners from easy to access supplies that can attach to a peristaltic pump to generate laminar flow. Furthermore, we discuss our parameters for eliciting flow responses, standardization/validation of this pump system, and downstream analysis techniques for studying cell biology using this platform.

Generation of dampeners to offset pump pulsation
Addition of a dampener to a peristaltic pump is one way to offset the pulsatile forces generated by the mechanical properties of the pump. To do this, we utilized easy to access and cost-effective supplies to custom make small footprint dampeners to attach to any flow system (Fig. 1). To create the body of the dampener, the barrel of a 10 ml syringe (Becton Dickinson) was cut 12 mm from the base of the barrel. A thin wall polypropylene tube (Beckman) was bonded to the syringe barrel by applying a thin layer of two-part urethane adhesive (J-B Weld Company), following the manufacturer's recommendation. Because the internal diameter of the syringe barrel is 13.34 mm and the external diameter of the polypropylene tube is 14.43 mm, to assemble the dampener body, both parts must be forced together, which along with the adhesive avoids air or liquid leaks. The tubing connection to the dampener was created with 2.5 cm silicone tube (FisherBrand) inserted directly to the syringe Luer-Lok tip. At the other end of the silicone tubing, a T-type connector (Nalgene) was inserted (Fig. 1B).
Two dampeners are connected to the system, one on the inlet side and one on the outlet side of the system between the pump head and its respective media reservoir (Fig. 1B). The dampeners must be oriented vertically (with the dome of the polyallomer tube toward the ceiling) to ensure proper functioning, as they work by allowing liquid displacement into a closed chamber subjected to atmospheric pressure. If the dampeners are not properly oriented, laminar flow will not be achieved as bubbles will form within the system. Furthermore, if the chamber is inverted (with the dome oriented toward the floor), liquid will fill the dampener preventing its function. The media reservoirs (Fig. 1A) were constructed using 30 ml high-density polyethylene bottles (Nalgene). Two holes 4 mm in diameter were drilled in the caps of the highdensity polyethylene bottles and two Elbow Luer Connector Male (ibidi) were bounded to the caps using the two-part urethane adhesive described previously. At the interior of the cap, two pieces of 2 and 4 cm silicone tubing were attached to the elbow connector as inlet and outlet ports, respectively.
Efficacy of the generation of laminar flow was confirmed two ways. First, we show via real-time visualization of quantum dots within the circulating media (Videos 1 and 3) that without the dampener (Video 1), there is a clear internal shift of the fluid/quantum dots as they transverse the flow chamber (flow is moving left to right across the video) from the peristaltic rollers stopping. This pulsation is virtually eliminated after addition of dampeners (Video 3). Second, we imaged the flow chambers using brightfield microscopy. Videos acquired at the level of the cells show a rhythmic pulsing of the fluid without the dampener (Video 2) that is abrogated by the addition of the dampener (Video 4).
To quantify the effectiveness of the dampeners in offsetting the inherent pulsatile nature of the pump, we installed a liquid flow sensor to the silicone tubing (SENSIRION) enabling us to determine directionality of flow (positive flow moving left to right across the sensor, negative if flow is moving right to left across the sensor) and the dynamic liquid flow rates of the system (Fig. 2). The sensor was installed before the inlet attachment point of the μ-Slide I 0.4 Luer slide of the flow circuit ( Fig. 2A). The system was evaluated under the following conditions: (1) the flow system without dampeners, using the original function of the pump (pulsatile, Fig. 2, A-C); (2) the flow system modified with dampeners installed at the inlet and outlet of the pump head (laminar, Fig. 2, D-F); and (3) the flow system modified with a commercial dampener (SENSIRION) installed before the sensor as is recommended by the manufacturer (laminar, Fig. 2, G-I). The system was placed under our standard experimental conditions (5% CO 2 , 37 C, and 95% humidity for 24 h) and 15 s of sensor measurements of fluid flow collected at 1 and 24 h of culture. The 1-and 24-h time points were chosen to demonstrate the stability of flow forces across the time course of experimentation. The nonmodified pump as purchased has an inherent rhythmic pulsation to it (Fig. 2B), which can be dramatically offset by the A B incorporation of dampeners into the flow circuit (Fig. 2E). As a direct comparison, we tested the performance of a commercial SENSIRION dampener and demonstrate that our homemade dampeners were able to outperform a commercial version and suppress pulsation more dramatically (Fig. 2, E and H). While there is still minimal pulsation within the system, it is reduced approximately sixfold.
To account for variance in location of the dampeners between the commercial and homebuilt setups, we analyzed three different variations of the original commercial setup (Fig. 2G). Our first modification added a flat bottom fluidic restrictor with an internal diameter of 0.5 mm (provided with the flow sensor)-we call this variation V.1 (Fig. S1, A-C). Second, we installed the flow sensor with the restrictor However, as shown, our dampeners suppressed pulsation to a much higher degree (commercial 1.4-fold from peristaltic; lab-built sixfold from peristaltic). I, average maximum and minimum flow forces generated after commercial dampener modifications of the peristaltic pump at 1 and 24 h of culture. The box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5 × above and below the interquartile range. All data points are shown as discrete dots (averages from each of the four individual pumps).
between the pump outlet and the media reservoir. The commercial SENSIRION dampener remained located between the media reservoir and the inlet of the ibidi chamber (μ-Slide I 0.4 Luer)-we call this variation V.2 (Fig. S1, D-F). Third, we moved the commercial SENSIRION dampener to be between the pump outlet and the media reservoir, as we have it installed with our homebuilt dampener, leaving the flow sensor with the restrictor between the outlet of the media reservoir and the inlet of the ibidi chamber slide-we call this variation V.3 (Fig. S1, G-I). As shown in Fig. S1, none of these modifications significantly improved the performance of the commercial SENSIRION dampener by more than approximately 1%.
The small improvement obtained with variation V.2 by the addition of the restrictor to the flow sensor (Fig. S1, D-F) prompted us to test this modification in line with our homemade dampeners and reassess flow pulsation (Fig. 2D). Following the addition of the flat bottom fluidic restrictor to the flow sensor on our homemade dampener system, we observed an additional 15% decrease in flow pulsation as compared with our system as originally conceived without the restrictor (Fig. S2). All data presented throughout the remainder of the article utilize our custom dampeners and system as shown in Fig. 2, as this platform reliably allowed us to observe phenotypic differences in EC behavior in response to various types of flow stimuli without the purchase of additional commercial products. However, if higher levels of pulse dampening are desired, addition of a SENSIRION restrictor to the system can improve overall function of our homebuilt dampeners.

Validation of cellular behavior in flow assays via live imaging analysis
To confirm that altered cellular behaviors are generated in response to differing flow stimuli from our modified pump system, we set up 2D live imaging assays to longitudinally track cellular motility and alignment over a 24-h period. Human umbilical vein ECs (HUVECs) or human aortic ECs were grown to confluence (2.5 × 10 5 cell/slide) on 1 mg/ml gelatincoated μ-Slide I 0.4 Luer slides in 1 × M199 media supplemented with 20% fetal bovine serum, 25 μg/ml of EC growth supplement, 0.01% heparin sodium salt, and 1 × antibioticantimycotic at 5% CO 2 , 37 C, and 95% humidity. The slides were then acclimated to a microscopy system containing a climate-controlled stage top incubator (EVOS M7000) to be cultured under conditions of constant laminar flow, at 12 to 15 dyn/cm 2 /s, or constant pulsatile flow, at 12 to 15 dyn/cm 2 /s and 60 RPM, for 24 h. To help maintain cell health, at the start of the experiment, the rate of flow was stepwise increased 10% every 10 min until reaching the desired final experimental flow rate.
Utilizing a microscope with a stage top incubator allows us to carry out multipoint time-lapse image acquisition. Images were acquired every 20 min for 24 h to follow dynamic changes in cell shape and alignment over time (Videos 5 and 6, Fig. 3, A  and B). At the end of the experiment, individual images were assembled into a video file to watch cellular behavior across the full 24 h of imaging. Angle of alignment (Fig. 3, C and D), cellular tracking (Fig. 3E), and total distance moved/linearity of movement (Fig. 3, F-H) amongst other features can be analyzed from these videos utilizing free ImageJ/Fiji software (38). Our results show that in response to laminar flow, ECs align to and move against the direction of flow over the 24 h imaging period (Fig. 3, D and E, Video 6), whereas ECs subjected to pulsatile flow largely remain perpendicular to the direction of flow and move in a more haphazard fashion over the 24-h imaging period (Fig. 3, C and E, Video 5). Under both flow conditions, HUVECs on average move the same total distance; however, under laminar flow, HUVECs exhibit a larger linear displacement from their starting point of origin and move more linearly (Fig. 3, F-H). This behavior mimics what is seen with both in vivo and in vitro flow models (5,6,(14)(15)(16)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31), validating that our system does indeed reliably model flow stimuli to elicit the expected changes in cellular behavior.
From a practical standpoint, use of this system does not require live imaging. If desired, the entire setup can be easily placed in a standard 37 C, 5% CO 2 , humidified tissue culture incubator. The peristaltic pump we adapted (Flocel (37)) has four pressure heads. When carrying out live imaging, stage top space prevents use of all four heads in tandem. Therefore, we often use the system in a standard incubator to allow conversion of two pressure heads to carry laminar flow, whereas two remain pulsatile to optimize throughput and pair culture conditions within a single experiment. At the end of the experiment, the cells can be collected for biochemical/molecular analysis to determine the downstream molecular impacts of individual flow stimuli: that is, rinsed with 1 × PBS and fixed with 4% paraformaldehyde (PFA) for immunostaining applications (discussed later), flushed with TRIZOL for mRNA collection, or lysed in sample buffer for protein collection (30).

Immunostaining of cultures following flow
Following termination of flow, our method enables analysis of protein localization, cellular signaling, or mRNA composition, among other parameters, to interrogate downstream signaling pathways that occur as a consequence of flow. Here, we will outline our immunostaining protocol to analyze protein accumulation and localization in response to flow forces.
After 24 h of exposure to flow stimuli, the cultured ECs were rapidly rinsed twice with 3 ml 1 × PBS by flushing the slides using a 3 ml syringe. The cells were fixed in 100 μl of 4% PFA for at least 4 h before starting the immunostaining protocol. For H, linearity of movement: "1" is a movement path that is completely linear, whereas "0" is a movement path that is completely nonlinear, that is, curved. The box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5 × above and below the interquartile range. All data points (individual cells; n > 8) are shown as discrete dots, with outliers shown above or below the whiskers. Data are representative of three independent replicates. HUVEC, human umbilical vein endothelial cell.
Adaptation of peristaltic pumps for laminar flow experiments morning, the primary antibody is removed, and the cells rinsed with 1 × PBS-Tx three times to decrease nonspecific antibody binding. Secondary antibody is then added at a 1:2000 dilution in 1% BSA/1 × PBS-Tx, and the slides were incubated for 1 h at room temperature. Finally, Hoechst dye is added at a 1:5000 dilution and incubated for 30 min at room temperature for staining of DNA/nuclei. Three washes with PBS-Tx were done at the end of all the steps to remove excess antibody and decrease nonspecific background staining prior to mounting the slides for imaging analysis. Images were acquired utilizing a 20, 40, and 60 × objectives on a Nikon Ti2 inverted microscope equipped with a CSU-W1 confocal spinning disk (Yokogawa).
As an example (Fig. 4), the EC junctional protein vascular endothelial (VE)-cadherin was labeled (1:100 dilution of primary antibody) and visualized using an Alexa Fluro-488 secondary antibody (1:2000 dilution, green). Nuclei (Hoechst, 1:5000 dilution) labeling is in blue. As shown, application of flow forces across ECs enhances the accumulation of VE-cadherin protein at junctions compared with no flow controls (Fig. 4, A-D). While no differences were noted in immunostaining intensity between laminar and pulsatile flow conditions (Fig. 4E), marked differences in orientation of VEcadherin "fingers" were noted at junctional planes parallel to the flow direction (Fig. 4, F and G)  ) under laminar flow conditions. Furthermore, phalloidin staining was done to assess orientation of actin stress fibers in response to various flow forces (Fig. 5). As shown in both the representative images and the quantification, under the no flow and pulsatile flow conditions, the actin stress fibers are oriented randomly, with a slight skew in stress fiber orientation toward the direction of flow under the pulsatile flow condition. Conversely, under laminar flow conditions, the actin fibers are oriented parallel to the direction of flow.

Discussion
In this work, we outline the validation and use of a peristaltic pump system that can be easily modified to concurrently deliver laminar or pulsatile flow to adjacent EC cultures in a highly reproducible manner. The system provides a costeffective easy to use alternative for laboratories looking to conduct in vitro flow modeling experiments. Moreover, the platform allows for real-time imaging and analysis of cellular responses to flow (Fig. 3), coupled with downstream molecular analysis (30) (Fig. 4 and 5).
Pulse dampeners are widely used to mitigate the interrupted pulsatile flow patterns, which are a consequence of rollers in peristaltic pumps (34,36,40,41). As part of this work, we compared the endogenous function of the peristaltic pump against our lab-built pulse dampeners and a commercial pulse dampener (Fig. 2, S Figs. 1 and 2). These studies demonstrated the ability of our lab-built dampeners to generate laminar type flow in contrast to the pulsatile nature of the endogenous pump function. In addition, our dampeners showed improved performance for generation of laminar nonpulsatile flow forces compared with the commercial dampener we tested. The analysis also confirms that our custom-built dampeners are stable across at least 24 h of assay at the temperature and humidity recommended for EC culture (Fig. 2, C, F, I), showing that this system is suitable for use in functional assays. Therefore, modification of a peristaltic pump to include dampeners into the system allows for an affordable and adaptable method to culture cells under laminar and/or pulsatile flow conditions for the reproducible study of shear stress-mediated cellular responses in vitro.
Functionally, the peristaltic pump utilized has four pressure heads, allowing for side-by-side comparison of cellular behavior under pulsatile versus laminar flow conditions. As shown by cellular tracking experiments, cells under pulsatile flow move haphazardly and tend to end up oriented perpendicular to the direction of flow (Fig. 3, A-C and E; Video 5), whereas cells under laminar flow rapidly orient parallel to and move against the direction of flow (Fig. 3, A, B, D, E, Video 6). These phenotypes are consistent with those published in vitro and in vivo under developmental and nonpathogenic flow settings (5, 6, 13-18, 21, 22, 24-29, 31, 37, 42-47), suggesting that this system is able to generate physiologically relevant flow forces. Finally, we confirm that ECs experiencing flow forces develop stronger junctions, as assessed by increased VEcadherin localization at junctions compared with no flow conditions (Fig. 4). While the intensity in staining is not significantly different between cells experiencing pulsatile versus laminar flow, localization of VE-cadherin fingers at the junction is differentially oriented (Fig. 4, C-G), with junctional fingers aligning to be parallel to flow under laminar flow conditions (Fig. 4F). Similar results were described previously in EC cultures, where disturbed flow regions in a flow chamber exhibited discontinuous/more "finger"-like VE-cadherin staining, whereas in areas where the flow was laminar, the VE-cadherin staining was continuous (39,48). Our findings reveal a similar phenomenon with actin stress fibers in response to flow, with the actin stress fibers oriented haphazardly following pulsatile flow, but oriented parallel to flow in the laminar flow conditions (Fig. 5).
Being able to reliably model physiological flow conditions allows for deeper mechanistic study of EC autonomous biology; however, it also opens up the possibility of identifying flow-regulated signals generated in ECs that alter physiology in a nonautonomous fashion. Embryonic and fetal hematopoietic stem cells (HSCs) and vascular mural cells (MCs) are two primary examples of this phenomenon. HSC development is dependent on blood flow via cell-intrinsic nitric oxide signaling (49). HSC development via the Yes-activated protein is also dependent on the mechanical forces of blood flow, as shown by Lundin et al. (50) using induced pluripotent stem cells grown in microfluidic culture devices. Beyond initial specification, HSCs extravasate into circulation and seed  (condition with dampeners). F and G, histogram (F) and radial histograms (G) depicting orientation angles of the VE-cadherin positive junctional "fingers" relative to the direction of flow (flow applied from left to right in images). No flow (gray), pulsatile flow (orange), and laminar flow (blue). The distribution of data shows that under no flow conditions, VE-cadherin fingers, if present, are randomly distributed. Under pulsatile flow conditions, these fingers skew to have a slightly more perpendicular orientation, when considering junctions parallel to the direction of flow (i.e., "top" and "bottom" of the cells). Under laminar flow conditions, the fingers have a parallel orientation, when considering the junctions parallel to the direction of flow (i.e., "top" and "bottom" of the cells). See Fig. S3 for schematic description. The alignment of the VE-cadherin fingers was carried out using the module for quantitative orientation measurement of the ImageJ plugin OrientationJ created by Daniel Sage (http://bigwww.epfl.ch/demo/orientation/). n = 45 cells, five images. Data plotted using a bin size of 45 and are representative of two independent experiments. F and G, represent varying presentation of the same dataset to demonstrate different visualization techniques, (F) the y-axis represents the proportion of cells that fall into each bin; (G) the radial axis represents the proportion of cells that fall into each bin. The box plots are graphed showing the median versus the first and third quartiles of the data (the middle, top, and bottom lines of the box, respectively). The whiskers demonstrate the spread of data within 1.5 × above and below the interquartile range. All data points (average intensity from individual images; n > 8) are shown as individual dots. p values are indicated above statistically significant datasets and were generated using one-way ANOVA. Data are representative of three independent replicates. EC, endothelial cell; VE, vascular endothelial. developmental niches where they complete their developmental program (reviewed by Horton et al. (51)). During this process, the HSCs are exposed to multiple vascular environments experiencing a variety of mechanical stresses; these stresses trigger signaling events in vascular cells that may play a role in determining ultimate hematopoietic cell fate, among other physiological processes. While there has been tremendous progress in understanding the effects of extracellular forces on HSC differentiation, our understanding of how these forces interface with the vascular niche to signal to HSCs traversing blood vessels has been limited by technical challenges.
Similarly, we have recently shown that differential sensing of blood flow forces can alter MC biology during development (52,53). Arteries are known to acquire much greater numbers of MCs than veins across development, and elevated mechanical forces felt by the arterial vasculature is predicted to be a key driver in this process. However, how these forces are sensed by ECs and subsequently communicated to MCs is still an active area of investigation. The transcription factor Klf2 is well known to play an essential role in vascular development in response to forces generated by blood flow (44,45,(54)(55)(56)(57)(58)(59)(60)(61)(62). In zebrafish and mice, during early vascular development, the expression of klf2a/Klf2 is significantly higher in veins compared with expression in the dorsal aorta, where the blood flow is pulsatory because of the dorsal aorta direct connection to the heart (43,52,53). We recently demonstrated that in klf2a-deficient zebrafish, there is a significant increase in association of MCs to the cardinal vein compared with wildtype siblings, suggesting that Klf2 might serve as a direct or indirect transcriptional repressor for MC recruitment cues (52).
These are just a few examples of anatomical adaptations that result as a response to the different types of flow forces experienced by vasculature during development or in disease.
Therefore, the development of low cost and easy to use devices to model blood flow forces, such as the one described in this article, has the potential to greatly expand our understanding of the physiological effects of EC intrinsic and extrinsic signaling events.

Calculation of shear stress
Shear stress is the stress/force felt by the endothelium related to hemodynamic forces in a blood vessel. Liquid flow through a cylindrical pipe, such as blood vessels, generates a force that is parallel to the pipe wall. The shear stress applied depends on the viscosity and velocity of the flowing fluid. Normally, viscous fluids will require more force to be transported inside of a pipe and will generate more stress. In addition, higher velocities produce more force on the pipe's walls than slow ones (63). For this platform, shear stress was calculated using the equations previously adapted by ibidi (Application Note 11. ibidi GmbH, version 6.0, February 14, 2022; https://ibidi.com/content/64-application-notes) as followed: where τ: shear stress (dyn/cm 2 ); η: dynamical viscosity (dyn⋅s/ cm 2 ); Φ: flow rate (ml/min); and κ: 136.6. This factor is calculated considering the viscosity of the liquid fluid-in this case, cell culture media supplemented with fetal bovine serum (0.0072 dyn⋅s/cm 2 ) and the shear rate-which indicates the velocity of the perfused fluid in a channel (64).

Microscopy and cell visualization
HUVEC cultures were stabilized at an initial concentration of 3000 cell/cm 2 and maintained in standard conditions: 37 C, 5% CO 2 , and 95% humidity until cells reached 80 to 90% confluence. Then the cells were trypsinized and reseeded on μ-Slide I 0.4 Luer slide at a final concentration of 2.5 × 10 5 cell/ cm 2 and incubated in standard conditions for an additional 24 h before starting the flow assays.
Flow assays were carried out in an EVOS M7000 microscope, equipped with an onstage incubator to maintain standard cell culture conditions. Time-lapse images were captured in bright field at 10 × magnification (Olympus; UPlanSApo objective), every 20 min for 24 h. The videos were created by stitching the 73 frames acquired together at a rate of five frames per second. The immunostaining images were acquired at 20 × (Nikon; CFI plan Apo Lambda 20 × objective) and 60 × (Nikon; CFI plan Apo Lambda 60 × objective) with a Nikon Ti2 microscope equipped with a W1 spinning disk.

Automated analysis of alignment parameters using ImageJ
Analysis of the effect of flow type on EC alignment was carried out using the ImageJ plugin "Directionality v2.3.0" created by Jean-Yves Tinevez (65), as follows: Images of 2048 pixels by 1536 pixels with a resolution of 50,300 pixels/inch were transformed into 8 bit files (3 MB). The plugin was applied using the Fourier component method of analysis, with data collected from −90º to +90º and plotted in the histogram as a proportion of events with a given alignment angle per bin. Bin sizes are set at 22.5 , and angles of alignment are measured against the horizontal axis of the image. For this analysis, 15 images at 0 and 24 h were evaluated for both pulsatile and laminar flow.
The same method was implemented to analyze the orientation of the F-actin filaments under the same flow conditions. For this assay, eight images at 24 h after flow were evaluated for pulsatile, laminar, and no flow.

Analysis of EC junctional markers under flow
Images of cells under different flow conditions and immunostained with VE-cadherin were evaluated using the ImageJ plugin "OrientationJ" created by Daniel Sage as follows: Images of the VE-cadherin immunostaining under no flow, laminar, and pulsatile flow were acquired with a size of 1074 microns by 629 microns with a resolution of 4.7 pixels/micron. The images were transformed to 32 bit (57 MB). VE-cadherin fingers at the upper and lower junctions of cells (Fig. S3), those parallel to the direction of the flow, were measured implementing the OrientationJ module "Quantitative Orientation Measurement" (66). The cells in each picture were selected randomly by creating a grid of squares with an area of 8337 μm 2 . A table of random numbers was then created to select the position of the cells to be measured. Using the rectangle tool in ImageJ, a Region of Interest was delimited at the points measured in each cell. The software automatically defines the orientation of the VE-cadherin fingers in the region of interest against the horizontal axis of the image. Using this method, we measured the orientation of junctional VE-cadherin fingers in 9 cells per picture, five pictures per condition.

Data availability
All data reported in this article are available from the corresponding author upon request.
Supporting information-This article contains supporting information.