All components for the MoCAP device were designed using a computer aided design software (Autodesk Inc., One Market, Suite 400, San Francisco, CA 94105). A bill of materials lists all the necessary hardware needed to build a fully functional MoCAP (Table 1). Components of MoCAP device were printed using a stereolithography Form 3+ 3D printer (Formlabs Inc., 35 Medford Street, Suite 201, Somerville, MA 02143) and high-temperature resin (RS-F2-HTAM-02). Low angle cones were coated with parylene-C using a parylene deposition machine (Specialty Coating Systems, 7,645 Woodland Drive, Indianapolis, IN 46278) to ensure biocompatibility (O’Grady et al., 2021). Commercially available bearings (McMaster-Carr Supply Co., 2,828 No Paulina St, Chicago, IL 60657) were purchased and utilized in the device. The gearbox within the device was lubricated utilizing a small amount of medical grade petroleum jelly (Covidiein, 15 Hampshire Street, Mansfield, MA 02048). The gearbox in the MoCAP device was driven by a NEMA 17 stepper motor (STEPPERONLINE Inc., 228 Park Ave S 79525, New York, NY 10003, United States) and held with four stainless steel screws (McMaster-Carr Supply Co., 2,828 No Paulina St, Chicago, IL 60657). The software for the MoCAP device was developed in-house utilizing a programming language (Python Software Foundation, 512 Lafayette Boulevard, Suite 2, Fredericksburg, Virginia 2,240). The software was based on a prior design used to control a spinning bioreactor for brain organoid culture (Romero-Morales et al., 2019).

All cells were grown in a standard humidified incubator (5% CO₂, 37°C). Primary BMECs (ACBRI 376) and immortalized BMECs (hCMEC/D3) (Weksler et al., 2005) were cultured and expanded with endothelial cell growth medium (R&D Systems) supplemented with Endothelial Cell Growth Supplement (R&D Systems), 10% fetal bovine serum (Thermo Fisher), GlutaMAX (Thermo Fisher), and gentamycin (25 μg/mL; Thermo Fisher). Primary and immortalized BMECs were seeded at 100,000 cells/cm² onto polyester Transwell^® filters (3,470; 0.4 µm pore size; Corning) coated with collagen IV (400 μg/mL; Sigma-Aldrich) and fibronectin (100 μg/mL; Sigma-Aldrich). Twenty-four hours later, the medium was changed to remove floating cells. Induced pluripotent stem cells (iPSCs; CC3 line) were cultured and seeded as previously published (Neal et al., 2019; Hollmann et al., 2017). Briefly, iPSCs were seeded onto Matrigel-coated plates with E8 medium supplemented with Y-27632 (Tocris) at a cell density of 15,600 cells/cm². Cells were differentiated 24 h after seeding by changing to E6 medium. E6 medium was replenished daily for 4 days. Then, cells were switched to human endothelial serum-free medium (hESFM; Thermo Fisher Scientific) supplemented with 10 µM retinoic acid (RA; Sigma-Aldrich), 20 ng/mL human basic fibroblast growth factor (bFGF; Peprotech), and B27 supplement (Thermo Fisher Scientific). Medium was not changed for 48 h. Then, cells were collected and seeded at 100,000 cells/cm² onto polyester Transwell^® filters (3,470; 0.4 µm pore size; Corning) coated with collagen IV (400 μg/mL; Sigma-Aldrich) and fibronectin (100 μg/mL; Sigma-Aldrich). Twenty-four hours after seeding, RA and bFGF were removed from the medium to induce barrier phenotype. For all experiments, media volumes were 200 μL and 600 µL in the apical and basolateral chambers of the Transwell^® insert, respectively.

To increase the cell media viscosity to 3 mPa (6.5% dextran w/w), we referenced data from two publications (Rouleau et al., 2010; Li et al., 2008), created a concentration curve with the data, and performed a non-linear regression analysis to calculate the concentration that would yield our desired media viscosity (Supplementary Figure S1). Dextran powder (40 kDa; Sigma-Aldrich) was dissolved in non-supplemented media utilizing a heat plate (40°C) and sterilized by vacuum filtration using a pre-heated filter (Sigma-Aldrich).

The MoCAP device was sterilized in an autoclave before each experiment. The motor is removed from the hardware and sterilized separately with 70% ethanol. Cells were cultured on 24-well Transwell^® inserts for 2 days, as detailed above, before exposure to flow. Cells were then continuously exposed to a defined CSS or PSS profile except for the brief daily period when TEER measurements were acquired. Culture medium (50 µL) was added to the apical chamber of filters each day to account for media loss during TEER measurements.

To calculate the shear stress induced by the MoCAP device, we utilized the analytical solution for a cone-and-plate system as previously described in the literature (Franzoni et al., 2016; Sucosky et al., 2008). The following formula was used to estimate shear stress and calibrate the angular of the velocity of the MoCAP device: τ = µ   ω   r h + r   tan ⁡ α  

Here, τ is shear stress, µ is the dynamic viscosity of the cell media (3 mPa), ω is angular velocity, h is the gap between the tip of the low angle cone and the cells (200 µm), r is the cone radius (0.29 cm), and α is the angle of the cone (2°). The angular velocity was adjusted to account for the dynamic viscosity of the cell media utilized during experimentation. Supplementary Figure S2A shows these parameters in the context of a single cone within a Transwell^® insert. While the constant distance between the cone and cells is assumed, it is understood that Transwell^® inserts are not perfectly flat and can move slightly within the cell culture well; thus, the shear stress is an estimate based on the constant distance assumption.

To calibrate the MoCAP device, an initial set of low angle cones with an arbitrary shaft length were printed and parylene coated. Plates were seeded with BMECs, and the MoCAP was placed on top of the plates and run overnight. The next day, the MoCAP device was removed, and the plates were inspected for cell monolayer damage under a phase contrast microscope. The experiment was then repeated with sequential additions of 100-micron shims (McMaster-Carr Supply Co.; 90214A111) until no scratches were detected in the BMEC monolayers. The low angle cones were then reprinted taking into the account the number of shims utilized to achieve zero damage to the cell monolayer. This final gap of 200 µm between the tip of the low angle cone and cells was utilized for all experimentation. Once fully assembled, reflective tape was placed on top of the shaft cap to verify that the input angular velocity of the software matched the output angular velocity of the motor utilizing a digital tachometer (NEIKO Tools, Taiwan).

To calculate the shear stress delay caused by gear backlash within the MoCAP device, we utilized the following formula: L a g   T i m e = ∆ θ   b a c k l a s h ω i n p u t

Here, ∆θ represents the gear backlash angle and ω i n p u t represents the angular velocity input in the MoCAP device. The gear backlash angle was measured to be 2.2° in the 3D printed gears utilized in the MoCAP device.

TEER measurements were taken daily during shear stress treatments. Prior to measurements, the MoCAP was removed from the filters and cells were equilibrated at room temperature for 10 min. Then, measurements were acquired with a commercially available electrode system (World Precision Instruments) with a chopstick configuration (STX2). Chopsticks were carefully inserted into the Transwell^® filters to avoid scratching the cell monolayer. The measurements were recorded after the signal had stabilized. The reported TEER (T_E) was determined with the following formula: T E = T M − T B × A r e a

The measured TEER from an endothelial monolayer (T_M) was subtracted by the measured TEER from a blank Transwell^® insert with no cells (T_B). This quantity was then multiplied by the surface area of the Transwell^® insert (0.33 cm²) to determine T_E. All TEER measurements in this study are reported as Ω × cm².

Following the mechanofluidic assays, cells were immediately fixed with 4% paraformaldehyde for 5 min. Cells were then washed three times with phosphate buffered saline (PBS). Once rinsed, cells were permeabilized with PBS containing 0.3% Triton X-100 for 5 min. After permeabilizing, the cells were blocked with PBS containing 10% goat serum for 60 min. Cells were then incubated with fluorescent-conjugated anti-GLUT1 antibody or phalloidin in PBS containing 10% goat serum overnight at 4°C on a shaker (R&D Systems FAB1418G, 1:250; Thermo Fisher Scientific A12379, 1:250). The next day, cells were rinsed three times with PBS and incubated with DAPI to label nuclei (Thermo Fisher Scientific 62248, 1:1,000), then rinsed three final times with PBS. The Transwell^® inserts were then transferred into a 12-well glass bottom plate (Cellvis). To improve imaging quality, a solution of 2.5 M fructose and 60% glycerol was used as the final imaging medium on the apical and basolateral sides of the Transwell^® inserts (Dekkers et al., 2019). Cells were imaged utilizing a Leica DMi8 epifluorescent microscope.

All experimental results are shown as mean ± standard error of the mean (SEM). Multiple comparisons between groups were analyzed by two-way ANOVA followed by a Bonferroni’s post hoc test. A two-tailed probability value p < 0.05 was considered statistically significant. An independent replicate for each experiment was considered as a singular Transwell^® insert. One MoCAP run refers to an experiment utilizing one MoCAP device applying shear stress across a plate of 24-well Transwell^® inserts.

We created a modular cone-and-plate (MoCAP) mechanofluidic device that is compatible with the 24-well Transwell^® insert system (Figure 1A). The MoCAP device consists of a nesting three-part housing body made up of a lid, an upper housing, and a lower housing (Figure 1B). A NEMA 17 stepper motor can be mounted to the lid in multiple column positions to allow for rapid reposition using M3 stainless steel screws. A key and lock geometry were utilized on custom made gears with a variety of gear ratios (1:1 and 2:1), as well as on the shaft of the low angle cones that were 3D printed, to allow users to quickly interchange components without the need for screws. A 2° low angle was utilized on the cones, as this design has been extensively characterized in the literature, (Chavarria et al., 2023; Spruell and Baker, 2013; Spencer et al., 2016) and the low angle cones were parylene coated to ensure biocompatibility (O’Grady et al., 2021). Bearings were press fitted into designed recesses on the lower and upper housings units to reduce friction during the rotation of the cones. All components of the MoCAP device (except the bearings and NEMA17 motor) were 3D printed with high-temperature resin, which allows for the device to be sterilized with an autoclave. Once fully assembled, the MoCAP device can be placed directly on top of a 24-well Transwell^® insert system, acting as a plate lid (Figure 1C). A final gap of 200 µm between the tip of the low angle cone and the cells was utilized to prevent any accidental scratching of the cell monolayer when operating and removing the MoCAP device from the Transwell^® inserts. The analytical evaluation of this gap configuration, taking into account the thicknesses of endothelial cells that are reported to range from 0.1–10 µm (Félétou, 2011), demonstrated a negligible difference in the shear stress produced by the MoCAP device at the cell surface creating similar shear stress profiles with a maximum shear stress value closer to the edges of the Transwell^® inserts and a dead point at the center of the Transwell^® insert (Supplementary Figure S2B). A shaft cap was printed and placed on top of the double shaft NEMA17 motor to visualize the rotation of the motor shaft (Figure 1C). Reflective tape was placed on top of the shaft cap to verify the input angular velocity of the software matched the output angular velocity of the motor utilizing a digital tachometer. The electronics and software were made in-house based on prior designs (Romero-Morales et al., 2019; O’Grady et al., 2018) and allow for the simultaneous operation of up to three MoCAP devices (Figure 1D).

The MoCAP device generates different shear stress magnitudes at each column by changing the gear ratio between the gears connecting adjacent columns. For example, the gears illustrated in gold in column C1 consist of a 1:1 gear ratio, resulting in the transfer of the same angular velocity from one gear to its neighboring connecting gear down the column (Figure 2A). The white gears illustrated on the diagram are staggered on height placement, allowing the gold and white gears to spin in different planes without interference (Figure 1B). The white gears connecting adjacent columns on the diagram have a 2:1 gear ratio, therefore multiplying the angular velocity from left to the right column by a factor of two. On the diagram shown, we start with an input of 28 rotations per minute (RPM), corresponding to 0.6 dyn/cm² from the NEMA17 motor that is doubled every column, which yields 32 times the initial angular velocity (896 RPM) at the last column in the MoCAP gearbox system (Figure 2A). This gear ratio principle allows the MoCAP device to achieve angular velocities of 28, 56, 112, 224, 448, and 896 RPM, with the corresponding shear stress values of 0.6, 1.3, 2.5, 5, 10, and 20 dyn/cm² at columns C1, C2, C3, C4, C5, and C6, respectively (Table 2). Thus, when the MoCAP device is run at a constant angular velocity, we can generate CSS ranging from 0.6 to 20 dyn/cm² by filling the entire gearbox with these 1:2 gear ratios between columns (Figure 2B). The maximum shear stress the MoCAP can generate using this configuration is 20 dyn/cm², as higher angular velocities exceed the maximum bipolar frequency of the stepper motor. An advantage of CAP systems is that they allow for the creation of intricate flow patterns. As such, in the MoCAP device, we can also generate PSS profiles with minimum to maximum shear stress values ranging from 0.4 to 0.6, 0.7 to 1.3, 1.4 to 2.5, 2.8 to 5, 5.6 to 10, and 11.2–20 dyn/cm² (Figure 2C). For experimental testing, we chose to focus on only shear stress profiles with a maximum shear stress value of 0.6 and 10 dyn/cm² for CSS and PSS as these shear stress values approximate commonly used values on previous in vitro models of the BBB ranging from 0.4–12 dyn/cm² (Meena et al., 2022; Santa-Maria et al., 2021; Suprewicz et al., 2022; Yeon et al., 2012; DeStefano et al., 2017; Peng et al., 2020; Harding et al., 2022). For PSS conditions, we chose to utilize an arbitrary frequency of 1 Hz (Hz) to mimic a normal adult resting heart rate (Nanchen, 2018), however, the frequency can be adjusted as needed for any system of interest.

To demonstrate the capabilities of the MoCAP, we conducted exploratory evaluations into the effects of CSS and PSS on BBB function across several common BMEC sources (primary, immortalized, and iPSC-derived). The MoCAP device was reconfigured with a 1 to 16 gear reduction ratio (Figure 3A). In this configuration, the motor drives the gold gears at an angular velocity of 28 RPM creating 0.6 dyn/cm² (Figure 3B). The white gears inside the top compartment of the MoCAP device then increase the input angular velocity and transfer it down to the black gears increasing the angular velocity to 448 RPM and thus generating 10 dyn/cm² of shear stress inside the Transwell^® inserts (Figure 3B). In this configuration, the application of shear stress to the cell culture monolayer is almost instantaneous, with a small millisecond delay (13 m) caused by the gear backlash within the MoCAP gearbox (Supplementary Figure S3). We exposed cells to CSS and PSS for 2 days, utilizing a high and low threshold described above (10 and 0.6 dyn/cm², respectively). We included dextran at a concentration of 6.5% (w/w), which increased media viscosity to 3 mPa based on rheology data from previous studies (Rouleau et al., 2010; Li et al., 2008). This allowed us to run the MoCAP device at a lower angular velocity to minimize the incorporation of bubbles into the cell media, while still achieving our desired shear stresses. Daily TEER measurements were collected throughout the course of the experiment for each cell line (Figure 3C). After 2 days of CSS and PSS acclimation, final TEER measurements were collected and compared across shear stress conditions within the same cell line. For the primary BMECs, there was a statistically significant decrease in TEER in the 0.6 dyn/cm² CSS, 10 dyn/cm² CSS, and 10 dyn/cm² PSS conditions when compared to the static control group (Figure 3D). There was a statistically significant increase in TEER for the immortalized BMECs exposed to 0.6 dyn/cm² CSS and 0.6 dyn/cm² PSS (Figure 3D). It is important to note that the magnitude of TEER differences in the primary and immortalized BMECs was very low, since these cells have poor passive barrier properties (Supplementary Figure S4). In contrast, for the iPSC-BMECs, which have comparably higher baseline TEER (Supplementary Figure S4), there were no statistically significant differences detected after 2 days of CSS or PSS acclimation. Upon visual inspection, approximately 65% of the filters had intact monolayers regardless of exposure to PSS or CSS, and we anticipate this number could be improved with additional optimization of media and culture conditions. Only filters with intact monolayers were utilized for downstream cellular and molecular analyses.

To determine whether CSS and PSS exposure affected nuclear morphology, we analyzed the three BMEC lines using a DAPI nuclear stain (Figure 4A). The number of cell nuclei per field of view (FOV) was counted and analyzed, which first revealed statistically significant differences in cell nuclei numbers between all cell lines (Supplementary Figure S5A), indicative of different packing densities. In terms of responsiveness to shear stress, we observed a statistically significant decrease in the number of cell nuclei per FOV for the primary BMECs acclimated to 0.6 dyn/cm² PSS, and 10 dyn/cm² PSS when compared to the static primary BMEC control (Figure 4B). The analysis also revealed a statistically significant decrease in cell nuclei count per FOV in all the immortalized BMECs acclimated to shear stress when compared to their corresponding static control (Figure 4B). Further, there was a statistically significant decrease in cell nuclei per FOV in all iPSC-BMECs acclimated to shear stress when compared to their respective static control (Figure 4B). These results suggest a change in cell density induced by different shear stress conditions within all the cell lines. We next analyzed the average nuclei area. Here, we observed a statistically significant decrease in cell nuclei area for primary BMECs acclimated to 0.6 dyn/cm² CSS when compared to the statically cultured primary BMECs (Figure 4C). There was also a statistically significant increase in cell nuclei area for primary BMECs acclimated to 0.6 dyn/cm² PSS and iPSC-derived BMECs acclimated to 0.6 dyn/cm² CSS when compared to their respective static control (Figure 4C). Cell nuclei area remained constant for all other shear stress conditions in all three BMEC lines when compared to their respective static control groups (Figure 4C), although we further note a statistically significant difference between the statically cultured primary, immortalized, and iPSC-BMEC average cell nuclei area (Supplementary Figure S5B), mirroring the differences in cell density between the lines. Overall, our results illustrate differential responses of the BMEC lines to shear stress, while all BMEC lines are resistant to shear stress induced nuclear shrinkage, which has been previously noted to occur in other cell types (Sahni et al., 2023; Jetta et al., 2019; Jin et al., 2020).

We investigated the effects of CSS and PSS on glucose transporter 1 (GLUT1) expression, as previous reports noted an upregulation in GLUT1 transporter expression when primary BMECs were exposed to 10 dyn/cm² of CSS, (Chavarria et al., 2023; Garcia-Polite et al., 2017), whereas GLUT1 expression in iPSC-BMECs is reported to be insensitive to CSS (DeStefano et al., 2017). We repeated the previously mentioned shear stress experiment and then performed immunofluorescent staining for GLUT1 (Figure 5A). The quantification of pixel intensity of GLUT1 showed a statistically significant decrease in primary BMECs exposed to 0.6 dyn/cm² CSS and 0.6 dyn/cm² PSS compared to their respective static control, as well as an increase in response to 10 dyn/cm² CSS that was not quite statistically significant (p = 0.0781) (Figure 5B). There was also a statistically significant decrease in GLUT-1 intensity in the immortalized BMECs exposed to 10 dyn/cm² CSS, 0.6 dyn/cm² PSS, and 10 dyn/cm² PSS compared to their respective static control (Figure 5B). No significant differences were noted in the iPSC-BMECs between any of the experimental conditions. Thus, our results are generally consistent with the published literature.

To examine the impact of shear stress on BMEC cytoskeleton morphology, we additionally performed immunofluorescent staining of the actin cytoskeleton utilizing a fluorescently labeled phalloidin after the cells had been exposed to CSS or PSS (Figure 6A). From the actin labeling, we manually measured cell width and length to calculate the inverse aspect ratio (IAR) and evaluate morphological changes in each BMEC line exposed to CSS or PSS. The IAR of each BMEC line remained constant irrespective of the shear stress condition it was exposed to (Figure 6B). The primary BMECs cultured under static conditions had significantly smaller IAR compared to immortalized BMECs and iPSC-BMECs (Supplementary Figure S6). This finding implies baseline differences in IAR between cell lines under static conditions. We then proceeded to calculate cell orientation based on the angle created between the flow vector and the length measured of each cell. This cell orientation measurement was not calculated on the static control groups as these cells were not exposed to flow conditions and did not have a flow vector. This analysis revealed that all cell lines, regardless of shear stress condition, align approximately perpendicular to the flow direction (∼90°; Figure 6C).