HUVECs (Lonza, C2519AS) were thawed and cultured in EGM-2 medium (Lonza, CC-3162). HUVECs were frozen at passage number 5 in EGM-2 + 10% FBS + 10% DMSO and stored at -150oC for further use. Peripheral blood mononuclear cells (PBMCs) were isolated from a Buffy Coat (Sanquin) as previously described by de Haan et al. (26) PBMCs were banked in 90% FBS + 10% DMSO and stored at –150°C for further use.

PBMCs were thawed and cultured in AIM-V medium (Gibco, 12055-091). T cell population in the PBMCs were stimulated for 48 hours by Dynabeads^® Human T-Activator CD3/CD28 (ThermoFisher Scientific, 11131D) according to the manufacturer’s instructions. Unstimulated PBMCs were cultured in AIM-V for 48 hours. Prior to adding PBMCs to endothelial tubules in an OrganoPlate, PBMCs (both unstimulated and stimulated) were labeled with CellTracker Orange CMRA (Invitrogen, C34551), as previously described in de Haan et al. (26) In short, CMRA stock of 5 mM in dimethylsulfoxide (DMSO, Sigma, D8418) was diluted to a working concentration of 2.5 µM in AIM-V medium. PBMCs were harvested and pelleted (300g for 5 min) before being resuspended in 2 ml of CMRA working solution. Cells were then placed in a humidified incubator at 37°C, 5% CO₂ for 30 minutes. After the incubation period, 10 mL of AIM-V medium was added, cells were counted, and cells were pelleted (300g for 5 min). The appropriate volume was added to the PBMCs to have a final concentration of 400,000 cells/ml.

OrganoReady Blood Vessel HUVEC 3-lane 64 plates (Mimetas B.V, MI-OR-BV-02) were cultured according to manufacturer’s instructions. Medium was changed on day of receiving to OrganoMedium HUVEC-BM (Mimetas B.V.). OrganoReady Blood Vessel HUVEC 3-lane 64 are ready to use HUVEC tubes in OrganoPlates, that follow a similar process for ECM seeding (rat tail collagen 1 at 4mg/ml), endothelial cell seeding and establishment of perfusion through the tubules as described by Duinen et al. for the OrganoPlate 3-lane 40 (41). Perfusion flow was maintained by placing the plate on OrganoFlow rocker (Mimetas B.V., MI-OFPR-L) set at 14 degrees with 8-minute intervals optimized for the 3-lane 64. This allows the perfusion of medium through the endothelial tube. On the second day after receiving, cultures were exposed to cytokines or PBMCs. Prior to TEER timelapse set up, the OrganoPlate was transferred to the ImageXpress Micro XLS microscope (Molecular devices) and every individual chip on the OrganoPlate 3-lane 64 was imaged for 4x magnification phase contrast images. OrganoPlate was placed back to 37°C incubator. TEER timelapse were started directly after addition of medium.

New electrode boards matching the layout of the 3-lane 64 (Figure 1), comprising 512 stainless steel electrodes were designed and manufactured as previously reported (28). Prior to measurements, the electrode board was connected to the OrganoTEER measurement unit using mezzanine connectors and inserted into the inlet and access wells of the OrganoPlate to provide one current-carrying loop and one voltage-sensing loop for each of the 64 chips through a total of 512 electrodes.

Prior to all experiments, the OrganoTEER electrode boards were disinfected by spraying the electrodes surface with 70% ethanol and allowed to dry in a sterile flow cabinet. Before the start of timelapse measurements, a single electrode board and measurement unit were assembled and inserted into a sterile single well plate. The stack was then placed in an incubator at 37°C for at least two hours to allow the OrganoTEER temperature to equilibrate.

The left channel inlet and outlet of each chip were filled with 50 μL of HBSS (for the cytokine exposure study) or basal medium (for the PBMC study) 1 hour prior to starting timelapse. For the PBMC study, the left 4 columns of chips had basal medium added to the left perfusion lane and the right 4 columns of chips had basal medium with 800 ng/ml of CLCX12 to the left perfusion lane. The incubated OrganoTEER was placed into the flow cabinet and the single well plate was replaced with the OrganoPlate. The OrganoTEER was placed back into the incubator and onto the OrganoFlow rocker for 30 minutes prior to the timelapse. The OrganoFlow rocker was set to a static horizontal position for 6 minutes to let the liquid levels within the wells equilibrate, after which the T_-1 TEER measurement was taken.

For the cytokine exposure, TNFα (R&D Systems, 210-TA-020) and INF-γ (R&D systems, 285-IF-100) were diluted in OrganoMedium HUVEC-BM to the appropriate concentrations (0.1, 1, 10 and 100 ng/ml). The medium was equilibrated to 37°C for at least 30 minutes before addition of 50 µl medium to inlets and outlets of each chip in the OrganoPlate.

For the experiments with PBMCs, PBMCs were harvested and prepared as mentioned above, for the addition of 20,000 cells/chip. For conditions with cytokines (10 ng/ml of TNFα and INF-γ), 2x concentration (20 ng/ml) of cytokines were added to medium. To ensure swift media change, the different medium conditions were pipetted in a 384 well plate according to the inlet and outlet layout of the OrganoPlate, and allowed to incubate for 15 minutes at 37°C. The conditions where arranged such that 50 µl of medium containing 20,000 PBMCs were added to the corresponding OrganoPlate inlets and 50 µl medium with or without cytokines were added to the corresponding OrganoPlate outlets.

To expose the tubules to their respective condition, the medium was aspirated from all inlet and outlet wells of the OrganoPlate. A multi-channel pipette was used to transfer 50 μL of medium containing cytokines or PBMCs from the 384 well plate to the corresponding wells of the OrganoPlate. To minimize temperature fluctuation, the OrganoTEER assembly was placed on the static rocker platform. The first TEER measurement was taken immediately after the medium change. The OrganoFlow was started 30 seconds after the completion of the first timepoint measurement (14 degrees changes every 8 minutes). To ensure synchronicity with the rocker cycle, TEER measurements were taken every subsequent 4 minutes for the first 44 minutes (16 points), then every 64 minutes for 43 hours. An endpoint measurement was taken between 44 to 48 hours. After the completion of the TEER timelapse, the OrganoPlate was removed from the TEER device and transferred to the microscope and phase contrast images of each chip were taken. After phase contrast imaging, the plates were fixed in 3.7% formaldehyde (Sigma, 252549) and stored at 4°C.

All measurements were composed of 121 frequency dependent impedance points equally distributed in a logarithmic scale from 1000 Hz to 1 MHz. Using a controlled voltage source linked to an operational amplifier, the OrganoTEER imposed a sinusoidal AC voltage of 100 mV across the current carrying electrodes. The resulting current was measured across voltage sensing electrodes via a transimpedance amplifier. Within the measurement unit, a set of 12 separate digital impedance analyzer units linked to multiplexing units was used to acquire impedance values of 12 chips at a time.

To minimize the impact of parasitic capacitance (See Supplementary Information 1), we built three compensation boards by replacing the electrodes of each chip connection by a resistor of defined value per board (24 kΩ, 30 kΩ and 36 kΩ respectively, E24 standard). Each compensation board was measured once in the incubator where measurements were performed. The compensation spectra and associated resistor values were used by the OrganoTEER software to compensate the loss of high frequency signal induced by parasitic capacitances. All compensated measurement spectra were then automatically fitted against an analog model using the OrganoTEER fitting algorithm (28). The extracted barrier resistance parameter, measured in Ω, was multiplied by an approximative value of the ECM-cell interface area (0.0057 cm²) to produce a TEER value per chip in Ω·cm².

After fixation, cultures in the OrganoPlate were stained for immunofluorescent markers. As described previously, in short, cells were permeabilized using a Triton X-100 solution for 10 min and blocked using a buffer containing FBS, bovine serum albumin, and Tween-20 for 45 min (42). Primary antibody was incubated for 1–2 hours or overnight, after which secondary antibody was incubated for 1 hour. The following primary antibodies were used to stain fixed cultures: Anti‐VE-Cadherin 1:500 (Abcam, ab33168), anti‐ICAM‐1 1:50 (R&D systems, BBA3), anti-human CD45 (R&D systems, MAB1430). The following secondary antibodies were used to stain fixed cultures: Goat anti‐rabbit IgG (H+L) Alexa Fluor 488 1:250 (Thermo Fischer Scientific, A11008), Goat anti‐mouse IgG (H+L) Alexa Fluor 647 1:250 (Thermo Fischer Scientific, A21428) and CF647 Goat anti‐mouse IgG (H+L) Alexa Fluor 647 1:250 (Biotium, 20040). Nuclei were stained using Hoechst (ThermoFisher, H3570). After staining, the OrganoPlate was transferred to a confocal high content imaging microscope for automated imaging (Micro XLS-C, Molecular Devices). Images were acquired at 10x magnification at 3 µm increments along the height of the microfluidic channel. Analysis was based on Sum Projection (ICAM expression) or Max projection (VE cadherin) images of the top and bottom 10 z-slices.

Quantification of PBMC dynamics and ICAM immunofluorescent staining was performed as previously reported by de Haan et al. in 2021 (26). Quantification of VE-Cadherin immunofluorescent staining was performed using machine learning based segmentation software Cellpose2 (43) and image processing software Fiji (44). The cell perimeter was segmented based on VE-Cadherin expression, using the built in livecell LC4 model. The segmented cell outlines were loaded into Fiji and roundness value for each outline was extracted using the built in Particle analysis tool.

Means of two or more groups were assessed using Brown-Forsythe and Welch ANOVA (Gaussian, heterogeneity of variance) or Kruskal–Wallis tests (non-Gaussian) for statistically significant differences. In the case of two or more factors, two-way ANOVA tests or three-way ANOVA tests were formed depending on the number of factors. Multiple comparisons were accounted for using Tukey’s or Dunnett’s tests. All data set were tested for normality using QQ-plots and residuals for normality. Statistical analyses were performed using GraphPad Prism v9.4 (GraphPad Software). Differences were considered significant when p < 0.05. For the number of replicates, each independent experiment is represented by a “N” and the individual chip is represented by a “n”. Total number of replicates is N x n.

The OrganoTEER platform was optimized for compatibility with the automation friendly OrganoPlate 3-lane 64 and sensitivity was improved to increase its dynamic range around typical TEER values of endothelial culture. We redesigned the OrganoTEER system for the measurement of single tubules in the OrganoPlate 3-lane 64. We built an Electrode board compatible with both the OrganoPlate 3-lane 64 and the OrganoTEER measurement unit (Figure 1B). The OrganoTEER software was rewritten to accommodate the setup and measurement of each 64 chips (Figures 1C, D). The system allowed 512 stainless steel electrodes to be inserted into each chip perfusion channels access wells. For each chip, one current-carrying one voltage-sensing loop was created from the inlet and outlet of one perfusion channel to the outlet of the opposite perfusion channel (Figure 2C). For both the current carrying and voltage sensing loop, each point of the 4-terminal setup was shorted to a pair of electrodes inserted in both inlet and outlet of the same perfusion channel. A typical measurement takes under 2 minutes for an entire plate. The Software allows timelapse measurements to be configured to accommodate long term exposure over several days.

The configuration of our TEER measurement setup increased the chip electrical resistance over the 3-lane 40 due to longer current path. The electrode pairs on the basolateral side were inserted in the perfusion inlet and outlet in the 3-lane 64. This induced a higher impact from parallel parasitic elements. To compensate for this, we built three separate electrical compensation boards using the PCB layout the 3-lane 64 electrode board. In place of the electrodes, we soldered surface mounted resistors electrically connecting the reference and counter side to the work and working sense side. The boards resistances were 24 kΩ, 30 kΩ and 36 kΩ. Prior to the exposure experiment, we measured the impedance spectra of each of the 3 compensation boards, which we then used to perform a multiple load compensation on all measured data prior to fitting using the OrganoTEER software (Supplementary Information Figures 1, 2).

OrganoReady Blood Vessel HUVEC 3-lane 64 contained 60 tubular structures were purchased. The endothelial cells in these tubules had a cobblestone like appearance that can be observed using phase contrast microscopy throughout the duration of the culture (Supplemental Figure 3A). These endothelial cells expressed VE-cadherin at the junctions between cells (Figure 2A). Using confocal microscopy, a 3D representative image was created showing that the endothelial cells formed a confluent 3-dimensional endothelial tubule (Figure 2B). Using the new TEER board set up described above, the lower resistance endothelial tubule was measured and disruptions in barrier due to Staurosporine could be detected (Figure 2D).

To model vascular inflammation, the endothelial vessels were exposed to different inflammatory cytokines for 48 hours. Phase contrast images showed that the vessels exposed to INF-γ had minimal changes in morphology. However, as the concentration of TNFα increased, the endothelial cells showed a less rounded and aligned phenotype (Figure 3A). These changes in morphology were more pronounced in the combination of cytokines compared to TNFα alone.

The effect of inflammatory cytokines on TEER was monitor over 44 hours. INF-γ showed a dose dependent slow decrease in barrier integrity over time (Figure 3B). TNFα induced a sharp decline in barrier integrity around 2 hours which recovered towards the end of the exposure (Figure 3C). The higher concentrations of TNFα at 10 and 100 ng/ml induced the same amplitude of barrier leakage with no significant difference, indicating that the increasing concentration of TNFα did not cause increased disruption beyond 10 ng/ml. The lower concentrations of TNFα (0.1 and 1 ng/ml) showed more dose dependent disruption. However endothelial barriers were fully recovered by the end of the exposure at which point their TEER was not statistically significantly different compared to medium control (Supplementary Figure 2B). The combination of INF-γ and TNFα had the largest effect on the barrier integrity of the endothelial vessel (Figure 3D). The combination of cytokines induced a dose dependent decrease in barrier integrity around 2 hours, similar to the changes in barrier due to TNFα only. However, the barrier did not recover for the remainder of the exposure as seen for the TNFα alone, except for the lowest concentration (0.1 ng/ml) which was not significantly different than the medium control at the final timepoint (Supplemental Figure 4C).

The two inflammatory cytokines also had a different effect on ICAM-1 expression. TNFα and combination of TNFα and INF-γ caused a significant dose dependent increase in ICAM-1 expression (Figures 4A, B). INF-γ only did not cause a significant increase in ICAM-1 expression. In addition to changing the ICAM-1 expression, TNFα and INF-γ affected the VE-cadherin expression differently. In the control conditions, the HUVEC cells showed a rounder and more cobblestone like appearance (Figure 4C). INF-γ did not have a strong effect on cell morphology and VE-cadherin expression. As the concentration of TNFα increased, the endothelial cell morphology was less rounded and slightly more aligned (Figures 4C, D). At the higher concentrations with TNFα and both cytokines, the endothelial tubes are not only less round, but the consistency of the VE-cadherin staining is reduced (Supplemental Figure 3B). The combination of cytokines increased these changes in VE- cadherin expression with decreased roundness of cells and less consistent VE-cadherin expression in a dose dependent manner.

To determine whether we could also study the effect of immune cells directly on vascular inflammation non stimulated and stimulated PBMCs were added in the lumen of the endothelial tubules. The addition of immune cells to the endothelial tubule did not interfere with the addition of the TEER electrodes as observed in the schematic (Figure 5A). Addition of unstimulated PBMCs did not influence the endothelial barrier when compared to the control conditions (Figure 5B). However, stimulated PBMCs had similar effect on the barrier as the 10 ng/ml of TNFα and INF-γ condition. In both conditions, there was a sharp decrease in the first 2 hours that remained for the rest of the exposure. Stimulated PBMCs adhered to the vessel wall in greater number than unstimulated PBMCs (Figure 5C).

The chemoattractant, CLCX12, was added to the empty channel to create a chemotactic gradient to induce immune cell migration (Figure 5D). The addition of CXCL12 did not result in a significant change in barrier integrity of the vessels that were perfused with stimulated or unstimulated PBMCs (Figure 5E). Addition of only CXCL12 did not result in significant increase in adhesion (Supplemental Figure 5B) or extravasation of PBMCs (Figure 5F). The combination of CLCX12, TNFα and INF-γ resulted in a significant increase in migration of both unstimulated and stimulated PBMCs out of the vessel into the ECM (Figure 5F). PBMC migration into the gel channel was confirmed by the overlapping of CD45 and CellTracker™ Orange (Figures 5G, H).

Adding stimulated PBMCs to the endothelial tubule caused a significant increase in ICAM-1 expression in the endothelial tubule when compared to the medium only or unstimulated PBMCs (p<0.00001) (Figure 5I and Supplemental Figure 5E) However when cytokines were added all conditions showed even higher ICAM-1 expression compared to stimulated PBMCs (Figure 5I). A significant effect on cell roundness by VE-cadherin expression was found by the addition of stimulated PBMCs (Figure 5J). The addition of cytokines did not have an added effect on cell roundness when compared to stimulated PBMC condition (Figure 5J and Supplemental Figure 5D). When both stimulated PBMCs and cytokines were added the VE-cadherin expression was almost gone (Supplemental Figure 5D).