The hCMEC/D3 cell line was obtained from Jennio Bio (Guangzhou, China). Cells were plated in cell culture flasks, incubated in a 5% CO₂ atmosphere at 37 °C, and culture medium (90% Dulbecco’s Modified Eagle Medium High Glucose + 10% fetal bovine serum; Gibco) changed every 2 days. Cells were used for experiments at passage 2–8. For intracellular Ca²⁺ [Ca²⁺]_i measurement and immunofluorescence experiments, cells were plated onto glass coverslips (8 mm diameter) and grown to 80% and 50% confluence, respectively.

Cells on glass coverslips were placed in a chamber perfused with normal Hanks’ buffered saline solution (HBSS) containing (in mM): 138 NaCl, 5 KCl, 0.3 KH₂PO₄, 4 NaHCO₃, 2 CaCl₂, 1 MgCl₂, 10 HEPES, and 5.6 glucose (pH 7.4). Mechanical stimulation of single cells was performed as described by Neuhaus et al. (2020). To mechanically stimulate single cells, glass micropipettes with closed and rounded tips (approximately 2 μm) were used to deflect the plasma membrane. Glass micropipette movement was controlled using a motorized MP-285 micromanipulator (Sutter Instruments, Novato, CA, United States); when the tip was close to the cell, the micropipette was dropped in 2 μm increments, to induce membrane deflection.

Cells on glass coverslips were perfused with a PE10 tube filled with HBSS. One end of the PE10 tube was connected to a 50 mL syringe pump (Genie touch, Kent Scientific Co.) and the other end was mounted on the probe of a patch-clamp recording platform (EPC10, HEKA Electronik, Lambrecht/Pfalz). The probe approached the cells at a 45° angle, controlled using a MP-285 micromanipulator (Sutter Instrument) and with the tip of the PE10 tube 30–50 μm away from the stimulated cells (Figure 1C). Flow SS was induced by changing the perfusion flow rate from 2 to 20 mL/min.

Ca²⁺ imaging experiment was conducted as described in our previous study (Wen et al., 2018). In brief, hCMEC/D3 cells on glass coverslips (n = 50–60 cells) were loaded with 1 µM fura-2/AM for 30–45 min at 37 °C in normal HBSS. Subsequently, cells were washed with HBSS to remove excessive external fura-2 dye. The glass coverslip was then mounted on a stage equipped with an inverted fluorescence microscope and HCMEC/D3 cells continuously perfused with HBSS driven by gravity. Cells were exposed to alternating 340- and 380-nm wavelength (bandwidth, 10 nm) ultraviolet light (1/s). Excitation time and image acquisition were controlled using Meta Fluor R (Molecular Devices, Sunnyvale, CA, United States). Data were analyzed using program C-Imaging (Compix). Background was subtracted to minimize camera dark noise. The ratio of fluorescence signal measured at 340 nm divided by the fluorescence signal measured at 380 nm was used to measure the increase of intracellular Ca²⁺. Baseline intracellular Ca²⁺ concentration was determined from the average of five to eight measurements obtained before drug application. Amplitudes of peak Ca²⁺ responses were computed as the difference between the peak value and the baseline value (Δ340/380). A significant increase in [Ca²⁺]_i was considered if the ratio changed by >0.1.

Total RNA was extracted from cultured cells using an RNA Simple Total RNA kit (YISHAN, Shanghai, China), according to the manufacturer’s instructions. An ultraviolet spectrophotometer was used to determine RNA concentrations. Reverse transcription was conducted using 1 µg RNA with the HiScript III RT SuperMix kit (Vazyme, Nanjing, China), and complimentary-DNA was used for RT-qPCR amplification. TRP and Piezo1 channel mRNA levels were measured by qPCR using SYBR qPCR Master Mix (Vazyme) and a QuantStudio™ 5 system (Thermo Fisher, Waltham, MA, United States), with the following cycling conditions: initial denaturation at 95 °C (2 min) followed by 40 cycles of denaturation at 95 °C (15 s) and combined annealing and extension at 60 °C (30 s). Primers for TRP and Piezo channels were generated by Sangon Biotech (Shanghai, China); the sequences are shown in Table 1.

hCMEC/D3 cells were fixed with 4% paraformaldehyde for 10 min at room temperature, then washed three times at room temperature using phosphate-buffered saline (PBS) and blocked with 5% normal goat serum for 30 min at room temperature. Cells were mixed with Piezo1 antibody (1:100; 15939-1-AP; Proteintech, China), TRPV4 (1:100; DF8624; Affinity Biosciences, China) and LAT1 (1:100; sc-374232; Santa Cruz Biotechnology, United States) overnight at 4 °C followed by 594-conjugated Goat Anti-Mouse IgG (H + L, 1:100; ABclonal, China) and Plus 488-Goat Anti-Rabbit Recombinant Secondary Antibody (H + L, 1:100; Proteintech, China) for 1 h at room temperature. Human brain tissue was obtained from one patient who underwent glioma resection surgery and written informed consent was provided by the patient. Brain tissue was fixed with 4% paraformaldehyde and 5 µm tissue slices were prepared for immunofluorescence staining as for hCMEC/D3 cells. The specificity of the antibodies for TRPV4 and Piezo1 have already been verified by the companies and the literature reports (Eisenhoffer et al., 2012). The immunofluorescence staining was analyzed using the Olympus BX53 positive fluorescence microscope (Olympus Corporation, Japan).

The total protein was extracted using RIPA lysis buffer (NCM Biotech, China) containing proteinase inhibitors. The protein concentration was determined using bicinchoninic acid (BCA) kit (Epizyme Biotech, China). The samples were separated by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to polyvinylidene fluoride (PVDF) membranes by wet transfer. The membranes were incubated overnight at 4 °C with primary antibodies against the following proteins: Piezo1 (1:1,000; A23380; ABclonal, China), TRPV4 (1:1,000; A5660; ABclonal, China) and β-actin (1:5,000, AC026; ABclonal, China). After incubation with specific secondary antibodies, proteins were detected with an enhanced chemiluminescence kit (Millipore, Germany).

A custom-designed pressure system was employed to apply pressure stimulation. Briefly, the hCMEC/D3 cells were seeded on 6-well plates reached > 90% confluence before pressure exposure. The cells were exposed to mechanical pressure of 100 cm H₂O at a frequency of 1 Hz (induced by pneumatic elements). Previous research from our lab confirmed that the pH, pO2, and pCO2 of the culture supernatants were not affected by the application of hydrostatic pressure at levels similar to those used in this study.

When cells cultured in 6-well plates reached > 90% confluence, they were stimulated with mechanical and pharmacological stimulation. Culture supernatants were obtained 2 min before and 1 min after stimulation.

NO concentration was determined with NO Colorimetric Assay Kit (Elabscience, Wuhan, China). Color developer (80 μL) was added to 160 μL sample, according to the manufacturer’s instructions. Samples were mixed and allowed to stand for 15 min and the absorbance values of each well at 550 nm measured following labelling with an enzyme marker. ATP concentration was determined with CellTiterGlo™ Luminescent Cell Viability Assay kit (Promega, Fitchburg, WI, United States). Samples were analyzed using a GloMax™ 20/20 luminometer (Promega).

The following drugs were used: Yoda1 (MCE, China), Dooku1 (TargetMOI, China), GSK (MCE, China), HC067047 (TargetMOI, China), PPADS (MCE, China), and apyrase (Sigma-Aldrich, China); drugs were stored at −20 °C before use. Dilutions were made with deionized water or DMSO on the day of experiments. The final concentrations of DMSO were <0.1%. In all experimental groups, at least one parallel control experiment was performed in the presence of the carrier, which had no pharmacological effect on either tension or agonist-induced contraction of the preparations. Chemicals were used in concentrations shown to be effective in preliminary experiments or in concentrations used in previously published studies (Swain et al., 2020; Zhao et al., 2021).

Continuous variables are expressed as mean ± SD. Differences between two groups were analyzed by paired or unpaired t-test. Excel™ (Microsoft, Redmond, WA, United States) and Prism 8.0.2 (GraphPad, San Diego, CA, United States) were used for analyses. Differences were considered statistically significant when p < 0.05.

To determine whether hCMEC/D3 cells are responsive to mechanical stimuli, we poked individual hCMEC/D3 cells with a glass micropipette (the tip of the pipette was indicated by the star in Figure 1A) and recorded the change in [Ca²⁺]_i using a Ca²⁺ imaging method. We found that poke stimulus evoked a rapid increase in [Ca²⁺]_i in stimulated cell (cell 1 in Figure 1B). Surprisingly, cells neighboring the stimulated cell also exhibited increased [Ca²⁺]_i (cell 2 and cell 3 in Figures 1A,B), and the [Ca²⁺]_i peak amplitude increase in neighboring cells reduced with distance from the stimulated cell (Figure 1B). These data suggest that elevated [Ca²⁺]_i in mechanically stimulated cells leads to release of a transmitter that acts on neighboring cells, thereby propagating a Ca²⁺ wave from one cell to the next.

hCMEC/D3 cells were also responsive to flow SS stimulation (Figures 1C,D); however, the amplitude of [Ca²⁺]_i increase induced by SS was relatively less than that provoked by poke stimulation (Figure 1B vs. 1D).

To examine which channels mediate the [Ca²⁺]_i increase triggered by mechanical stimuli in hCMEC/D3 cells, the expression levels of several mechanical sensing channels were first investigated in cultured hCMEC/D3 cells at the mRNA level by RT-qPCR. RT-qPCR experiments showed that the relative mRNA expression levels of PIEZO1 and TRPV4 were highest, while those of TRPV2 and Piezo2 were relatively low (Figure 2D). Accordingly, immunofluorescence and western bolt experiments revealed prominent staining for PIEZO1 and TRPV4 channels in cultured hCMEC/D3 cells (Figures 2A,B) stained with L-type amino acid transporter 1 (LAT1), one specific marker of HBMECs. To note, prominent staining of Piezo1 and TRPV4 channels were also found on microvascular endothelial cells of the human brain tissue (Figure 2C).

Functional expression of Piezo1 and TRPV4 channels was examined by Ca²⁺ imaging using specific selective agonists (Figure 3). Treatment of cultured hCMEC/D3 cells (n = 180) with the Piezo1-specific agonist, Yoda 1 (30 µM), induced a significant increase in [Ca²⁺]_i (Figure 3A), and this effect of Yoda 1 on [Ca²⁺] _i was significantly diminished following pretreatment of cultured hCMEC/D3 cells with DooKu1 (10 µM), a Piezo1 antagonist (Evans et al., 2018) (Figures 3B,C). Application of GSK1016790A (GSK, 500 nM), a specific agonist of the TRPV4 channel, to cultured hCMEC/D3 cells (n = 165), induced a significant increase in [Ca²⁺]_I (Figure 3D), which was significantly blocked after pretreatment of cells with the TRPV4 specific antagonist, HC-067047 (1 µM) (Figures 3E,F). Together, the above findings suggest that Piezo1 and TRPV4 channels may be responsible for sensing mechanical stimuli in cultured hCMEC/D3 cells.

To investigate the involvement of Piezo1 and TRPV4 in hCMEC/D3 cell mechanical responses, the effects of Piezo1 and TRPV4 antagonists on poke- or SS-induced [Ca²⁺]_i were analyzed. As expected, poke-induced [Ca²⁺]_i increases was significantly reduced in the presence of Piezo1 (DooKu1, 10 μM) and TRPV4 (HC-067047, 1 μM) antagonists (Figures 4A–D) in both stimulated and neighboring cells. SS-induced increase in [Ca²⁺]_i were also significantly reduced in the presence of Piezo1 and TRPV4 antagonists (Figures 4E,F).

As demonstrated by the results presented in Figure 1, cells surrounding poked cells also showed increases in [Ca²⁺]_i, suggesting the release of a transmitter from the poked cell that influences the surrounding cells in a paracrine manner.

NO is considered a transmitter in cerebrovascular endothelial cells, and is released in response to [Ca²⁺]_i increase (Boedtkjer et al., 2011; Kawano et al., 2007; Lin et al., 2013). To determine whether NO was involved, we measured NO concentrations in culture supernatants before and after TRPV4 and piezo1 agonist stimulation. Unexpectedly, no significant change in NO concentration was detected after stimulation with the Piezo1 agonist, Yoda1 (30 μM), or the TRPV4 agonist, GSK (500 nM) (Figure 5A).

ATP is another transmitter involved in endothelial cell communication (Subauste, 2019). To assess ATP involvement in the observed phenomenon, we next measured ATP concentrations in culture supernatants before and after TRPV4 and Piezo1 agonist stimulation. As expected, application of the Piezo1 agonist, Yoda1 (30 µM), and the TRPV4 agonist GSK (500 nM), resulted in a significant increase in ATP concentration (Figure 5B).

To further confirm our findings, we conducted mechanical stress experiments. We measured the concentrations of ATP and NO in the culture supernatants before and after mechanical stimulation. Consistent with our pharmacological stimulation results, mechanical stimulation also led to an increase in ATP concentration (Figure 5D), while NO levels remained unchanged (Figure 5C).

As previously demonstrated in peripheral vessel endothelial cells (Lillo et al., 2018), ATP also induced an increase in [Ca²⁺]_i in cultured hCMEC/D3 cells, and this ATP-evoked [Ca²⁺]_i increase could be reduced using the non-selective P2X and P2Y receptor blocker, pyridoxal-6-phenyl-2′,4′-disulfonic acid (PPADS). To investigate whether released ATP contribute to poke-induced [Ca²⁺]_i increase in an autocrine manner, the ATP-diphosphate hydrolase, apyrase, which can degrade extracellular ATP to ADP, and PPADS were applied. Pretreatment with apyrase (10 U/mL) and PPADS (10 μM) significantly inhibited the poke-induced increase in [Ca²⁺]_i in both poked and surrounding cells (Figure 6).