Mechanosensation by endothelial PIEZO1 is required for leukocyte diapedesis

The endothelial cell layer is a tight barrier for cells in the circulation. However, during inflammation, leukocytes can transmigrate the endothelium and extravasate in the perivascular space, a process that involves a well-coordinated cascade of events. This includes initial leukocyte capture and rolling, firm adhesion, and crawling, which are then followed by breaching of the endothelial barrier and the extravasation.^1-3 The molecular mechanisms that control and mediate the initial interactions between leukocytes and endothelial cells are well characterized and involve interactions between endothelial selectins and glycoproteins of leukocytes during capture and rolling steps, whereas arrest, firm adhesion, and crawling are mediated mainly by integrins on leukocytes that bind to endothelial ICAM-1 and VCAM-1 and induce their clustering.^2,4-6 How these initial processes are linked to the opening of the endothelial barrier, which requires the remodeling of endothelial adherens junctions and endothelial cell contraction,^7-12 is, however, poorly understood.

Opening of endothelial junctions and endothelial cell contraction during leukocyte transmigration require activation of endothelial signaling pathways, and several studies have shown that leukocytes induce an increase in the cytosolic Ca²⁺ concentration in endothelial cells.^13-18 This calcium signal is not necessary for leukocyte adhesion but is required to induce transendothelial migration.^13-15 ICAM-1 has been shown to be involved in lymphocyte-induced Ca²⁺ transients in endothelial cells,¹⁵ and, more recently, the transient receptor potential channel C6 (TRPC6) has been shown to be required for endothelial calcium transients induced by neutrophils and for transendothelial migration,¹⁹ but how leukocytes induce endothelial Ca²⁺ transients is still unclear.

The Piezo proteins PIEZO1 and PIEZO2 are mechanically activated cation channels that form homotrimeric complexes,^20-22 which are sufficient to mediate mechanically induced currents.²⁰ PIEZO1 has been shown to be gated directly by changes in membrane tension^23-25 and to mediate multiple cellular processes including endothelial flow sensing.^26-28 In this study, we found that PIEZO1 is required for leukocyte extravasation by coincidentally sensing increased membrane tension induced by flow and ICAM-1 clustering. The subsequent activation of signaling pathways then results in the localized opening of the endothelial barrier.

Cells were lysed in cell lysis buffer (9803; Cell Signaling) containing 1% triton X-100 or in radioimmunoprecipitation assay (RIPA) buffer (9806; Cell Signaling) supplemented with protease and phosphatase inhibitors (5872; Cell Signaling). Lysates were centrifuged at 10 000g at 4°C for 10 minutes. Supernatants were then subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were probed with primary and horseradish peroxidase–conjugated secondary antibodies (8884 and 7076, respectively; Cell Signaling) and were developed using the enhanced chemiluminescence (ECL) detection system (Pierce).

For the determination of the intracellular Ca²⁺ concentration, endothelial cells were loaded with 5 μM Ca²⁺-sensitive dye Fluo-4 AM (F14201; Molecular Probes) or 5 μM Fura-2 AM (F1221; Molecular Probes) in Hanks balanced salt solution supplemented with 20 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES) for 30 minutes at 37 °C and were then washed with Hanks balanced salt solution 3 times at room temperature. Live-cell images were acquired with an IX81 microscope (Olympus) at a frequency of 1 Hz. Fluo-4 fluorescence was measured by using excitation at 488 nm and emission collected at 500 to 550 nm. Fura-2 was monitored by digital fluorescence 340/380-nm ratio imaging. In experiments in which polymorphonuclear leukocytes (PMNs) were added to endothelial cells, only endothelial cells to which PMNs were attached were used for the analysis.

A total of 1.5 × 10⁴ human umbilical venous endothelial cells (HUVECs) were seeded per well of a collagen-coated transwell plate (3-μm pore size; Corning) and were cultured with daily medium changes until reaching confluency. For knockdown experiments, 8000 cells were transfected using Lipofectamine RNAiMAX with the indicated small interfering RNAs (siRNAs). For permeability assay, the medium of the upper insert was removed and replaced with medium containing 250 µg/mL fluorescein isothiocyanate (FITC)-conjugated dextran (relative molecular mass, 40 kDa; Molecular Probes). The permeability was determined by passage of FITC-dextran through the endothelial monolayer into the lower chamber using FlexStation-3 (Molecular Devices).

Whole cell patch-clamp recordings were performed at room temperature using an EPC10/2 amplifier with Pulse software (HEKA Electronik GmbH, Lambrecht, Germany). Pipette resistance was between 3 and 4 MΩ, and membrane potential was clamped at −80 mV. Normal external solution contained 140 mM NaCl, 5 mM KCl, 1.8 mM CaCl₂, 1mM MgCl_2, 10 mM HEPES, and 10 mM glucose, pH 7.4. The intracellular pipette solution contained 95 mM l-aspartate, 40 mM CsCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, and 0.1 mM guanosine triphosphate, pH 7.2. Local low flow at indicated shear rate was generated with a multichannel microperfusion system (LEAD-2, LONGER). Mechanical stimulation was applied to cells using a fire-polished glass pipette with a 3- to 5-μm tip diameter. Pipette movement was controlled by PCS-5000 Patch-Clamp micromanipulator (Burlegh). The micromanipulator uses the solid-state stability of piezoelectric technology to maintain smooth and predictable pipette motion without drift. Whole cell current was analyzed with IGOR Pro software (WaveMetrics). Experiments were performed with dissociated single cells. In experiments in which PMNs were added to endothelial cells, only endothelial cells to which PMNs were attached were used for the analysis.

We routinely applied 2 × 10⁵ PMNs per milliliter to cells. For antibody-mediated clustering of ICAM-1, sheep anti-mouse IgG-coupled dynabeads (M280; Invitrogen) were coated with mouse anti-human ICAM-1 antibody (BBIG-I1; R&D Systems) or immunoglobulin IgG1 (IgG1) control (MAB002; R&D Systems) overnight at 4°C according to the manufacturer’s protocol. To induce clustering, 1.5 × 10⁶ antibody-coated beads/mL was added to a tumor necrosis factor α (TNFα)-stimulated HUVEC monolayer cultured in a 6-well dish and incubated for 15 minutes. For some experiments, antibody-coated beads (5 × 10⁵/mL) were injected into the ibidi perfusion system containing HUVECs to induce ICAM-1 clustering on HUVECs under physiologic flow conditions. Alternatively, ICAM-1 was ligated with 15 μg/mL mouse monoclonal antibodies (R&D Systems, catalog no. BBIG-I1) for 30 minutes, followed by washing and ICAM-1 cross-linking with 50 μg/mL mouse secondary antibody (R&D Systems, catalog no. AF007) for 20 minutes at 37°C. Live cell imaging of membrane tension and intracellular Ca²⁺ during ICAM-1 clustering were performed using a Leica SP8 or an Olympus IX81 microscope (see above). For immunoblot analysis, beads were isolated using a magnetic holder (Miltenyi Biotec), and cells were lysed with RIPA buffer as described above.

EC-Piezo1-knockout (KO) and control mice were injected intraperitoneally with 200 µL phosphate-buffered saline (PBS) alone or containing 1 mg/kg body weight lipopolysaccharide  (LPS) (L4391; Sigma). After 4 hours, animals were euthanized, and lungs were collected for immunostaining with anti-MPR14 and anti–platelet endothelial cell adhesion molecule 1 (PECAM-1) antibodies. For quantification of extravasating PMNs, cryosections were analyzed in XYZ views on a Leica SP8 confocal microscope by the following criteria: PMNs directly surrounded by PECAM-1 staining (ie, blood vessel) were scored as intravascular, whereas cells outside blood vessels were scored as extravascular.

Wild-type or EC-Piezo1-KO mice were injected intraperitoneally with 100 µL PBS containing no or 500 ng TNFα prewarmed to 37°C. After 60 minutes, animals were euthanized, and cells in the peritoneal cavity were collected by flushing with 5 mL ice-cold PBS. Peritoneal cells were filtered using a 70-μm strainer and analyzed by flow cytometry (BD FACS Canto II). The following antibodies were used: FITC conjugated anti-mouse CD11b (101205; BioLegend) and allophycocyanin (APC)-conjugated anti-mouse Ly6G (127614; Biolegend).

Yoda1 (5586) was from Tocris Bioscience. Cytochalasin D (C8273), Blebbistatin (B0560), and PF431396 (PZ0185) were from Sigma. PP2 (529576) was from Merck Chemicals. Anti–PIEZO1-antibody was from Proteintech (15939-1-AP). Anti-GAPDH (catalog no. 5174), anti-protein tyrosine kinase 2 (PYK2) (catalog no. 3292), anti–p-PYK2 (Tyr402; 3291), antisarcoma (SRC) (2109), anti–p-SRC (Tyr416; 6943), anti-MLC (3672), and anti–p-MLC (Ser19; 3675) were from Cell Signaling Technology. Anti–endomucin antibody was from Santa Cruz (sc-65495). Gd³⁺ (439770) was from Sigma, and GsMTx4 (4912) was from Tocris.

Trial experiments or experiments done previously were used to determine sample size with adequate statistical power. Samples were excluded in cases where RNA/cDNA quality or tissue quality after processing was poor (below commonly accepted standards). Data are presented as means ± standard error of the mean (SEM). Comparisons between 2 groups were performed with the unpaired, 2-tailed Student t test, and multiple group comparisons at different time points were performed by 1- or 2-way analysis of variance (ANOVA). A value of P ≤ .05 was considered statistically significant.

In a screen to identify endothelial transmembrane proteins involved in the transendothelial migration of leukocytes, we identified the mechanosensitive cation channel PIEZO1 (Figure 1A; supplemental Table 1, available on the Blood Web site). The siRNA-mediated knock down of PIEZO1 in HUVECs or in the mouse brain endothelial cell line bEnd.3 strongly reduced endothelial transmigration of PMNs and peripheral blood mononuclear cells (Figure 1B-D). Similarly, PMN transmigration through mouse lung endothelial cells (MLECs) from mice with endothelium-specific loss of Piezo1 (EC-Piezo1-KO)²⁸ was strongly reduced compared with wild-type MLECs (Figure 1E; supplemental Figure 1A-B). Basal and TNFα-induced expression of endothelial adhesion molecules was not affected by loss of PIEZO1 (supplemental Figure 1C). Both PMN transmigration of human and murine endothelial cells could be stimulated by Yoda1, an activator of PIEZO1, and this effect was not seen after knockdown of PIEZO1 in endothelial cells (Figure 1D-G). PMN rolling on and adhesion to endothelial cells was not affected by loss of endothelial Piezo1 expression (Figure 1B-C), and endothelial barrier function analyzed by measuring the electrical impedance of the endothelial cell layer in vitro or by determining the permeability of the endothelial layer for FITC-labeled dextran in vivo was normal after loss of PIEZO1 (Figure 1H-I).

To study leukocyte extravasation in vivo, we injected TNFα into the peritoneal cavity and determined the number of CD11b⁺/Ly6G⁺ myeloid cells in the peritoneal cavity 6 hours later. Whereas TNFα induced a significant influx of cells into the peritoneal cavity of wild-type mice compared with untreated controls, the effect of TNFα was strongly reduced in EC-Piezo1-KO mice (Figure 2A). We then studied the role of endothelial PIEZO1 in a model of acute dermatitis of the ear by applying croton oil to the ear surface and found that the total number of neutrophils seen in sections of ears from wild-type and EC-Piezo1-KO mice was identical (supplemental Figure 1D). However, when analyzing postcapillary venules characterized by a diameter of 20 to 30 µm, the primary site of leukocyte extravasation, we found that most neutrophils had completed extravasation and were found in the perivascular space in wild-type mice, whereas about 25% to 30% of the leukocytes were found in the lumen of vessels (Figure 2B-D). However, in EC-Piezo1-KO mice, a significantly reduced portion of leukocytes had completed extravasation, and the majority, about 70% of cells, showed arrest at the luminal surface of the endothelium (Figure 2B,D), suggesting that they adhered to the endothelium but were not able to initiate the process of endothelial transmigration. Similarly, LPS-induced extravasation of neutrophils into the lung parenchyma was strongly reduced in EC-Piezo1-KO mice (Figure 2E-F). Staining and intravital microscopy of the cremaster of EC-Piezo1-KO mice revealed a reduced extravasation of neutrophils compared with controls after intrascrotal injection of interleukin 1β (IL-1β; Figure 2G-I). Hemodynamic parameters were similar in both mouse types, and there was no significant different in leukocyte rolling and adhesion within venules (supplemental Figure E-I). Also, basal extravasation of Evans blue and extravasation after subcutaneous injection of histamine or vascular endothelial growth factor (VEGF) were indistinguishable between wild-type and EC-Piezo1-KO mice (Figure 2J), indicating that vascular permeability was unchanged. Expression of genes encoding proteins involved in endothelial functions was not changed in endothelial cells from EC-Piezo1-KO mice (supplemental Figure 1J-K).

Because increases in [Ca²⁺]_i are involved in the initiation of leukocyte transendothelial migration and because leukocyte diapedesis occurs in the presence of low flow in vivo, we studied leukocyte-induced increases in endothelial cytosolic Ca²⁺ in the absence and presence of flow at a low shear rate (1.2 dynes/cm²). In control HUVECs loaded with the [Ca²⁺]_i indicators Fluo-4 or Fura-2, low flow alone or addition of human neutrophils alone had only a small effect on the cytosolic [Ca²⁺] (Figure 3A; supplemental Figure 2A). Similar but less pronounced effects were seen with the even lower shear rates of 0.4 and 0.8 dynes/cm² (supplemental Figure 2B). However, when neutrophils were given together with flow, we observed a strong increase in the endothelial cytosolic Ca²⁺ concentration (Figure 3A; supplemental Figure 2A). This synergistic effect was rarely seen during rolling or initial arrest of leukocytes but during crawling and during the transmigration phase (supplemental Figure 2C), and it depended on PIEZO1 (Figure 3B; supplemental Figure 2D). Chelation of intra- and extracellular Ca²⁺ by 1,2-bis(o-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid (BAPTA) and EGTA, respectively, blocked Ca²⁺ transients (supplemental Figure 2E-F). Whole cell patch-clamp recordings of MLECs showed characteristic inward currents when cells were exposed simultaneously to low flow and PMNs, which were not seen in MLECs from EC-Piezo1-KO animals (Figure 3C-D; supplemental Figure 2G-H). In contrast, exposure of MLECs to flow alone or PMNs alone induced only small currents (Figure 3C-D).

We then tested the potential involvement of PIEZO1 in the induction of downstream signaling events mediating leukocyte-induced opening of endothelial junctions. Whereas low flow alone had hardly any effect on the phosphorylation of PYK2, SRC, and the myosin light chain (MLC) in endothelial cells, addition of PMNs had as small but significant effect. However, application of both flow and PMNs synergistically induced endothelial PYK2, SRC, and MLC phosphorylation to a degree significantly higher than each of the 2 stimuli alone, and this effect was strongly reduced after knockdown of PIEZO1 (Figure 3E-F). The effect of PMNs and flow was mimicked by application of Yoda1, and this effect was again blocked after knockdown of PIEZO1 (Figure 3G-H). Inhibition of endothelial PYK2 or SRC by PF431396 or PP2, respectively, reduced basal transmigration and blocked Yoda1-induced increases in PMN transmigration (Figure 3I). These data strongly indicate that PMNs and low flow synergistically induce downstream signaling events through endothelial PIEZO1, resulting in the opening of endothelial junctions and leukocyte transmigration. Consistent with this, we also observed synergism in the ability of flow and PMNs to induce internalization of vascular endothelial (VE)-cadherin, an effect strongly inhibited after siRNA-mediated suppression of Piezo1 expression (Figure 3J).

Because engagement of endothelial ICAM-1 by leukocyte β₂ integrins is essential for induction of increases in [Ca²⁺]_i and diapedesis,^13,15,29,30 we suppressed expression of endothelial ICAM-1 and found that this strongly inhibited PMN-induced Ca²⁺ transients and PYK2, SRC, and MLC phosphorylation (Figure 4A-B; supplemental Figure 3A-B). Clustering of ICAM-1 using beads coated with anti–ICAM-1 antibodies mimicked the effect of PMNs and induced Ca²⁺ transients and phosphorylation of PYK2, SRC, and MLC synergistically with low flow (Figure 4C-D; supplemental Figure 3C), whereas beads coated with a control IgG had no effect (supplemental Figure 3D-E). The effects of ICAM-1 clustering were inhibited after knockdown of PIEZO1 and ICAM-1 (Figure 4C-D). Similar results were obtained when ICAM-1 clustering was induced by cross-linking of bound anti–ICAM-1 antibodies (supplemental Figure 3F-I). When beads coated with anti–ICAM-1 antibodies were given together with low flow, an inward current was induced, which was sensitive to Gd³⁺ and the PIEZO1 inhibitor GsMTx4, whereas beads and flow alone had hardly any effect (Figure 4E-F). This strongly indicates that clustering and activation of ICAM-1 by leukocytes in the presence of low flow results in PIEZO1-mediated downstream signaling leading to the opening of endothelial junctions.

ICAM-1 clustering and adhesion of leukocytes to endothelial cells have been shown to induce stiffening of the endothelial surface and to induce traction stress.^31-35 We therefore determined membrane tension in response to ICAM-1 clustering and PMNs using the fluorescent lipid tension sensor FliptR³⁶ and the membrane stress sensor (MSS) biosensor³⁵ (supplemental Figure 4A-B). We found that clustering of ICAM-1 or addition of PMNs leads to a small increase in endothelial membrane tension (Figure 5A-D). Low flow, which by itself had no significant effect on endothelial membrane tension, when given together with ICAM-1 clustering agents or PMNs, resulted in a very strong increase in plasma membrane tension (Figure 5A-D). This indicates that low flow and ICAM-1 clustering synergistically increase endothelial membrane tension.

Because ICAM-1 clustering has been shown to induce localized actin polymerization, MLC phosphorylation, and actomyosin contractility, which promote junctional opening,^18,31 we analyzed the effect of cytochalasin D and blebbistatin on membrane tension and on phosphorylation of PYK2, SRC, and MLC induced by ICAM-1 clustering. Both agents blocked ICAM-1–dependent changes in membrane tension and downstream signaling (Figure 6A-D). We then tested whether increased membrane tension and downstream signaling induced by ICAM-1 clustering involves the actin adapter proteins α-actinin-4 and cortactin, which have been shown to be recruited after clustering of ICAM-1 and be required for ICAM-1–mediated actin filament branching and for ICAM-1–dependent transendothelial migration of neutrophils.^37-39 As shown in Figure 6E-G and supplemental Figure 5A-B, siRNA-mediated knockdown of the RNAs encoding α-actinin-4 and cortactin blocked the effect of ICAM-1 clustering on membrane tension and downstream signaling. These data suggest that actin polymerization and actomyosin contractility of the cortical cytoskeleton induced by ICAM-1 clustering and leading to increased cortical tension^40,41 directly affect plasma membrane tension⁴¹ and thereby induce PIEZO1 activation.

We report that the mechanosensitive cation channel PIEZO1 plays a critical role in transendothelial migration of leukocytes in vitro and in vivo by integrating low levels of fluid shear stress and mechanical signals induced by clustering of ICAM-1 and thereby mediating increases in [Ca²⁺]_i and a localized opening of the endothelial barrier (Figure 6H). It has been reported that TRPC6 is critically involved in leukocyte-induced increases in endothelial [Ca²⁺]_i during leukocyte transendothelial migration.¹⁹ In expression analyses, we were not able to observe expression of TRPC6 in endothelial cells (supplemental Figure 1J). Nevertheless, it could well be that both PIEZO1 and TRPC6 operate in parallel under in vivo conditions or that PIEZO1 is involved in the initiation of leukocyte extravasation by sensing both flow and leukocyte adhesion, whereas TRPC6 mediates increases in [Ca²⁺]_i mainly at later stages of diapedesis.

ICAM-1 is a central endothelial adhesion receptor that functions as a ligand for β2 integrins on leukocytes and promotes leukocyte spreading, migration, and transmigration.^29,42 Engagement of ICAM-1 leads to the clustering of ICAM-1 molecules and its redistribution into ring-like structures around adherent leukocytes, which is a requirement for efficient downstream signaling.^43,44 ICAM-1 clustering induces cytoskeletal changes such as actin polymerization, MLC phosphorylation, and actomyosin contractility, which promote junctional opening.^18,31,42,45 ICAM-1 also promotes increase in [Ca²⁺]_i levels,^13,46 which has been shown to lead to activation of SRC via protein kinase C.¹⁵ ICAM-1–mediated activation of SRC and PYK2 has been shown to be required for VE-cadherin–dependent leukocyte transendothelial migration.³⁰ This involves direct phosphorylation of VE-cadherin^30,47,48 and indirect regulation of VE-cadherin through VE-PTP⁴⁹ or by phosphorylation of β-catenin.⁵⁰ How ICAM-1 clustering induces activation of these downstream signaling events resulting in junctional opening and transendothelial migration is poorly understood. Our data indicate that downstream signaling through ICAM-1 requires coactivation of PIEZO1 by fluid shear stress and ICAM-1–induced reorganization of the cortical cytoskeleton.

Various mechanical stimuli acting on cellular membranes have been shown to be able to activate PIEZO1. These include exposure to fluid shear stress, mechanical indentation of the cell surface, cell migration, compression of the cell membrane, or forces generated at the cell–cell or cell–matrix interface.^21,51 Our data show that low-level fluid shear stress and interaction of leukocytes with the endothelial surface act in a synergistic manner to activate endothelial PIEZO1 and to initiate leukocyte transendothelial migration. In postcapillary venules, the place where leukocyte extravasation mainly takes place, the shear stress exerted by the flowing blood is relatively low at about 1 to 2 dynes/cm,^2,52,53 a shear rate hardly able to induce PIEZO1-mediated signaling.^27,28 Consistent with this, we saw only very small increases in [Ca²⁺]_i and no significant increase in the phosphorylation of PYK2, SRC, or MLC in response to fluid shear stress of 1.2 dynes/cm². Similarly, when ICAM-1 clustering was induced in TNFα-pretreated endothelial cells by PMNs or anti–ICAM-1 antibodies, only small increases in [Ca²⁺]_i and phosphorylation of PYK2, SRC, and MLC could be observed, which were further reduced after suppression of Piezo1 expression. However, when endothelial ICAM-1 clustering was induced while exposing cells to low flow, downstream signaling was strongly activated, and this effect was inhibited by knockdown of PIEZO1. This raised the question as to how ICAM-1 clustering promotes PIEZO1 activation. Both ICAM-1 clustering and adhesion of leukocytes to endothelial cells have been shown to induce stiffening of the endothelial surface and to induce traction stress.^31-34,54 These endothelial responses are caused by increased actin polymerization and actomyosin contractility of the cortical cytoskeleton, which lead to increased cortical tension^40,41 and require recruitment of the actin adapter proteins α-actinin-4 and cortactin.^31,37 Because the plasma membrane and the underlying cortical cytoskeleton are closely interconnected,^40,41 changes in the actomyosin cortical tension directly affect plasma membrane tension⁴¹ and therefore are likely to regulate PIEZO1 activity. Consistent with this, we found that inhibition of actin polymerization and myosin activity and siRNA-mediated knockdown of α-actinin-4 and cortactin blocked ICAM-1–mediated increases in membrane tension and PIEZO1-dependent downstream signaling required for leukocyte transendothelial migration. We therefore also think that it is the combined effect of low flow and leukocyte-induced ICAM-1 clustering on plasma membrane tension that induces PIEZO1 activation. This is consistent with a series of biophysical studies showing that changes in plasma membrane tension are sufficient to regulate the open probability of PIEZO1 and that no intra- or extracellular interactions are required.^23-25 However, we cannot exclude that additional mechanisms are involved. For instance, it has recently been suggested that Piezo1 can be linked to the actin cytoskeleton through members of the cadherin family including VE-cadherin,⁵⁵ and this mechanism may contribute to Piezo1 activation induced by ICAM-1 clustering and subsequent actin polymerization.

Recent data indicate that changes in plasma membrane tension are restricted to subcellular domains of endothelial cells as local increases in membrane tension lead only to local activation of mechanosensitive ion channels such as PIEZO1.⁵⁶ The finding that leukocyte-induced endothelial downstream signaling and diapedesis require PIEZO1 and flow is consistent with earlier observations, which showed that fluid shear stress promotes transendothelial leukocyte migration.^57-60 Our data identify a novel synergism of local hemodynamic forces and initial endothelial leukocyte adhesion to induce plasma membrane tension and endothelial signaling events that promote leukocyte extravasation. The discovery of a novel mechanosensing and mechanosignaling process required for the initial phase of leukocyte diapedesis may also lead to new anti-inflammatory therapeutic approaches.

The authors thank Svea Hümmer for secretarial help; Yin Hao (Instrument Analysis Center of Xi’an Jiaotong University) for assistance with fluorescence-lifetime imaging microscopy; Shuya Liu, Martina Finkbeiner, Ulrike Krüger, and Claudia Ullmann for technical help; and Bo Liu (Dalian University of Technology, Dalian, China) for donating a plasmid encoding the MSS sensor.

This work was supported by the National Key R&D Program of China (2021YFA1301200 [Z.Y.]), the National Natural Science Foundation of China (grants 81870220 and 82122008 [S.W.] and grants 92049203 and 81941005 [Z.Y.]), the Shaanxi Natural Science Fund for Distinguished Young Scholars of China (S2020-JC-JQ-0239 [S.W.]), the Collaborative Research Centre 834 of the German Research Foundation (S.O.), and the Collaborative Research Center 1348 of the German Research Foundation (D.V.).

Contribution: S.W. initiated and designed the study, performed experiments, analyzed data, and wrote the manuscript; Y.S., B.W., and C.W. performed in vitro experiments; T.M. and R.I.S. performed in vivo experiments; B.S. helped with in vitro and in vivo experiments; T.L. and Z.Y. helped with in vitro experiments; N.W. supervised part of the study and discussed data; D.V. supervised part of the in vivo experiments and analyzed and discussed data; S.O. designed and supervised the study, discussed data, and wrote the manuscript; and all authors commented on the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: ShengPeng Wang, Department of Cardiology, First Affiliated Hospital, Cardiovascular Research Center, Xi'an Jiaotong University, No. 277 W Yanta Rd, Yanta District 710061, Xi’an, China; e-mail: shengpeng.wang@xjtu.edu.cn; and Stefan Offermanns, Max Planck Institute for Heart and Lung Research, Department of Pharmacology, Ludwigstr 43, 61231 Bad Nauheim, Germany; e-mail: stefan.offermanns@mpi-bn.mpg.de.