Transendothelial cell macroaperture (TEM) tunnels control endothelium barrier function and are triggered by several toxins from pathogenic bacteria that provoke vascular leakage. Cellular dewetting theory predicted that a line tension of uncharacterized origin works at TEM boundaries to limit their widening. Here, by conducting high-resolution microscopy approaches we unveil the presence of an actomyosin cable encircling TEMs. We develop a theoretical cellular dewetting framework to interpret TEM physical parameters that are quantitatively determined by laser ablation experiments. This establishes the critical role of ezrin and non-muscle myosin II (NMII) in the progressive implementation of line tension. Mechanistically, fluorescence-recovery-after-photobleaching experiments point for the upstream role of ezrin in stabilizing actin filaments at the edges of TEMs, thereby favouring their crosslinking by NMIIa. Collectively, our findings ascribe to ezrin and NMIIa a critical function of enhancing line tension at the cell boundary surrounding the TEMs by promoting the formation of an actomyosin ring.

Actin filaments assemble into a variety of networks to provide force for diverse cellular processes [1]. Tropomyosins are coiled-coil dimers that form head-to-tail polymers along actin filaments and regulate interactions of other proteins, including actin-de polymerizing factor (ADF)/cofilins and myosins, with actin [2-5]. In mammals, >40 tropomyosin isoforms can be generated through alternative splicing from four tropomyosin genes. Different isoforms display non-redundant functions and partially non-overlapping localization patterns, for example within the stress fiber network [6, 7]. Based on cell biological studies, it was thus proposed that tropomyosin isoforms may specify the functional properties of different actin filament populations [2]. To test this hypothesis, we analyzed the properties of actin filaments decorated by stress-fiber-associated tropomyosins (Tpm1.6, Tpm1.7, Tpm2.1, Tpm3.1, Tpm3.2, and Tpm4.2). These proteins bound F-actin with high affinity and competed with a-actinin for actin filament binding. Importantly, total internal reflection fluorescence (TIRF) microscopy of fluorescently tagged proteins revealed that most tropomyosin isoforms cannot co-polymerize with each other on actin filaments. These isoforms also bind actin with different dynamics, which correlate with their effects on actin-binding proteins. The long isoforms Tpm1.6 and Tpm1.7 displayed stable interactions with actin filaments and protected filaments from ADF/cofilin-mediated disassembly, but did not activate non-muscle myosin Ila (NMIIa). In contrast, the short isoforms Tpm3.1, Tpm3.2, and Tpm4.2 displayed rapid dynamics on actin filaments and stimulated the ATPase activity of NMIla, but did not efficiently protect filaments from ADF/cofilin. Together, these data provide experimental evidence that tropomyosin isoforms segregate to different actin filaments and specify functional properties of distinct actin filament populations.