LXN deficiency regulates cytoskeleton remodelling by promoting proteolytic cleavage of Filamin A in vascular endothelial cells

Abstract Endothelial cells (ECs) respond to blood shear stress by changing their morphology is important for maintaining vascular homeostasis. Studies have documented a relationship between endothelial cell shape and the stress flow, and however, the mechanism underlying this cytoskeletal rearrangement due to shear stress remains uncertain. In this paper, we demonstrate that laminar shear stress (LSS) significantly reduces latexin (LXN) expression in ECs. By using siRNA and cell imaging, we demonstrated that LXN knockdown results in the morphologic change and F‐actin remodelling just like what LSS does in ECs. We further demonstrate that LXN interacts with Filamin A (FLNA) and regulates FLNA proteolytic cleavage and nuclei translocation. By constructing LXN‐/‐ mice and ApoE‐/‐LXN‐/‐ double knockout mice, we evaluated the effect of LXN knockout on aortic endothelium damage in mice. We found that LXN deficiency significantly improves vascular permeability, vasodilation and atherosclerosis in mice. Our findings provide confident evidence, for the first time, that LXN is a novel regulator for morphological maintenance of ECs, and LXN deficiency has a protective effect on vascular homeostasis. This provides new strategies and drug targets for the treatment of vascular diseases.

shear stress in vivo. When ECs sense shear stress, they change the cell shape, functions and gene expressions and involved in both physiologic and pathophysiologic vascular biology. 5,6 In this regard, two types of shear stress were reported in large vessel: one is laminar shear stress, and the other is disturbed shear stress; laminar shear stress (LSS) is vascular protective and founded in straight arterial regions (such as thoracic aorta). However, disturbed shear stress (DSS) is vascular harmful and founded in branched or curved regions (such as aortic arch). Interestingly, studies have shown that vascular permeability increases where the branches of blood vessels are located, along with an increased risk of atherosclerosis. [7][8][9] Mechanotransduction by shear stress is complex. It can be roughly determined that shear stress activates various conduction pathways through membrane molecules, including calcium channel, G protein, tyrosine kinase receptor, adhesion protein and cytoskeleton. [10][11][12][13] Under shear stress, the most intuitive phenotype of ECs is morphological change: ECs undergo a transition from polygonal cobblestone-like uniform sheets to uniform spindle-like monolayer.
This phenomenon is thought to be partly the result of F-actin stress fibre induction, 14,15 and however, the mechanism underlying this cytoskeletal rearrangement due to shear stress is ill-defined.

Latexin (LXN), a carboxypeptidase inhibitor with 222 amino acids
in length, was first identified in the lateral neocortex of rats and acts as a marker of neurons in the lateral neocortex of developing rat brain. 16,17 Due to the characteristics of its protease inhibitor, it is thought that LXN might be involved in protein degradation and metabolism. 18 LXN is also implicated in inflammation because it is expressed in macrophage and mast cells and can be induced by lipopolysaccharide. 19 However, the function of LXN in ECs is unknown.
In the present study, we found that laminar shear stress induces the down-regulation of LXN in ECs, which results in the morphologic change and F-actin remodelling just like what LSS does in ECs.
Logically, we hypothesize LXN is a novel regulator that involves in the process of endothelial cell morphologic change and LXN deficiency has a protective effect on vascular homeostasis.

| Cells and reagents
HUVECs were purchased from ATCC and cultured in EBM-2 medium

| Shear stress
Laminar shear stress was applied to cells by using parallel plate flow chambers (GlycoTech) set in series in a closed circulating system with 5% CO 2 at 37℃. The parallel plate flow chamber comprised a polycarbonate plate attached to rectangular gaskets (Thickness: 0.010 in; Flow path width: 1.0 cm). HUVECs were cultured on a fibronectin coated slice. When cells grow to 90% confluence, attached the slice to flow chamber. The shear stress flow was initiated with EBM-2 medium. Shear stress (15 dyne/cm 2 ) was applied for indicated times. Cell morphology was demonstrated by Phase-contrast photomicrograph.

| Immunostaining
For en face staining, the whole aorta was cut open longitudinally, permeabilized with 0.1% Triton X-100 in PBS for overnight and blocked with 10% normal goat serum in Tris-buffered saline containing 2.5% Tween-20 for 24 hours at 4℃. Next, aortas were incubated with primary antibody in blocking buffer 48 hours at 4℃. After rinsing with washing solution (Tris-buffered saline con-  Carbamidomethylation of cysteines and oxidation of methionines were allowed during the search of peptides. The maximum number of missed cleavages was set to 1, with trypsin as the protease.

| Subcellular fractionation
Subcellular fractions were prepared by differential centrifugation of cell homogenates according to the protocol of cytosol and nuclear protein extraction kit (Beyotime Biotechnology).

| Calpain activity assay
Cells were grown to confluence and lysed with radioimmunoprecipitation assay lysis buffer. The lysate was cleared of debris and used for the Calpain-Glo protease assay (Promega) according to the manufacturer's protocol.

| Statistical analyses
Data are expressed as means ± SEM. The statistical significance of differences was assessed by Student's t tests; a value of P < .05 was considered statistically significant.

| Laminar shear stress down-regulated LXN in ECs
HUVECs were subjected to laminar shear stress of 15 dynes/cm 2 for 12 hours ( Figure 1). As expected, we observed that laminar shear stress induces cell shape change ( Figure 1A), up-regulates vasoprotective genes (such as CPY1B1, THBD and NOS3) and, however, down-regulates pro-atherosclerosis genes (such as VCAM1, CD36, and ANGPT2) ( Figure 1B). Interestingly, we found that laminar shear stress significantly inhibited the expression of

| LXN knockdown mimics LSS-induced ECs shape change and cytoskeleton remodelling
To explore the functional role of LXN in endothelial cell, function-  Figure 2F-ii, G-ii). Interestingly, we found that ECs display large and frequent lamellipodia formation in control conditions ( Figure 2G-i), and however, the lamellipodia formation were decreased in LXN knockdown cells (Figure 2G-ii). Taken together, these data clearly demonstrate that LXN, at least, involves in endothelial cell shape change and actin cytoskeleton re-organization in ECs.

| LXN interacts with FLNA in ECs
To explore the mechanism how LXN regulates endothelial morphology, we have undertaken a proteomic screen to identify intracellular targets of LXN in ECs. To this end, HUVECs were lysed, and immunoprecipitation was performed with anti-LXN antibody. LXN complex was resolved by SDS-PAGE, and protein bands were cut into gel slices for trypsin digestion followed by liquid chromatography tandem mass spectrometry. FLNA, a scaffolding protein, was identified as a potential partner for LXN ( Figure 3A). We further validated this interaction by immunoprecipitation and Western blot. As shown in Figure 3B and FLNA, co-immunoprecipitation and immunostaining assays were conducted. We validated that endogenous LXN exactly formed a physical complex with FLNA in HUVECs ( Figure 3C). The interaction was further confirmed by immunostaining ( Figure 3D). Together, our data strongly suggest that FLNA might be involved in binding to LXN and form a complex in ECs.

| LXN knockdown regulates proteolytic cleavage of FLNA and promotes FLNA nucleus translocation in ECs
FilaminA, a well know F-actin crosslink protein, which regulates signalling events involved in cell shape change and motility. 21 Since LXN regulates cell shape, changing and actin cytoskeleton remodelling in ECs has been observed in our study (Figures 1 and 2). The interaction of LXN with FLNA promoted us to evaluate whether LXN effects FLNA protein levels in ECs. To test this hypothesis, LXN function-loss was performed in ECs. As shown in Figure 4, we found that LXN knockdown did not affect the mRNA level of FLNA ( Figure 4A), however, significantly decreased the protein level of FLNA (280 kD) in ECs ( Figure 4B). Interestingly, we found a new band about 190 kD, which can also be recognized by FLNA antibodies, suggesting FLNA may be degraded ( Figure 4B). FLNA is highly susceptible to proteolysis by calpain, and calpain-induced cleavage products of 190 kD for FLNA have been reported. 22 Therefore, we speculate whether LXN regulates calpain activity in endothelial cells. Using the luminescence Calpain-Glo assay, we observed that calpain activity was increased in LXN knockdown ECs compared with control cells ( Figure 4C). Therefore, we treated

| LXN regulates cytoskeleton anchoring and focal adhesion formation through FLNA in ECs
FilaminA has been reported to involve in cytoskeleton remodelling and cytoskeleton anchoring. 24,25 We further asked whether LXN knockdown affects the stability of cytoskeleton proteins. We first tested some cytoskeleton related proteins, such as integrinβ, paxillin, Arp2/3, α-actinin, FAK and vinculin in HUVECs treated with siCTL or siLXN. Western blot showed that the expression of integrinβ increased slightly after knocking down LXN, while the expression of paxillin, Arp2/3 and α-actinin was decreased significantly ( Figure 4H). Because paxillin has been reported to implicate in the formation of focal adhesion (FA) and cytoskeleton anchoring, we next investigated whether FLNA has effect on cytoskeleton anchoring in ECs. To this end, control or LXN knockdown cells were fixed, and paxillin and F-actin were immunofluorescently probed by using confocal microscopy. As shown in Figure 4I,F-actin filaments, which are randomly aligned in control cells, are linked to focal adhesions at their ends ( Figure 4I-i). However, in LXN knockdown cells, the association between actin filaments and focal adhesion was disrupted with the disappearance of focal adhesion such as structures and thus resulted in the distribution of F-actin filaments in longitudinal direction of the cell, as well as the morphologic change ( Figure 4I-ii). We next knockdown FLNA in ECs ( Figure 4J). We found that the focal adhesion such as plaques was impaired and resulted

| LXN deficiency modulates endothelial cell shape change and improves vascular function in mice
To evaluate the role of LXN in regulating endothelial cell morphology in vivo, we examined the cell shape change and distribution of cytoskeleton using enface staining of endothelial cell lining the aorta arch of WT and LXN -/mice. Confocal microscopy revealed that LXN exists in the endothelium layer in aortic ( Figure 5). The level of LXN in the endothelium of the aortic arch was significantly higher than that in the thoracic aorta in WT mice ( Figure 5A-ii,v-i, iv), CD31 staining showed that the cell displayed a polygonal shape ( Figure 5Ai,iv). However, in LXN -/mice, the endothelial cells were elongated and displayed fusiform shape ( Figure 5B-i, iv). FLNA was found in both the perinuclear and the cytoplasm of the endothelial cells in WT mice, and the colocalization of LXN and FLNA was observed ( Figure 5C). However, FLNA was observed to transfer to the nucleus of the endothelial cells in LXN -/mice. This is consistent with the results observed in the cells in vitro ( Figure 4E). Together, these results showed that LXN modulates endothelial cell shape change and actin cytoskeleton re-organization in vivo.
Next, we evaluated the effect of LXN deficiency on vascular function in vivo. We first performed the miles assay in WT and LXN -/mice by Evans blue injecting. Vascular permeability within the aorta was assessed by measuring leakage of Evans blue dye into the vascular wall. We found that VEGF ( Figure 5D) and TNFα ( Figure 5E

| D ISCUSS I ON
In the present study, we discovery a novel function of LXN in ECs. We demonstrate an interesting phenomenon that laminar Vascular ECs, which line the inner surface of blood vessels, are important in maintaining structural and functional roles of blood vessels. 26 It is well known that ECs are continuously exposed to shear stress in vivo, and the ECs respond to shear stress by changing their morphology and function, which are important for maintaining cellular homeostasis in vascular system. [26][27][28][29] In this regard, two types of shear stress were reported in large vessel: one is LSS, which is vascular protective and founded in straight arterial regions; and the other is DSS, which is vascular harmful and founded in branched or curved regions. 29-31 LSS induces vascular ECs morphologic change has been reported, and these morphologic changes are accompanied by cytoskeleton remodelling, as well as actin filaments becoming rearranged into bundles of stress fibres and aligned in the direction of the shear stress. 5,32 We found that LXN was highly expressed in ECs and down-regulated by LSS ( Figure 1). Interestingly, we observed that LXN knockdown resulted in the morphologic change and F-actin filaments remodelling just like what LSS does in ECs (Figure 2). In this regard, our findings provided compelling evidence, demonstrating that LXN might be a novel molecular in response to shear stress of ECs.
To reveal the mechanism underline LXN regulating endothelial cell morphology, we have undertaken a proteomic screen to identify intracellular targets of LXN in ECs, and FLNA was identified as partner of LXN in ECs (Figure 3). FLNA is a 280 kDa dime with two N-terminal actin-binding domains. 33 Dimerization of FLNA forms V-shaped molecules that crosslink actin filaments into orthogonal networks and maintains endothelial cell morphology. 34 A large number of literatures have reported that FLNA plays a key role in the cytoskeleton and determines cell shape and locomotion. 21,35 Indeed, FLNs can be regulated through cleavages by calpains. 22 It has been reported that FLNA can be cleaved by calpain proteases, generating 190-and 90-kDa fragments. 36 Importantly, FLNA fragments can be transferred to the nucleus to regulate gene expression. 21,23 We further showed that LXN knockdown leads to activation of calpain ( Figure 4C), and results in FLNA degradation and subcellular translocation ( Figure 4D-F). It was important that siLXN-induced FLNA degradation and stress fibre formation could be inhibited by calpain specific inhibitors ( Figure 4D and G). These results suggest that LXN can maintain the stability of FLNA in the cytoplasm in ECs, thus maintaining normal ECs morphology. In addition, filamin has been shown to interact with integrin β cytoplasmic domain constructs and depletion of filamin results in integrin activation. 37 We found that LXN knockdown up-regulates the protein level of integrin β1, and however, destroys protein level of cytoskeleton proteins as paxillin, Arp2/3 and α-actinin ( Figure 4H). Arp2/3 is very important for maintenance of cell morphology. 24,25 Activation of the Arp 2/3 complex could promote the formation of branched actin networks, especially, F I G U R E 4 LXN knockdown modulates FLNA proteolytic cleavage and subcellular localization in ECs. A and B, HUVECs cultured in plates were treated with siCTL or siLXN for 72 h. The expression of FLNA was determined by qPCR (A) and Western blot (B). C, Control (siCTL) or LXN siRNA (siLXN) HUVECs were grown to confluence, and total lysates were used for the luminescent Calpain-Glo protease assay. The relative luminescence units (RLU) were averaged and normalized to the amounts of proteins in cell lysates. D, HUVECs were transfected with siCTL or siLXN for 24 h, and then, cells were untreated or treated with calpeptin (10, 40 μg/mL) for 48 h. Cells were lysed and analysed by Western blot with antibodies as indicated. E, HUVECs cultured in plates were transfected with siCTL or siLXN for 72 h. Cells were washed, fixed, and immunofluorescence stained with anti-LXN and anti-FLNA antibody. Scale bars = 20 μm. F, HUVECs cultured in plates were treated with siCTL or siLXN for 72 h. The cytoplasmic (C) and nuclear (N) fractions of HUVECs were detected by Western blot with antibodies as indicated. G, HUVECs cultured in plates were pro-treated calpeptin for 2 h and treated with siLXN for 72 h. HUVECs were stained with Phalloidin. Scale bars = 20 μm. H, Western blot shows the level of proteins as indicated in siCTL-and siLXN-treated HUVECs. I, HUVECs cultured in plates were treated with siCTL or siLXN for 72 h. HUVECs were immunofluorescence stained with Phalloidin and anti-paxillin antibody. High magnification images of the areas denoted with a dashed box were showed. Scale bars = 20 μm. J, Immunofluorescence staining shows the F-actin associated with FLNA in HUVECs. siCTL or siFLNA-treated HUVECs were immunofluorescence stained with Phalloidin and anti-FLNA antibody. High magnification images of the areas denoted with a dashed box were showed. Scale bars = 20 μm | 6825 HE Et al.
the formation of transverse arcs in lamellipodia area by combining short myosin filaments and actin filaments. 38,39 Alpha-actinin is a cytoskeletal actin-binding protein that crosslinks actin filaments by forming an anti-parallel rod shaped dimer. 40 Functionally, α-actinin plays multiple important roles in the cell involved in linking the cytoskeleton to transmembrane proteins, regulating the activity of receptors and serving as a scaffold to connect the cytoskeleton to diverse signalling pathways. 40 We found that LXN loss could induce the parallel distribution of F-actin stress fibres in ECs (Figure 2), as well as down-regulation of α-actinin ( Figure 4H). Together, these data support our hypothesis that LXN is involved in ECs morphologic change via the regulation of cytoskeleton remodelling.
Finally, and most importantly, our results show that LXN deficiency has protective effects on vascular homeostasis. Disturbed shear stress, hypertension and diabetes can damage vascular endothelium, leading to the occurrence and development of atherosclerosis. [41][42][43] Laminar shear stress has been reported to improve vascular function, including reducing endothelial permeability, increasing vasodilation and the expression of cytoprotective genes. 44,45 However, laminar shear seems to be a necessary condition for endothelial cell integrity, and laminar shear stress is a 'survival' factor rather than a 'growth' factor of endothelial cells. 3 In animal models, we demonstrate that LXN deficiency inhibits VEGF and TNFα-induced vascular permeability ( Figure 5D,E), as well as significantly improves F I G U R E 5 LXN deficiency improves vascular function in mice. A and B, en face immunostaining shows the ECs morphology in aortic arch (i-iii) or thoracic aorta (iv-vi) in WT (A) and LXN -/-(B) mice. ECs were stained with CD31 and LXN antibody. Scale bars = 100 μm. C, en face immunostaining showed the localization of LXN and FLNA in ECs of thoracic aorta of WT and LXN -/mice. D and E, Vascular permeability induced by VEGF (D) and TNFα (E) in Miles assay with different doses in the skin of WT and LXN -/mice. Quantification of VEGF or TNFαinduced Evans blue (EB) incorporation in the skin (n = 6). The results shown are the mean ± SEM F I G U R E 6 LXN deficiency improves vasodilation capacity and atherosclerosis in HF-induced ApoE -/mice. A, Relaxation responses to the cumulative addition of Ach for aorta arteries contracted with 1µm NE for ApoE -/-LXN +/+ and ApoE -/-LXN -/mice feed with high-fat diet (n = 4). B, Representative images of Oil Red O staining of en face preparations of aortas and quantification of the atherosclerotic surface area of the entire aorta of ApoE -/-LXN +/+ and ApoE -/-LXN -/mice (n = 6). C, HE staining of aortic arch root of ApoE -/-LXN +/+ and ApoE -/-LXN -/mice (n = 6). The results shown are the mean ± SEM vasodilation and atherosclerosis in mice fed with high-fat diet ( Figure 6).
In summary, our results suggest that LXN deficiency has a protective effect on blood vessels. Our data support LXN as an important regulator of ECs morphology through forming a complex with cytoskeleton proteins. LXN directly interacts with FLNA in vascular ECs, and this interaction is functional, because LXN loss enhances FLNA proteolytic cleavage and subcellular localization; and it is likely physiological, as it suppresses actin filaments anchoring with focal adhesion and results in cell morphologic change which is similar with that of laminar shear stress does on ECs. Importantly, we show that LXN deficiency significantly improves vascular permeability, vasodilation and atherosclerosis in mice, which indicated LXN deficiency has a protective effect on vascular homeostasis, and provides new targets for the treatment of vascular diseases.

F I G U R E 7
Schematic presentation of the cell shape and stress fibre network in static and laminar shear stress or LXN deficiency ECs. In normal condition, transverse arcs are generated from Arp2/3, α-actinin and FLNA crosslinked actin filaments in normal ECs. Under laminar shear stress, LXN loss resulted in the proteolytic cleavage and subcellular localization of FLNA, down-regulation of Arp2/3, α-actinin and paxillin in ECs. As a result, ECs were elongated from the typical cobblestone pattern to uniformly fusiform and showed the distribution of F-actin filaments in longitudinal direction of the cell