SALM4 regulates angiogenic functions in endothelial cells through VEGFR2 phosphorylation at Tyr1175

Angiogenesis depends on VEGF‐mediated signaling. However, the regulatory mechanisms and functions of individual VEGF receptor 2 (VEGFR2) phosphorylation sites remain unclear. Here, we report that synaptic adhesion‐like molecule 4 (SALM4) regulates a specific VEGFR2 phosphorylation site. SALM4 silencing in HUVECs and Salm4 knockout (KO) in lung endothelial cells (ECs) of Salm4−/− mice suppressed phosphorylation of VEGFR2 tyrosine (Y) 1175 (Y1173 in mice) and downstream signaling upon VEGF‐A stimulation. However, VEGFR2 phosphorylation at Y951 (Y949 in mice) and Y1214 (Y1212 in mice) remained unchanged. Knockdown and KO of SALM4 inhibited VEGF‐A‐induced angiogenic functions of ECs. SALM4 depletion reduced endothelial leakage, sprouting, and migratory activities. Furthermore, in an ischemia and reperfusion (I/R) model, brain injury was attenuated in Salm4−/− mice compared with wild‐type (WT) mice. In brain lysates after I/R, VEGFR2 phosphorylation at Y949, Y1173, and Y1212 were induced in WT brains, but only Y1173 phosphorylation of VEGFR2 was reduced in Salm4−/− brains. Taken together, our results demonstrate that SALM4 specifically regulates VEGFR2 phosphorylation at Y1175 (Y1173 in mice), thereby fine‐tuning VEGF signaling in ECs.—Kim, D. Y., Park, J. A., Kim, Y., Noh, M., Park, S., Lie, E., Kim, E., Kim, Y.‐M., Kwon, Y.‐G. SALM4 regulates angiogenic functions in endothelial cells through VEGFR2 phosphorylation at Tyrll75. FASEB J. 33, 9842–9857 (2019). www.fasebj.org

labyrinth formation (11). Synaptic adhesion-like molecules (SALMs) are novel axon guidance molecules containing LRRs that are involved in synapse development and functions, including synaptic transmission and plasticity (5). Five members of the SALM family have been identified. SALM1-5 have similar domain organization, with 6 LRRs, an Ig domain, and a fibronectin type III domain on the extracellular side, followed by a transmembrane domain and a cytoplasmic region that ends with a PDZ domainbinding motif (5). VEGF signaling depends on scaffolding proteins, such as synectin, that bind to PDZ domains (12). SALM4 and SALM5 do not contain PDZ-binding domains, in contrast to SALM1-3 (5). SALM4 regulates neurite branching through mechanisms that involve lipid raftassociated proteins (13). Furthermore, the hippocampal CA1 region of the Salm4 knockout (KO) mouse has an increased number of excitatory and inhibitory synapses (14). The role of SALM4 in ECs remains unknown but must be elucidated to understand guidance by tip cells in ECs.
Vascular sprouting and permeability are highly dependent on VEGFs and their receptors (VEGFRs), which regulate EC functions, such as proliferation, migration, and viability. VEGF-A binds VEGFR1 and VEGFR2 in ECs. Although the affinity of VEGF-A is higher for VEGFR1 than for VEGFR2, VEGFR2 has higher tyrosine kinase activity (15). Therefore, VEGFR2 is regarded as the most important receptor for VEGF-A effects in ECs. The major phosphorylation sites in VEGFR2 are tyrosine (Y) 951 in the kinase-insert domain and Y1175 and Y1214 in the C-terminal domain. The VEGFR2 signaling cascade includes Y951-SRC kinase, Y1175-ERK, Y1175-PI3K-AKT-eNOS, and Y1214-p38 MAPK (16). Regulation of VEGFR2 phosphorylation is critical for angiogenesis and vascular permeabilityrelated diseases. Nevertheless, the regulatory mechanisms of some VEGFR2 phosphorylation sites and pathways remain poorly understood.
In the present study, we determined that SALM4 is expressed in ECs and involved in angiogenic functions through VEGFR2 phosphorylation. In addition, we investigated fine-tuned potential regulators of VEGFR2 signaling in pathologic conditions using a model of acute brain ischemia and reperfusion (I/R).

Isolation and culture of umbilical cord blood mononuclear cells and HUVECs
Umbilical cord blood mononuclear cells (UCB-MNCs) were isolated from human umbilical cord blood as previously described (9). Briefly, blood samples were collected from placentae with attached umbilical cords by gravity centrifugation. This procedure was approved by the Ethics Committee at Yonsei University.

Fertility assessment
Mouse fertility was measured as previously described in Kuchmiy et al. (18). WT or Salm4 2/2 10-wk-old female mice were mated with WT or Salm4 2/2 10-wk-old male mice for 16 wk. The time intervals for each litter and the litter size for each pair were recorded. The mean number of total litters/female within the 16-wk period was calculated. Pups were counted as soon as litters were found to minimize underestimation of litter size. Pups were weaned at 21 d.

EC migration assay
A wound-healing assay was performed by scratching confluent HUVECs on 35-mm dishes with micropipette tips. Medium containing 1% FBS and 40 ng/ml human VEGF-A were used, with images captured at 0 and 8 h after wounding. For quantitative analysis, 5 fields/plate were photographed, and the areas between the front lines were measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Each assay was repeated 3 times.
The chemotactic motility of HUVECs was assayed using Transwell chambers (3422; BD Biosciences) with polycarbonate filters (8-mm pore size, 6.5-mm diameter). In brief, the lower surface of the filters were coated with 0.1% gelatin, and M199 medium containing 1% FBS and 40 ng/ml VEGF-A were added to the lower wells. The HUVECs were trypsinized and resuspended in M199 containing 1% FBS to a final concentration of 1 3 10 6 cells/ml. A 100-ml aliquot of cell suspension was added to each of the upper wells and incubated at 37°C for 2 h. Cells were then fixed and stained with hematoxylin and eosin. Nonmigrating cells were removed from the upper surface of filters using cotton swabs. Chemotaxis was quantified by counting the cells that migrated to the lower sides of the filters using optical microscopy (3200 magnification). Ten fields were counted/assay. Each sample was assayed in triplicate, and the assays were repeated 3 times.

In vitro tube formation assay
Tube formation assay was performed as previously described in Choi et al. (19). Briefly, 250 ml growth factor-reduced Matrigel (354230; BD Biosciences) was added to 16-mm diameter tissue culture wells and allowed to polymerize for 20 min at 37°C. HUVECs were incubated in complete M199 medium. After trypsinization, harvested cells were resuspended in 1% FBS and 40 ng/ml human VEGF-A and plated on layers of Matrigel (1.5 3 10 5 cells/ml). Matrigel cultures were incubated at 37°C and photographed at various time points (3200 magnification). The area covered by the tube network was determined using ImageJ software.

Isolation of lung ECs from neonatal mice
Mouse lung ECs (MLECs) were isolated from 3 postnatal (6-8 d old) WT and Salm4 2/2 mice for seeding in a 60-mm plate as previously described in Sobczak et al. (20). Mice were anesthetized by intraperitoneal injection of tribromoethanol (Avertin, 2.5%) at 125 mg/kg for cardiac perfusion with PBS. The lungs were excised, minced, and digested with 250 U/ml collagenase type 2 and 4 (LS004188; Worthington Biochemical) in PBS for 45 min. The digest was homogenized through a 14-gauge needle and then filtered through a 70-mm cell strainer (352350; BD Biosciences). The cell suspension was isolated by immunoselection with CD31-conjugated magnetic beads (11035; Thermo Fisher Scientific). Medium was changed the following day and then every other day. Fully enriched cells were further sorted using ICAM2 (553326; BD Biosciences)-conjugated magnetic beads to achieve .90% purity. VE-cadherin staining was used to confirm ECs. MLECs were cultured in endothelial basal medium 2 (CC-3156; Lonza, Basel, Switzerland) containing endothelial growth medium 2 kit (CC-4176; Lonza) and 10% FBS.
For immunoprecipitation of endogenous proteins, HUVECs were lysed in 1 ml NP-40 lysis buffer (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 1% NP-40, and protease inhibitor cocktail). Cell lysates were centrifuged at 14,000 rpm for 15 min. The supernatants were immunoprecipitated with antibodies against NRP1, NRP2, VEGFR1, VEGFR2, or IgG at 4°C overnight, followed by the addition of protein A agarose beads ; MilliporeSigma) at 4°C for 2 h. Immunoprecipitates were washed 3 times with lysis buffer, resuspended in SDS-PAGE sample buffer containing 2-ME, and analyzed by Western blotting.
Membranes were incubated with primary antibodies (1:1000) overnight at 4°C and with secondary antibodies (1:3000) for 45 min at room temperature. The primary antibodies, phosphory-

Matrigel plug angiogenesis assay
Six-week-old mice were subcutaneously injected with growth factor-reduced Matrigel containing 200 ng/ml mouse VEGF-A (K0921632; Koma Biotech) or FGF-2 (PMG0035; Thermo Fisher Scientific) with 10 U/ml heparin. The injected Matrigel rapidly formed a single solid gel plug. After 7 d, the skin of the mouse was pulled back to expose the Matrigel plug, which remained intact. To identify infiltrating ECs, immunostaining was performed on slices of the Matrigel plug using an anti-mouse CD31 antibody.

Aortic ring assay
The aortic ring assay was performed as previously described in Baker et al. (21). Briefly, aortic rings were isolated from 6-wk-old mice and cut to a fixed size in PBS. Growth factor-reduced Matrigel was spread into each well of a 96-well plate, which was incubated at 37°C for 20 min to allow matrix polymerization. Aortic rings were then placed on the matrices (1/well). Each embedded ring was cultured in Opti-MEM medium (31985070; Thermo Fisher Scientific) supplemented with 2.5% FBS and 50 ng/ml mouse VEGF-A. Medium was changed the following day and then every other day.

mRNA sequencing and analysis
Total RNA was isolated using an RNeasy Kit (Qiagen, Germantown, MD, USA). The mRNA sequence libraries were prepared using the TruSeq Stranded mRNA Sample Prep Kit (RS-122; Illumina, San Diego, CA, USA). The protocols followed the TruSeq Stranded mRNA Sample Preparation Guide (Part 15031047 Rev. E). Sequencing was performed using an Illumina HiSeq 4000 sequencer (101-bp paired-end runs), the HiSeq 3000 4000 System User Guide Document 15066496 protocol, and TruSeq 3000 4000 Sequencing by Synthesis kit v.3 sequencing reagents. After removing low-quality and adapter sequences, sequence reads were aligned to the University of California-Santa Cruz, mouse genome reference sequence Mus musculus using Hierarchical Indexing for Spliced Alignment of Transcripts (HISAT) v.2.0.5 (22).

Statistical analysis of gene expression
StringTie v.1.3.3b was used to estimate gene abundance (23). Abundance was measured in fragments per kilobase of exon per million fragments mapped; any values of 0 were discarded. To establish log 2 transformation, 1 was added to each abundance value of filtered genes, and quantile normalization was performed. Statistical significance of differential expression data was determined using independent Student's t tests. For the differentially expressed gene set, hierarchical clustering analysis was performed using complete linkage and Euclidean distance as a measure of similarity (absolute fold change, $2). The permeabilities of HUVECs were assayed using Transwell chambers (3460; BD Biosciences) with polyester filters (0.4-mm pore size, 12-mm diameter). The upper wells were coated with 1% gelatin. HUVECs were trypsinized and resuspended in endothelial basal medium 2 containing endothelial growth medium 2 kit and 10% FBS to a final concentration of 8 3 10 4 cells/ml. A 1-ml aliquot of cell suspension was added to each upper well and incubated at 37°C until cells formed junctions. Cells were then starved for 2 h in serum-free medium and treated with 50 ng/ml human VEGF-A for 30 min. Two electrodes were used for electrical measurements: one placed in the upper well and the other in the lower well, separated by the cellular monolayer.

Miles vascular permeability assay
Six-to-eight-week-old mice were intravenously injected with PBS containing 100 ml 1% Evans blue (EB; E2129; Milli-poreSigma). After 10 min, 50 ml mouse VEGF-A (50 ng/ml) or histamine (200 nmol/ml; H7125; MilliporeSigma) were intradermally injected into the skin of a shaved back. PBS and VEGF-A or histamine were injected into the dorsum of each mouse. Twenty-eight (4 groups of 7) mice were used in the assay. After 30 min, the back skin was biopsied using a 6-mm punch and then incubated in formamide solution overnight at 56°C to extract the dye. The absorbance of the supernatant was measured on a spectrophotometer at 620 nm. The dye content in the skin was calculated against a standard curve and expressed as mg/g.

I/R model through middle cerebral artery occlusion
Eight-to-ten-week-old mice were anesthetized by intraperitoneal injection of tribromoethanol (Avertin, 2.5%). A heating pad (JD-OT-03106; Jeung-Do Bio and Plant, Seoul, South Korea) was used to maintain a body temperature of 37°C during surgery. I/R was induced by middle cerebral artery occlusion on the left side as previously described in Zhang et al. (24). Briefly, the left common carotid artery (CCA) was exposed and separated from the vagus nerve. The left CCA was ligated at the distal side of its intersection. The left external carotid artery was dissected free and ligated at the distal side of the CCA intersection. A 6-0 silicone rubber-coated monofilament suture (602156PK10; Doccol, Sharon, MA, USA) was inserted into the left external carotid artery stump and gently advanced 8 mm into the internal carotid artery. The suture proceeded from the internal carotid artery to the middle cerebral artery to occlude the middle cerebral artery. After 3 h of ischemia, reperfusion was produced by withdrawal of the suture. In the sham-treated group, the same surgical procedures were performed without insertion of the suture.

Neurologic scores
Neurologic defects were evaluated after 16 h reperfusion as previously described in Zhang et al. (24). The scores were as follows: 0, no observable neurologic deficit; 1, flexion of contralateral torso and forelimbs upon lifting the whole animal by the tail; 2, circling to the contralateral side when held by the tail with feet on the floor; 3, spontaneous circling to the contralateral side; 4, no spontaneous motor activity. For consistency, scoring was repeated 3 times.

Measurement of infarction volume
Infarction volume was measured as previously described in Zhang et al. (24). Briefly, after 16 h reperfusion, mice were anesthetized by intraperitoneal injection of tribromoethanol (Avertin, 2.5%). Brains were harvested and cut into 2 mm coronal slices using a brain matrix. Each slice was incubated with 2% 2,3,5-triphenyltetrazolium chloride (TTC) at 37°C for 20 min. After the TTC reaction, infarction was marked by unstained areas, whereas normal areas were stained red. Infarction was quantified using ImageJ software. To eliminate the effects of edema in the ipsilateral hemisphere, the infarction volume was calculated as follows: [(contralateral hemisphere 2 undamaged ipsilateral hemisphere) 4 (contralateral hemisphere 3 2) 3 100] %.
Determination of blood-brain barrier permeability and brain-water content Blood-brain barrier (BBB) permeability was assessed by EB extravasation using a method previously described in Zhang et al.

Statistical analysis
Data are presented as means 6 SD or SEM. Multiple group means were compared by 1-way ANOVA with Dunnett's or Turkey's corrections, followed by the 2-tailed Student's t test for pairwise comparisons.

SALM4 is expressed in ECs
We previously performed Affymetrix gene chip analysis of UCB-MNCs, which were characterized as hematopoietic monocytes, during their differentiation into outgrowth ECs (OECs) (Gene Expression Omnibus accession GSE12891) (25). The analysis revealed that SALM4 was highly expressed in the OEC stage compared with the UCB-MNC stage, which was confirmed by RT-PCR (Fig. 1A). We also examined the expression of other SALM family members in HUVECs, human umbilical artery ECs, and lymphatic ECs. SALM4 was expressed in all 3 cell types. Human embryonic kidney 293 transformed cell lines were used as a positive control for SALM125 expression (Fig. 1B). To determine the expression of Salm4 in mouse ECs, we examined Salm4 expression in organs of Salm4 2/2 mice. Frozen mouse organs were sectioned and treated with DAPI as a nuclear marker. Tissues were immunostained for CD31 as an EC marker and b-galactosidase for Salm4 expression. Salm4 was expressed in the ECs of the brain, liver, and lungs (Supplemental Fig. S1), suggesting potential functions in ECs.

SALM4 regulates EC survival, migration, and capillary-like tube formation
To investigate whether SALM4 has important roles in endothelial function, we used 3 types of siSALM4. SALM4 mRNA and protein expression levels were inhibited in siSALM4-transfected HUVECs ( Fig. 1C and Supplemental Fig. S2A-C). We then evaluated the effects of SALM4 knockdown on EC properties. In a previous study, when SALM4 was overexpressed, knockdown of flotillin-1 prevented neurite branching by SALM4 (13). Thus, in the present study, we examined this function in ECs. Cell viability was evaluated by MTT assay and cell counting. siSALM4 decreased the viability of HUVECs treated with VEGF-A ( Fig. 1D and Supplemental Fig. S2D). Knockdown of SALM4 also inhibited wound-healing migration in the presence of VEGF-A stimulation (Fig.  1E, F). Reduced migration was observed with 3 types of siSALM4-transfected HUVECs, indicating that this function is not an off-target effect of siRNA (Supplemental   S2E, F). Chemotactic motility was inhibited in SALM4-depleted HUVECs upon VEGF-A stimulation (Fig. 1G, H). Furthermore, a well-organized tube-like formation was observed in siCon-transfected HUVECs, but it was severely impaired in SALM4-silenced HUVECs with VEGF-A treatment (Fig. 1I-L). These results indicate that SALM4 can modulate VEGF-A-induced angiogenic functions in ECs.

VEGF-A-induced angiogenic sprouting is inhibited in Salm4 2/2 mice
We analyzed the phenotype of Salm4 2/2 mice to examine the angiogenic effects of SALM4 deficiency in vivo. Body weight, fertility, development, and organ vasculature were not affected by Salm4 KO (Supplemental Fig. S3A-F). We then performed an in vivo Matrigel plug assay. Before performing this experiment, we confirmed that Salm4 KO did not alter vessel density in the skin (Supplemental Fig.  S4A, B). In the VEGF-A (2) group, no significant difference was observed in Matrigel plugs from WT and Salm4 2/2 mice. In the VEGF-A (+) group, VEGF-A-induced neovascularization was inhibited in plugs from Salm4 2/2 mice compared with WT mice (Fig. 2A). This finding was confirmed by hemoglobin contents (Fig. 2B). In addition, fewer CD31-positive cells were observed in confocal images of plugs from Salm4 2/2 mice compared with WT mice (Fig.  2C, D). To determine whether the phenotype of Salm4 2/2 mice is VEGF-A dependent, we examined FGF-2-induced neovascularization. Similar to VEGF-A, FGF-2 induces survival, proliferation, and migration of ECs (26,27). In the FGF-2 (2) and FGF-2 (+) groups, no significant differences were observed in the Matrigel plugs from WT and Salm4 2/2 mice (Supplemental Fig. S4C-F). We further examined the sprouting activity of ECs in Salm4 2/2 mice using an ex vivo aortic ring assay. In the VEGF-A (2) group, no significant differences were observed in aortic rings from WT and Salm4 2/2 mice. In the VEGF-A (+) group, EC sprouting areas and length were inhibited in aortic rings of Salm4 2/2 mice compared with those of WT mice (Fig. 2E-G). Taken together, these results suggest that loss of SALM4 inhibits angiogenesis in response to VEGF-A.

VEGF-A-induced vascular permeability is attenuated in SALM4-knockdown HUVECs and Salm4 2/2 mice
VEGF-A is a vascular permeability factor that plays an essential role in physiologic and pathologic angiogenesis (28). As shown by the in vitro trans-endothelial/epithelial resistance assay, VEGF-A markedly induced vascular permeability in the siCon-transfected HUVECs. By contrast, siSALM4-transfected HUVECs did not exhibit VEGF-A-induced permeability (Fig. 3A). To identify whether this effect also occurred in vivo, the Miles vascular permeability assay was performed. In the VEGF-A (2) group, no significant difference was observed in skin from WT and Salm4 2/2 mice. In the VEGF-A (+) group, VEGF-A strongly induced vascular permeability in WT mouse skin, as shown by increased EB leakage. Salm4 2/2 mouse skin showed reduced leakage of EB compared with WT mouse skin (Fig. 3B). Reduced extravasation of EB in Salm4 2/2 mouse skin was also confirmed by quantification of EB contents (Fig. 3C). Because histamine also induces vascular permeability (29), we examined whether Salm4 KO affects histamine-induced permeability. In the histamine (2) and histamine (+) groups, no significant differences were observed in the skin from the WT and Salm4 2/2 mice (Supplemental Fig.  S4G, H). VEGF-A disrupts junction proteins and alters actin filament distribution (30). Therefore, we assessed these effects in SALM4-knockdown HUVECs. The VEGF-Auntreated group retained the polygonal shape and linear pattern of VE-cadherin at cell borders, confirming junction formation. VEGF-A-treated HUVECs transfected with siCon showed disrupted junctions and breakage of the linear VE-cadherin pattern. However, knockdown of SALM4 prevented these effects. Furthermore, VEGF-Ainduced stress fiber formation was increased, and cortical actin ring structure was decreased in control HUVECs using F-actin staining. By contrast, reduced VEGF-Ainduced stress fiber formation and increased cortical actin ring structure were observed in the absence of SALM4 (Fig. 3D). Collectively, these results indicate that depletion of SALM4 inhibits EC permeability in response to VEGF-A.

Angiogenesis-related gene transcription is inhibited in Salm4 2/2 MLECs stimulated with VEGF-A
To investigate transcriptional changes elicited by VEGFR2 activation, we performed mRNA sequencing analysis of isolated lung ECs from WT and Salm4 2/2 mice. MLECs were immunostained for VE-cadherin to confirm these were ECs before performing analysis (Supplemental Fig. S7A). Because well-known VEGF-A-VEGFR2 signaling target genes, such as regulator of calcineurin 1, ESM1, and angiopoietin 2, are up-regulated after 4 or 8 h of VEGF-A treatment, we stimulated with VEGF-A in MLECs at 0, 4, and 8 h (35). We identified 23 significantly up-regulated genes at 4 h and 10 significantly up-regulated genes at 8 h in WT MLECs (absolute fold change, .2; P , 0.01). Because these genes were not induced in Salm4 2/2 MLECs (Supplemental Fig. S7B), we proposed that these 33 genes are targets of VEGFR2 phosphorylation at Y1173. Consistent with previous reports, ESM1 (36) and PTGS2 (37), which is positively regulated downstream of VEGF-A-VEGFR2 signaling, were induced by VEGF-A treatment at 4 h in WT MLECs but not in Salm4 2/2 MLECs (Supplemental Fig. S7B, C). COL6A3, which is positively regulated downstream of PI3K-AKT signaling in ECs (38), was induced at 8 h in WT MLECs but not in Salm4 2/2 MLECs (Supplemental Fig. S7B, C). These results suggest that Salm4 KO inhibits angiogenic gene transcription induced by VEGFR2-Y1173 phosphorylation.
Salm4 2/2 mice show ameliorated brain damage after I/R As mentioned above, SALM4-depleted ECs showed attenuated permeability through VEGFR2 phosphorylation at the Y1175-PI3K-AKT-eNOS signaling axis, which plays a role in acute brain stroke (39). Acute stroke increases vascular permeability, and leaky blood vessels aggravate BBB disruption and edema (40). Thus, we used the I/R model to identify SALM4 functions in a pathologic condition (Supplemental Fig. S8A). In the sham-treated group, no neurologic deficits were observed 16 h after reperfusion in WT and Salm4 2/2 mice. In the I/R group, severe neurologic deficits were present in WT mice but not Salm4 2/2 mice (Supplemental Fig. S8B). In the sham-treated group, TTC staining indicated that there were no areas of infarction in WT or Salm4 2/2 mice. In the I/R group, there were fewer infarcted regions in Salm4 2/2 mice compared with WT mice (Fig. 5A, B). No significant differences in BBB permeability, assessed by EB extravasation, were observed in the sham-treated WT and Salm4 2/2 mice. In the I/R group, EB extravasation was observed in WT mice. However, Salm4 2/2 mice did not show ischemia-induced EB extravasation (Fig. 5C-E). Next, we measured water content 16 h after reperfusion by the wet-dry method. In the sham-treated group, no significant differences were observed in WT and Salm4 2/2 mice. In the I/R group, the brain-water content was increased in WT mice compared with Salm4 2/2 mice (Supplemental Fig. S8C, D). To identify which effects of SALM4 on VEGFR2 phosphorylation are due to I/R, we examined VEGFR2 phosphorylation sites in brain lysates. In the sham-treated group, no significant differences were observed in WT and Salm4 2/2 brains. In the I/R group, VEGFR2 phosphorylation was induced at Y949, Y1173, and Y1212 in WT brains. However, Salm4 2/2 brains showed reduced VEGFR2 phosphorylation only at Y1173 (Fig. 5F-L). Therefore, Salm4 2/2 brains can ameliorate I/R damage through inhibition of VEGFR2-Y1173 in ECs. Salm4 2/2 mice show attenuated BBB disruption after I/R The tight junction-related proteins ZO-1, occludin, and claudin-5 were examined 16 h after reperfusion by conjunction with CD31, an EC marker. In the sham-treated group, ZO-1 and CD31 were nearly merged. In the I/R group, WT brains showed greatly reduced ZO-1 expression, indicative of a disrupted BBB, compared with ZO-1 of Salm4 2/2 brains. A superimposed lining of ZO-1 and CD31 was observed in Salm4 2/2 brains in the I/R group, suggesting that BBB destruction was attenuated (Fig. 6A, B). Similar results were observed for occludin and claudin-5 ( Fig. 6C-F). These results further demonstrate that Salm4 2/2 brains exhibit attenuated disruption of the BBB, potentially through inhibition of VEGFR2-Y1173 in ECs.

Salm4 2/2 mice show suppressed expression of adhesion molecules and activation of glial cells after I/R
The neuroinflammatory response may be an important factor in VEGF-A-mediated BBB disruption during brain ischemic stroke (41,42). In the sham-treated group, ICAM1 and VCAM1 were barely expressed in WT and Salm4 2/2 brains. In the I/R group, ICAM1 and VCAM1 were significantly decreased in Salm4 2/2 brains compared with WT brains 16 h after reperfusion (Fig. 7A-D).
Activation of glial cells, such as astrocytes and microglia, are induced by cerebral ischemia (43,44). Based on B-E) Blots were assessed using ImageJ software; n = 3 independent experiments. F ) Effects of Salm4 KO on signaling pathways induced by VEGF-A (20 ng/ml) in MLECs. G-J) Quantification of blots using ImageJ software; n = 3 independent experiments. Error bars represent means 6 SD. Blots incubated with phospho-specific antibodies were probed only once from the different gels. Other blots were stripped and reprobed. P-, phosphorylated. *P , 0.05, **P , 0.01, ***P , 0.001 by paired, 2-tailed Student's t test.
this report, we investigated the number of astrocytes and microglia in the I/R model. In the sham-treated group, no significant changes were observed in WT and Salm4 2/2 brains. In the I/R group, WT brains showed activation of GFAP-positive astrocytes and CD11b-positive microglia.

DISCUSSION
This study identified SALM4 as a novel regulator of VEGF-A-induced VEGFR2-Y1175 phosphorylation and associated angiogenic functions, including migration, tube-like formation, EC recruitment, and vascular permeability (Supplemental Fig. S9). It also established SALM4 as a fine controller of brain ischemic reperfusion injury.
VEGFR2 signaling should be precisely regulated, because aberrant VEGFR2 phosphorylation causes abnormal EC sprouting and permeability (1). Based on these results, in vivo studies of VEGFR2 phosphorylation sites were performed. VEGFR2-Y949F and Y1212F knock-in mice are viable and fertile (16). However, Li et al. (45) reported that VEGFR2-Y949F knock-in mice demonstrate reduced glioblastoma permeability, B16F10 Figure 6. Brains of Salm4 2/2 mice prevent disruption of tight junction-related proteins after I/R. A) Immunostaining for ZO-1 and CD31 in ischemic brain sections of sham-treated and I/R groups. Merged images of ZO-1 and CD31 staining are also shown. Square: enlarged image of the region. B) Quantification of ZO-1-positive blood vessels. C) Immunostaining for occludin and CD31 in ischemic brain sections. Merged images of occludin and CD31 staining are shown. Square: enlarged image of the region. D) Quantification of occludin-positive blood vessels. E) Immunostaining for claudin-5 and CD31 in ischemic brain sections. Merged images of claudin-5 and CD31 staining are shown. Square: enlarged image of the region. Scale bars, 20 mm. F) Quantification of claudin-5-positive blood vessels; n = 5/group. Ns, not significant. Error bars represent means 6 SD. **P , 0.01, ***P , 0.001 by paired, 2-tailed Student's t test.
tumor vascular leakage, and metastasis. VEGFR2-Y1173F knock-in mice show embryonic lethality between 8.5 and 9.5 d, similar to the phenotypes of Vegfr2 and Vegf-a KO mice. These results suggest that Y1175 has essential roles in VEGFR2 activation (46). The present study showed that Salm4 2/2 mice have attenuated VEGFR2 phosphorylation at Y1173, but not at Y949 or Y1212, following brain I/R injury. This finding indicates that SALM4-mediated potentiation of VEGFR2 phosphorylation at Y1173 plays critical roles in neurologic impairments, BBB disruption, and neuroinflammation following I/R damage.
We used several strategies to delineate the mechanism of SALM4-mediated VEGFR2 signaling. The strength of VEGFR2 signaling is tightly regulated at Immunostaining for ICAM1 in ischemic brain sections of sham-treated and I/R groups (A) and quantification of relative ICAM1 levels (B). C, D) Immunostaining for VCAM1 in ischemic brain sections (C ) and quantification of relative VCAM1 levels (D). E, F ) Immunostaining for GFAP in ischemic brain sections (E ) and quantification of GFAP expression levels (F ). G, H ) Immunostaining for CD11b in ischemic brain sections (G) and quantification of CD11b-positive cells (H ). Ns, not significant. Error bars represent means 6 SD; n = 5/group. Scale bars, 50 mm. **P , 0.01 by paired, 2-tailed Student's t test. numerous levels, including by PTPs, coreceptors, and receptor internalization (16). There are 2 categories of PTPs that dephosphorylate VEGFR2: receptor type PTPs, such as VE-PTP and PTP receptor type J, and intracellular PTPs, such as PTP1B, T-cell PTP, Src homology 2 domaincontaining proteins 1 and 2, PTP-non-receptor type 9, and low MW-PTP (44). Among these, VE-PTP and PTP1B are well characterized. The Ve-ptp KO mouse phenotype is embryonic lethal at 9.5 d, suggesting that VE-PTP is required for endothelial development (33,34). VE-PTP dephosphorylates all tyrosine residues of VEGFR2 (47). PTP1B dephosphorylates VEGFR2 at Y1175 and decreases ERK activation in early endosomes (48). VE-PTP and PTP1B bind to VEGFR2 and elicit conformational changes that detach phosphorus from VEGFR2 (16). Because SALM4 did not bind to VEGFR2, we examined whether SALM4 could bind to PTPs. However, we did not find evidence that SALM4 binds directly to VE-PTP or PTP1B. Therefore, SALM4 does not seem to affect recruitment of PTPs. Next, NRP1, an indispensable coreceptor of VEGFR2, binds to VEGFR2 only in the presence of VEGF-A stimulation (49). SALM4-depleted ECs formed a VEGF-A-NRP1-VEGFR2 complex as siCon-transfected ECs. Because SALM4 did not bind to NRP1, SALM4 does not regulate VEGFR2 phosphorylation through this mechanism. Similar to SALM4, Zhang et al. (50) reported that roundabout guidance receptor 4 also inhibits VEGFR2 phosphorylation. Roundabout guidance receptor 4 binds UNC5B, and the UPA domain of UNC5B reduces VEGFR2 phosphorylation at Y951 but not at Y1175. These possibilities suggest that SALM4 regulates the folding of VEGFR2-Y1175 by binding to an unidentified molecule. Another possibility is that VEGFR2 signaling is regulated by receptor internalization. The VEGF-A-NRP1-VEGFR2 complex is internalized and binds to synectin-myosin-VI. The internalized VEGFR2 complex then undergoes small GTPase of the Ras superfamily 11(Rab11)dependent recycling. Because internalized VEGFR2 is phosphorylated at Y951 and Y1175, the downstream kinases SRC and ERK are phosphorylated, respectively (51). SALM4 depletion in ECs did not alter ERK phosphorylation compared with siCon-transfected ECs, suggesting that SALM4 is unlikely to affect VEGFR2 internalization. SALM4 was previously reported to regulate neurite branching through interactions with lipid raft-associated proteins (13). In the present study, we demonstrated the vascular role of SALM4 in VEGFR2 signaling in ECs. Thus, SALM4 is likely to act as a modulator of receptor-mediated signal transduction in the membranes of neuronal cells and ECs. Although we could not identify the exact interacting protein and mechanism by which SALM4 regulates VEGFR2 phosphorylation, its action in ECs appears specific to VEGFR2-Y1175 and not Y951 or Y1214.
Reduced VEGF-A-VEGFR2-Y1175-PI3K-AKT-eNOS signaling in SALM4 inhibited ECs, suggesting that VEGF-A-induced disruption of endothelial junction is reduced in SALM4-silenced ECs. Such signaling is important for regulating vascular permeability. Acute brain I/R damage can increase VEGF-A expression in neurons, astrocytes, macrophages, and ECs. Increased VEGF-A influences vascular permeability through PI3K-AKT phosphorylation, which activates eNOS (52)(53)(54). Phosphorylation of VEGFR2, a major receptor for VEGF-A, leads to degradation of endothelial junction molecules after stroke (55). Consequently, ischemic brains have vascular leakage, causing the infiltration of inflammatory molecules into the cerebral blood flow. Behavioral defects, areas of infarction, albumin extravasation, edema, BBB disruption, neuroinflammation, and glial cell activation were reduced after I/R in the brains of Salm4 2/2 mice. Furthermore, VEGFR2 phosphorylation at Y949, Y1173, and Y1212 was induced in WT brains after I/R, but only VEGFR2 phosphorylation at Y1173 was reduced in Salm4 2/2 brains. The current treatment for acute ischemia is largely dependent on tissue plasminogen activator-mediated thrombolysis. However, this treatment exacerbates the risk of hemorrhage, which is related to BBB breakdown (56). Because BBB stabilization increases the efficacy of tissue plasminogen activator treatment, BBB stabilization is critical for reducing poor outcomes in patients with acute ischemia. Based on our findings, the VEGFR2-Y1175-PI3K-AKT-eNOS signaling axis is a potentially novel therapeutic target for vascular permeability-related diseases, such as brain ischemic reperfusion and tumor progression.
In summary, we suggest that SALM4 induces conformational changes in the proximity of VEGFR2 phosphorylation at Y1175 by binding an unidentified mediator in the lipid raft, which is the platform for signaling machinery in the plasma membrane (57). Future research is needed to fully elucidate the role of SALM4 in VEGFR2 phosphorylation at Y1175.