Interference With ESAM (Endothelial Cell-Selective Adhesion Molecule) Plus Vascular Endothelial-Cadherin Causes Immediate Lethality and Lung-Specific Blood Coagulation

Primary ECs from lungs (mouse lung microvascular endothelial cells [MLMVECs]) of wild-type (WT) or ESAM gene inactivated mice were isolated and cultured as previously described.⁹ COS-7 (CV-1 in origin with SV40 genes) cells were cultured and transfected as described.⁴⁵ WT Chinese hamster ovary (CHO) cells, CHO cells expressing mouse JAM-A or ESAM were cultured as described.^37,46

The following antibodies were used: rabbit polyclonal antibodies (rab pAb) against mouse MAGI-1 (membrane-associated guanylate kinase, WW and PDZ domain containing 1; VE-18) and pAb against the extracellular domain of mouse ESAM (VD3) were raised and purified as described⁴⁷ and,³⁷ respectively. Rab pAb VE19 and rat mAb 1G8 both against ESAM,³⁷ rat mAb BV13 (Thermo Fisher Scientific), rab pAb VE42,⁴⁸ and goat pAb AF1002 (R&D Systems) against mouse VE-cadherin, rat mAb 1G5.1 and rat mAb 5D2.6 against PECAM-1,⁴³ rab pAb against JAM-A,⁴⁹ rat mAb against TER-119 (TER-119: BioLegend), rat mAb against GP (glycoprotein)Ib-β subunit of the mouse platelets (X649, Emfret Analytics), rab pAb against ZO-1 (zonula occludens-1; 40-2200, Thermo Fisher Scientific), rab pAb against Mesothelin (PA5-79698, Thermo Fisher Scientific), isotype control Ab (Rat IgG1, Thermo Fisher Scientific), mAb against α-tubulin (B-5-1-2, Sigma-Aldrich), and rab pAb against claudin-5 (34-1600 Thermo Fisher Scientific). Alexa Fluor 488–, Alexa Fluor 568–, Alexa Fluor 594–, and Alexa Fluor 647–coupled secondary antibodies were purchased from Invitrogen. All other secondary antibodies were purchased from Dianova.

The following mouse lines were used: 8- to 12-week-old C57BL/6JRj WT mice (Janvier Labs), ESAM deficient mice,⁴³ PECAM-1 deficient mice,⁵⁰ and JAM-A deficient mice⁵¹ were used. ESAM^−/− Cdh5^iECKO or control ESAM^−/− Cdh5^lox/lox mice were generated by mating ESAM^−/− mice⁴³ with Cdh5^lox/lox mice⁹ either expressing PDGFbiCre or not. ESAM-ΔSH3DBM and ESAM-mCherry knockin mice were generated by homologous recombination using a genetically modified targeting construct. The genomic sequence of ESAM was cloned from BAC-clone RP23-383h20 (Thermo Fisher Scientific). The knockin construct comprised an 8 kb genomic sequence from Exon 2 to Exon 7 (including the 1.3 kb mutagenesis region of Exon 7), an FRT (flippase recognition target)-site flanked hygromycin resistance gene, a sequence for polymerase chain reaction-genotyping and a homologous region of 1.7 kb containing the 3′ untranslated region and Exon1 of the adjacent gene Vsig2. The ESAM-ΔSH3DBM mutation was generated using site-directed mutagenesis to delete the 3 potential SH3-binding motifs A₃₂₇APPRP₃₃₂, P₃₄₉RL₃₅₁, and M₃₈₃VP₃₈₅ in the cytoplasmic region of ESAM. To generate the ESAM-mCherry mutation, mCherry was fused to the C terminus of ESAM with the short linker AGLGG, thereby replacing the stop-codon of ESAM. The targeting construct was transfected into mouse embryonic stem cells of C57BL/6 mice using electroporation. Hygromycin-resistant colonies were screened by a polymerase chain reaction and confirmed by Southern blotting. Positive ES cell clones were injected into blastocysts of C57BL/6 mice to generate chimeras, which were mated with C57BL/6 mice. Genotyping was performed to verify the presence of the mutation. All animal studies were approved by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen. Animal studies were performed exclusively with male mice since they were abundant because of the use of female mice in another project.

mCherry was fused to the C terminus of ESAM with the linker GAGG and cloned into pcDNA3 vector for expression in COS-7. MAGI-1-FLAG expression vector was described.⁴⁷

Samples were prepared as previously described⁹ with minor modifications. Mice were anesthetized with ketamine/xylazine, and tracheas were cannulated with an intravenous catheter tube. Lungs were perfused with PBS, followed by fixative buffer (2% paraformaldehyde and 2% glutaraldehyde in 0.1 mol/L cacodylate buffer, pH 7.2), then filled with fixative buffer via the trachea. Lungs were further fixed and processed as described.⁹

Staining of cultured cells and tissue vibratome sections were performed as described under Methods in the online-only Data Supplement.

Mice were consecutively intravenously injected with rat IgG1 or BV13 antibodies and Evans blue and were euthanized 30 minutes later. For lung permeability assays, 6.5 hours after antibody application, Evans blue was administrated for 15 minutes. The body circulation was perfused, organs of interest were removed, and Evans blue was extracted with formamide for 5 days. The concentration of the dye was measured at 620 nm with a spectrophotometer (Shimadzu).

Assays were performed with ESAM and JAM-A transfected cells as described under Methods in the online-only Data Supplement.

MLMVECs were seeded onto L-cysteine-reduced, fibronectin-coated 8W10E electrodes (Applied Biophysics). Electrical impedance was measured at 4000 Hz in real-time at 37°C and 10% CO₂ using the ECIS ΖΘ system (Applied Biophysics). Confluent cells were treated with 50 µg/mL of control antibodies, BV13 or VD3 for 2 hours.

Coimmunoprecipitations and immunoblotting were performed as previously described⁵² except that the 4.5 μm Dynabeads (Thermo Fisher Scientific) were used instead of protein A/G–Sepharose.

Tamoxifen (100 mg) was mixed with 100 µL ethanol and 5 mL peanut oil and dissolved for 30 minutes in a water bath at 37°C using ultrasound. One hundred microliters of this solution was injected intraperitoneally daily for 5 consecutive days into ESAM^−/− Cdh5^lox/lox mice either expressing PdgfbiCre or not, and mice were analyzed 2 or 3 days later.

Transplantations were performed as described.⁵³ Platelets were isolated as described,⁵⁴ then lysed with 2X SDS sample buffer and subjected to Western blotting.

Cell isolation and flow cytometry were performed as described under Methods in the online-only Data Supplement.

Data sets were tested for normality (Shapiro-Wilk) and equal variance (F test or Brown-Forsythe). Comparisons between 2 groups (1-factorial experiments) were performed using 2-tailed Student t- test or Wilcoxon signed-rank test for normal and non-normally distributed data, respectively. Comparisons among 3 groups (1-factorial experiments) were performed using 1-way ANOVA followed by Tukey multiple comparison test or Welch ANOVA followed by Dunnett multiple comparison test for normally distributed data with either equal or unequal variances, respectively. Two-way ANOVA followed by Tukey multiple comparison test was used for 2-factorial experiments. GraphPad Prism8 software was used for this analysis. P- values are indicated by asterisks: *P<0.05, **P<0.01, ***P<0.001, and ****P≤0.0001. Results are shown as means ± SEM. Immunoblot signals were quantified using the software Image Studio (LI-COR) or Fiji (version 1.52f).

We have previously shown that induced EC-specific deletion of the adhesion molecule VE-cadherin in adult mice (Cdh5^iECKO mice) increases vascular permeability selectively in the heart and lung.⁹ Despite this strong increase in vascular permeability in Cdh5^iECKO mice, ultrastructural analysis by electron microscopy revealed that endothelial junctions were not physically ruptured, suggesting that additional adhesion molecules maintain the junction structure in the absence of VE-cadherin.

This prompted us to investigate, which junctional molecules are relevant for the maintenance of junctional architecture despite the loss of VE-cadherin. As candidates, we analyzed the relevance of ESAM, JAM-A, and PECAM-1. To this end, we administered the anti–VE-cadherin antibody BV13⁸ or an isotype-matched control antibody intravenously into WT mice and ESAM^−/−, JAM-A^−/−, or PECAM-1^−/− mice, immediately followed by intravenous injection of Evans blue. After 30 minutes, mice were euthanized and extravasation of the albumin-adsorbed dye was measured in the lung, heart, skin, and brain. When comparing the different genotypes treated with control antibody, we found that the lack of PECAM-1 and JAM-A did not lead to any changes in basal permeability of any of the 4 organs (Figure 1A). In contrast, the absence of ESAM caused a 3-fold increase in vascular permeability in the lung and had no effect in the heart, skin, and brain (Figure 1A). When analyzing the effect of the anti–VE-cadherin antibody in WT mice, we found that vascular permeability was only increased in the lung and heart, but not in the skin and brain, in agreement with previous results.^8,9 Accessibility of the intravenously injected antibody to VE-cadherin could be demonstrated as shown for brain capillaries (Figure 1B). The absence of JAM-A or PECAM-1 did not enhance the effect of BV13 in the lung or heart and did not enable the antibody to interfere with vascular integrity in the skin or brain. In contrast, gene inactivation of ESAM further enhanced the permeability-inducing effect of BV13 in the lung but had no destabilizing effect in other organs even in the presence of the anti–VE-cadherin antibody (Figure 1A).

We planned initially to administer the antibody BV13 and measure permeability after 2 hours, since effects on permeability are increasing between 1, 2, and 7 hours.⁸ However, we observed that between 35 and 45 minutes after injection of the antibody, ESAM^−/− mice showed strong dyspnea symptoms and died shortly afterwards. No such effects were seen in any of the other genotypes, clearly highlighting the synergistic function of ESAM and VE-cadherin, which together are essential for survival.

Although the vascular permeability enhancing effect in the lung of ESAM^−/− mice was slightly higher than the effect observed after 30 minutes BV13 treatment of WT mice, we do not conclude that ESAM is of similar importance for junction stability as VE-cadherin. First of all, ESAM^−/− mice are viable and seem to be able to compensate somehow for the higher vascular permeability in their lungs.^41,43 Second, VE-cadherin function is far from being fully blocked at 30 minutes after BV13 injection, since the antibody takes some time to reach binding equilibrium.⁸ When we applied the antibodies for 6.5 hours, lung permeability in BV13 treated WT mice was almost 4 times higher than in control-IgG treated ESAM^−/− mice, showing that the loss of VE-cadherin function affects vascular integrity more severely than the loss of ESAM (Figure 1C). Nevertheless, our results clearly show that in the absence of ESAM, even mild, partial inhibition of VE-cadherin is sufficient to cause immediate lethality. This effect is not seen upon blocking VE-cadherin if ESAM is expressed at junctions.

Enhanced vascular permeability is one way of affecting junctional integrity, yet this is not necessarily accompanied by loss of physical intactness and rupture. We observed previously that Cdh5^iECKO mice showed strongly enhanced vascular permeability in the lung, but no signs of rupture of endothelial junctions.⁹ To investigate the physical state of the junctions in the different knockout mice challenged with BV13, we analyzed lung endothelial junctions by electron microscopy. We found that lung endothelial junctions of WT, JAM-A^−/−, and PECAM-1^−/− mice treated with isotype control antibodies were intact and even those of ESAM^−/− mice showed no signs of rupture or even partial damage (Figure 2A). Similar results were obtained when we blocked VE-cadherin with BV13 for 30 minutes in WT, PECAM-1^−/−, and JAM-A^−/− mice. Intriguingly, blocking VE-cadherin in ESAM^−/− mice led to the disruption of lung EC junctions, observed as gaps or open spaces between otherwise intact tails or trailers of ECs (Figure 2A). For quantification, randomly taken pictures of each genotype or treatment group were analyzed and intact junctions (electron-dense areas of EC contacts) plus gaps/open spaces between ECs (damaged junctions) were counted and set to 100%. This revealed that 58% of all junctions were damaged in BV13 treated ESAM^−/− mice, whereas no damage was found in the other groups (Figure 2C).

To exclude any potential side effects of the anti–VE-cadherin antibody, we generated mice double deficient for ESAM and VE-cadherin by mating ESAM^−/− mice with Cdh5^lox/lox mice expressing tamoxifen-inducible PdgfbiCre (Cdh5^iECKO). Mice expressing no Cre were used as negative controls. ESAM^−/− Cdh5^iECKO or ESAM^−/− Cdh5^lox/lox mice were stimulated with tamoxifen daily for 5 consecutive days and 2 to 3 days later, we collected the lungs and analyzed them by electron microscopy. In line with our blocking antibody data, we found that ablation of VE-cadherin together with ESAM led to disruption of endothelial junctions (Figure 2B). Quantification (Figure 2D) showed that 56% of the junctions were damaged. In agreement with this, double gene-deficient mice did not survive >3 days after the tamoxifen injection period.

Collectively, our results show that ESAM is able to maintain the structure of endothelial junctions in the absence of VE-cadherin. Only when both proteins are absent from EC junctions, the junctional structure cannot be sustained, which leads to rupture and gaps between ECs.

Since the combination of ESAM ablation and blocking or deleting VE-cadherin in ECs leads to lethality, rupture of junctions in the lung and exposure of the basement membrane, we investigated whether thrombus formation could be the cause for lethality. To this end, we injected control or BV13 antibodies into WT or ESAM^−/− mice. Thirty minutes later, mice were euthanized, the lung circulation was perfused and the tissue was fixed. Staining of lung vibratome sections for TER-119 and GPIb-V-IX revealed that erythrocytes and platelets accumulated in the lungs of BV13 treated ESAM deficient mice, but not in lungs of BV13 treated WT mice (Figure 3A). We validated the observation by electron microscopy, which showed clear signs of thrombosis at ruptured junctions which exposed freely accessible regions of the basement membrane (Figure 3B). Different stages of junction disruption and coagulation were found. Sites of loosened junctions without the typical electron-dense structure were probably representing early stages of junction rupture (first panel), whereas gaps between ECs (second panel) and finally formation of a thrombus containing activated platelets and erythrocytes (third panel) represented later stages of junction rupture and thrombus formation. Similar results were obtained from ESAM^−/− Cdh5^iECKO mice (Figure 3C).

We found previously that ESAM is not only expressed by ECs but also by platelets³⁷ and platelet ESAM was reported to limit thrombus growth and stability.⁵⁵ We, therefore, investigated whether the absence of ESAM on platelets contributed to the thrombus formation we observed in BV13 treated ESAM^−/− mice. To this end, we transplanted WT bone marrow into ESAM^−/− mice and into WT mice for controls. Platelets of the resulting mice were analyzed by immunoblotting for ESAM expression, documenting efficient repopulation of ESAM^−/− mice with WT platelets (Figure IC in the online-only Data Supplement). The transplanted mice were treated with either BV13 or control antibody, followed by staining of vibratome sections of the lungs, as described above. We found that BV13 induced thrombus formation in ESAM^−/− mice transplanted with WT bone marrow similarly as we had found before for ESAM^−/− mice, whereas no thrombus formation was found in WT mice transplanted with WT bone marrow (Figure IA in the online-only Data Supplement). Furthermore, we repeated Evans blue-based permeability assays (as in Figure 1A) and found similar results with ESAM^−/− mice transplanted with WT bone marrow as before with ESAM^−/− mice (Figure IB in the online-only Data Supplement). We conclude that blocking or deleting VE-cadherin in ESAM^−/− mice induced lethal lung thrombus formation and enhanced vascular permeability in the lung independently of ESAM on platelets.

Whereas blocking or deleting VE-cadherin selectively enhances vascular permeability in the lung and heart, deletion of ESAM only has this effect in the lung. To understand the reason for this, we analyzed the expression levels of ESAM in the lung and heart by immunoblotting. WT mice were euthanized, the vasculature was perfused and lungs and hearts were lysed and subjected to Western blotting. As shown in Figure 4A and quantified in Figure 4B, expression levels of ESAM (normalized to VE-cadherin) in the lung were almost 2-fold higher than in the heart.

To verify these results by another independent technique, we stained vibratome sections of lungs and hearts from perfused mice for ESAM and VE-cadherin (Figure 4C). In agreement with the previous immunoblot data, ESAM expression levels (normalized to VE-cadherin) in lungs were ≈2 times higher than in hearts (Figure 4D). In the course of these immunofluorescence experiments, we found that ESAM was not only restricted to ECs but was also expressed in mesothelial cells which surround the outer surface of inner organs such as lungs and hearts (Figure II in the online-only Data Supplement). Therefore, all images used to quantify ESAM expression in hearts and lungs were captured in the parenchyma of those organs to avoid the mesothelium.

As a third method, we compared the expression level of ESAM in the vasculature of lungs and hearts by fluorescence-activated cell sorter (FACS) analysis of primary isolated ECs. We determined ESAM expression levels by antibody staining of ECs which were gated as VE-cadherin⁺ and CD31^high cells. Quantification of mean fluorescence intensity revealed again a 2-fold higher expression level of ESAM on lung EC compared with heart ECs (Figure 4E and 4F). The higher expression level of ESAM in lung endothelium may partially explain why ESAM contributes more to the stability of endothelial junctions in the lung than in the heart.

Since endothelial adhesion molecules such as VE-cadherin support the expression of claudins,⁵⁶ we tested whether ESAM gene inactivation might affect the expression of other junctional molecules, which could be the reason for enhanced vascular permeability in the lung. To test this, we compared lung primary ECs (MLMVECs) isolated from WT and ESAM^−/− mice for the expression of VE-cadherin, JAM-A, claudin-5, and ZO-1 by immunoblots and immunofluorescence staining. As shown in Figure 5A and 5B, immunoblots showed no difference between these cells in the expression levels of these proteins. Likewise, their expression levels at cell junctions were unaffected by the disruption of the ESAM gene (Figure 5C).

To verify these results with primary isolated ECs (avoiding culturing of these cells), we next investigated the overall expression levels of additional junctional molecules at both the mRNA and protein level. We freshly isolated ECs from lungs of WT and ESAM^−/− mice by FACS sorting. Immediately after isolation, RNA was isolated from these cells and mRNA levels of 11 junctional molecules (JAM-A, -B, and -C; claudin-1, -3, -5, and -12; occludin; ZO-1; PECAM-1; and Cdh5) were analyzed by quantitative real-time polymerase chain reaction. In line with our results above, mRNA levels of each of these molecules were similar for both genotypes (Figure 5D). Furthermore, staining of lung vibratome sections showed that in situ distribution and expression levels of VE-cadherin and claudin-5 were indistinguishable between WT and ESAM^−/− mice (Figure 5E).

Thus, we found no evidence for a reduction of junctional proteins in lung endothelium devoid of ESAM. This suggests that the loss of junction integrity we observed in the lung of ESAM^−/− mice is directly due to the absence of this adhesion molecule.

ESAM is a homophilic cell adhesion molecule.^36,37 In addition, it contains putative binding motifs for SH3DBM (SH3-domains) and PDZDBM (PDZ-domains) in its cytoplasmic part.^36,47 To test whether any of these sites might be relevant for endothelial junction integrity, we generated knockin mice, carrying mutations affecting these putative signaling sites. For the mutation of the 3 presumptive SH3DBM (Figure 6A), coding regions for the respective amino acids were deleted in a construct for homologous recombination (Figure 6C) which was used to generate knockin mice (see Materials and Methods section). The PDZDBM of ESAM is located at the C-terminal end of the protein and binds to the PDZD protein MAGI-1.⁴⁷ This binding motif becomes masked if the fluorescent protein mCherry is fused to its C terminus (Figure 6A), as we tested in coprecipitation experiments with COS-7 cells transfected with MAGI-1 and either full-length ESAM or ESAM-mCherry (Figure 6B). Since mCherry blocked the function of the PDZDBM, we decided to generate ESAM-mCherry knockin mice by homologous recombination (see Materials and Methods section) and analyze them for the relevance of the PDZDBM for vascular integrity.

The 2 knockin mouse lines ESAM-ΔSH3DBM and ESAM-mCherry were healthy and viable. Performing immunofluorescence staining for ESAM of vibratome sections of the lung, heart, skin, and brain, of WT mice and the 2 knockin mouse lines, we found that the expression levels and junction localization of the 2 modified versions of ESAM were indistinguishable from WT ESAM (Figure III in the online-only Data Supplement). To test whether the SH3DBM or PDZDBM contributes to endothelial junction integrity in lung, heart, skin, and brain, we performed Evans blue-based vascular permeability assays in these mice as described above. As shown in Figure 6D, none of the ESAM-ΔSH3DBM and ESAM-mCherry mutant mice showed a different level of vascular permeability than WT mice. We conclude that the increased vascular permeability we observed in ESAM^−/− mice is not due to impaired signaling functions of ESAM via its PDZ domain-binding site or its putative SH3DBM motifs.

Next, we tested whether ESAM mediated adhesion could contribute to endothelial junction integrity. ESAM and JAM-A belong to the same subfamily of the large class of IG-SF membrane proteins, which share a similar domain organization with 2 Ig-like domains, a single transmembrane domain and a cytoplasmic tail that ends in a canonical PDZ domain-binding sequence.¹² Both proteins were described as homophilic adhesion molecules, with ESAM mediating cell aggregation³⁶ and JAM-A being involved in the reformation of endothelial junctions after dissociating them by Ca²⁺ chelating reagents.^31,57 Since we found that only ESAM^−/− mice but not JAM-A^−/− mice exhibited an increase in lung vascular permeability, we decided to directly compare the adhesion properties of both adhesion molecules in cell aggregation assays. To this end, we stably transfected CHO cells with ESAM or JAM-A and isolated cell clones expressing either ESAM or JAM-A at similar levels (Figure 6E). After 1 hour of incubation, ESAM-CHO cells were significantly aggregated when compared with untransfected CHO cells (Figure 6F). In contrast, JAM-A-CHO cells aggregated only slightly, but insignificantly better than CHO cells (Figure 6F). Thus, ESAM mediates cell contact formation and supports cell contacts more efficiently than JAM-A.

If ESAM directly contributes to EC contact stability by mediating homophilic cell adhesion, anti-ESAM antibodies should be able to interfere with endothelial junction integrity. Indeed, we found that electrical resistance of MLMVEC monolayers was reduced by 50 μg/mL polyclonal antibodies (incubated for 2 hours) against the extracellular part of ESAM, while negative control antibodies from the corresponding preserum and antibodies against endomucin had no such effect (Figure 6G). In line with our in vivo results (Figure 1C), the mAb BV13 against VE-cadherin reduced electrical resistance across MLMVEC monolayers more strongly than the anti-ESAM antibodies (Figure 6H).

Collectively, our results suggest that ESAM could support endothelial junction integrity directly by its adhesive function. Although its contribution to junction stability is less dominant than the contribution of VE-cadherin (Figures 1C and 6H), the absence of ESAM causes immediate lethality even if VE-cadherin is only partially blocked.