Endothelial CD47 promotes Vascular Endothelial-cadherin tyrosine phosphorylation and participates in T-cell recruitment at sites of inflammation in vivo

Mice

CD47 knockout (CD47−/−) mice (C57BL/6 strain) have been previously described (21) and were obtained from Dr. Eric Brown (Genentech Inc., San Francisco, CA). WT C57BL/6 mice were purchased from Charles River Laboratories (Wilmington, MA) and a breeding colony was established in our facility for use as WT control animals. Mice were maintained in a specific pathogen-free barrier unit at our institution. Animal care and experimentation were in accordance with the National Institute of Health Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee.

Reagents and antibodies

Recombinant human TNF-α and CXCL12 were from PeproTech (Rocky Hill, NJ). Recombinant murine TNF-α was purchased from BioLegend (San Diego, CA). The Src kinase inhibitor PP2 and the p38 MAP kinase inhibitor SB203580 were from Calbiochem (San Diego, CA). Pertussis toxin (PTX) was purchased from Sigma-Aldrich (St. Louis, MO). The following mAbs were used as purified IgG: mAb B6H12 is a function blocking mAb to human CD47 and mAb TS1/22 is a function blocking mAb against αLβ2 integrin (American Type Culture Collection, Manassas, VA); mAb 2D3 is a non-blocking anti-CD47 mAb (28) obtained from Dr. Eric Brown; unlabeled and FITC-labeled rat anti-mouse CD47 function blocking mAb (miap301) was from BioLegend (San Diego, CA); anti-VE-cad mAb Hec-1 (29) was from Dr. William Muller (Northwestern University, Chicago, IL); anti-HLA Class I (Class 1 MHC) mAb W6/32, anti-ICAM-1 mAB Hu5/3, and function blocking anti-JAM-A mAb 2A9 were as previously described (30). The blocking mAb to α4 HP2/1 was from Immunotech, Inc (Glendale, CA). The anti-phosphotyrosine specific mAb 4G10 was from Millipore (Billerica, MA). Phospho-specific polyclonal Abs to VE-cad (pY658) and (pY731), and Pyk2 (pY402) were from Biosource International (Camarillo, CA). Phospho-specific and polyclonal anti-Src and anti-p38 MAPK Abs were from Cell Signaling Technology, Inc. (Danvers, MA). FITC-conjugated markers for CD3 (145-2cl), monocytes and macrophages (APC-conjugated anti-F4/80), neutrophils (APC-conjugated GR-1 antibody), and isotype control mAbs were from BD Biosciences (San Jose, CA). Alexa-488- or FITC-conjugated F(ab)₂ goat anti-mouse IgG and phalloidin-Alexa 488 were from Invitrogen (Carlsbad, CA).

In vivo transmigration assay (subcutaneous air pouch model)

Air pouches were created in the dorsal portion of the back of 8–10 week old C57BL/6 WT and CD47−/− male mice as previously described (31). PBS or PBS containing murine recombinant TNF-α (500ng) was injected in each air pouch, and 4 or 24 h later cell infiltrates were harvested by lavage with repeated PBS washes. The recovered volume was measured and the number of recovered cells was determined by hemocytometer. The frequency of CD3+ T-cells, neutrophils and monocyte/macrophages were determined by staining with primary labeled mAb specific for these cell types and acquired on a FACScalibur flow cytometer (BD Biosciences). FACS data were analyzed using FlowJo software (Tree Star Inc., Ashland, OR).

Bone marrow (BM) transplantation protocol

Donor BM cells were harvested aseptically from femurs and tibias of 8–10 weeks old WT mice. Recipient CD47−/− and WT male mice (8 weeks old) received a lethal dose of whole body irradiation with a cesium source (1200 rads in 2 doses, 4 h apart) as described previously (32). Donor BM cells (2×10⁶ cells in 0.3 ml, 3% FBS in PBS) were transfused by tail vein injected into irradiated WT or CD47−/− recipient mice. Recipient mice received normal chow and water containing antibiotics (Sulfatrim) 1 week prior to and 6 weeks after transplantation. Air pouch leukocyte recruitment studies were performed 8–9 weeks post-BM transplantation. To confirm BM engraftment, leukocytes from chimeric and WT mice were stained with FITC rat anti-mouse CD47. Expression levels were acquired on a FACScalibur flow cytometer and analyzed using FlowJo software.

Intravital microscopy (IVM) of leukocyte recruitment during microvascular inflammation in vivo

IVM of leukocyte transmigration of postcapillary venules of the mouse cremaster muscle was performed at 2 hr following intrascrotal injection of mouse TNF-α (500 ng in saline/mouse) as initially described (33). Mice were anesthetized, and surgical exteriorization of the cremaster muscle was performed as we have recently described (34). Microvessel data and physiology indices were obtained using a specialized Olympus FV 1000 intravital microscope (Olympus, Center Valley, PA) fitted with an Olympus 403 water immersion objective (Olympus) (34). Leukocyte transmigration events were recorded from eight to twelve vessels per mouse using an Olympus DP71 CCD video camera and Olympus FluoView 1000 imaging software (Olympus). Videos were analyzed offline with the National Institutes of Health software package ImageJ (National Institutes of Health, Bethesda, MD). The number of transmigrated leukocytes per vessel was determined by counting the average number of perivascular cells in a 50 × 100 μm area adjacent to the vessel wall, as previously described (34,35).

Measurement of leukocytes in murine peripheral blood

Peripheral blood from mice was collected in EDTA-coated vials (Sarsdedt, Germany) at 4 and 24 h after injections of PBS or PBS-TNF-α. A complete blood count with cellular differential was assessed by automated multispecies hematology instrument (Hemavet 950FS, Drew Scientific, Oxford, CT).

Isolation of human endothelial cells and T-cells and TEM under flow conditions

Human CD3⁺ T-cells (>95% purity) were isolated by negative selection from anticoagulated whole blood obtained from healthy volunteers as previously described (17). Blood was obtained from volunteer donors according to Brigham and Women’s Hospital IRB approved protocols for protection of human subjects, and all volunteer subjects gave informed consent, in accordance with the Declaration of Helsinki. Pooled HUVEC were isolated and cultured as described (17). Confluent HUVEC monolayers were used in all biochemical and TEM studies. Confluent HUVEC monolayers on fibronectin-coated glass coverslips were stimulated with TNF-α (4 h, 25 ng/ml) followed by treatment with CXCL12 (50 ng/ml) for 15 min, a step that promotes T-cell TEM, prior to insertion into the flow chamber thermostated to 37°C. T-cells (1×10⁶ in 0.1 ml) were drawn across HUVEC monolayers at 0.8 dynes/cm² and the accumulation/mm² and TEM was measured as described (17). The percent of TEM was calculated as: total transmigrated T-cells/[total adhered+transmigrated T-cells] × 100.

RNA interference

Predesigned siRNA against CD47 was obtained from Qiagen (Valencia, CA). The silencing siRNA target sequences (catalog #SI04894582 and #SI03117401) were CCCTTACGTGATTGTTAGTTA and TGAAACGATCATCGAGCTAAA. Both reduced surface expressed CD47 by 70–80%. A predesigned siRNA oligonucleotide sequence targeting CD47 but with no inhibitory effect on CD47 expression (target sequence, CACGATAAGTTTACTCCCCCA (cat# SI00028168) was used as a negative control. HUVEC were transfected with siRNAs (100 nmol/L) using Oligofectamine in Optimem sera free culture medium (Invitrogen). After 92 h, HUVEC were stimulated with TNF-α (25 ng/ml, 4 h) and used in studies.

Immunofluorescence microscopy of CD47 expression and phalloidin staining

HUVEC monolayers transfected with different siRNAs were fixed, stained with B6H12 anti-CD47 mAb, and primary mAb was detected by staining with Alexa-488 conjugated goat anti-mouse IgG as previously described (36). Confluent resting or 4 h TNFα-activated HUVEC on fibronectin-coated coverslips were treated with anti-CD47 mAb (B6H12) (30 μg/ml, 30 min), washed, and then goat anti-mouse secondary mAb (10 μg/ml) was used to crosslink CD47 mAb for 0, 5, 10, 20 and 30 min. As a control, HUVEC were incubated with secondary mAb only. HUVEC were fixed, permeabilized and endogenous F-actin was detected by staining with Alexa-568-phalloidin according to the manufacturer’s protocol (Invitrogen, Inc.). Actin stress fiber formation was quantified in >100 cells in multiple fields at the 30 min time point. Cells with 5 or more stress fibers were considered positive. Results are representative of 4 or more separate experiments.

Flow cytometry and analysis

Confluent resting or 4 h TNF-α stimulated HUVEC monolayers were trypsinized and stained for expression of CD47 and ICAM-1 using B6H12 and Hu5/3 mAbs, respectively, followed by FITC-tagged secondary F(ab′)₂ mAb as previously described (17). Expression levels were acquired on a FACScalibur flow cytometer and analyzed using FlowJo software.

CD47 antibody crosslinking studies and western blotting

Twenty-four h TNF-α stimulated HUVEC in 35mm diameter wells (6-well culture plate, Corning Inc. Life Sciences, Lowell, MA) were serum starved for 2 hr in OPTI-MEM I (Gibco/Invitrogen, Inc) and then incubated for 10 min at 37°C with Dynabeads (Invitrogen Dynal AS, Oslo, Norway) prebound to mAb to CD47 (B6H12), ICAM-1 (Hu5/3), Class 1 MHC, or murine IgG Ab (10 μg/ml), using a ratio of 10 beads per cell. The HUVEC were washed twice in ice cold PBS to remove non-adherent beads and immediately lysed in 250 μl of boiling sample buffer under reducing conditions. Cell lysates were diluted to allow immunoprecipitation with anti-phospho-tyrosine mAb 4G10 as described previously (8,11). Recovered proteins were separated by SDS-PAGE, transferred to nitrocellulose membranes and western blotted for phospho-Src (Tyr416), phospho-Pyk2 (Tyr402), phospho-VE-cad (Tyr658) and phospho-VE-cad (Tyr731) and detected by an enhanced chemiluminescence (Pierce Biotechnology, Rockford, IL) following incubation with appropriate HRP conjugated secondary antibodies. Band intensity was quantified by NIH Image J software.

Statistical Analysis

Data are expressed as mean ± SEM and were compared using Student’s t test for experiments with two groups or one-way ANOVA with Tukey multiple-comparison posttest for experiments with three or more groups by Prism software v5.00 for Windows (GraphPad, San Diego, CA). Data were considered statistically significant at P < 0.05.

CD47^−/− mice exhibit defective leukocyte recruitment to TNF-α in a dermal air pouch model of inflammation

CD47 has been shown previously to play a role in T-cell recruitment in vitro under physiologically relevant shear flow conditions (17,37). Here, the involvement of CD47 in regulating TNF-α induced leukocyte recruitment was investigated using a skin air pouch model used by other investigators to examine the contribution of adhesion molecules and chemokines to leukocyte recruitment (31,38). In WT mice, TNF-α triggered increased leukocyte accumulation in air pouches at 4 and 24 h (Figure 1A). Flow cytometric analysis showed that recruited leukocytes consisted of neutrophils, T-cells, and monocyte/macrophages (Figure 1B–D). In contrast to WT mice, recruitment of each leukocyte type in CD47−/− mice was dramatically reduced at both 4 and 24 h. The peripheral blood leukocyte counts at 0, 4, 24 h were similar in both lines of mice (Figure 1E). Consistent with a previous study the percent of lymphocytes in blood was reduced by 22% in CD47−/− compared to WT as shown in Table I (21,39). We conclude that CD47 plays an important role in the recruitment of blood T-cells, neutrophils and monocytes in this murine model of inflammation.

We next addressed the role of BM and host expressed CD47 in leukocyte recruitment in the air pouch model by comparing CD47−/− and WT mice reconstituted with WT BM (WT BM → CD47−/−; WT BM → WT mice). The reciprocal reconstitution of CD47−/− BM → WT is not feasible because transfused CD47−/− BM cells are cleared rapidly from the circulation by splenic macrophages in WT animals, and these mice do not survive (40). Reconstituted WT BM → CD47−/− mice had comparable levels of CD47 expression on leukocytes as WT BM → WT animals. Analysis of TNF-α induced leukocyte accumulation in these animals showed that reconstitution of CD47−/− mice with WT BM (WT BM → CD47−/−) does not rescue the defect in recruitment of neutrophils, T-cells or monocytes in response to TNF-α at both 4 and 24 h (Figure 2A–D) Since WT BM → CD47−/− chimeric animals have CD47+/+ hematopoietic cells but lack CD47 in parenchymal cells, these data indicate host cell CD47, presumably the vascular endothelium, is necessary for leukocyte recruitment in vivo.

IVM studies in CD47−/− BM chimera mice support a role for endothelial CD47 in leukocyte TEM

To directly visualize leukocyte TEM in vivo, we performed IVM experiments in 2 h TNFα-activated cremaster muscle microcirculation of WT BM→ CD47−/− chimera mice. Hemodynamic parameters and estimated wall shear rates were similar among the animals after TNF-α treatment (data not shown). As shown in Figure 2E the number of emigrated leukocytes in WT BM → CD47−/− chimera animals were significantly less as compared to the WT BM→WT chimera animals. Representative images of a segment of venules in WT and CD47−/− animals depicting transmigrated leukocytes are shown (Figure 2F,G). The results in the cremaster model indicate that endothelial cell CD47 contributes directly to leukocyte TEM in vivo.

Both human endothelial and leukocyte CD47 are required for T-cell TEM

Human CD3⁺ T-cells (95 ± 11 % purity) arrest uniformly on 4 or 18 h TNF-α or IL-1β activated HUVEC and transmigrate at cell junctions in the presence of exogenously added apical chemokine CXCL12 under shear flow conditions (17,41). Pretreatment of TNF-activated HUVEC with blocking mAb to CD47 had no significant effect on T-cell accumulation (Figure 3A), but significantly reduced T-cell TEM (Figure 3B) as compared with control mAb to MHC Class I. Function blocking mAb to ICAM-1 in HUVEC (Figure 3B), or to αLβ2 integrin, its ligand on T-cells also significantly reduced T-cell TEM whereas anti-α4 integrin blocking mAb did not (Supplemental Figure 1), which is consistent with a previous report (41). Next, the effect of CD47 blockade in T-cells was examined in parallel. Treatment of T-cells with anti-CD47 mAb significantly reduced T-cell transmigration without altering T-cell adhesion. mAb blocking of CD47 in both HUVEC and T-cell was no more effective than blocking either cell type alone. These data demonstrate that both endothelial and T-cell CD47 play an important role in T-cell TEM in vitro.

siRNA knockdown of CD47 in HUVEC reduces T-cell TEM under flow

To corroborate CD47 mAb blocking of TEM in Figure 3, siRNA silencing of CD47 in HUVEC was performed. Silencing of CD47 by two different siRNA caused a 70 – 80% reduction (n = 4 exp) in surface CD47 staining in resting (data not shown) and 4h TNF-α activated HUVEC (Figure 4A,B) as compared to control (non-silencing) siRNA targeting of CD47. As observed with CD47 mAb blocking, CD47 siRNA significantly reduced T-cell TEM and also did not alter T-cell accumulation (Figure 4C,D). However, HUVEC transfected with control targeting CD47 siRNA but with no inhibitory effects supported T-cell adhesion and robust TEM (Figure 4C,D). Live cell DIC imaging of TEM showed that adherent T-cells polarize and transmigrate in control siRNA-treated TNF-activated HUVEC whereas treatment of HUVEC with CD47 specific siRNA inhibited T-cell TEM (Figure 4E, arrowheads identify non-transmigrated T-cells; arrows, transmigrated T-cells). To control for off-target effects of siRNA, we observed that silencing CD47 siRNA and non-silencing control CD47 siRNA treated HUVEC showed no difference in the expression of ICAM-1 (Figure 4F), or in surface expression of VE-cad, VCAM-1, or E-selectin in 4 h TNF-α activated HUVEC (data not shown).

Pertussis toxin treatment of endothelial cells does not inhibit T-cell TEM under shear flow conditions

CD47 has been reported to signal through a heterotrimeric G_i coupled receptor complex in a variety of cell types (14,15,42). To examine whether CD47 signals through the G_αi pathway in endothelium during TEM, we preincubated HUVEC with different concentrations of pertussis toxin (PTX) for 5 h to inactivate G_αi subunits and assessed T-cell adhesion and TEM (Figure 5A,B). PTX pretreatments of HUVEC had no inhibitory effect on T-cell adhesion or TEM, whereas PTX treatment of T-cells significantly blocked transmigration (Figure 5C), consistent with earlier reports (41,43,44). Videomicroscopy of the TEM assay revealed that adherent T-cells flatten, polarize, and transmigrate in PTX-treated HUVEC (Figure 5D, left panel; arrowheads). In contrast, pretreatment of T-cells with PTX abrogated T-cell TEM and most adherent T-cells did not flatten or migrate on the HUVEC monolayer (Figure 5D, right panel; arrowheads) Because PTX treatment of rodent endothelium has variable inhibitory effects, ranging from no effect in T-cell TEM of rat HEV (43) to 80% inhibition of T-cell TEM of murine brain EC in vitro (45) or neutrophil and eosinophil TEM of endothelial monolayers isolated from G_αi −/− mice (46), we tested the effects of PTX in endothelium derived from human dermal microvessels (HDMEC) and saphenous vein (HSVEC). Consistent with the results in HUVEC, T-cell adhesion or TEM across either HDMEC or HSVEC was not inhibited by PTX pretreatment (Supplemental Figure 2 A, B). Treatment of HUVEC with PTX had no effect on the TNF-α-induced increase in ICAM-1, E-selectin or VCAM-1 (data not shown) and the treated and control monolayers maintained the same “normal” cobblestone morphology throughout the study. In addition, PTX-treatment of HUVEC had no effect on neutrophil TEM under flow conditions (data not shown). These results suggest a CD47 dependent mechanism that is independent of G_αi protein signaling in human endothelium during TEM.

Cross-linking of endothelial cell CD47 triggers robust cytoskeletal remodeling and tyrosine phosphorylation of VE-cad

Rapid remodeling of the endothelial actin cytoskeleton and increased phosphorylation of tyrosine residues 658 and 731 in the cytoplasmic domain of VE-cad are necessary events in T-cell and neutrophil TEM (8,10,11). Our lab has previously reported (9) that mAb crosslinking of ICAM-1 leads to a rapid increase in actin cytoskeletal remodeling in HUVEC. To determine if engaging endothelial CD47 triggers remodeling of the actin cytoskeleton, we preincubated HUVEC with an anti-CD47 mAb for 10 min, followed by a secondary mAb for different periods of time (Figure 6A). Crosslinking of anti-CD47 mAb triggers robust stress fiber formation, as assessed by phalloidin staining, as early as 5 min with most cells responding by 30 min (Figure 6A, right panel). In the absence of CD47 crosslinking, or cross-linking of endothelial Class I MHC mAb (control) HUVEC showed few stress fibers (data not shown).

ICAM-1 crosslinking by anti-ICAM-1 mAb or mAb coated polystyrene beads are well characterized approaches that mimic ICAM-1 binding by leukocytes. We next assessed the effects of incubating 24 h TNF-α activated HUVEC with 3 μm Dynal beads coated with either anti-CD47 or anti-ICAM-1 mAbs and found a 2 to 3-fold increase in phosphorylation of tyrosine 658 and 731 in the VE-cad cytoplasmic tail (Figure 6B,C) as compared to medium or anti-Class I MHC mAb coated beads in both cases. Class I mAb coated beads bind equally as well as CD47 mAb coated beads to HUVEC monolayers, but did not trigger increased phosphorylation of VE-cad tyrosine 658 or 731. Previous studies have shown that engagement of ICAM-1 activates Src and Pyk2 family kinases, which lead to tyrosine phosphorylation of several endothelial cell proteins including VE-cad (8,10,36). To determine whether engaging CD47 also induced Src and Pyk2 kinase activation, we challenged CD47 and ICAM-1 or as control Class I MHC with appropriate antibody-coated beads. Western blot analysis of cell lysates with phospho-specific Abs directed against activation induced phospho-tyrosine pY416 of Src or pY402 of Pyk2 were performed. Both anti-CD47 and anti-ICAM-1 mAb coated beads triggered a 2-fold increase in phosphorylation of both Src and Pyk2 (Figure 7A). In contrast, Class I mAb beads did not alter phosphorylation of either Src or Pyk2.

We next used a pharmacological approach to corroborate involvement of Src and Pyk2 family kinases or pertussis toxin to block G_αi coupled signaling. Since ICAM-1 cross-linking by mAbs is known to induce MAP kinase activation (reviewed in (47)), we also used a p38 MAP kinase inhibitor to test its effects on ICAM-1- and CD47-induced VE-cad phosphorylation. Interestingly, ICAM-1- and CD47-induced VE-cad phosphorylation was not prevented by pretreatment of HUVEC with either PTX or the p38 MAP kinase inhibitor, SB203580 (Figure 7B). However, pre-treatment of cells with Src kinase inhibitor, PP2 significantly inhibited both ICAM-1- and CD47-induced VE-cad phosphorylation at tyrosines 658 and 731 (Figure 7B).

CD47−/− animals have defects in leukocyte recruitment

A murine dermal air pouch was employed as a model of TNF-α induced skin inflammation to study the role of CD47 in leukocyte extravasation (31,38). The data show for the first time that injection of TNF-α into the dorsal site of WT mice induced a time-dependent accumulation of neutrophils, T-cells and monocytes in the air pouch exudates whereas recruitment of leukocytes in CD47−/− mice was strikingly reduced and nearly absent. Analysis of bone marrow chimera mice indicates that endothelial cell CD47 plays a significant role in leukocyte recruitment. In addition, the analysis of inflamed cremaster muscle microcirculation in WT and CD47−/− BM chimera animals by intravital microscopy revealed an important role of endothelial cell CD47 in TEM. These in vivo observations are consistent with our in vitro human T-cell TEM results that show function blocking mAb to CD47 on T-cells or human endothelial monolayers significantly reduced T-cell TEM (Figures 3, 4, and supplemental Fig 2E and 5F). Silencing of CD47 by siRNA in HUVEC strongly reduced T-cell TEM but not adhesion under flow conditions in vitro and thus corroborate the function blocking mAb blocking results. We conclude that endothelial CD47 plays a key role in the transmigration of multiple leukocyte types to sites of TNF-α induced skin inflammation and in vitro models of TEM. CD47 on T-cells appears to also contribute to TEM in vitro, but the in vivo relevance of this could not be addressed since the reciprocal bone marrow reconstitution of WT mice with BM from CD47−/− animals is not feasible due to clearance of these cells by splenic macrophages (40).

Our data demonstrate an important role for CD47 in TEM by both the T-cell and endothelium. In T-cells CD47 provides several stimulatory signals (reviewed in (14)). In particular CD47 was reported to participate in TCR-mediated Jurkat T-cell spreading, and in Jurkat T-cell adhesion to an endothelial cell line (EA.hy926) or VCAM-1 under shear flow or static assay conditions through regulation of VLA-4 integrins (37,55). The precise role that T-cell CD47 plays in adhesion and TEM in vivo or in vitro, however, is as yet unexplored and hence will require future in depth studies. Interestingly, the combined mAb blocking of T-cell and HUVEC CD47 was not additive and this was unexpected. The reason for this result is not immediately apparent but may be due to the complexity of CD47 interactions with its endothelial and leukocyte ligands TSPs and SIRPs and with its well characterized regulation of multiple integrins. Future studies will be required to address these questions.

Two other studies have evaluated the contribution of CD47 to T-cell recruitment in a dermal contact hypersensitivity (CHS) model of inflammation. The level of inflammation in CD47−/− animals as compared to WT mice was similar (no reduction) in one CHS study (56) and significantly elevated and also protracted in a second study (24). The CHS, however, is complex and neutrophils and NK cells as well as T-cells are present (57). While this might seem at odds with our air pouch data, one explanation is that the mechanism(s) of leukocyte recruitment depend on the nature of the inflammatory stimulus (i.e., induction of different chemokines or cytokines) and/or that recruited T-cells may be different in the CHS site versus the dermal air pouch, and one subset is more dependent on CD47 than the other.

Human CD47 binds to thrombospondins, SIRPα and SIRPγ. The interaction of human endothelial CD47 with SIPRγ expressed on T-cells is important for T-cell TEM in vitro assays using HUVEC. In rat, endothelial CD47 has been shown to interact with SIRPα-expressing monocytes and this pathway was critical for monocyte adhesion and TEM. Currently a rat or mouse SIRPγ ortholog has not yet been described or predicted from searches of genomic or EST sequence databases. However, there is no evidence that rodent T-cells lack receptors for CD47. To the contrary, functional assays, as those described herein this study, indicate the presence of a CD47 ligand on T-cells whose identity remains to be determined.

Endothelial CD47 signaling in leukocyte TEM is G_αi independent

CD47 has been shown to participate in the regulation of cell-cell adhesion and cell migration through reorganization of the actin cytoskeleton in epithelial cells, a process mediated by activation of the MAP kinase pathway (58). Several studies reported that “cis” interactions between CD47 and integrins recruit a pertussis toxin-sensitive G_αi protein-thrombospondin complex (reviewed in (14)). Of relevance to our study, treatment of rodent endothelium with PTX resulted in reduced T-cell transmigration; the level of inhibition was variable and ranged from no effect in a rat HEV system (43) to an 80% reduction in murine brain endothelial cell migration system (45). In G_αi−/− animals, leukocyte recruitment in an allergen-induced lung model was significantly impaired and was dependent on endothelial cell G_αi dependent signaling (46). In these models and in models of leukocyte chemotaxis (41), PTX pretreatment essentially blocked >80% of leukocyte migration indicating a critical role of T-cell G_αi proteins. However, the role of G_αi proteins in TEM of human endothelial cells has not been reported. Here we observed that PTX pretreatment of endothelium isolated from three different human vascular beds had no effect on T-cell adhesion or TEM, indicating that G_αi proteins in human endothelial cells are not involved in TEM.

CD47 engagement induces actin stress fibers and activates Src and Pyk2 to phosphorylate VE-cad

Previous studies have shown that ICAM-1 cross-linking triggers robust actin stress fiber formation in HUVEC by activating Rho A (59) and induces tyrosine phosphorylation of cortactin (9,36,60,61) and VE-cad (8,10,11), and inhibition of these events significantly reduces leukocyte TEM (reviewed in (6,7)). Engagement of ICAM-1 also activates Pyk2, which collaborates with Src protein kinases to phosphorylate tyrosine 658 and 731 in the cytoplasmic tail of VE-cad (8). Tyrosine phosphorylation of VE-cad is widely thought to facilitate dissociation of junctional proteins during paracellular migration of leukocytes. Previous studies by our lab showed that inhibition of Src leads to reduced leukocyte TEM (36). Our results show that like ICAM-1, CD47 ligation triggers actin cytoskeletal remodeling and Src and Pyk2 mediated VE-cad tyrosine phosphorylation in HUVEC. CD47-induced VE-cad phosphorylation was significantly inhibited by a Src inhibitor, but was not affected by PTX or by a p38 MAPK inhibitor. Taken together, Src can act as the main regulator of CD47- and ICAM-1-induced phosphorylation events required for TEM in HUVEC.

In conclusion our in vivo and in vitro studies show for the first time a crucial role for endothelial CD47 in T-cell transmigration and in recruitment of T-cells, neutrophils, and monocytes to extravascular sites of inflammation. We also identify signaling pathways mediated by CD47 that contribute to leukocyte TEM. Uncontrolled recruitment of leukocytes and increased endothelial cell permeability is observed in various pathologies, such as psoriasis, autoimmune diseases, multiple sclerosis, and rheumatoid arthritis. We propose that CD47 and/or the CD47 pathway downstream could serve as potential future targets for the treatment of such inflammatory conditions.