CD31 signals confer immune privilege to the vascular endothelium

Abstract

Constitutive resistance to cell death induced by inflammatory stimuli activating the extrinsic pathway of apoptosis is a key feature of vascular endothelial cells (ECs). Although this property is central to the maintenance of the endothelial barrier during inflammation, the molecular mechanisms of EC protection from cell-extrinsic, proapoptotic stimuli have not been investigated. We show that the Ig-family member CD31, which is expressed by endothelial but not epithelial cells, is necessary to prevent EC death induced by TNF-α and cytotoxic T lymphocytes in vitro. Combined quantitative RT-PCR array and biochemical analysis show that, upon the engagement of the TNF receptor with TNF-α on ECs, CD31 becomes activated and, in turn, counteracts the proapoptotic transcriptional program induced by TNF-α via activation of the Erk/Akt pathway. Specifically, Akt activation by CD31 signals prevents the localization of the forkhead transcription factor FoxO3 to the nucleus, thus inhibiting transcription of the proapoptotic genes CD95/Fas and caspase 7 and de-repressing the expression of the antiapoptotic gene cFlar. Both CD31 intracellular immunoreceptor tyrosine-based inhibition motifs are required for its prosurvival function. In vivo, CD31 gene transfer is sufficient to recapitulate the cytoprotective mechanisms in CD31(-) pancreatic β cells, which become resistant to immune-mediated rejection when grafted in fully allogeneic recipients.

Keywords: endothelium; immunology; inflammation; lymphocytes; transplantation.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

CD31 protects from extrinsic apoptosis. (A and B) WT and CD31-KO ECs were exposed to TNF-α (50 ng/mL) for 6 h, and apoptosis was measured by TUNEL assay. (C) Alternatively, WT ECs were pretreated with either a blocking anti-CD31 or an isotype control antibody before exposure to TNF-α. (D and E) Male-derived murine ECs (pretreated with IFN-γ for 48 h to up-regulate MHC molecule expression) were incubated for 6 h with HY-specific CTLs (EC:CTL ratio 1:5), which were removed at the end of the incubation by gentle washing with warm PBS. EC death was measured by TUNEL assay and TpB exclusion assay. (F) ECs also were cultured in 10% (vol/vol) human serum as a source of antibody and complement, and cell death was measured by TpB exclusion assay. Representative images are shown in A and D. (Scale bars: 100 μm.) The mean percentages of apoptotic (B, C, and E) and necrotic (F) cells in three independent experiments (±SD) are shown. ***P < 0.001, ****P < 0.0001. NT, not treated.

Fig. S1.

Characterization of an anti-CD31 blocking antibody. (A) Subconfluent WT and CD31-KO EC monolayers were fixed in 4% paraformaldehyde (pH 7.4) for 15 min at room temperature. Following permeabilization and blocking, the slides were coincubated with an anti-CD31 antibody (rat anti-mouse CD31; clone 390; eBioscience) for 4 h. After three washings with PBS, the slides were incubated with anti-mouse Alexa Fluor 488 in the dark for 2 h. Nuclei were counterstained with DAPI. (Scale bars: 20 μm.) (B) Immunoprecipitation of CD31 molecules from ECs exposed to PBS and TNF-α (50 ng/mL) for 20 min at 37 °C followed by immunoblotting with an anti-phosphotyrosine antibody. An isotype control (IsC) antibody or the rat anti-mouse CD31 blocking antibody clone 390 was added at a concentration of 10 μg/mL for 30 min at room temperature before incubation with TNF-α.

Fig. 2.

Antiapoptotic pathways activated by CD31 in ECs. (A) Immunoprecipitation of CD31 molecules from ECs exposed to TNF-α for 20 min followed by immunoblotting with an anti-phosphotyrosine antibody and an anti-SHP2 antibody. (B–D) Erk (B and C) and Akt (B and D) activation in WT and CD31-KO ECs exposed to TNF-α; time points are indicated. (E) WT ECs were treated with an Akt inhibitor (3 μM) or with a MEK 1/2 inhibitor (10 μM) for 4 h at 37 °C before incubation with TNF-α. EC apoptosis was measured after 6 h by TUNEL assay. (F) Akt activation in WT ECs pretreated with a blocking anti-CD31 or an isotype control antibody and exposed to TNF-α for 30 min. In C–E the mean percentage of apoptotic cells in three independent experiments (±SD) is shown. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. S2.

Apoptosis-related gene profile of WT and CD31-KO ECs before and after exposure to TNF-α. (A and B) Volcano plots showing the expression of apoptosis-related genes by WT (A) and CD31-deficient (B) ECs following exposure to TNF-α. Plots were created using the RT2 profiler apoptosis array as described in Methods in the main text. mRNA expression levels were normalized to untreated control cells. Based on pairwise comparisons of TNF-α–treated WT and CD31-KO ECs (see Gene Array Dataset_CD31KO and _WT), the TNF family DR CD95/Fas, the executioner caspase-family member caspase 7, and the antiapoptotic gene cFlar were selected for further investigation (circled). (C) Summary of differentially regulated genes in WT and CD31-KO ECs stimulated with TNF-α as compared with untreated counterpart ECs. Fold-changes greater than 2 are indicated in red; fold-change values less than 0.5 are indicated in blue. ✔ Indicates genes chosen for further analysis. *Fas up-regulation was significantly higher in CD31-KO ECs (see Gene Array Dataset_CD31KO and _WT).

Fig. 3.

CD31 modulates the proapoptotic transcriptional program downstream of DR signaling. (A–C) Expression of Fas mRNA and surface receptor by WT and CD31-KO ECs either untreated or exposed to TNF-α for 6 h as quantified by qRT-PCR (A) and flow cytometry (B and C). (D–F) Expression of caspase 7 mRNA and caspase 3/7 and caspase 8 activity by WT and CD31-KO ECs, either untreated or exposed to TNF-α as measured by qRT-PCR (D), ApoTox-Glo Triplex Caspase 3/7 (E), or Caspase 8 Reagent (F). In A and C–F cumulative data from three independent experiments (±SD) are shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. MFI, mean fluorescence intensity.

Fig. S3.

CD31 does not affect the expression of PDL-1 or Spi6 or antioxidant activity by ECs. (A) Surface expression of the inhibitory coreceptor PDL-1 by untreated or TNF-α–treated WT and CD31-KO ECs was assessed by flow cytometry. (B) Expression of Spi6 mRNA by WT and CD31-KO ECs, either untreated or exposed to TNF-α, was quantified by qRT-PCR. Cumulative data from three independent experiments (±SD) are shown. ***P < 0.01, ****P < 0.0001. (C) EC monolayers were incubated with 25 μM 2′,7′-dichlorofluorescein diacetate (DCF-DA) for 45 min at 37 °C in the dark. The cells subsequently were incubated with TNF-α (50 ng/mL) for 3 h. The level of intracellular ROS was evaluated by measuring DCF-DA oxidation using the DCF-DA Cellular ROS Detection Assay Kit (ab113851; Abcam) according to the manufacturer’s instructions. Oxidation was measured as fluorescence intensity using a microplate reader (ARVO-X5 2030 Multilabel Reader; Perkin-Elmer) with excitation at 485 nm and emission at 535 nm. Data are indicated as the mean relative ratio of intensity (stimulated with TNF-α vs. unstimulated) ± SD measured in two independent experiments of identical design. **P < 0.01.

Fig. 4.

The antiapoptotic gene cFlar is instrumental to CD31-mediated cytoprotection. mRNA was isolated from CD31-KO or WT ECs either untreated or stimulated with TNF-α (50 ng/mL) for 6 h. (A and B) cFlar mRNA quantification (A) and protein expression (B) by WT and CD31-KO ECs after incubation with TNF-α. (C and D) cFlar expression by ECs following shRNA knockdown was evaluated by qRT-PCR (C) and immunoblotting (D). (E) Apoptosis of ECs exposed to TNF-α following cFlar knockdown measured by TUNEL assay. In some cFlar-silenced cells, CD31 was stimulated by antibody ligation. pLKO.1 indicates cells transduced with the empty vector backbone. (F) WT ECs were treated with an Akt inhibitor (3 μM) or a MEK1/2 inhibitor (10 μM) before incubation with TNF-α. Expression of cFlar was measured after 6 h. (G and H) EC cultures were fixed, and FoxO3a intracellular localization was assessed by immunofluorescence staining. Nuclei were stained with DAPI. Representative images are shown in G. (Scale bars: 50 μm.) H shows the quantification of immunofluorescence staining patterns for endogenous FOXO3 (n = 100 cells per experiment). The data shown in A, C, E, and H are the mean value (±SD) of three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 5.

Both CD31 ITIMs are required for its prosurvival activity. CD31 gene constructs with mutations leading to the loss-of-function amino acid substitutions Y663F and Y686F in the ITIMs were generated and expressed by lentiviral transduction into CD31-KO ECs (CD31Y663F and CD31Y686F). As a control, CD31-KO ECs were transduced with a WT CD31 gene construct or an empty plasmid (mock). (A) Surface expression of CD31Y663F and CD31Y686F molecules following transduction of CD31-KO ECs. As a control, WT CD31 and empty plasmid (mock) constructs were used. (B and C) Apoptosis of mock-, WT-, CD31Y663F-, and CD31Y686F-transduced CD31-KO ECs either exposed to TNF-α for 6 h or cocultured with antigen-specific CTLs for 6 h, as measured by TUNEL assay. (D) Complement-mediated lysis of the indicated EC populations. (E–G) Erk and Akt activation in ECs transduced with the indicated constructs following exposure to TNF-α. (H–M) CD95/Fas surface expression (H and I), caspase 7 expression and activity (J and K), caspase 8 activity (L), and cFlar protein expression (M) in the indicated EC populations, either untreated or exposed to TNF-α for 6 h. In B–D, F, G, and I–L the mean value of data measured in three independent experiments (±SD) is shown. **P < 0.01, ***P < 0.001, ****P < 0.0001. (N and O) Mock-, WT-, CD31Y663F- and CD31Y686F-transduced ECs were fixed, and FoxO3a intracellular localization was assessed by immunofluorescence staining. (N) Nuclei were stained with DAPI, and representative images were taken by deconvolution wide-field immunofluorescence microscopy. (Scale bars: 50 μm.) (O) Quantification of immunofluorescence staining patterns for endogenous FoxO3 (n = 100 cells per experiment). Data represent the mean value (±SD) of three independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. S4.

Spi6 gene expression is induced by proapoptotic stimuli irrespective of CD31 signals. Mock-, WT-, CD31Y663F-, and CD31Y686F-transduced ECs were exposed to TNF-α (50 ng/mL) for 6 h or were left untreated. Expression of Spi6 mRNA was quantified by qRT-PCR. The mean value of data in three independent experiments (±SD) is shown. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 6.

Enhanced rejection of CD31-deficient skin grafts. WT female mice received either a WT or CD31-KO male-derived skin graft. As a control, female WT and CD31-KO female-derived skin was used. (A) Graft survival (n = 6). ****P < 0.0001. (B–D) In some recipients, WT or CD31-KO grafts were harvested 14 d after transplantation. Representative tissue sections of grafts from female WT (B) and CD31-KO (D) controls and male WT (C) and CD31-KO (E) mice stained with H&E. 1, epidermis; 2, mononuclear inflammatory cells; 3, hair follicles; 4, dermis vessels. (F and G) Allograft-draining lymph nodes were harvested 7 d posttransplantation, and the presence of HY-specific CD8 T cells was evaluated by tetramer staining. (F) Representative dot-plots. (G) The mean percentage of CD8+ tetramer+ T cells from six animals.

Fig. S5.

Donor endothelium is maintained in skin allografts. WT female mice received either a WT or CD31-KO male-derived skin graft. As a control, female WT and CD31-KO female-derived skin was used. Skin grafts were excised 7 d posttransplantation and were observed by fluorescence wide-field microscopy for expression of von Willebrand factor (VWF, green fluorescence) expressed by WT and CD31-KO ECs and CD31 (red fluorescence) expressed only by WT graft endothelium. ECs of recipient tissue expressed both markers. Shown are representative images of tissue sections taken from the graft area indicated in the scheme above each set of images. The yellow line shows the limit between graft and recipient s.c. tissue. (A) WT female (WTf) skin grafted onto WT male (WTm) recipients. (B) CD31-KO female (CD31KOf) skin grafted onto WT male (WTm) recipients. (C) WT female (WTf) skin grafted onto WT female (WTf) recipients. (D) CD31-KO female (CD31KOf) skin grafted onto WT female (WTf) recipients. (Magnification: 10×.) (Scale bars: 100 μm.) (E) Image showing a direct anastomosis between vessels of CD31-KO skin origin and WT skin origin. (Magnification: 40×.) The white line follows the margin of the vessel. (Scale bar: 10 μm.)

Fig. 7.

CD31-transduced pancreatic β cells are protected from extrinsic apoptosis. (A) Surface expression of CD31 by mock- and CD31-transfected MIN6 cells. (B) Insulin secretion by untreated and TNF-α–exposed mock- and CD31-transfected starved MIN6 cells incubated in DMEM containing 2.8 mM glucose for 30 min at 37 °C, measured 1 h after reconstitution with 25 mM glucose. (C–F) Cell death of mock- and CD31-transfected MIN6 cells either exposed to 50 ng/mL TNF-α for 6 h (C and D) or cocultured with allospecific CTLs for 6 h (E and F) as measured by TUNEL and TpB assays. Representative images are shown in C and E. (Scale bars: 100 μm.) (G) Apoptosis of mock- and CD31-transfected MIN6 cells cultured with allospecific CTLs for 6 h and pretreated with a DR ligand (DRL) inhibitor (5 μM) and/or concanamycin A (5 μg/mL) for 1 h. (H) FoxO3a intracellular localization in MIN6 cells was assessed by immunofluorescence staining following coculture with CTLs (MIN6:CTL ratio 1:5) for 6 h. Quantification of immunofluorescence staining patterns for endogenous FOXO3 is shown (n = 100 cells per experiment). (I) Apoptosis of CD31-transfected MIN6 cells cultured with CTLs for 6 h following treatment with a blocking anti-CD31 antibody or isotype control. Mock-transduced cells are included for comparison. Data in B, D, and F–I are the mean percentage (±SD) of apoptotic cells measured in three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 8.

Antiapoptotic pathways activated in CD31-transduced pancreatic β cells. (A–F) Erk and Akt activation (A), Fas surface expression (B), caspase 7 expression and activity (C and D), caspase 8 activity (E), and cFlar protein expression (F) by mock- and CD31-transfected MIN6 cells exposed to TNF-α for 6 h. (G) Nuclear localization of FoxO3 in MIN6 cells exposed to TNF-α for 6 h. Quantification of immunofluorescence staining patterns for endogenous FOXO3 is shown (n = 100 cells per experiment). Data in C–E and G are the mean value (±SD) of three independent experiments. **P < 0.01.

Fig. S6.

Spi6 gene expression is induced by proapoptotic stimuli in pancreatic β cells independently of CD31 expression and signaling. Expression of Spi6 mRNA by WT and CD31-transduced MIN6 pancreatic β cells either untreated or exposed to TNF-α (50 ng/mL) for 6 h was quantified by qRT-PCR. The mean value of data in three independent experiments (±SD) is shown. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 9.

CD31-expressing pancreatic β cells become resistant to T-cell–mediated allograft rejection. (A and B) The function of mock-transduced and CD31 MIN6 islets was evaluated by measuring blood glucose levels in grafted syngeneic (C57BL/6) (A) or allogeneic (Balb/C) (B) mice rendered diabetic by STZ injection. (C and D, Left) Representative images of a kidney that received either mock- or CD31-transduced MIN6 β cells and was stained to detect insulin production. (Right) Magnified sections of the grafts processed as indicated. (Scale bars: 100 μm.) (E) At the time mice were killed, T cells were isolated from the spleen and lymph nodes of recipients and labeled with carboxyfluorescein succinimidyl ester (CFSE). Cell division in response to an ex vivo rechallenge with C57BL/6 splenocytes by T cells from naive mice (recipients of mock-transduced or CD31 MIN6 cells) was measured by flow cytometry. Representative histograms show percent division, division indices, and proliferation indices as determined by FlowJo. **P < 0.01, ***P < 0.001.

Fig. S7.

CD31-induced prosurvival pathway. (1) Upon DR triggering with their cognate ligands (DR-L), the DISC forms and initiates the extrinsic pathway of apoptosis; also, the transcription factor FoxO3 is induced to translocate into the nucleus where it promotes the transcription of proapoptotic genes and represses the transcription of antiapoptotic genes, including cFlar. (2) DR signals also induce the phosphorylation of the intracellular tail of CD31 and recruitment of SHP2 (3), events that are required for the downstream phosphorylation and activation of the Akt/Erk pathways (4). Upon CD31-dependent activation, Akt phosphorylates FoxO3 in the nucleus and promotes its relocalization to the cytoplasm, thereby blocking proapoptotic transcriptional programs and de-repressing transcriptional inhibition of cFlar. (5) cFlar becomes available to counteract the extrinsic apoptotic pathway by directly blocking the assembly of the DISC.