Selective expression of the human class Ib HLA molecule HLA-G in immunologically protected sites and its function in the inhibition of NK and T-cell effector functions support an important role of this molecule in immunoregulation. Here, we demonstrate that HLA-G is constitutively expressed in the endocrine compartment of the human pancreas. Surface expression of this HLA determinant in endocrine cells is regulated in response to growth and inflammatory stimuli. Furthermore, we provide evidence that HLA-G expressed in this tissue may associate with a subset of insulin-containing granules and may be shuttled to the cell surface in response to secretory stimuli. Thus, HLA-G presentation by endocrine cells may be regulated in concert with their secretory activity. These results identify the expression of a major histocompatibility complex locus with putative regulatory functions in human pancreatic islets, a finding with potentially important implications for the progression of autoimmunity as well as for the establishment of transplant tolerance to this tissue.
Class I HLA molecules have diversified into two distinct, though structurally related, families of proteins. Class Ia HLAs, namely HLA-A, -B, and -C, are heterodimers comprising a 45-kDa polymorphic heavy chain noncovalently associated with an invariant light chain, β2-microglobulin. Their ubiquitous expression regulates many cellular immune responses, from thymic selection to presentation of endogenous peptides, for efficient eradication of virus-infected or tumor cells (1). While the high polymorphism of these proteins ensures immune recognition of a variety of self-antigens and viral peptides, it also represents a major barrier to allo-transplantation.
Unlike class Ia HLA molecules, class Ib HLA-E, -F, and -G antigens display very limited or no polymorphism within the human population, may function in presenting limited sets of peptides, and/or exhibit a restricted tissue expression (2). The latter feature characterizes the expression of HLA-G. Thus, high levels of HLA-G were described in the blastocyst and the fetal cytotrophoblast during the 1st trimester of pregnancy, the epithelial cells of the anterior chamber of the eye, the testis, and the fetal liver (3). In addition, we have reported a restricted expression of HLA-G within the thymus in a select subset of medullary and subcortical epithelial cells (4). Interestingly, these tissues represent immunologically protected sites (the fetal trophoblast), sites of immune privilege (the anterior chamber of the eye and the testis), or lymphocyte selection (the thymus), suggesting a role for HLA-G in tissue-specific immunoregulation. In support of this hypothesis, it was demonstrated that presentation of HLA-G on target cells downregulates effector functions in NK cells, antigen-specific CD8+T cells, and monocytes through engagement of inhibitory receptors expressed on effector cells (5–9). Thus, unlike polymorphic class Ia HLA molecules, which have evolved as efficient activators of immune responses, presentation of HLA-G may have specialized to increase the activation/effector thresholds of T-, NK, and antigen-presenting cells for the immune protection of the semiallogeneic fetus and certain autologous tissues.
We hypothesize that the expression of HLA-G as an immunomodulatory molecule may be relevant at sites of organ-specific autoimmunity, such as pancreatic islets. Indeed, studies in animal models indicate that immune tolerance to these organelles may involve mechanisms of immune privilege. Hence, it was shown that transgenic expression of viral antigens within pancreatic islets does not elicit an immune response even in the face of circulating autoreactive T-cells (10). Tolerance to islet allografts devoid of professional antigen-presenting cells may also involve immunological ignorance (11), suggesting immunoregulatory mechanisms intrinsic to this endocrine tissue that may even overcome immune recognition of HLA disparities between the host and the graft. These observations prompted us to investigate whether expression of HLA-G could be related to such phenomena in human islets. Our studies demonstrate that HLA-G is a determinant of the HLA repertoire of pancreatic islets and provide evidence for a regulated presentation of this antigen at the surface of endocrine cells in response to growth, inflammatory, and secretory stimuli.
RESEARCH DESIGN AND METHODS
Tissues and cell culture.
Human pancreas was embedded in OCT and snap frozen or embedded in paraffin. Human islets were obtained from Juvenile Diabetes Research Foundation islet distribution centers, purified by dithizone staining and hand picking, and used within 24–48 h. For the generation of islet cell monolayers, islets were dissociated into small clusters corresponding to the β-cell cores, collected by gravity sedimentation, and cultured on HTB-9 matrix, as described (12). Immunostaining for the islet transcription factor PDX-1 (pancreatic duodenal homeobox-1) and cytokeratin 19, a marker of exocrine and ductal cells, confirmed that >95% of the cells in the monolayers were PDX-1+CK19−. The human choriocarcinoma cell line JEG (ATCC) was cultured in RPMI/10% FCS. The endocrine β-cell lines INS (kind gift of Chris Rhodes, Pacific Northwest Research Institute, Seattle, WA) and MIN-6 were cultured as described (13,14). In secretagogue stimulation assays, islets were starved for 2 h at 37°C in RPMI/0.1% BSA/2.8 mmol/l glucose/100 μm diazoxide and cultured in RPMI/0.1% BSA containing either 2.8 mmol/l glucose or 16 mmol/l glucose/100 μm tolbutamide for 1 h at 37°C. Islets were then dissociated into single cells using a nonenzymatic medium (Sigma), incubated with the anti–HLA-G mAb IB8 or the pan-HLA mAb W6/32 followed by RPE-secondary antibodies (Caltag), and analyzed at a FACScan (Becton Dickinson).
Cryostat sections of human pancreas were fixed in 2% paraformaldehyde, permeabilized in PBS/0.1% TX-100, and blocked and incubated in the presence of the anti–HLA-G mAbs IB8 or 4H84 (15) or the HLA class I mAbs W6/32 or HC10 (kind gift of Dr. D. Schust, Harvard medical School, Boston, MA) in combination with sheep anti-insulin IgGs (The Binding Site) and rabbit anti-glucagon (Chemicon). After washings, sections were incubated with species-specific biotin-, LCRC and Cy5 secondary antibodies (The Jackson Laboratories). Binding of the biotin-labeled antibody was revealed using a tyramide-based amplification system (NEN). For simultaneous staining of class I HLAs, insulin and platelet-endothelial cell adhesion molecule (PECAM)-1 sections were first incubated with the anti-HLA mAbs listed above, followed by goat anti-mouse Fab’ fragments and a biotin–anti-goat antibody. Slides were then incubated with an anti–PECAM-1 mAb (clone MBC78.1; Caltag) and a guinea pig anti-insulin antibody (Santa Cruz Biotechnologies) followed by fluorescein isothiocyanate and Cy5 secondary antibodies (The Jackson Laboratories). The slides were viewed on a microscope equipped with a laser scanning confocal attachment (MRC-1024; Bio-Rad).
Immunoprecipitation and immunoblotting.
Single cells were dissociated from pancreatic islets or cell monolayers by a nonenzymatic treatment, counted, and surface biotinylated in PBS-LC-NIH-biotin (100 μg/ml) (Pierce). Labeled cells were lysed in 20 mmol/l Tris-1% Tx-100, pH 7.4, and proteases inhibitors. Proteins from equal numbers of cells or 10–30 μg total proteins were immunoprecipitated using avidin-conjugated Sepharose beads (Pierce), resolved by SDS-PAGE, and transferred onto polyvinylidine fluoride membranes (Millipore). In other experiments, 30–100 μg total proteins were immunoprecipitated at 4°C with the W6/32 antibody, a rabbit anti-human β2m antibody (Chemicon) or control mouse or rabbit IgGs. Immunocomplexes were adsorbed to protein G-Sepharose (Invitrogen), resolved by SDS-PAGE, and transferred to polyvinylidene fluoride membranes. After blocking, the membranes were probed using the anti–HLA-G mAb 4H84, an anti HLA-A,B–specific mAb (clone 174.1; kind gift of Dr. Juan Scornik, University of Florida, Gainsville, FL) or mouse IgG, followed by a peroxidase anti-mouse IgG antibody and chemoluminescence (Kirkegaard and Perry Labs).
PCR amplification of HLA-G transcripts, construction of tetracystein-tagged HLA-G, cloning into lentiviral vectors, and expression in endocrine cells.
cDNA was transcribed from poly-A+ RNA of human islets using Superscript II and oligo dT primers (Invitrogen) and used as template for PCR of HLA-G transcripts, as previously described (4). An HLA-G1 cDNA, amplified by PCR from human placental mRNA, was cloned into pcDNA 2.1 (Invitrogen) and sequenced. Next, using the Expand High Fidelity PCR System (Roche), nested PCR were performed to engineer an HLA-G cDNA devoid of the endogenous stop codon and with the COOH-terminal fused to the tetracystein-containing peptide 5′ ADPPVATMPCCPGCCGC-3′. This product was ligated into the HIV vector ROVER downstream of a human cytomegalovirus promoter. To construct ROVER, a plasmid pCR-XL-CSPre was first assembled by cloning the MluI-ApaI fragment from the SIN (self-inactivating) HIV vector pHIV-SINPre into the pCR-XL Topo backbone (Invitrogen). The cytomegalovirus promoter from pcDNA3.1/V5-His-TOPO was then cloned into the BamHI site of pCR-XL-CSPre. Then, the human cytomegalovirus promoter, the internal ribosome entry site sequence, enhanced green fluorescent protein, and a scaffold attachment region were cloned between the BamHI and SacII sites of pCR-XL-CSPre. This HIV vector was VSV-G pseudotyped by packaging in HEK-293T cells transfected with the transgene plasmid and the packaging plasmids pMD.G (VSV-G), pMDLg/p.RRE (gag and pol), and pRSV-Rev (rev). Viral supernatants were harvested after 48 h, concentrated by ultracentrifugation and viral titers, defined as transducing units/ml or multiplicity of infection, estimated by transduction of 293T-cells and flow cytometry of GFP+ cells. INS cells were transduced with the HLA-G lentivirus at 40 multiplicity of infection and HLA-G expression verified by Western blotting after 72 h of culture. To evaluate the function of the tetracystein tag, i.e., its ability to bind the biarsenical derivative ReASH-EDT2, the transduced cells were labeled for 1 h at 37 C with ReASH-EDT2 and imaged by confocal microscopy (16).
Photoconversion, immunogold labeling, and electron microscopy.
To localize tetracysteine-tagged HLA-G at the ultrastructural level, HLA-G–transduced INS cells were labeled with ReASH-EDT2 and then fixed in 2% glutaraldheyde/0.1 mol/l sodium cacodylate buffer, pH 7.4 (16). For photoconversion, 1 mg/ml diaminobenzidine in oxygenated 0.1 mol/l sodium cacodylate buffer was added to the dish, and the cells were exposed to a 585-nm light from a xenon lamp until a brownish reaction product appeared. After postfixing in 1% osmium tetroxide, cells were dehydrated, embedded in Epon 812, cut, and ultrathin sections examined at a Jeol-1200 electron microscope (Jeol USA, Peabody, MA). For immunogold labeling, HLA-G–transduced INS or MIN-6 were grown on glass coverslips, fixed in 4% paraformaldehyde/0.05% glutaraldheide, 1% l-lysine, and 0.25% sodium metaperiodate in 0.04 mol/l sodium cacodylate buffer, pH 7.4, and permeabilized with 0.1% saponin. After washing and blocking, samples were incubated overnight at 4°C with primary antibody (mAb 4H84, 2.5 μg/ml) followed by nanogold-labeled (1.4 nm gold conjugate) goat anti-mouse F(a)b’ fragments (Nanoprobes). After washings, samples were fixed in 1.2% glutaraldehyde in 0.1 mol/l sodium cacodylate buffer, pH 7.4, with 5% sucrose, and gold nanoparticles were enhanced using Gold Enhance LM. After postfixation in 2% osmium tetroxide and 2% potassium ferricyanide, samples were dehydrated, embedded in Epon 812, sectioned, and stained with lead citrate. For morphometric analysis, we evaluated the volume density of secretory granules by planimetry using the Image Pro Plus software and scored the number of gold particles over secretory granules and the whole cell. The relationship between the number of gold particles in granules and the volume density of these organelles was analyzed by regression analysis using the Statistical Package for Social Science (SPSS, Chicago, IL). The theoretical distribution of gold particles over granules as predicted by random labeling was compared with the distribution actually observed using a χ2 test.
HLA-G expression in the human pancreas is restricted to insulin, glucagon, and ductal cells.
To investigate the expression of HLA-G in the human pancreas, pancreatic sections were stained by three-color immunofluorescence for HLA-G, insulin, and glucagon and analyzed by confocal microscopy. Expression of HLA-G was detected in insulin- and glucagon-positive cells (Fig. 1A–D) as well as in intra-acinar (Fig. 1D, arrows) and interlobular pancreatic ducts (Fig. 1E) but not in acinar cells. Two-color immunostaining for PECAM-1 and HLA-G failed to reveal any HLA-G–specific immunoreactivity in blood vessels, using the anti–HLA-G–specific antibodies IB8 and 4H84 (Fig. 2A). Unlike HLA-G’s pattern of expression, the antibody HC10, reactive with free HLA-B and C heavy chains but not HLA-G (17,18) and the antibody W6/32, specific for class Ia and class Ib HLA heavy chains/β2-microglobulin heterodimers, showed strong reactivity with pancreatic vessels as well as labeling of exocrine and endocrine cells (Fig. 2B).
Membrane-bound and HLA-G isoforms generated by alternative splicing of a single gene have been reported in human trophoblast, thymic epithelium, and cell lines (3). To investigate the isoform(s) expressed in pancreatic islets, mRNA from pancreatic islets was analyzed by RT-PCR using locus-specific primers. This analysis revealed the presence of 1,200-, 900-, and 500-bp DNA fragments (Fig. 3A, left panels) consistent with transcripts for the membrane-bound HLA-G1, -G2, and -G3 isoforms, respectively. To determine which of these transcripts were translated into protein, total protein extracts from purified pancreatic islets were analyzed by Western blotting using the anti–HLA-G mAb 4H84. Because of its specificity for a peptide sequence within the α 1 domain, conserved in all HLA-G isoforms, this antibody is a suitable tool for the detection of HLA-G1, -G2, and -G3 and soluble proteins. Indeed control lysates of K562 cells transfected with HLA-G1, -G2, and -G3 cDNAs confirmed the antibody reactivity to all three isoforms (Fig. 3B, right panels). Similar analysis of islet cell protein extracts revealed expression of a single 39-kDa protein, consistent with the molecular weight previously reported for HLA-G1 (Fig. 3A, right panels).
These results identify HLA-G as a determinant of the HLA repertoire of the human pancreas uniquely restricted to the endocrine and ductal compartments of this organ, where it coexists with class Ia HLA molecules. Expression of HLA-G proteins in pancreatic islets appears to be limited to the HLA-G1 isoform.
Differential expression of HLA-G and HLA-A,B antigens in pancreatic islets in response to growth and inflammatory stimuli.
The balance between surface expression of HLA-G and class Ia HLAs in the trophoblast has been proposed to play an important role in the maintenance of maternal immune tolerance to the fetus. In this tissue, the highest levels of HLA-G are detected during the proliferative stages of the blastocyst and throughout the invasive stages of throphoblast development (3,15), suggesting that HLA-G expression may be associated with cell growth.
To investigate the expression of HLA-G and class Ia HLA-A,B antigens in resting versus cycling islet cells, single cells from pancreatic islets 2 days postisolation, or from islet monolayers cultured on the HTB-9 matrix, a substrate supporting islet cell proliferation (12), were surface biotinylated and lysed. Surface and cytoplasmic proteins from the same number of cells were then analyzed by Western blotting. Figure 3B shows the detection of HLA-G and HLA-A,B antigens expressed on the cell surface (Fig. 3, upper panels) or retained in the cytoplasm (Fig. 3, lower panels) of freshly isolated islet cells. Low levels of HLA-G are expressed at the surface of islet cells compared with HLA-A,B molecules, whereas relatively higher levels of HLA-G are detected in the cytoplasm. Interestingly, cells grown on the HTB9 matrix demonstrated significant upregulation and enhanced translocation of HLA-G at the cell surface (Fig. 3C). Thus, compared with islet cell clusters, the surface expression of HLA-G on a per cell basis increases ∼10-fold in cells cultured on the matrix, whereas that of classical HLA class I molecules increases only ∼3-fold (Fig. 3D). To investigate whether this result was accounted by the increase in the total protein content per cell, possibly accompanying cell activation/growth in the monolayers, we performed surface expression studies using identical amounts of total proteins from islet cell clusters and monolayers (n = 2). We found that cells cultured as monolayers had approximately two- to threefold higher protein content than islet cell clusters. Nevertheless, analysis of surface proteins from lysates normalized for protein content and β-actin revealed a fivefold increase in the surface expression of HLA-G but not HLA-A,B in cells cultured as monolayers versus islet clusters, supporting the conclusion of a relative increase of surface HLA-G over HLA-A,B in islet cell monolayers (online appendix Fig. 1 [available at http://diabetes.diabetesjournals.org]).
Assembly of the heavy chain with β2-microglobulin regulates translocation and stability of class I HLA molecules at the cell surface. To determine whether HLA-G expressed in pancreatic islet cells is bound to β2-microglobulin, cell lysates were immunoprecipitated with the W6/32 antibody. Western blotting analysis revealed that only a small fraction of HLA-G expressed in islet cell clusters (i.e., <5%) is associated with β2-microglobulin (Fig. 3E). In contrast, culture of islet cells on the HTB9 matrix results in increased expression of HLA-G/β2-microglobulin heterodimers, as recognized by both the W6/32 antibody and a polyclonal anti–β2-microglobulin antibody (Fig. 3E). In this condition, HLA-G/β2m complexes represent ∼20% of total HLA-G expressed in islet cells.
The heterogeneous HLA context presented by pancreatic islets implies that differential regulation and ultimately the balance between HLA-G and polymorphic class I HLA antigens at the cell surface may be important to the outcome of immune responses to this tissue in a inflammatory setting. Among proinflammatory cytokines known to regulate class I major histocompatibility complex (MHC) antigen expression are γ-interferon (IFN-γ) and interleukin (IL) 10 (19,20). To investigate the expression of HLA-G and HLA-A,B antigens in response to these cytokines, pancreatic islets were cultured in the presence of IFN-γ (500 units/ml) or IL10 (10 ng/ml) for 72 h. We found that IFN-γ upregulates both HLA-G and HLA-A,B at the cell surface (Fig. 3E). In contrast, consistent with a previous report in trophoblast and monocytes (21), IL10 upregulates HLA-G but not HLA-A,B at the surface of islet cells (Fig. 3E). Thus, inflammatory cytokines may increase the fraction of HLA-G at the cell surface and differentially affect the ratio of HLA-G and HLA-A,B antigens presented by pancreatic islet cells.
Regulation of HLA-G expression during insulin secretion: evidence for shuttling of HLA-G to the cell surface through a subset of secretory granules.
Pancreatic β-cells adapt to constant variations of nutrients and metabolites present in the extracellular milieu by modifying their secretory function. Interestingly, there is evidence that expression of MHC products can be regulated in concert with the secretory activity in certain cell types. Thus, it was shown that class I MHC products are present in the synaptosomes of developing neurons (22). In addition, class II MHC antigens were shown associated with secretory granules in mast cells and demonstrated to be shuttled to the cell surface upon cell degranulation (23).
To investigate whether modifications of the secretory activity of pancreatic β-cells could influence surface expression of HLA-G, pancreatic islets were incubated in serum-free medium containing either low (1.2 mmol/l) or high (16.2 mmol/l) glucose plus the insulin secretagogue tolbutamide. Fluorescence-activated cell sorter analysis revealed a similar surface expression of total class I HLA molecules in either cultures, as detected by the antibody W6/32 (Fig. 4A, left panels). In contrast, cells cultured in high glucose plus tolbutamide increased surface expression of HLA-G compared with the levels detected in low glucose–containing medium (Fig. 4A, right panels). This effect was apparent as a shift of HLA-G mean intensity of fluorescence and/or increase in the percentage of positive cells cultured in high versus low glucose ([means ± SE] mean intensity of fluorescence = 43.16 ± 11 vs. 29.8 ± 12; percent of HLA-G+ cells in high versus low glucose = 28.3 ± 9.9 vs. 19.6 ± 8.8, n = 3). Western blotting of cells cultured in either conditions revealed no significant differences in the expression of total HLA-G protein (data not shown). These results indicated that cell surface expression of HLA-G in islet cells is regulated in concert with glucose stimulation and/or hormone secretion and suggested that the intracellular pool of HLA-G may associate with the regulated secretory pathway in β-cells.
To investigate the distribution of HLA-G at the subcellular level, we engineered a recombinant HLA-G fused to a tetracystein tag and expressed this construct in the β-cell lines INS or MIN-6. The reactivity of tetracysteine motifs with biarsenical derivatives and the precipitate that they form during photo-oxidation has been shown previously to allow precise subcellular localization of molecules by electron microscopy (16). This approach confirmed a predominant intracellular location of the protein in the endoplasmic reticulum, the main cellular compartment where class I MHC molecules are synthesized and assembled (Fig. 4B). A patchy expression of HLA-G was also detected at the basolateral but not at the apical side of the cell membrane (Fig. 4B, arrow and Fig. 4C, arrowheads), indicating targeting of this molecule to distinct membrane microdomains of β-cells. Interestingly, the photo-oxidation studies also revealed membrane labeling of some insulin-containing granules (Fig. 4E–H, arrowheads). To confirm the specificity of this labeling, we detected HLA-G by immunogold staining. Labeling of a subset of mature insulin-containing granules, mostly localized at the membrane of the granules, was detected in HLA-G–transduced (Fig. 5A, arrowheads) but not mock-infected cells (Fig. 5B). Morphometric analysis demonstrated that HLA-G+ secretory granules represent 13.8 ± 3.9% of the total granules in MIN-6 cells (n = 58 cells). There was no correlation between levels of HLA-G expression in the cell and frequency of HLA-G+ granules. Within individual cells, the proportion of gold particles over secretory granules (2.3 ± 0.3%, n = 75) was larger than the relative volume occupied by these organelles in the cells (0.92 ± 0.03%, n = 67). In addition, we found no significant correlation between the volume density of secretory granules and the amount of gold labeling of these organelles (r2 = 0.005, P = 0.58), indicating that the gold labeling did not merely reflect the relative abundance of granules in the cells. In contrast, we observed a highly significant difference between the distribution of granule immunolabeling and that computed assuming a random staining (P < 0.001), indicating that the gold labeling in granules was specific.
Enhanced MHC class I expression is a prominent early feature of pancreatic islets in autoimmune diabetes and islet transplantation. However, the nature of class I MHC loci potentially critical to the onset and progression of insulitis in humans remains poorly defined. Our findings reveal a previously unrecognized complexity of islet immunogenicity, to include MHC determinants such as HLA-G with specialized immunoregulatory functions. Indeed, the long-lasting, self-contained, peri-islet and periductal lymphoid infiltrates, hallmark of the early stages of autoimmune diabetes and the evidence for T- and NK cell–dependent regulatory networks in those lymphoid infiltrates (24,25), indicate that destructive processes are counterbalanced by immunoregulatory mechanisms at sites of islet immunity. Our observation that pancreatic islets and ducts express HLA-G constitutively and upregulate this MHC determinant in response to proinflammatory cytokines raises the possibility that mechanisms of immune regulation activated by this MHC molecule may be involved.
We demonstrate that in pancreatic endocrine cells, HLA-G is expressed constitutively at low levels, as β2-microglobulin–free heavy chain, and is mainly intracellular. A significant intracellular retention of HLA-G was previously reported in trophoblast and transfected cell lines and explained by an unusually extensive recycling of this molecule from the golgi to the endoplasmic reticulum driven by signaling sequences of its cytoplasmic tail (26,27). Interestingly, while this internal recycling significantly prolongs the intracellular lifespan of HLA-G relative to that of other MHC molecules, it also allows interaction of HLA-G with multiple low-affinity peptides (27). Since low-affinity peptides may act as autoantigens (28), it is possible that intracellular HLA-G may contribute to islet immunologic ignorance by functioning as a “buffering” system, preventing the presentation to the cell surface of self-antigens likely not to have elicited clonal deletion.
Consistent with the high levels of HLA-G reported in the invasive cytotrophoblast in vivo and at sites of contact with extracellular matrices in vitro (29), we observed a significant upregulation of HLA-G in islet cells cultured on an extracellular matrix supporting cell replication. Unlike the fetal cytotrophoblast, however, we find that expression of HLA-G and class Ia HLA-A,B antigens in pancreatic islets is not mutually exclusive. Furthermore, we provide evidence that the relative expression of these two classes of HLA antigens can be differentially regulated by inflammatory cytokines such as IFN-γ and IL10. This suggests the potential for immune cell interactions distinct from those occurring at the fetal/maternal interface. In this regard, it was shown that expression of HLA-G in association with class Ia MHC molecules is essential for the protection of target cells from polyclonal NK cell lysis (30). Under polyclonal immune cell activation, class Ia MHC molecules are not very effective, likely due to the low frequency of NK and cytotoxic T-cells carrying inhibitory receptors for a given class Ia HLA (31).
The preferential basolateral expression of HLA-G identified by our ultrastructural studies indicates a distinct membrane cell targeting of this molecule in endocrine cells. It is likely that this localization represents a feature common to polarized epithelia rather than a characteristic of the HLA-G molecule per se, since a similar expression pattern of class I MHC molecules has been reported in the epithelium of the endometrium and small intestine (32). Intriguingly, we find that HLA-G can be detected in a population of secretory granules and be upregulated at the cell surface of primary islet cells stimulated to secrete insulin. These findings suggest that HLA-G may be exported to the surface of islet cells not only through the constitutive secretory pathway but also through the regulated pathway by which insulin is secreted. Whether this phenomenon reflects a default routing of the MHC in cells with secretory activity or has another biological function is currently unknown. However, many autoantigens in islet immunity are components of secretory granules (33–35). Thus, it is possible that sites of insulin exocytosis may represent subcellular domains where a high density of potentially immunogenic ligands become exposed. Since the activation of autoreactive T-cells depends upon the surface density of antigen/MHC complexes, this may lead to the activation of low-affinity cytotoxic T-cells. Local clustering of immunoregulatory molecules, such as HLA-G, at sites of granule exocytosis may prevent this unwanted activation. Interestingly, a similarly immunoregulatory class Ib MHC molecule, Qa-1b, was shown to present distinct insulin epitopes in mouse β-cells (36). Thus, the expression of class Ib MHC molecules in pancreatic endocrine cells may include the regulation of immune responses to their secreted products and/or components of their secretory machinery during hormone exocytosis.
Additional information for this article can be found in an online appendix at http://diabetes.diabetesjournals.org.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a Juvenile Diabetes Foundation International Career Development Award and National Institutes of Health (NIH) Grant RO1 AI446723 to L.C. V.C. was supported by Juvenile Diabetes Research Foundation (JDRF) Grant no. 1-2004-13, a Network Research Grant from The Larry L. Hillblom Foundation, NIH RO1 DK98183, and RO1 DK063443. P.M. was supported by grants from the Swiss National Fund (310000-109402), the JDRF, and NIH RO1-DK63443. Confocal and electron microscopy was performed at the National Center for Microscopy and Imaging Research supported by NIH 5P41RR004050-17 to Dr. Mark Ellisman. The isolation of pancreatic islets obtained from C.R. was supported in part by the NIH National Institute of Diabetes and Digestive and Kidney Diseases, the JDRF, and the National Center for Research Resource’s Islet Cell Resource Islet Distribution Program.
We thank Dr. Guido Gaietta and Dr. Mark Ellisman for insightful advice in the performance of photoxidation studies and Dr. Mario Tchan for the construction of the Rover lentiviral vector. This study was also part of the Geneva Program for Metabolic disorders (GeMet). This is TSRI manuscript number 17407-MEM.