A cryptic non-GPI-anchored cytosolic isoform of CD59 controls insulin exocytosis in pancreatic β-cells by interaction with SNARE proteins

Abstract

CD59 is a glycosylphosphatidylinositol (GPI)-anchored cell surface inhibitor of the complement membrane attack complex (MAC). We showed previously that CD59 is highly expressed in pancreatic islets but is down-regulated in rodent models of diabetes. CD59 knockdown but not enzymatic removal of cell surface CD59 led to a loss of glucose-stimulated insulin secretion (GSIS), suggesting that an intracellular pool of CD59 is required. In this current paper, we now report that non-GPI-anchored CD59 is present in the cytoplasm, colocalizes with exocytotic protein vesicle-associated membrane protein 2, and completely rescues GSIS in cells lacking endogenous CD59 expression. The involvement of cytosolic non-GPI-anchored CD59 in GSIS is supported in phosphatidylinositol glycan class A knockout GPI anchor-deficient β-cells, in which GSIS is still CD59 dependent. Furthermore, site-directed mutagenesis demonstrated different structural requirements of CD59 for its 2 functions, MAC inhibition and GSIS. Our results suggest that CD59 is retrotranslocated from the endoplasmic reticulum to the cytosol, a process mediated by recognition of trimmed N-linked oligosaccharides, supported by the partial glycosylation of non-GPI-anchored cytosolic CD59 as well as the failure of N-linked glycosylation site mutant CD59 to reach the cytosol or rescue GSIS. This study thus proposes the previously undescribed existence of non-GPI-anchored cytosolic CD59, which is required for insulin secretion.-Golec, E., Rosberg, R., Zhang, E., Renström, E., Blom, A. M., King, B. C. A cryptic non-GPI-anchored cytosolic isoform of CD59 controls insulin exocytosis in pancreatic β-cells by interaction with SNARE proteins.

Keywords: CD59 isoforms; VAMP2; diabetes; glycosylphosphatidylinositol; insulin secretion.

Figure 1

Characterization of rat CD59 mutants. N79W mutant lacks a GPI anchor, which is required for antibody detection by Western blot. A) Diagram representing the studied mutations in rat CD59 in this paper. TM, transmembrane. B) RT-PCR expression of rat CD59 mRNA from each mutant within transfected CHO cells. First lane represents untransfected CHO cells; n = 3. GAPDH, glyceraldehyde 3-phosphate dehydrogenase. C) Western blot using anti-rat CD59 (TH9) and anti–FLAG tag of rat CD59 mutants in CHO cells; n = 4. D, E) Cell surface expression of rat CD59 mutants in CHO cells using anti-FLAG (D) and anti-CD59 TH9 (E) (lower panel) antibodies; n = 3. Statistical difference calculated in comparison with untransfected CHO cells. F) Anti–FLAG tag immunoprecipitates of rat CD59 mutants from transfected INS-1 cells were treated or not with PLC, showing that removal of the GPI anchor abrogates recognition by monoclonal anti-rat CD59 antibodies (TH9); n = 3. Ab, immunoprecipitating antibody (heavy chain and light chain). G) Human CD59 was immunoprecipitated from A549 and RBCs incubated with/without PLC, and immunoblotted with anti-human CD59 antibodies. H) CHO cells overexpressing WT but not N79W CD59 show increased fluorescence when stained with GPI-binding fluorescent protein FLAER; n = 3. I) Anti–FLAG tag staining of cells expressing N79W or WT CD59. PLC significantly removed WT but not N79W CD59, showing that N79W is not GPI anchored; n = 3. J) Immunoblotting for FLAG tag of supernatants of INS-1 CD59 mutants treated or not with PLC. N79W was not released, confirming lack of GPI anchor on this mutant; n = 3. K) Immunoprecipitation of rat CD59 mutants with FLAG tag antibodies incubated with or without deglycosylation enzymes, confirmed expected size of N16G mutant; n = 3. gMFI, geometric mean fluorescence intensity; ns, not significant. Statistical analysis: 1-way ANOVA (D, E), 2-way ANOVA (H, I). *P < 0.05, **P < 0.01, ***P < 0.01.

Figure 2

Reintroduction of WT and N79W CD59 rescues insulin secretion in CD59 knockdown INS-1 cells. A–G) GSIS was carried out after CD59 siRNA transfection (48 h) in INS-1 cells stably expressing various synonymously encoded CD59 mutants, as indicated in each panel. CD59 knockdown decreased insulin secretion upon stimulation with 16.7 mM glucose (glc) in INS-1 untransfected cells (top panel), and reintroduction of WT CD59 or N79W, W40E, and C64Y mutants could rescue insulin secretion; n = 5. H) Basal levels of insulin secretion (2.8 mM glc, no knockdown) were significantly enhanced only in N79W transfected cells, as compared with WT INS-1 cells; n = 5. Ns, not significant; UT, untransfected. Statistical analysis by 1-way ANOVA (H) and 2-way ANOVA (A–G). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3

Characterization of cytoplasmic rat N79STOP-CD59 mutant verifying that GPI anchor is not required for CD59 effect on insulin secretion. A) RT-PCR expression of exogenous rat CD59 cDNA from N79STOP, WT, and control (Ctrl) untransfected (UT) INS-1 cells. B) Anti-FLAG Western blot for immunoprecipitated rat CD59 mutants in INS-1 cells. Lc, immunoprecipitation antibody light chain. C) Cell surface expression of rat CD59 mutants in INS-1 cells detected using anti–FLAG tag antibodies. N79STOP mutant is not present on the cell surface; n = 3. D) GSIS from N79STOP or INS-1 UT cells. N79STOP mutant rescues insulin secretion in CD59 knockdown INS-1 cells; n = 4. E, F) PIGA KO INS-1 cells stain negative for GPI-binding FLAER, with untreated WT and PLC-treated WT cells as Ctrl; n = 3. G) GSIS from PIGA KO or INS-1 Ctrl clones shows that PIGA KO clones are able to secrete insulin upon stimulation with glucose, which is decreased after silencing of CD59; n = 5. H) CD59/FLAG tag Western blot of subcellular fractions from INS-1 clones: TH9 recognizes endogenous and WT CD59 only, and anti-FLAG detects only introduced variants of CD59. n = 3. I) Confocal microscopy shows intracellular localization of N79STOP mutant in INS-1 cells, shown by immunofluorescent detection of FLAG tag (red) in cells expressing tagged WT or N79W CD59. Insulin is shown in green; n = 3. gMFI, geometric mean fluorescence intensity. Statistical analysis by 2-way ANOVA (D, G) and 1-way ANOVA (C, F). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 4

CD59 exploits the ERAD pathway to enter the cytoplasm, where WT and N79STOP but not N16G interact with VAMP2 in a glucose-dependent manner. A) Examples of confocal images showing that colocalization of WT (upper panels) and N79W (lower panels) CD59 with VAMP2 increases in high glucose (right panels) compared with low glucose (left panels), as demonstrated by PLA, using antibodies against VAMP2 and FLAG tag. Fluorescent dots are generated when the 2 proteins detected by antibodies are closer than 40 nm; n = 3. B) Quantification of data from PLA assay as demonstrated in A, n = 3. C) Binding of WT and N79STOP to VAMP2 in INS-1 cells increased upon stimulation with glucose as measured by ELISA; n = 4. D) Immunoprecipitation with FLAG tag antibodies in INS-1 cells. VAMP2 was coimmunoprecipitated with CD59 in WT and N79STOP-expressing clones but not in negative control INS-1 cells; n = 4. Input, total lysate before immunoprecipitation; IP, immunoprecipitation; Lc, immunoprecipitation antibody light chain; ns, not significant; UT, untransfected. Statistical analysis by 2-way ANOVA (B, C). *P < 0.05, ****P < 0.0001.

Figure 5

A) Western blot of INS-1 subcellular fractionation for FLAG-tagged CD59 and control (Ctrl) proteins. The majority of N79STOP/N16G and N16G mutants is present in a membrane/organelle fraction (fract.), whereas N79STOP is found in both fracts.; n = 3. B) Quantification of signal from A, n = 3. C) GSIS was carried out after CD59 siRNA transfection (48 h) in parental INS-1 cells or those stably expressing synonymously encoded N79STOP/N16G mutant and Ctrl. CD59 knockdown decreased insulin secretion in INS-1 untransfected (UT) cells (left site), and reintroduction of N79STOP but not N16G or N79STOP/N16G double mutant rescued this phenotype; n = 3. D) Tunicamycin, an inhibitor of the ERAD pathway, blocks N79STOP mutant retrotranslocation; n = 3. E-cad., E-cadherin. E) Exotoxin A blocked N79STOP mutant retrotranslocation form ER into cytosol, seen by slight accumulation of N79STOP in a membrane and organelle fract. (presumably ER) and substantial loss of protein expression in cytosol; n = 3. F) Quantification of signal from E; n = 3. Ns. not significant. Statistical analysis by 2-way ANOVA (B, C, F). **P < 0.01, ***P < 0.01.