MCL-1 Is a Key Antiapoptotic Protein in Human and Rodent Pancreatic β-Cells

Secretion of proinflammatory cytokines by immune cells contributes to β-cell death in type 1 diabetes (T1D) (1), and exposure of human islets in vitro to the proinflammatory cytokines interleukin-1β (IL-1β) and interferon-γ (IFN-γ) induce global changes in gene expression (2) that are remarkably similar to those observed in laser-captured islet cells from patients with T1D (3). One of the mechanisms involved in cytokine-mediated β-cell killing is the induction of endoplasmic reticulum (ER) stress and the consequent activation of the intrinsic (mitochondrial) apoptotic pathway (4,5). The intrinsic apoptotic pathway is regulated by the B-cell lymphoma 2 (BCL-2) family of proteins, which is divided into three groups: the antiapoptotic members (e.g., MCL-1 and BCL-2), the proapoptotic BH3-only proteins (e.g., BIM, DP5/HRK, and PUMA), and the multi-BH (BCL-2 homology) domain proapoptotic proteins BAX and BAK (5,6).

MCL-1 prevents apoptosis induced by BIM and PUMA, two BH3-only proteins that contribute to ER stress and inflammatory cytokine–induced β-cell killing (5,7–9). MCL-1 protein is expressed in rodent β-cells, and cytotoxic insults that induce β-cell apoptosis, such as inflammatory cytokines, reduce MCL-1 protein expression in these cells (7,10,11). Importantly, MCL-1 overexpression protects primary rat β-cells against diverse cytotoxic conditions (7,11). Moreover, overexpression of microRNA-29 (miR-29) promotes β-cell death by decreasing MCL-1 protein expression (10). Interestingly, abnormally reduced expression of MCL-1 is present in islets from patients with T1D infected with a diabetogenic enterovirus, suggesting that defects in MCL-1 expression play a role in the development of human diabetes (12).

We have previously shown that MCL-1 protein expression is reduced in inflammatory cytokine–treated β-cells due to the combined actions of ER stress–mediated translation arrest and c-Jun N-terminal kinase (JNK)-induced MCL-1 phosphorylation, which was reported to prime this protein for ubiquitination and proteasomal degradation (11). Given the importance of MCL-1 in the survival of β-cells, the aim of the current study was to clarify its role in human β-cells, as well as in an in vivo murine model of diabetes and to determine the mechanisms involved in the posttranslational regulation of MCL-1 protein levels in β-cells. We found that inflammatory cytokine–mediated MCL-1 downregulation contributes to the death of human β-cells and that this can be partially inhibited by ectopic MCL-1 overexpression. We generated a β-cell–specific Mcl-1 knockout mouse line (βMcl-1KO) and observed that these mice were more hyperglycemic and had lower insulin content after multiple low-dose streptozotocin (MLDSTZ) treatment. Mechanistic studies in rat β-cells indicated that the kinase GSK3β, the E3 ligases MULE (MCL-1 ubiquitin ligase) and βTrCP (β-transducin repeat–containing protein), as well as the deubiquitinase USP9x (ubiquitin-specific peptidase 9 X-linked) regulate inflammatory cytokine–mediated MCL-1 reduction in β-cells. Overall, our results identify MCL-1 as an important guardian against β-cell death and clarify the mechanisms behind its regulation by proinflammatory cytokines.

The cytokine concentrations used were based on prior studies (13–15) and are described in Supplementary Table 1. The GSK3α/β inhibitors SB216763 (SB) and BIO (Sigma-Aldrich, Diegem, Belgium) were dissolved in DMSO (1:1,000) and used at 5 or 1 µmol/L, respectively. For immunoprecipitation experiments, cells were treated with the proteasome inhibitor MG-132 (Sigma-Aldrich) at 1 µmol/L.

The human β-cell line EndoCβH1 and the rat insulinoma cell line INS-1E were cultured as previously described (14). Human islets were isolated from six organ donors without diabetes (age 74 ± 3 years, BMI 25 ± 1 kg/m², insulin-positive cells 63 ± 8%) in Pisa, Italy, with the approval of the local human research ethics committee. These islets were cultured and treated as previously described (16).

Conditional Mcl-1 knockout mice were generated as previously described (17). Mcl-1^fl/fl mice were crossed with RIP-Cre transgenic mice (18) to generate the βMcl-1KO line. Both lines are on the C57BL/6 genetic background, and wild-type (WT) littermates were used as controls. βMcl-1KO mice were born at the expected normal Mendelian ratio. The nonfasted glycemia and body weight were followed in male and female βMcl-1KO mice and their respective WT littermates from 6 to 24 weeks (data not shown and Supplementary Fig. 2). An intraperitoneal glucose tolerance test was performed in these animals at 12 and 24 weeks of age. Mice were injected with 2 g/kg body weight glucose after 6 h of fasting. At 24 weeks, mice were sacrificed and their pancreas collected for measuring the total pancreatic insulin content (19) or for immunofluorescence analysis (see below). Mice were housed and handled according to the Belgian Regulations for Animal Care and with permission from the local ethics committee.

For ex vivo experiments, mouse islets were isolated and cultured as previously described (20). Glucose-stimulated insulin secretion was performed in freshly isolated islets (19,21). Insulin was quantified using the Ultra Sensitive Mouse Insulin ELISA Kit (Crystal Chem, Downers Grove, IL). The glucose-stimulated insulin secretion experiments were performed and measured in triplicate.

Nonfasted male mice aged 7–8 weeks were injected intraperitoneally for five consecutive days with either 42.5 mg/kg body weight streptozotocin (STZ) (Sigma-Aldrich) dissolved in citrate buffer (100 mmol/L, pH ≤4.5, made freshly) or citrate buffer alone. Blood glucose levels were measured on days −3, 1, 2, 3, 4, and 5 pre- and postinjection and later weekly during 10 weeks after the last injection in nonfasting conditions using a glucometer (Accu-Chek; Roche, Basel, Switzerland) (22). Hyperglycemia was defined as nonfasting blood glucose levels >200 mg/dL in two sequential measurements. At the end of the experiment, the animals were sacrificed and the pancreas collected for histological analysis or for measuring the insulin content.

Pancreatic tissues were collected and fixed overnight in 4% formaldehyde (BDH-Prolabo, Radnor, PA). Subsequently, the tissues were embedded in paraffin and at least three different sections were stained for analysis as previously described (23). Antibodies used are listed in Supplementary Table 2. Semiquantitative analysis of insulin and glucagon staining was performed by analyzing at least three different pancreas sections of three different animals per genotype, with a minimum of 15 islets examined per animal.

The small interfering RNAs (siRNAs) (30 nmol/L) used are listed in Supplementary Table 3 and transfections were performed as previously described (13,16,24). MCL-1 overexpression was achieved by transient transfection with a plasmid encoding rat MCL-1 (16) or by adenoviral infection (11). Control adenoviruses encoding luciferase or β-galactosidase were obtained from SIRION Biotech (Martinsried, Germany). The expression vector for Flag-tagged rat MCL-1 (Agilent Technologies, Santa Clara, CA) was provided by Eminy Lee (Institute of Biomedical Sciences, Taipei, Taiwan). Sequences encoding nontagged rat MCL-1 were cloned into the pExpress plasmid (Express Genomics, Frederick, MD). The constructs encoding the phosphorylation site mutants of rat MCL-1 were generated using the Q5 Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA) using the WT rat Flag-MCL-1 vector as a template.

The percentages of viable, apoptotic, and necrotic cells were determined using the DNA binding dyes propidium iodide (PI; 5 µg/mL) (Sigma-Aldrich) and Hoechst 33342 (HO; 5 µg/mL) (Sigma-Aldrich), as previously described (11). In all experiments, the numbers of necrotic cells were low (between 1.8 and 2.1%) and did not change between the different conditions. For mouse islets, the percentages of dead cells were evaluated in a minimum of 10 islets per condition. All assessments were performed by two independent researchers, one of whom was unaware of the identity of the samples.

Poly(A)⁺ mRNA was isolated and reverse transcribed as previously described (16). The real-time PCR amplification reaction was performed using SYBR Green and compared with a standard curve (25). All primers used are listed in Supplementary Table 3.

For total cellular extracts, cells were lysed in Laemmli buffer, and for immunoprecipitation, 1× RIPA buffer and the lysates were processed as previously described (26). Western blot analysis was performed as previously described (11). The antibodies used are listed in Supplementary Table 2. Images were acquired by Chemidoc (Bio-Rad, Temse, Belgium) and analyzed using ImageJ 1.49 software (27).

Single mouse islet cell preparations were obtained as previously described (20). To exclude dead cells, cells were stained using the LIVE/DEAD Fixable Dead Cell Stain Kit (Life Technologies, Ghent, Belgium). APC–anti-human CD4 (mouse, clone RPA-T4) antibody was used to assess Mcl-1 gene deletion (28,29). Cells were acquired on a FACSCanto II flow cytometer (BD Biosciences), and the data were analyzed using the FlowJo software (TreeStar, Ashland, OR) (29). The sorting of β-cell–enriched cell populations was performed as previously described (30).

Data are presented as means ± SEM. Comparisons were performed by two-tailed paired Student t test or by ANOVA followed by paired Student t test with Bonferroni correction for multiple comparisons. A P value of <0.05 was considered as statistically significant.

Exposure of human EndoCβH1 cells to IL-1β+IFN-γ or TNF+IFN-γ for 16 h induced a significant increase in β-cell apoptosis that was further slightly increased at 24 h (Supplementary Fig. 1A). By 24 h, these inflammatory cytokines also caused a significant reduction of the expression of MCL-1 protein (up to 50%) in EndoCβH1 cells (Fig. 1A). We next exposed EndoCβH1 cells to different concentrations of cytokines for 24 h. In all conditions, a significant reduction of MCL-1 expression (Fig. 1B) and increased apoptosis of EndoCβH1 cells (Fig. 1C and D) was observed. Moreover, knockdown of Mcl-1 further increased the killing of EndoCβH1 cells induced by these cytokine combinations (Fig. 1C and D and Supplementary Fig. 1B and C). A second siRNA confirmed the potentiating effect of Mcl-1 knockdown on cytokine-mediated apoptosis of EndoCβH1 cells (Supplementary Fig. 1B–D). In converse experiments, overexpression of rat MCL-1 using an adenovirus system (Ad-MCL-1) diminished cytokine-mediated apoptosis in both EndoCβH1 cells and primary human islet cells (Fig. 1E and F and Supplementary Fig. 1E and F).

To explore the role of MCL-1 in islet cells in vivo, we generated the βMcl-1KO strain. The expression of MCL-1 protein in islets was significantly reduced in βMcl-1KO mice as determined by Western blot analysis (Fig. 2A). In βMcl-1KO mice, a human CD4 reporter is subjugated to Mcl-1 promoter/enhancer elements as previously described (17). Therefore, β-cells in which Mcl-1^fl has been deleted expressed the hCD4 protein that can be detected by FACS using fluorochrome-labeled anti–human-CD4 antibodies. We observed that 91.3 ± 0.5% of the β-cell–enriched population from βMcl-1KO mice expressed the hCD4 protein, whereas only 3.9 ± 0.3% of the β-cell–depleted population expressed this marker (Fig. 2B). As expected, hCD4 expression was undetectable in β-cells from WT mice (Fig. 2B). These results indicate an efficient recombination process and Mcl-1^fl deletion in βMcl-1KO mice.

β-Cell death induced by IL-1β+IFN-γ or TNF+IFN-γ was exacerbated in islets from βMcl-1KO mice (Fig. 2C and D). Neither mRNA nor protein expression of BCL-2 and BCL-XL were altered in islets from βMcl-1KO mice (Fig. 2E–H). The βMcl-1KO mice were born at the expected Mendelian ratio and showed no obvious abnormalities up to 24 weeks (data not shown). Intraperitoneal glucose tolerance tests showed no difference between WT and βMcl-1KO mice (Supplementary Fig. 2A–D). There were also no differences in fed blood glucose or body weight between WT and βMcl-1KO mice (Supplementary Fig. 2E and F). Moreover, the total insulin content of the pancreas, islet morphology, and β-cell response to glucose were normal in the βMcl-1KO mice (Supplementary Fig. 2G–K).

We next evaluated whether βMcl-1KO mice are more susceptible to diabetes induced by MLDSTZ, a treatment that induces a combined toxic autoimmune form of diabetes in mice (31,32). As expected, both WT and βMcl-1KO mice developed hyperglycemia 10 days after the last injection of STZ (Fig. 3A). The pancreatic insulin content was decreased by 60–90% in animals of both genotypes as compared with control buffer–injected mice (Fig. 3C and D). Interestingly, MLDSTZ-induced hyperglycemia was more severe in βMcl-1KO mice, with a twofold increase in the area under the curve (AUC) calculated based on the glycemic values as compared with WT littermates (Fig. 3A and B). In line with this, the insulin content of pancreata from βMcl-1KO mice was >50% lower than in WT mice after MLDSTZ treatment (Fig. 3C). Analysis of the pancreatic sections of both βMcl-1KO and WT mice showed a clear decrease in the numbers of islets as well as insulin staining in the MLDSTZ-treated animals of both genotypes compared with the vehicle-injected mice (Fig. 3E). Islets from the MLDSTZ-treated βMcl-1KO mice displayed reduced insulin staining compared with their WT counterparts (Fig. 3E and Supplementary Fig. 2K). In line with these observations, an increased percentage of α-cells was observed in MLDSTZ-treated mice as compared with their vehicle-treated counterparts (Fig. 3E and Supplementary Fig. 2L). Collectively, these results demonstrate that β-cells from βMcl-1KO mice are abnormally sensitive to stress.

We have previously shown that the cytokine-mediated MCL-1 protein decrease in rat β-cells is a posttranscriptional process since these cells actually displayed increased Mcl-1 mRNA levels (11). We have now made similar observations in human β-cells, since cytokines cause an increase in MCL-1 mRNA expression in EndoCβH1 cells (Supplementary Fig. 3A). Studies in other cell types showed that MCL-1 protein stability can be regulated through phosphorylation by GSK3β, which primes MCL-1 for ubiquitination and proteasomal degradation (33). Inhibition of GSK3 using specific compounds did not affect MCL-1 protein expression under basal condition, but it diminished the cytokine-mediated reduction in MCL-1 protein expression (Fig. 4A). β-Catenin, a known target of GSK3 kinase, was reported to be modulated by the GSK3 inhibitors. Inhibition of GSK3β activity induced a significant protection from cytokine-induced cell death in both rat INS-1E cells and human EndoCβH1 cells (Fig. 4B and C).

GSK3β-targeted phosphorylation sites are conserved between human and rat MCL-1 (33) (Fig. 5A). We therefore generated expression vectors for rat phosphorylation site mutant MCL-1 proteins in which serine 139 or 142 or threonine 143 were substituted for by the phospho-null residue alanine. Treatment with cytokines downregulated the levels of both the overexpressed and exogenous WT MCL-1 protein with similar kinetics as the endogenous MCL-1 protein in INS-1E cells (Supplementary Fig. 3B). The expression levels of all MCL-1 protein forms were comparable in untreated cells (Supplementary Fig. 3C). The T143A mutation diminished the cytokine-mediated decrease in MCL-1 protein expression at 16 h but not at 24 h (Fig. 5B). Conversely, the S139A mutation resulted in more substantial protection against IL-1β+IFN-γ–induced MCL-1 degradation (Fig. 5C). Finally, the S142A mutation did not affect MCL-1 protein turnover (Fig. 5D). A significant decrease in cytokine-mediated caspase-3 activation was observed in INS-1 cells overexpressing the S139A mutant form of MCL-1 (Fig. 5B–E). These results show that GSK3β-mediated phosphorylation controls the levels of MCL-1 protein in cytokine-treated pancreatic β-cells and thereby controls their survival.

Exposure of INS-1E cells to IL-1β+IFN-γ did not affect the general protein ubiquitination pattern (Fig. 6A, left, input), but it increased the ubiquitination of overexpressed WT-Flag-MCL-1 (Fig. 6A, right, anti-Flag). In certain tumor cells, MCL-1 was shown to be ubiquitinated by the E3 ubiquitin ligases F-box and WD repeat domain–containing 7 (FBW7), βTrCP, and MULE (34). We found that FBW7 was downregulated in β-cells upon treatment with IL-1β+IFN-γ (14), and conversely, the levels of MULE and βTrCP were increased under these conditions (Fig. 6B and C). Knockdown of FBW7 with a previously validated siRNA (14) increased MCL-1 expression under basal conditions but had no impact on the cytokine-mediated decrease of this protein (Fig. 6D). Conversely, knockdown of MULE or βTrCP, using two different siRNAs for each target, did not alter MCL-1 protein levels under basal conditions, but it substantially reduced the cytokine-mediated loss of MCL-1 in INS-1E cells (Fig. 6E and F and Supplementary Fig. 4A–D).

USP9x was shown to deubiquitinate MCL-1 in certain tumor cells, thereby protecting MCL-1 from proteasomal degradation (35). Treatment with IL-1β+IFN-γ decreased the expression of USP9x in INS-1E cells (Fig. 6G), and knockdown of USP9x using two different siRNAs potentiated the IL-1β+IFN-γ–mediated decrease in MCL-1 levels (Fig. 6H and Supplementary Fig. 4E and F). These results show that proinflammatory cytokines increase the expression of the E3 ligases βTrCP and MULE but decrease the expression of the deubiqutinase USP9x. These processes together conspire to cause a reduction in MCL-1 protein.

MCL-1 is an antiapoptotic member of the BCL-2 protein family that protects pancreatic cells against apoptosis by sequestering the multi-BH domain proapoptotic BAX/BAK proteins and the proapoptotic BH3-only proteins, such as BIM and PUMA (7,33). We have previously shown that a variety of cytotoxic stimuli, such as proinflammatory cytokines, the ER stressor thapsigargin, the saturated acid palmitate, and the viral mimetic double-stranded RNA, promote apoptosis by decreasing MCL-1 protein levels in rat β-cells (7,11). We observed here that inflammatory cytokines also decrease MCL-1 protein expression in human β-cells and that this contributes to the apoptosis of these cells. Interestingly, dose response experiments indicated that relatively low levels of cytokines were already able to decrease MCL-1 levels and induce death of EndoCβH1 cells. These results are in line with a recent report revealing decreased expression of MCL-1 in islets from patients with T1D infected with diabetogenic enterovirus (12). Collectively, these findings suggest that decreased levels of MCL-1 in β-cells may contribute to the development of diabetes.

We report here that MCL-1 overexpression partially protects human β-cells against the proapoptotic effect of proinflammatory cytokines. To evaluate the protective role of MCL-1 in immune-mediated β-cell death in vivo, we developed the βMcl-1KO mouse strain. βMcl-1KO mice were born at the expected Mendelian frequency and showed no obvious metabolic abnormalities even in adulthood. This is remarkable, because loss of MCL-1 in other essential cell types, such as neurons, cardiomyocytes, or hematopoietic stem/progenitor cells, causes death of mice (36–39). It is therefore possible that during development, β-cells are less dependent on MCL-1 than these other cell types, suggesting that the survival of β-cells may be safeguarded by other prosurvival BCL-2 family members (e.g., BCL-XL) or that several of these prosurvival proteins function in these cells in an overlapping manner.

In agreement with findings from Mcl-1 gene knockdown studies in rodent and human β-cells (Allagnat et al. [11] and present data), islets from the βMcl-1KO mice were abnormally susceptible to cell death induced by both IL-1β+IFN-γ and TNF+IFN-γ treatment. MLDSTZ treatment induces a slow and progressive decrease in insulin levels due to direct toxicity and islet inflammation with consequent β-cell killing (31,32). The development of diabetes in our experiments followed the usual time course in this model, with hyperglycemia appearing 10 days after the end of STZ treatment, indicating that the disease was mainly due to islet inflammation. Interestingly, MLDSTZ treatment induced a more pronounced hyperglycemia in the βMcl-1KO mice compared with their WT littermates, as reflected by a twofold increase in the AUC calculated based on the glycemic levels. In line with these data, there was a >50% decrease in insulin content in βMcl-1KO mice as compared with their WT littermates. Moreover, histological analysis of the pancreas showed a reduced percentage of insulin-positive cells and an increased percentage of glucagon-positive cells in the βMcl-1KO islets. The increased glucagon staining probably reflects the stronger depletion of insulin-positive cells in the KO islets. These results indicate that MLDSTZ treatment causes increased β-cell destruction in the βMcl-1KO mice. This could be due to increased mononuclear cell infiltration in βMcl-1KO mice, since islets from these mice are probably more susceptible to STZ- and immune-induced cell death than WT islets, and this could lead to increased inflammation (40). Overall, our results establish the crucial role for MCL-1 in protecting mouse as well as human β-cells from apoptosis in vitro and during diabetes development in mice in vivo.

Our next aim was to unravel the regulatory mechanisms that drive inflammatory cytokine–mediated MCL-1 downregulation in pancreatic β-cells. Because of its short half-life (∼30 min, compared with ∼20 h for BCL-XL and BCL-2), MCL-1 is disproportionally susceptible to alterations in protein translation (33). MCL-1 is structurally unique among other antiapoptotic proteins of the BCL-2 family, since it contains an N-terminal domain with several phosphodegron motives that are associated with its rapid turnover (33,34). JNK-mediated phosphorylation decreases MCL-1 protein stability, a process that can be prevented by JNK inhibitors (11). Phosphorylation by GSK3β can also promote MCL-1 protein degradation, at least in cancer cells (33). We have now found that inhibition of GSK3β, using two specific compounds, diminishes IL-1β+IFN-γ–mediated MCL-1 protein degradation. This was accompanied by a significant protection against cytokine-induced apoptosis in both rat and, importantly, human β-cells.

We hypothesized that mutations of the GSK-3β–targeted phosphodegrons might disrupt MCL-1 phosphorylation and thereby increase MCL-1 protein stability. In rat MCL-1, threonine 143 (corresponding to Thr163 in human MCL-1) acts as a critical primary phosphorylation site that then primes MCL-1 for subsequent phosphorylation at other sites, as shown in different cell types (41,42). Surprisingly, in our studies, the T143A mutation only delayed cytokine-mediated MCL-1 degradation. This suggests that phosphorylation of T143 contributes to, but is not an absolute prerequisite for, GSK3β-mediated phosphorylation of other sites in MCL-1 and its degradation in INS-1E cells (33,34). This indicates that cell type–specific differences exist in the posttranslational modifications of MCL-1 and the control of its turnover. Furthermore, our studies revealed that serine 139 in rat MCL-1 (corresponding to S159 in human MCL-1) is critical for MCL-1 degradation in INS-1E cells. Indeed, the S159A mutation substantially protected MCL-1 against cytokine-induced degradation in INS-1E cells, and this decreased caspase-3 activation, indicating protection against apoptosis. The S159 residue was shown to be phosphorylated only by GSK3β, reducing the antiapoptotic function of MCL-1 by priming it for ubiquitination and proteasomal degradation (33,34). Accordingly, we found that treatment with IL-1β+IFN-γ induces prominent ubiquitination of MCL-1 in INS-1E cells.

Several E3 ligases, including MULE, βTrCP, and FBW7, contribute to the ubiquitination-mediated proteasomal degradation of MCL-1 in cancer cells after phosphorylation by GSK3β (34). Interestingly, we observed that proinflammatory cytokines modulate the mRNA expression of these three E3 ligases in INS-1E cells. Whereas MULE and βTrCP are upregulated, FBW7 expression is downregulated after exposure to IL-1β+IFN-γ (Fig. 7). In HeLa cells, MULE was shown to ubiquitinate MCL-1 and thereby prime it for proteasomal degradation, whereas, conversely, knockdown of MULE was found to increase MCL-1 protein levels in sarcoma cells and thereby protected them from apoptosis induced by DNA damage (34). In A293T cells, overexpression of βTrCP induced MCL-1 ubiquitination, followed by a decrease in its levels. The opposite was seen when βTrCP expression was silenced (34). In agreement with these reports, we observed that knockdown of both MULE and βTrCP increased the levels of MCL-1 protein even when the INS-1E cells were exposed to IL-1β+IFN-γ. This suggests that proinflammatory cytokine–induced MCL-1 degradation depends on ubiquitination mediated by these E3 ligases. Surprisingly, knockdown of FBW7 in INS-1E cells increased MCL-1 expression only under control (no treatment) conditions, suggesting that this E3 ligase regulates MCL-1 steady-state protein levels but has no role in MCL-1 degradation induced by cytotoxic stressors, at least in this cell type. This may be explained by the fact that depletion of this E3 ligase enhances nuclear factor-κB (NF-κB)–mediated NO generation (14), which has previously been shown to induce ER stress and a consequent decrease in MCL-1 (11) (Fig. 7).

Besides the effects of E3 ligases, MCL-1 turnover is also affected by the opposing action of the deubiquitinase USP9x. USP9x antagonizes K48-linked polyubiquitination and thereby stabilizes the MCL-1 protein, allowing it to inhibit apoptosis (35). Interestingly, exposure to proinflammatory cytokines decreased expression of USP9x in INS-1E cells, and knockdown of USP9x significantly decreased MCL-1 expression in both untreated and cytokine-treated INS-1E cells. Collectively, these findings suggest that the deubiquitinase USP9x contributes to MCL-1 stabilization by antagonizing its K48-linked polyubiquitination and thereby functions as an inhibitor of apoptosis in β-cells (Fig. 7).

In conclusion, the data presented here provide evidence that MCL-1 exerts a critical antiapoptotic role in human β-cells in vitro, and its deficiency affects diabetes development in mice. Moreover, we clarified the mechanisms that regulate MCL-1 protein turnover in rat β-cells when exposed to a proinflammatory milieu (Fig. 7). Prevention of MCL-1 downregulation in β-cells might be achieved via multiple ways: inhibition of GSK3β kinase, mutation of critical residues in the MCL-1 phosphodegron motives, or modulation of ubiquitin-editing enzymes, such as overexpression of the deubiquitinase USP9x. This information might serve as a basis for the development of new strategies to prevent β-cell death in T1D.

Acknowledgments. The authors thank S. Mertens for expert technical assistance, as well as A. Musuaya, M. Pangerl (all from Center for Diabetes Research, Université Libre de Bruxelles, Brussels, Belgium), and Rudi Beyaert (VIB-UGent, Ghent, Belgium) and his team for helpful advice regarding ubiquitination/immunoprecipition and Dr. Philippe Bouillet (Walter and Eliza Hall Institute of Medical Research) for Mcl-1 gene–targeted mice.

Funding. N.M.V. was the recipient of a PhD scholarship from the São Paulo Research Foundation (2015/01237-0). P.M. and D.L.E. received funding from the European Union’s Horizon 2020 research and innovation programme project T2DSystems (667191) and from the Innovative Medicines Initiative 2 joint undertaking (INNODIA) (115797). This joint undertaking receives support from the Union’s Horizon 2020 research and innovation programme and the European Federation of Pharmaceutical Industries and Associations, JDRF, and The Leona M. and Harry B. Helmsley Charitable Trust. The research of the A.S. group was supported by the National Health and Medical Research Council (NHMRC) (1016701) and the Leukemia & Lymphoma Society of America (Specialized Center of Research, 7001-13). A.S. is a recipient of an NHMRC Senior Principal Research Fellow Fellowship (1020363). The research of D.L.E. was supported by the National Funds from Scientific Research (FNRS) (T.0036.13). Work in the A.K.C. group was supported by JDRF (1-2011-589), Actions de Recherche Concertées de la Communauté Française (ARC-Belgium, 20063), and FNRS (Belgium, T.0107.16).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.M. and N.M.V. contributed to the study concept and design; acquired, analyzed, and interpreted the data; and wrote and edited the manuscript. M.F., V.D., and N.P. acquired the data. L.M. and P.M. contributed reagents/materials/analytical tools. A.S. contributed to the study concept and design, analyzed and interpreted the data, contributed reagents/materials/analytical tools, and wrote and edited the manuscript. D.L.E. and A.K.C. contributed to the study concept and design, analyzed and interpreted the data, and wrote and edited the manuscript. All authors revised the article and approved the final version. A.K.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.