PTPN2 Regulates Metabolic Flux to Affect β-Cell Susceptibility to Inflammatory Stress Free

Type 1 diabetes (T1D) is a chronic autoimmune disease that results in the irreversible loss of functional β-cells (1). Many of the genetic factors of T1D have been identified in genome-wide association studies (GWAS), but the molecular mechanisms by which most risk alleles increase susceptibility to T1D remain unclear. Interestingly, >60% of the GWAS candidate genes are strongly expressed in human islets (2), suggesting there are β-cell features that contribute to their own destruction. Together, these data suggest that β-cell dysfunction in T1D-susceptible individuals may be at least partly due to maladaptive β-cell–intrinsic defense mechanisms in response to inflammatory stimuli.

Protein tyrosine phosphatase N2 (PTPN2), also known as T-cell protein tyrosine phosphatase (TC-PCP), is a candidate gene for T1D (1,2). The association of PTPN2 with T1D risk remains incomplete since PTPN2 is expressed in many tissue types, including T cells and β-cells (2,3). It has been determined that many of the risk variants of PTPN2 are noncoding (rs2542151, rs1893217, rs478582, and rs189217) and result in reduced expression (1,4,5). In mice, whole-body Ptpn2-knockout (KO) animals suffer premature lethality (6), and mice lacking Ptpn2 in T cells die of systemic inflammation (7). T-cell–specific Ptpn2-deficient NOD mice showed accelerated development of T1D and autoimmune comorbidities (8). In β-cells, PTPN2 has been shown to have an important role in modulating interferon (IFN) signaling (9–11). Inhibition of Ptpn2 expression in β-cell lines increased activation of the signal transducer and activator of transcription (STAT) signaling pathway and augmented apoptosis induced by IFN-α, IFN-β, and IFN-γ and subsequent triggering of the mitochondrial cell death pathway (9–11). Furthermore, two recent studies of human stem cell-derived β-cells lacking PTPN2 demonstrated the importance of PTPN2 in regulating IFN signaling networks and modulating the endoplasmic reticulum response to promote autoimmune T-cell reactivity in the mutant β-cells (12,13). In a mouse model where Ptpn2 was deleted from all cells of the developing embryonic pancreas using a Pdx1:Cre allele, the mutant mice did not display any discernible phenotypes under normal chow diet conditions but became glucose intolerant when fed a prolonged high-fat diet (14).

Although several studies have explored the β-cell role of PTPN2 in the context of type 2 diabetes or an acute inflammatory response, PTPN2 in mouse β-cells has not been examined in vivo in the context of a chronic or low-grade immune response, as might be seen in the progression of T1D. To investigate whether loss of PTPN2 adversely affected β-cell function and/or survival in this context, we generated β-cell–specific Ptpn2-KO mice (henceforth, PTPN2-βKO) using the INS2:Cre allele (15) and subjected them to T1D-like stress conditions. Similar to previous reports (11), PTPN2-βKO mice displayed only subtle phenotypes under basal conditions but were hyperglycemic due to impaired glucose-stimulated insulin secretion (GSIS) under several T1D-like stress conditions. Similar to previous studies, we also observed alterations in the Janus kinase (JAK)/STAT signaling pathway and increased expression of MHC2-related genes (9,11–13). Notably, however, the mutant islets also displayed reduced oxidative consumption rates and defects in mitochondrial function, suggesting for the first time that PTPN2 regulates β-cell susceptibility by altering cellular fitness over time.

Generation of Ptpn2^fl/fl control mice was previously described (16). To generate β-cell–specific PTPN2-βKO mice, Ptpn2^fl/fl control mice were mated to INS2-Cre mice (15). All animal studies and protocols were approved by the Institutional Animal Care and Use Committee at the Columbia University Medical Center and University of Colorado Anschutz Medical Campus.

Islets were isolated from control and PTPN2-βKO mice (n = 4 mice per genotype) and incubated with and without proinflammatory cytokines for 22 h (1 ng/mL INF-γ, 0.1 ng/mL tumor necrosis factor [TNF]-α, and 0.01 ng/mL interleukin [IL]-1β). Size-matched islets (n = 10 per group) were treated with and without proinflammatory cytokines and in the different glucose conditions: 2.8 mmol/L glucose, 20 mmol/L glucose, or KCl with 2.8 mmol/L glucose. Islet insulin was measured with the mouse insulin ELISA jumbo kit (ALPCO). Secreted insulin was calculated as the percentage of total insulin content.

Islets were isolated and pooled from three mice per group. Size-matched islets (n = 10) were transferred to the XFe96 Spheroid plate well and serum starved at 37°C in a non-CO₂ incubator for 1 h. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were measured sequentially on a Seahorse XFe96 analyzer. Conditions applied were basal (2.8 mmol/L glucose, three times), glucose-stimulated (20 mmol/L glucose, six or seven times), ATP-synthase inhibitor (4 μmol/L oligomycin, three times), uncoupler (4 μmol/L carbonylcyanide-4-(trifluoromethoxy)-phenylhydrazone [FCCP], three times), and complex 1 inhibitor (4 μmol/L rotenone, three times). The Agilent WAVE program was used to analyze the results.

Islets were isolated from mice (n = 3 per genotype) at 10 weeks and 16 weeks of age. Samples fixed with glutaraldehyde/paraformaldehyde were processed at the Boulder Electron Microscopy facility, as previously described (17). Thin sections (60–80 nm) were cut using a Leica UCT ultramicrotome and collected on Formvar-coated transmission electron microscopy (TEM) slot grids. Samples were imaged using a Tecnai T12 Spirit TEM with an AMT charge-coupled device digital camera. Three islets were randomly selected from each mouse for analysis. For each islet, 10 pictures from 10 different regions were assessed. Thirty pictures per each mouse were used to count mitochondria and assess cristae structures.

Islets were isolated from control and PTPN2-βKO mice. After overnight incubation, islets were harvested and lysed with RIPA lysis buffer. A total of 15 µg of protein per sample was electrophoresed on 4–20% Mini-PROTEAN TGX gels (Bio-Rad) under denaturing conditions and blotted onto a polyvinylidene fluoride membrane (Amersham). The blots were blocked in 5% nonfat dry milk in Tris-buffered saline with Tween and incubated for the following primary antibodies overnight at 4°C: STAT3 total (Santa Cruz, SC-482), phosphoSTAT3 (Tyr-705, Cell Signaling, no. 9131), and histone H3 (Abcam, ab176842). Goat anti-rabbit IgG horseradish peroxidase (Abcam, ab205718) secondary antibody was used for the quantification of protein expression with West Pico Plus chemiluminescent substrate (Thermo Fisher Scientific).

Islets were isolated from mice at 12 weeks, and proteins were processed as previously described (18,19). Phosphopeptide enrichment was done in immobilized metal affinity chromatography (IMAC) tips following published protocols (20,21). Samples were analyzed by liquid chromatography (LC)-tandem mass spectrometry (MS/MS) using the same settings as those of a previous study (18). LC-MS/MS data were processed with MaxQuant (22,23) and peptides identified by searching MS/MS spectra against the human Swiss-Prot database (13 April 2017 with 20,198 sequences). The false discovery rate was set to 1% at the level of proteins, peptides, and modifications; no additional filtering was included. The intensities of all 10 MS/MS tag reporter ions were extracted from MaxQuant outputs and analyzed by Perseus (24) for statistical analysis.

Islets were isolated from control and PTPN2-βKO mice at 16 weeks of age. Islets from three control and three PTPN2-βKO mice were treated with proinflammatory cytokines. Total RNA was extracted using the RNeasy Micro Kit (QIAGEN). RNA samples all had RNA integrity number values >8.0. Libraries were constructed using 0.5 µg total RNA. Libraries were prepared using the Universal Plus mRNA-sequencing (Seq) library preparation kit, with NuQuant sequenced by the University of Colorado Anschutz Medical Campus Genomics Core using the NovaSEQ 6000 for paired-end sequencing (2 × 150) from polyA-selected total RNA. Differential expression analysis and data normalization were performed using DESeq2-1.30.0 with an adjusted P value threshold of 0.05 within an R 4.0.3 environment.

Islets were isolated from control and PTPN2-βKO mice at 16 weeks of age (n = 3 per genotype). Islets were starved for 1 h in Krebs-Ringer bicarbonate buffer with 2.8 mmol/L glucose and proinflammatory cytokines and then transferred to noncoated 12 wells containing [¹³C]glucose and the final volume of 500 µL Krebs-Ringer bicarbonate buffer for 30 min. Islet samples were analyzed via ultra-high-pressure LC coupled to MS (UHPLS-MS)-based metabolomics at the University of Colorado School of Medicine Metabolomics Core, where metabolites were extracted and analyzed as previously described (25–28). Metabolites were manually annotated and integrated with Maven (Princeton University) in conjunction with the Kyoto Encyclopedia of Genes and Genomes database. Peak quality was determined using blanks, technical mixes, and [¹³C] abundance (27). [¹³C₂] peak areas were corrected for natural abundance of [¹³C₂]. Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software, Boston, MA), GENE-E (Broad Institute, Cambridge, MA), and MetaboAnalyst 4.0 (29).

All RNA-Seq raw data generated for this study are publicly available at Gene Expression Omnibus accession GSE21582. The proteomics raw data are available upon request. All resources, including mouse strains, generated for this study are available upon request.

Mice specifically lacking PTPN2 in β-cells (Ptpn2^fl/fl; INS2-Cre; henceforth, PTPN2-βKO) were generated by breeding Ptpn2^fl/fl mice (16) with INS2-Cre (rat insulin 2 promoter) mice (15). In the PTPN2-βKO islets, ∼81% of Ptpn2 transcripts were deleted for the floxed exons (Supplementary Fig. 1A), which correlated with a significant decrease in PTPN2 protein expression (Supplementary Fig. 1B and D). The remaining PTPN2 protein likely represents its expression in the nonislet β-cells. Importantly, PTPN2 expression was not altered in the hypothalamus where ectopic INS2-Cre allele expression has been reported (Supplementary Fig. 1C and D). As expected from previous studies using a Pdx1-Cre allele to delete Ptpn2 in all cells of the pancreas during embryonic development (14), PTPN2-βKO animals were born at the expected Mendelian ratios, were viable and fertile, and did not display alterations in body weight at any age tested (Fig. 1A). Furthermore, there were no gross alterations in islet size or composition (Fig. 1B and Supplementary Fig. 1E). Mice carrying the individual INS2:Cre, Ptpn2^fl/fl, and Ptpn2^fl/+ alleles were phenotypically indistinguishable from wild-type mice and therefore used as controls for subsequent analyses. Intraperitoneal glucose tolerance tests (2 g/kg glucose) indicated that male PTPN2-βKO mice had normal glucose tolerance throughout life compared with littermate controls (Fig. 1C). At 11 weeks of age, the PTPN2-βKO mice also did not display any significant defects in in vitro GSIS or in vivo plasma insulin (Supplementary Fig. 1F and G). The mutant mic also displayed normal insulin tolerance at 12 weeks (Supplementary Fig. 1H). By 22 and 26 weeks of age, PTPN2-βKO male mice displayed slightly worsening glucose intolerance, although this did not reach significance (Fig. 1C).

Previous studies have suggested that PTPN2 functions in β-cells to either protect against IFN-induced apoptosis or regulate compensatory insulin secretion when mice are fed a high-fat diet (10,11,14). Although these studies proposed relatively different roles for PTPN2 in β-cells, they both demonstrated PTPN2 deficiency led to enhanced tyrosyl phosphorylation of STAT3, which we were able to confirm in the PTPN2-βKO mice (Supplementary Fig. 1I and J). Despite the shared observation that PTPN2 regulated STAT3 phosphorylation, the previous two studies identified nonoverlapping sets of downstream gene changes in immortalized rat and mouse cell lines using a candidate gene approach. To gain a more unbiased view of the downstream pathways affected by the β-cell disruption of Ptpn2, we performed RNA-Seq analysis on islets isolated from adult PTPN2-βKO animals compared with controls. Consistent with the lack of an overt glucose intolerance phenotype, only a small number (126) of genes were significantly differentially expressed, and there were no changes in insulin, endocrine hormones, and islet transcription factor gene expression (Fig. 2A, Supplementary Fig. 2, and Supplementary Table 2). Interestingly, however, the dysregulated gene set contained a disproportionate number of genes encoding proteins involved in mitochondrial function and related metabolic pathways, including PTEN-induced kinase 1 (PINK1), a protein that regulates mitochondrial quality control (Fig. 2B) (30). This analysis suggests that although PTPN2-βKO islets do not display an overt phenotype in basal conditions, they may harbor underlying metabolic defects.

To determine whether changes in mitochondrial gene expression corresponded to overt defects in β-cell mitochondria, we performed TEM of β-cells from wild-type and PTPN2-βKO islets at 10 and 16 weeks of age. This analysis demonstrated that there were no changes in mitochondrial number and cristae structure in 10-week-old mice (3,881 control mitochondria vs. 3,953 βKO mitochondria) (Fig. 2C and Supplementary Fig 3); however, by 16 weeks of age, mitochondria in the PTPN2-βKO β-cells displayed abnormal cristae structure and an ∼20% increase in mitochondrial numbers (4,789 control mitochondria vs. 5,957 βKO mitochondria) (Fig. 2D and Supplementary Fig. 3). The changes in mitochondrial gene expression, morphology, and numbers supported the idea that PTPN2-βKO mice harbor subtle mitochondrial defects that could cause increased β-cell susceptibility to exogenous environmental stressors such as observed in T1D. To test this possibility, we assessed islet mitochondrial function in basal (Fig. 3A–D and Supplementary Fig. 4A and B) and low-dose proinflammatory cytokine treatment conditions (IFN-γ: 1 ng/mL, TNF-α: 0.1 ng/mL, and IL-1β: 0.01 ng/mL) for 24 h (Fig. 3E–H and Supplementary Fig. 4C and D). The dosage of cytokines was titrated down to a level that would induce a stress response but would not cause extensive apoptosis. In 16- to 17-week-old mice, at the age when significant changes in mitochondrial morphology and numbers were first observed, OCRs were also significantly decreased in the cytokine-treated PTPN2-βKO islets (Fig. 3A vs. E and Supplementary Fig. 4A vs. C), and the defect became more pronounced with age (Fig. 3G and Supplementary Fig. 4D). Alternatively, at 16–17 weeks, there were no apparent changes in the ECAR of islets from wild-type or mutant islets in basal or cytokine-treated conditions (Fig. 3B vs. F and Supplementary Fig. 4B vs. D). By 24 weeks of age, the PTPN2-βKO islets displayed a moderate decrease in the ECAR, although it did not reach significance (Fig. 3H). This analysis suggests that PTPN2 is necessary for optimal mitochondrial respiration and that deletion of Ptpn2 results in impaired β-cell metabolic responses to cytokine-induced stress.

To determine whether the mitochondrial defects associated with PTPN2-βKO islets impaired β-cell response to external stress in vivo, we treated PTPN2-βKO mice with a single high-dose (150 mg/kg) of streptozotocin (STZ) to mimic an acute inflammatory environment. A single high dose of STZ was sufficient to induce severe hyperglycemia in both mutant and control mice within 2 days of injection (Fig. 4A); however, survival of the PTPN2-βKO mice (n = 17) was impaired to a greater extent than the controls (n = 16) (Fig. 4B). Because the rapid lethality precluded more in-depth phenotypic analyses, we assessed mice treated with multiple low doses of STZ (50 mg/kg). Under these conditions, the PTPN2-βKO mice displayed consistently elevated ad libitum blood glucose levels by 2 weeks posttreatment and were significantly glucose intolerant by 24 days postinjection (Fig. 4C and D).

To begin to identify the molecular changes associated with β-cell depletion of Ptpn2, we moved to an in vitro experimental paradigm to assess β-cell function under temporally controlled stress conditions. Consistent with the phenotypes observed in vivo, cytokine-treated PTPN2-βKO islets displayed impaired insulin secretion upon glucose or KCl stimulation at ages that corresponded to overt changes in mitochondria (Fig. 4E–G). RNA-Seq on the 16-week-old wild-type and mutant islets exposed to low-dose proinflammatory cytokines for 24 h (n = 3) identified a small but significant number (659) of genes that were differentially expressed (adjusted P < 0.05) in cytokine-treated wild-type versus PTPN2-βKO. Gene ontology (GO) term analysis revealed that most of the downregulated gene changes were associated with the insulin and hormone secretion pathways, which corresponds with the observed glucose tolerance defects (Fig. 5A and Supplementary Table 2). Not surprisingly, a large number of proinflammatory genes were induced in both control and mutant cohorts of mice in response to cytokine treatment; however, they were increased to a greater extent in the PTPN2-βKO cohort, which is consistent with a decreased cellular fitness that cannot adapt appropriately to metabolic demand (Fig. 5B and Supplementary Table 3). Assessment of the gene expression changes associated with the different GO term categories revealed changes in the expression of transcriptional regulators known to confer β-cell identity and function (Fig. 5C), a small number of genes essential for optimal insulin secretion (Fig. 5D), and a large number of β-cell functional genes (Fig. 5E). These findings are consistent with studies suggesting that compromised metabolic flux can impact β-cell identity and function (31,32). Finally, many genes involved in mitochondrial function were also dysregulated, including the significant downregulation of Ldhb (lactate dehydrogenase B) and Idh2 (isocitrate dehydrogenase 2 and NADP⁺) (Fig. 5F–H).

To determine whether the altered genetic programs in the cytokine-treated PTPN2-βKO islets were conserved in human β-cells, we compared the differential expression data from the cytokine-treated PTPN2-βKO islets to the published data sets from EndoC-βH1 cells that were depleted for PTPN2 and treated with IFN-γ (12). This analysis revealed that cytokine-challenged mouse and human β-cells depleted for PTPN2 shared 3,146 differentially expressed genes, representing ∼64% of the genes changed in the PTPN2-knockdown human β-cells (P = 1.31 E−39), suggesting PTPN2 has conserved functions between the two species (Fig. 5I).

The cumulative in vitro and in vivo analysis of the PTPN2-βKO mice suggested that the mutant islets had suppressed metabolic flux that was exacerbated upon stress induction. To determine which aspects of mitochondrial function and glucose metabolism were compromised in the cytokine-treated PTPN2-βKO islets, we used [¹³C]glucose tracing to detect potential alterations in glycolytic intermediates, members of the pentose phosphate pathway, tricarboxylic acid cycle (TCA) metabolites and several amino acids upon 20 mmol/L glucose stimulation (Supplementary Fig. 5A). Glucose stimulation of control islets induced increases in phosphoenolpyruvate, citrate, and TCA-derived amino acid pools (Fig. 6), as previously reported in β-cell lines and mouse islets (33,34). Compared with wild-type islets, however, pools of unlabeled citrate and cis-aconitate were significantly decreased in the PTPN2-βKO islets (Fig. 6F and G). Furthermore, upon glucose stimulation, the mutant islets had reduced [¹³C] incorporation into citrate and succinate, which corresponded to the observed reduction in oxidative metabolism. Similar effects on oxidative phosphorylation were observed when PTPN2 was deleted from cytokine-treated human embryonic stem cell–derived β-like cells, again suggesting a conservation of PTPN2 function in stress conditions (13). There were no discernable changes in components of the pentose phosphate pathway (Supplementary Fig. 5A and B).

Although it has been shown that PTPN2 inhibits the activation of STAT3 in β-cells (10,11), we sought to identify novel substrates of PTPN2 that could more completely explain the observed metabolic phenotype. Myc epitope-tagged versions of the wild-type and substrate-trapping form of PTPN2 were expressed in the murine MIN6 cell line. Coimmunoprecipitation, followed by MS, identified 48 proteins that potentially interacted with PTPN2 in two independent experiments (Fig. 7 and Supplementary Table 4). Not surprisingly, many of these proteins function to regulate glycolysis, the TCA cycle, and gluconeogenesis pathways, including two key enzymes involved in regulating metabolic function: ATP-citrate synthase (ACLY) and malate dehydrogenase (MDH2) that are known to be tyrosyl phosphorylated. To determine whether deletion of PTPN2 corresponded to altered protein phosphorylation status, islets were isolated from control and PTPN2-βKO mice and subjected to global phosphoproteomics. This analysis identified 80 peptides with statistically significant changes in their phosphorylation status (Fig. 7 and Supplementary Table 5), of which ACLY and MDH2 were identified as putative interacting proteins in the MS and contained peptides with altered tyrosyl phosphorylation. Notably, ACLY tyrosyl phosphorylation was significantly increased in the absence of Ptpn2, suggesting that ACLY is an important PTPN2 substrate that contributes to the dysregulation of mitochondrial function in the mutant islets.

GWAS originally identified PTPN2 as a T1D candidate gene. Although the function of PTPN2 in T cells has been extensively characterized (35–37), its in vivo role in pancreatic β-cells has been less well explored, primarily focusing on known inflammatory signaling pathways. In the current study, we generated mice carrying a β-cell deletion of Ptpn2 to specifically assess the β-cell function of PTPN2 in the context of low-level cytokine-induced stress that would more accurately reflect chronic stress conditions. Similar to the panc-PTPN2-KO mice in which Ptpn2 was deleted from the entire embryonic pancreas (14), PTPN2-βKO mice did not exhibit significant metabolic defects in basal conditions; however, blood glucose levels trended upward with age. Consistent with the idea that PTPN2-βKO mice harbored underlying β-cell defects, the mutant β-cells began to display abnormal cristae formation by 4 months of age. The corresponding increase in mitochondria numbers and the fact that the mutant mice remained relatively euglycemic suggested that the mutant β-cells could adequately compensate under basal conditions. However, in response to in vivo STZ administration and in vitro low-dose proinflammatory cytokine treatment, PTPN2-βKO islets were unable to functionally compensate, displaying decreased OCRs and impaired insulin secretion (Fig. 8).

Our transcriptome analysis suggested that in addition to the abnormal cristae structure in the PTPN2-βKO mutant β-cells in basal conditions, there were already alterations in the expression of a small number of mitochondrial- and metabolism-related genes. In particular, there was a reduction in the expression of the Pink1 gene that encodes a protein that protects cells from stress-induced mitochondrial dysfunction (30), which would also be consistent with an inability for the mutant cells to withstand cytokine-induced stress. Cytokine treatment led to additional differential gene expression changes, including the downregulation of Idh2 mRNA. A recent study by Zhang et al. (38) demonstrated that IDH2 catalyzes the reductive carboxylation of 2-ketoglutarate to isocitrate, resulting in the impairment of glucose-stimulated TCA cycle flux, the lowering of NADPH levels, and inhibition of insulin secretion (38). Furthermore, Idh2 mRNA and protein expression were shown to be decreased in two independent mouse models of type 2 diabetes that displayed impaired insulin secretion due to mitochondrial defects (39). Finally, Campbell et al. (40) demonstrated that the existence of a counter-clockwise TCA cycle flux via IDH2 was critical for glucose- and glutamine-stimulated insulin secretion. This pathway also serves as a source of citrate and isocitrate (40), which could lead to the reduced citrate flux and resulting defects in insulin secretion observed in the PTPN2-βKO islets (41). Although we also observed changes in IFN signaling networks, as previously reported (9–12), these genes were altered to similar extents in both wild-type and PTPN2-βKO islets. This may suggest they were primarily induced by the cytokine treatment and had a lesser response to the loss of Ptpn2.

Although PTPN2 has been previously shown in several different organs and cancers to dephosphorylate nonreceptor protein tyrosine kinases such as JAK1, JAK2, JAK3, Src family kinases, STAT1, STAT3, and STAT6, which likely contribute to the altered β-cell phenotype, it has also been shown to regulate additional cell-specific substrates, as previously reviewed (3). The identification of ACLY as a potential substrate of PTPN2 is of particular interest given the importance of this enzyme in regulating several aspects of cellular metabolism and gene expression. ACLY is critical for catalyzing the formation of cytosolic acetyl-CoA from citrate and, as a consequence, de novo lipogenesis. ACLY also mediates glucose-responsive histone acetylation to influence gene expression in many different cell types (42). ACLY has been shown to be phosphorylated at several different sites within the protein, and increased serine phosphorylation has generally been associated with higher enzyme activity (43–45). Specifically, in vitro studies demonstrated that phosphorylation at serine 455 (Ser455) increases ACLY activity by sixfold (44) and that AKT-dependent phosphorylation at Ser455 has been shown to sustain acetyl-CoA production and enable high levels of histone acetylation in low-glucose conditions (46). Furthermore, several studies have identified roles for phosphorylation of ACLY at Ser455 during innate immune responses, as previously reviewed (47). Studies in cancer cells have also determined that acetylation of ACLY at lysine residues 540, 546, and 554 under high-glucose conditions increases ACLY protein stability by blocking ubiquitylation and subsequent degradation (48). Although there are many studies exploring how serine and lysine posttranslational modifications regulate ACLY activity, a role for ACLY tyrosine phosphorylation has been less well explored. However, a recent study by Basappa et al. (45) identified six tyrosine residues of ACLY that are phosphorylated by the Src-family kinase Lyn in acute myeloid leukemia patient-derived cells. Inhibition of Lyn reduced ACLY enzyme activity, supporting the importance of a tyrosine kinase-mediated pathway for ACLY activation (45). Studies to characterize the function of ACLY in pancreatic β-cells have produced conflicting results—showing both positive and negative consequences of ACLY activity on β-cell functions, which may reflect the importance of posttranslational modifications (49–51). In this current study, it appears that increased tyrosyl phosphorylation of ACLY causes detrimental effect on metabolic activity and insulin secretion that could be attributed to both its role in the generation of excess acetyl-CoA and increased fatty acid oxidation to impact downstream mitochondrial function and regulation of gene expression, and the resulting changes could be compounded by the reduction in citrate production. Ultimately, the defects associated with deletion of Ptpn2 and increased phosphorylation of ACLY are subtle but cause the metabolic set point to be sufficiently altered to compromise the ability of β-cells to adequately respond to stresses associated with an autoimmune attack (Fig. 8).

In summary, we have demonstrated a novel role for PTPN2 in regulating pancreatic β-cell fitness through the modulation of mitochondrial function and insulin secretion. Although the deletion of Ptpn2 does not result in overt β-cell dysfunction in basal conditions, we demonstrated that the mutant β-cells are more susceptible to extrinsic stressors. The increased cell susceptibility is consistent with the fact that T-cell–specific inactivation of PTPN2 in mice caused less severe autoimmune phenotypes than observed in global KO mice (6,7). It is likely that mutation of Ptpn2 in T cells initiates an autoimmune process that when combined with Ptpn2-deficiency in the target cells makes them more vulnerable to the resulting cytokine-induced inflammation. Future studies of PTPN2 function in other autoimmune-targeted organs will reveal whether underlying mitochondrial defects also contribute to their reduced ability to withstand the pathogenic effects of inflammatory cytokines.

Acknowledgments. Electron microscopy was done at the University of Colorado, Boulder Electron Microscopy Services Core Facility in the Molecular, Cellular & Developmental Biology Department, with the technical assistance of facility staff. The authors thank the University of Colorado School of Medicine Metabolomics Core for their advice and technical support for the glucose flux assays.

Funding. This work was supported by National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases grants R01 DK12536, R01 DK82590, and P30 DK116073 to L.S. and by P30 DK048520 to the University of Colorado School of Medicine Metabolomics Core. Y.K.K. was supported by a Diabetes Research Connection Fellowship. Y.R.K. had institutional support from Columbia University MD/PhD Program and a Sara & Arnold P. Friedman Award. E.S.N. performed the phosphoproteomics in the Environmental Molecular Sciences Laboratory, a national scientific user facility sponsored by the Department of Energy and located at Pacific Northwest National Laboratory, which is operated by Battelle Memorial Institute for the Department of Energy under contract DE-AC05-76RL01830.

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

Author Contributions. Y.K.K. performed most of the experiments and assisted with data analysis and manuscript preparation. Y.K.K. and L.S. wrote the manuscript. Y.R.K. initiated the study and characterized the mouse model (Supplementary Fig. 1) and performed the STZ experiments described in Fig. 4A–D and the proteomics analysis in Supplementary Table 3. D.S. performed the co-IP and Western blot analyses. K.L.W. analyzed the RNA-Seq data and generated the heat maps. M.G. processed the tissue and extracted the protein for the phosphoproteomics analysis. T.Z. and G.K. generated the Ptpn2 floxed allele and shared the mutant animals prior to publication. C.-F.T. and E.S.N. performed the phosphoproteomics MS and analysis. L.S. conceived the study, guided the project, and assisted with interpreting data. Y.K.K., Y.R.K., and L.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Islet Biology Keystone meeting, Santa Fe, NM, 27–31 January 2020; the Cold Spring Harbor Laboratories Metabolism virtual meeting, 26–29 October 2021; Human Islet Research Network 2023 Investigator Meeting, Washington, DC, 19–21 September 2023; and the 13th International Congress of Diabetes and Metabolism, Gyeongju, Korea, 19–21 October 2023.