Sex differences in islet stress responses support female β cell resilience

Objective Pancreatic β cells play a key role in maintaining glucose homeostasis; dysfunction of this critical cell type causes type 2 diabetes (T2D). Emerging evidence points to sex differences in β cells, but few studies have examined male-female differences in β cell stress responses and resilience across multiple contexts, including diabetes. Here, we address the need for high-quality information on sex differences in β cell and islet gene expression and function using both human and rodent samples. Methods In humans, we compared β cell gene expression and insulin secretion in donors with T2D to non-diabetic donors in both males and females. In mice, we generated a well-powered islet RNAseq dataset from 20-week-old male and female siblings with similar insulin sensitivity. Our unbiased gene expression analysis pointed to a sex difference in the endoplasmic reticulum (ER) stress response. Based on this analysis, we hypothesized female islets would be more resilient to ER stress than male islets. To test this, we subjected islets isolated from age-matched male and female mice to thapsigargin treatment and monitored protein synthesis, cell death, and β cell insulin production and secretion. Transcriptomic and proteomic analyses were used to characterize sex differences in islet responses to ER stress. Results Our single-cell analysis of human β cells revealed sex-specific changes to gene expression and function in T2D, correlating with more robust insulin secretion in human islets isolated from female donors with T2D compared to male donors with T2D. In mice, RNA sequencing revealed differential enrichment of unfolded protein response pathway-associated genes, where female islets showed higher expression of genes linked with protein synthesis, folding, and processing. This differential expression was physiologically significant, as islets isolated from female mice were more resilient to ER stress induction with thapsigargin. Specifically, female islets showed a greater ability to maintain glucose-stimulated insulin production and secretion during ER stress compared with males. Conclusions Our data demonstrate sex differences in β cell gene expression in both humans and mice, and that female β cells show a greater ability to maintain glucose-stimulated insulin secretion across multiple physiological and pathological contexts.


INTRODUCTION
Pancreatic b cells make and secrete insulin, an essential hormone required to maintain whole-body glucose homeostasis. Emerging evidence from multiple species points to biological sex as an important, but often overlooked, factor that affects b cell biology [1e6]. Large-scale surveys of gene expression in mice and humans show that differences exist between the sexes in the pancreas [7e9], in islets [10], and in b cells specifically [4,11]. Humans also have a sex-specific b cell gene expression response to aging [12], and show male-female differences in pancreatic b cell number [6]. With respect to b cell function, most data from rodent and human studies suggests glucose-stimulated insulin secretion is higher in females than in males [5,10,13e16]. While male-female differences in peripheral insulin sensitivity [15,17e27] may contribute to these differences, sex-biased insulin secretion in humans persists in the context of equivalent insulin sensitivity between males and females [5]. Whether sex differences in other aspects of b cell gene expression and function are similarly independent of insulin sensitivity in rodents and humans remains unclear, as insulin sensitivity is not routinely monitored across datasets showing sex differences in b cell biology. Biological sex also affects the risk of developing T2D. Across many population groups, men are at a higher risk of developing T2D than women [28e31]. Some of this differential risk is explained by lifestyle and cultural factors [31e33]. Biological sex also plays a role; however, as the male-biased risk of developing diabetes-like phenotypes is conserved across multiple animal models [22,34e39]. Despite a dominant role for b cell function in T2D pathogenesis [40,41], T2Dand stress-associated changes to b cell gene expression and function this important cell type. Our data show male-female differences in islet and b cell gene expression and stress responses in both humans and mice. These differences contribute to sex differences in b cell resilience, where we find female b cells show a greater ability to maintain glucose-stimulated insulin secretion in response to stress and T2D in mice and humans, respectively. Given that an insulin tolerance test indicated similar insulin sensitivity between the male and female mice used in our study, our findings suggest biological sex is an important variable to consider in studies on islet and b cell function.

Animals
Mice were bred in-house or purchased from the Jackson Laboratory. Unless otherwise stated, islets were isolated from C57BL/6J mice aged 20e24 weeks. Animals were housed and studied in the UBC Modified Barrier Facility using protocols approved by the UBC Animal Care Committee and in accordance with international guidelines. Mice were housed on a 12-hour light/dark cycle with food and drinking water ad libitum. Mice were fed a regular chow diet (LabDiet #5053); 24.5% energy from protein, 13.1% energy from fat, and 62.4% energy from carbohydrates.

Islet isolation, culture, dispersion and treatment
Mouse islet isolations were performed by ductal collagenase injection followed by filtration and hand-picking, using modifications of the protocol described by Salvalaggio [50]. Islets recovered overnight, in islet culture media (RPMI media with 11.1 mM D-glucose supplemented with 10% vol/vol fetal bovine serum (FBS) (Thermo: 12483020) and 1% vol/vol Penicillin-Streptomycin (P/S) (GIBCO: 15140-148)) at 37 C with 5% CO 2 . After four washes with Minimal Essential Medium [L-glutamine, calcium and magnesium free] (Corning: 15-015 CV) islets were dispersed with 0.01% trypsin and resuspended in islet culture media. Cell seedings were done as per the experimental procedure (protein synthesis: 20,000 cells per well, live cell imaging: 5,000 cells per well). ER stress was induced by treating islets with the SERCA inhibitor thapsigargin. For assays less than 24 h, we used (11.1 mM D-glucose RPMI, 1% vol/vol P/S). For assays greater than 24 h we used (11.1 mM D-glucose RPMI, 1% vol/vol P/S, 10% vol/ vol FBS).

Analysis of protein synthesis
Dispersed islets were seeded into an optical 96-well plate (Perkin Elmer) at a density of approximately 20,000 cells per well in islet culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S, 10% vol/ vol FBS). 24 h after seeding, treatments were applied in fresh islet culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S). After incubation, fresh culture media was applied (11.1 mM D-glucose RPMI, 1% vol/vol P/S), then supplemented with 20 mM OPP (Invitrogen) and respective drug treatments. The assay was performed according to manufacturer's instructions. Cells were imaged at 10Â with an ImageXpress MICRO high-content imager and analyzed with MetaXpress (Molecular Devices) to quantify the integrated staining intensity of OPP-Alexa Fluor 594 in cells identified by NuclearMask Blue Stain.

Live cell imaging
Dispersed islets were seeded into 384-well plates (Perkin Elmer) at a density of approximately 5,000 cells per well. Islet viability was measured with the TC20 Automated Cell Counter (Bio-Rad: 1450102) with Trypan Blue (Bio-Rad: 1450021). Islets were allowed to adhere for 48 h in islet culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S, 10% vol/vol FBS). Cells were stained with Hoechst 33342 (Sigmae Aldrich) (0.05 mg/mL) and propidium iodide (SigmaeAldrich) (0.5 mg/ mL) for 1 h in islet culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S, 10% vol/vol FBS) prior to the addition of treatments and imaging. 384-well plates were placed into the environmentally-controlled (37 C, 5% CO2) ImageXpress MICRO high content imaging system. To measure cell death, islet cells were imaged every 2 h for 84 h. Met-aXpress software was used to quantify cell death, defined as the number of Propidium Iodide-positive/Hoechst 33342-positive cells. To measure Ins2 gene activity, dispersed islets from Ins2 GFP/WT mice aged 21e23 weeks were used [51]. Islet cells were imaged every 30 min for 60 h. MetaXpress analysis software and custom R scripts were used to perform single-cell tracking of Ins2 GFP/WT b cells as previously described [51].
1 mM Tg was added to the 3 mM low glucose pre-incubation buffer 4 h prior, 2 h prior, or at the start of the low glucose incubation period. Islets were incubated in KRB with 3 mM glucose then 20 mM glucose for 45 min each. Supernatant was collected after each stimulation. Islet insulin and proinsulin content was extracted by freeze-thawing in 100 mL of acid ethanol, then the plates were shaken at 1200 rpm for 10 min at 4 C to lyse the islets. Insulin was measured by Rodent Insulin Chemiluminescent ELISA (ALPCO: 80-INSMR) and proinsulin by Original Article Rat/Mouse Proinsulin ELISA (Mercodia: 10-1232-01). Measurements were performed on a Spark plate reader (TECAN).

Blood collection and in vivo analysis of glucose homeostasis and insulin secretion
Mice were fasted for 6 h prior to glucose and insulin tolerance tests. During glucose and insulin tolerance tests, tail blood was collected for blood glucose measurements using a glucometer (One Touch Ultra 2 Glucometer, Lifescan, Canada). For intraperitoneal (i.p.) glucose tolerance tests, the glucose dose was 2 g glucose/kg of body mass. For insulin tolerance tests, the insulin dose was 0.75U insulin/ kg body mass. For measurements of in vivo glucose-stimulated insulin secretion, femoral blood was collected after i.p. injection of 2 g glucose/kg body mass. Blood samples were kept on ice during collection, centrifuged at 2000 rpm for 10 min at 4 C and stored as plasma at À20 C. Plasma samples were analysed for insulin using Rodent Insulin Chemiluminescent ELISA (ALPCO: 80-INSMR). Glucose-stimulated insulin secretion from an independent large cohort of Insr f/f :Ins1Cre À/À :nTnG þ/À mice reared in our facility was replotted from a published Johnson lab study [53]. Blood glucose was monitored using control and insulin-reduced mice. Control mice were Ins1 À/À :Ins2 f/f :mTmG þ tamoxifen and Ins1 À/À :Ins2 f/f : Pdx1Cre ERT :mTmG þ corn oil [54]. Because we confirmed there were no significant differences in blood glucose between control genotypes, the data were combined in our analysis. Insulin-reduced mice were generated by injecting Ins1 À/À :Ins2 f/f :Pdx1Cre ERT :mTmG mice with tamoxifen (3 mg/40 g body weight, dissolved in corn oil, for 4 consecutive days) at 6e8 weeks [54].

RNA sequencing
To assess basal transcriptional differences, islets from male and female mice (n ¼ 9M, 8F) were snap-frozen and stored at À80 C until RNA extraction. To assess Tg-induced transcriptional changes, islets from each mouse were treated with DMSO or 1 mM Tg for 6-or 12hours in culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S). Four groups per sex (eight groups total) were analyzed: 6 h DMSO, 6 h Tg, 12 h DMSO and 12 h Tg, n ¼ 3e4 per group, each n represents pooled islet RNA from two mice. Islets were frozen at À80 C in 100 mL of RLT buffer (Qiagen) with beta mercaptoethanol (1%). RNA was isolated using RNeasy Mini Kit (Qiagen #74106) according to manufacturer's instructions. RNA from 43 to 62 islets was pooled from two mice and 19e150 ng of RNA was sequenced per pooled sample. RNA sequencing was performed at the UBC Biomedical Research Centre Sequencing Core. Sample quality control was performed using the Agilent 2100 Bioanalyzer System (RNA Pico LabChip Kit). Qualifying samples were prepped following the standard protocol for the NEBNext Ultra II Stranded mRNA (New England Biolabs). Sequencing was performed on the Illumina NextSeq 500 with Paired End 42bp Â 42bp reads. Demultiplexed read sequences were then aligned to the reference sequence (UCSC mm10) using STAR aligner (v 2.5.0b) [55]. Gene differential expression was analyzed using DESeq2 R package [56]. Pathway enrichment analysis were performed using Reactome [57]. Over-representation analysis was performed using NetworkA-nalyst3.0 (www.networkanalyst.ca) [58].

Proteomics
Islets were treated with DMSO or 1 mM Tg for 6 h in islet culture media (11.1 mM D-glucose RPMI, 1% vol/vol P/S). Two groups per sex (four groups total): 6 h DMSO and 6 h Tg, n ¼ 5e7 per group, each n represents 200e240 islets pooled from two mice. Islet pellets were frozen at À80 C in 100 mL of SDS lysis buffer (4% SDS, 100 mM Tris, pH 8) and the proteins in each sample were precipitated using acetone. The University of Victoria proteomics service performed non-targeted quantitative proteomic analysis using data-independent acquisition (DIA) with LC-MS/MS on an Orbitrap mass spectrometer using 1 mg of protein. A mouse FASTA database was downloaded from Uniprot (http://uniprot.org). This file was used with the 6 gas phase fraction files from the analysis of the chromatogram library sample to create a mouse islet-specific chromatogram library using the EncyclopeDIA (v 1.2.2) software package (Searle et al., 2018). This chromatogram library file was then used to perform identification and quantitation of the proteins in the samples again using EncyclopeDIA with Overlapping DIA as the acquisition type, trypsin used as the enzyme, CID/HCD as the fragmentation, 10 ppm mass tolerances for the precursor, fragment, and library mass tolerances. The Percolator version used was 3.10. The precursor FDR rate was set to 1%. Protein abundances were log2 transformed, imputation was performed for missing values, then proteins were normalized to median sample intensities. Differential expression was analyzed using limma in Perseus [59].
2.10. Analysis of the transcriptome and partial proteome Tg-induced changes to gene expression and protein levels were compared 6 h post-treatment. Log2-transformed fold change values were used to assess the congruence between our proteomics data and RNAseq data. Genes and proteins that were concordantly altered by Tg treatment at both the mRNA and protein level were searched in 2.12. Statistical analysis Statistical analyses and data presentation were carried out using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA) or R (v 4.1.1). Correlation plots were generated using the corrplot R package (v 0.92) with default settings [60]. All R codes are published on github (https://github.com/johnsonlabubc/ER-Stress-in-Mouse-Beta-Cells-Data-Analysis). Student's t-tests or two-way ANOVAs were used for parametric data. A ManneWhitney test was used for non-parametric data. Statistical tests are indicated in the figure legends. For all statistical analyses, differences were considered significant if the p-value was less than 0.05. *: p < 0.05; **p < 0.01; ***p < 0.001. Data were presented as means AE SEM with individual data points from biological replicates.

Sex differences in b cell transcriptional and functional
responses in ND and T2D human islets Gene expression studies on human pancreas and islets identify significant sex differences in gene expression [7,10]. Indeed, >1500 genes expressed in the pancreas show sex-biased expression [7]. While our dataset contained fewer individuals, we also note sex dif- 1). Given this differential gene expression, we wanted to define b cellspecific gene expression changes in T2D in each sex. We therefore used a recently-compiled meta-analysis of publicly available scRNAseq datasets from male and female human islets [61]. Our goal was to use sex-based analysis to determine whether b cell gene expression changes in T2D are shared between the sexes. In line with prior reports [12], b cells from non-diabetic (ND) and T2D donors showed significant transcriptional differences. In b cells isolated from female T2D donors, mRNA levels of 127 genes were significantly different from ND female donors (77 downregulated, 50 upregulated in T2D) (Figure 1Ae C). In b cells isolated from male T2D donors, 462 genes were differentially expressed compared with male ND donors (138 downregulated, 324 upregulated in T2D) (Figure 1AeC). Of the 660 genes that were differentially regulated in T2D, 71 were differentially regulated in both males and females (15 downregulated, 56 upregulated in T2D) (Figs. S1AeC); however, the fold change for these 71 shared genes was different between males and females ( Fig. S1A; Supplementary file 2). This suggests that for shared genes, the magnitude of gene expression changes in T2D was not the same between the sexes. Beyond shared genes, we observed that the majority of differentially expressed genes in T2D (589/660) were unique to either males or females (Figs. S1B,C; Supplementary file 2). Indeed, the most prominent gene expression changes in T2D were found in genes that were unique to one sex (Figs. S2A,B; Supplementary file 2). While these data do not address the reasons for the sex-biased risk of T2D, and could reflect differences in medication [62e65], age [12,16,66e68], and body mass index [15,68], our data suggest biological sex influences b cell gene expression in T2D.
To determine which biological pathways were altered in b cells of T2D donors from each sex, we performed pathway enrichment analysis.
Genes that were upregulated in b cells isolated from T2D donors included genes involved in Golgi-ER transport and the unfolded protein response (UPR) pathways ( Figure 1DeF; Supplementary file 2). While these biological pathways were significantly upregulated in T2D in both males and females, w75% of the differentially regulated genes in these categories were unique to each sex ( Table 1). Genes that were downregulated in b cells from T2D donors revealed further differences between the sexes: biological pathways downregulated in b cells from female T2D donors included cellular responses to stress and to stimuli ( Figure 1E; Supplementary file 2), whereas b cells from male T2D donors showed downregulation of pathways associated with respiratory electron transport and translation initiation ( Figure 1F; Supplementary file 2). Our analysis therefore suggests that sex-biased b cell gene expression responses to T2D may influence different cellular processes in males and females.
The sex-biased b cell transcriptional response in T2D prompted us to compare glucose-stimulated insulin secretion in each sex from ND and T2D human islets using data from the Human Pancreas Analysis Program database [69]. In ND donors, islets from males and females showed similar patterns of insulin secretion in response to various stimulatory media ( Figure 1G,H). In donors with T2D, we found that insulin secretion was impaired to a greater degree in islets from males than in females ( Figure 1GeK). This difference cannot be fully attributed to a sex difference in disease severity, as our analysis of donor characteristics revealed no significant correlation between sex and HbA1c (Fig. S3A). Indeed, in male but not female islets, insulin secretion was lower in donors with T2D following stimulation with both high glucose and IBMX ( Figure 1I,J), which potentiates insulin secretion by increasing cAMP levels to a similar degree as the incretins [70]. Human islets from female donors with T2D therefore show a greater ability to maintain glucose-stimulated insulin secretion than islets from males with T2D ( Figure 1K). Indeed, while diabetes status was the main donor characteristic that correlated with changes in insulin secretion (Fig. S3B), we noted that in T2D sex and age were two donor characteristics showing trends toward an effect on insulin secretion (Fig. S3A). Combined with our b cell gene expression data, these findings suggest that b cell transcriptional and functional responses in T2D are not shared between the sexes.

Sex differences in UPR-associated gene expression in mouse islets
Our unbiased analysis of human b cell gene expression and function in T2D revealed differences between male and female donors with T2D.
Because human b cell gene expression and function can be affected by factors such as peripheral insulin sensitivity, disease processes, and medication [31,33], we investigated sex differences in b cell gene expression and function in another context. We generated a wellpowered islet RNAseq dataset from 20-week-old male and female C57BL/6J mice. We used an insulin tolerance test (ITT) to show that insulin sensitivity was similar between the sexes at this age (Fig. S4A); however, we acknowledge that the ITT may not be as sensitive as a hyperinsulinemic-euglycemic clamp in detecting modest sex differences in insulin sensitivity. Principal component analysis and unsupervised clustering clearly separated male and female islets on the basis of gene expression (Figure 2A; Fig. S5A). We found that 17.7% (3268/18938) of genes were differentially expressed between the sexes (1648 upregulated in females, 1620 upregulated in males), in line with estimates of sexbiased gene expression in other tissues [71,72]. Overrepresentation and pathway enrichment analysis both identified UPR-associated pathways as a biological process that differed significantly between the sexes, where the majority of genes in this category were enriched in female islets (Figure 2B,C; Supplementary file 3). Additional genes that were enriched in female islets were those associated with the gene ontology term "Cellular response to endoplasmic reticulum stress" (GO:0034976), which included many genes involved in regulating protein synthesis ( Figure 2D). For example, females showed significantly higher levels of most ribosomal protein genes ( Figure 2E). Further genes enriched in females included those associated with protein folding, protein processing, and quality control ( Figure 2D). Given that protein synthesis, processing, and folding capacity are intrinsically important for multiple islet cell types [73e76], including b cells [77,78], this suggests female islets may have a larger protein production and folding capacity than male islets.
3.3. Female islets are more resilient to endoplasmic reticulum stress in mice The burden of insulin production causes endoplasmic reticulum (ER) . ER stress is associated with an attenuation of mRNA translation [82], and, if ER stress is prolonged, can lead to cell death [83e85]. Given that female islets exhibited higher expression of genes associated with protein synthesis, processing, and folding than males, and higher expression of genes associated with the UPR, which is activated in response to ER stress [86], we examined global protein synthesis rates in male and female islets under basal conditions and during ER stress. We incubated islets with O-propargyl-puromycin (OPP), which is incorporated into newly-translated proteins and can be ligated to a fluorophore. Using this technique, we monitored the accumulation of newly-synthesized islet proteins with single-cell resolution (Fig. S6A). In basal culture conditions, male islet cells had significantly greater protein synthesis rates compared with female islet cells (Fig. S6B). To investigate islet protein synthesis under ER stress in Original Article each sex, we treated islets with thapsigargin (Tg), a specific inhibitor of the sarcoplasmic/endoplasmic reticulum Ca 2þ -ATPase (SERCA) that induces ER stress and the UPR by lowering ER calcium levels [83,87]. At 2 h post-Tg treatment, protein synthesis was repressed as expected in both male and female islet cells ( Figure 3A,B; Fig. S6C). At 24 h post-Tg treatment, we found that protein synthesis was restored to higher-than basal levels in female islet cells, but not in male islet cells ( Figure 3A,B; Fig. S6C). Importantly, a two-way ANOVA showed that recovery from protein synthesis repression was significantly different between males and females (sex:treatment interaction p < 0.0001). This suggests that while protein synthesis repression associated with ER stress was transient in female islets, this phenotype persisted for longer in male islets. Because insulin biosynthesis accounts for approximately half the total protein production in b cells [88], one potential explanation for the sex-specific recovery from protein synthesis repression is a sex difference in transcriptional changes to insulin. To test this, we quantified GFP levels in b cells isolated from male and female mice with GFP knocked into the endogenous mouse Ins2 locus (Ins2 GFP/WT ) [51,89]. While ER stress induced a significant reduction in Ins2 gene activity, this response was equivalent between the sexes. This suggests Ins2 transcriptional changes cannot fully explain the sex difference in recovery from protein synthesis repression during ER stress (Figs. S7AeC). Given the prolonged protein synthesis repression in males following ER stress, we next quantified cell death, another ER stress-associated phenotype [86], in male and female islets. Using a kinetic cell death assay, we observed clear sex differences in Tg-induced cell death at 0.1 mM and 1.0 mM Tg doses throughout the time course of the experiment ( Figure 3C,D). Notably, viability prior to the assay was not different between males and females (Fig. S8). After 84 h of Tg treatment, no significant increase in female islet cell apoptosis was observed with either 0.1 mM or 1.0 mM Tg treatment compared with controls ( Figure 3E). In contrast, cell death was significantly higher at these data is that female islets are resistant to Tg-induced cell death, we found a significant increase in apoptosis in both female and male islet cells treated with 10 mM Tg ( Figure 3G,H, sex:treatment interaction p ¼ 0.0996 [0.1 mM]; data graphed separately due to different DMSO control). This suggests female islets were more resilient to mild ER stress caused by low-dose Tg than male islets. To determine whether this increased ER stress resilience was caused by differential UPR signaling, we monitored levels of several protein markers of UPR activation including binding immunoglobulin protein (BiP), phosphorylated inositol-requiring enzyme 1 (pIRE1), phosphorylated eukaryotic initiation factor alpha (peIF2a), and C/EBP homologous protein (CHOP) [90,91] after treating male and female islets with 1 mM Tg for 24 h. We found no sex difference in UPR protein markers between male and female islets without Tg treatment (Figs. S9AeD) and observed a significant increase in levels of pIRE1a and CHOP in islets from both sexes and BiP in female islets after a 24-hour Tg treatment (Figs. S9AeD). Lack of a sex difference in protein markers suggests UPR activation by Tg treatment was similar between male and female islets at 20 weeks of age. This finding differs from the male-biased UPR activation reported in the KINGS mouse model of endogenous ER stress [37]. While one potential explanation for this discrepancy is that Tg treatment induces acute ER stress in contrast to the chronic ER stress in KINGS mice, further experiments will be needed to confirm this possibility. Of note, we reproduced the malebiased induction of BiP in islets isolated from 60-week-old male and female mice (Figs. S9EeG), suggesting that age contributes to the sex difference in UPR activation. Together, our data indicate that despite equivalent UPR activation in male and female islets treated with Tg, significant sex differences exist in ER stress-associated protein synthesis repression and cell death.

Female islets retain greater b cell function during ER stress in mice
We next examined glucose-stimulated insulin secretion in islets cultured under basal conditions and during Tg treatment ( Figure 4A). In all conditions tested, high glucose significantly stimulated insulin secretion in both sexes (Fig. S10A); however, we identified sex differences in how well islets sustained glucose-stimulated insulin secretion during longer Tg treatments ( Figure 4B,C, Fig. S10A,B). Female islets, in both low and high glucose, maintained robust insulin secretion during Tg treatment ( Figure 4B). Specifically, we observed a significant increase in insulin secretion after short Tg treatment (0 and 2 h post-Tg), with a return to basal secretion levels 4 h post-Tg ( Figure 4B). Because Tg is a drug that depletes ER calcium stores, it may induce an acute rise in cytosolic calcium that could explain this acute increase in high glucose-stimulated insulin secretion in Tgtreated samples compared with vehicle [83]. In contrast, male islets showed no significant increase in insulin secretion after short Tg treatment, and there was a significant drop in insulin secretion at 4 h post-Tg treatment ( Figure 4C). This suggests female islets sustained insulin secretion for a longer period than male islets during ER stress. Given that insulin content measurements showed insulin content significantly increased during the 4-hour Tg treatment in female islets, but not male islets ( Figure 4D, Fig. S10C), our data suggest one reason female islets maintain insulin secretion during ER stress is by augmenting islet insulin content. Proinsulin secretion followed similar trends to those we observed with insulin secretion (Figs. S10D,E), but Tg treatment reduced proinsulin content to a greater degree in male Original Article islets ( Figure 4E). There was no sex difference in the ratio of proinsulin:insulin content at any timepoint (Fig. S10G). This suggests that in addition to a greater ability to maintain glucose-stimulated insulin secretion during ER stress, female islets also show a larger increase in insulin content and a smaller decrease in proinsulin content in this context.
To determine whether female islets have improved b cell function under ER stress in other contexts, we next monitored glucose-stimulated insulin secretion and glucose tolerance in mice at 20 weeks, an age where we demonstrated that insulin sensitivity was similar between the sexes (Figure 4FeH; Fig. S4). We found that fasting plasma insulin levels were higher in males ( Figure 4F) and that the sexes showed similar glucose one-way ANOVA followed by Tukey HSD test), significantly higher after a 24-hour Tg treatment compared to a 2-hour Tg treatment (p < 0.0001; one-way ANOVA followed by Tukey HSD test) and recovered to a significantly higher level than control levels p < 0.0001; one-way ANOVA followed by Tukey HSD test). (B) In male islet cells, protein synthesis was significantly lower after a 2-and 24-hour Tg treatment compared to control (p < 0.0001 for both treatments; one-way ANOVA followed by Tukey HSD test) and was not significantly different after a 24-hour treatment compared to a 2-hour Tg treatment p ¼ 0.3022; one-way ANOVA followed by Tukey HSD test    (B) and male (C)  tolerance ( Figure 4H). To determine the magnitude of the acute glucosestimulated insulin secretion response in each sex, we normalized glucose-stimulated insulin secretion to basal insulin secretion levels. We found that females showed a trend toward higher acute glucosestimulated insulin secretion response in C57BL/6J mice ( Figure 4G), and significantly higher glucose-stimulated insulin secretion in 10-and 22-week-old Insr f/f :Ins1Cre À/À :nTnG þ/À mice (Figs. S11AeD; data replotted from a prior Johnson lab study [53]). These findings align with data showing higher glucose-stimulated insulin secretion from prior studies in humans and rodents [5,10,16,92,93]. We also found that fasting blood glucose levels in female mice were more resilient to the near-total insulin gene knockout in Ins1 À/À ;Ins2 fl/fl ;Pdx1CreER mice given tamoxifen ( Fig. S11E; replotted from published [54] and unpublished Johnson lab data). We cannot rule out all potential factors that may contribute to the sex differences in blood glucose levels following near-total loss of insulin gene function (e.g. RNA stability, translation efficiency); however, given that the burden of insulin production [54] leads to ER stress even in normal physiological conditions, our data add further support to a model in which b cells in female mice show a greater ability to maintain glucose-stimulated insulin production and secretion during ER stress.

Sex differences in islet transcriptional and proteomic responses to ER stress in mice
To gain insight into the differential ER stress-associated phenotypes in male and female islets, we investigated global transcriptional changes after either a 6-or 12-hour Tg treatment in each sex. Principal component analysis and unsupervised clustering shows that islets clustered by sex, treatment, and treatment time ( Figure 5A; Fig. S12A). The majority of the variance was explained by treatment ( Figure 5B), and pathway enrichment analysis confirms the UPR as the top upregulated pathway in Tg-treated male and female islets at both 6-and 12-hours after treatment (Figs. S13A,B; Supplementary file 4). While some UPR-associated genes differentially regulated by Tg treatment were shared between the sexes (6-hour: 29/36, 12-hour: 25/31), biological sex explained a large proportion of variance in the gene expression response to ER stress. This suggests the transcriptional response to ER stress was not fully shared between the sexes. Indeed, after a 6-hour Tg treatment, 32.6% (2247/4655) of genes that were differentially expressed between DMSO and Tg were unique to one sex (881 to females, 1376 to males). After a 12-hour Tg treatment, 29% (2259/7785) were unique to one sex (1017 to males, 1242 to females).
To describe the transcriptional response of each sex to Tg treatment in more detail, we used a two-way ANOVA to identify genes that were upregulated, downregulated, or unchanged in male and female islets between 6-and 12-hours post-Tg (Supplementary file 5). By performing pathway enrichment analysis, we were able to determine which processes were shared, and which processes differed, between the sexes during Tg treatment. For example, we observed a significant increase in mRNA levels of genes corresponding to pathways such as cellular responses to stimuli, stress, and starvation in both male and female islets between 6-and 12-hour Tg treatments ( Figure 5C; Supplementary file 4), suggesting Tg has similar effects on genes related to these pathways in both sexes. Similarly, the Tg-induced changes in mRNA levels of genes related to apoptosis were largely shared between the sexes (Figs. S14A,B). While this suggests that the sex-specific regulation of genes related to apoptosis does not fully account for the susceptibility of male islets to low-dose Tg-induced cell death, in-depth studies of b cell apoptosis will be needed to confirm this point.
In contrast to these non-sex-specific changes in gene expression, there was a male-specific increase in mRNA levels of genes associated with translation during Tg treatment ( Figure 5C; Supplementary file 4). In females, there was a decrease in mRNA levels of genes associated with b cell identity, such as Pklr, Rfx6, Hnf4a, Slc2a2, Pdx1, and MafA (Fig. S15A), and in genes linked with regulation of gene expression in b cells ( Figure 5C). Neither of these categories were altered between 6and 12-hour Tg treatments in males ( Figure 5C; Fig. S15B). While our data suggests some aspects of the gene expression response to ER stress were shared between the sexes, we found that many genes corresponding to important cellular processes were differentially regulated during Tg treatment in only one sex. Beyond sex-specific transcriptional changes following Tg treatment, ER stress also had a sex-specific effect on the islet proteome. Although the majority of proteins were downregulated by Tg treatment due to the generalized repression of protein synthesis under ER stress ( Figure 5D), we identified 47 proteins (35 downregulated, 12 upregulated in Tg) that were differentially expressed in female islets and 82 proteins (72 downregulated, 10 upregulated in Tg) that were differentially expressed after Tg treatment in male islets (Supplementary Table 1). Proteins downregulated only in females include proteins associated with the GO term 'endoplasmic reticulum to Golgi vesiclemediated transport' (GO:0006888) (BCAP31, COG5, COG3, GOSR1), whereas proteins downregulated only in males include proteins associated with GO terms 'insulin secretion' (GO:0030073) (PTPRN2, CLTRN, PTPRN) and 'lysosome pathway' (KEGG) (NPC2, CTSZ, LAMP2, PSAP, CLTA). Importantly, only seven differentially expressed proteins were in common between the sexes ( Figure 5D). This suggests that as with our phenotypic and transcriptomic data, the proteomic response to Tg treatment was largely not shared between the sexes. To integrate our islet transcriptome and partial islet proteome data, we assessed the direction of changes to mRNA and protein levels following Tg treatment. In females, the number of differentially expressed islet proteins with concordant mRNA changes was 43% (20/ 47), whereas the number of differentially expressed islet proteins in males with concordant mRNA changes was 49% (40/82) ( Fig. S16; Supplementary file 6). This data suggests that many genes with differential mRNA expression during Tg treatment show congruent changes in protein abundance. When we next asked whether the islet genes with concordant changes in mRNA and protein levels during Tg treatment were shared between the sexes, we found that only 7% (4/ 56) of these genes were differentially expressed in both sexes during Tg treatment (Tmem27, Emb, Fkbp9, Pdcd4). The remaining 93% (51/ 55) of islet genes with concordant changes in mRNA and protein levels during Tg treatment were differentially expressed in only one sex. Thus, when taken together, our islet transcriptome and partial proteome data suggest that male and female islets show distinct responses to ER stress.
Of the islet genes with concordant changes in mRNA and protein levels, several have been linked with b cell and/or islet function. For example, Tmem27 plays a role in enhancing GSIS [94], and Pdcd4 expression is associated with b cell death under stressed conditions [95,96]. Atp6ap2 and Lamp2 have important roles in autophagy [97e99], Ptprn2 is required for the accumulation of insulin granules [100], and both Ptprn2 and Chga are involved in glucose-stimulated insulin secretion [101,102]. Chga is also important for maintaining islet volume and islet cell composition [103], and studies show that Tbc1d1 influences insulin secretion and b cell mass in rodents [104,105]. Because the Tbc1d1 study [104,105], and others [94e103], used single-or mixed-sex models, or cell lines, future studies will need to address whether these effects on b cell and/or islet function are shared between the Original Article

DISCUSSION
Emerging evidence shows biological sex affects many aspects of b cell gene expression and function. Yet, many studies on b cells do not include both sexes, or fail to analyze male and female data separately.
To address this gap in knowledge, the goal of our study was to provide detailed information on sex differences in islet and b cell gene expression and function in multiple contexts. In humans, we used a large scRNAseq dataset from ND and T2D donors to reveal significant male-female differences in the magnitude of gene expression changes, and in the identity of genes that were differentially regulated, between ND and T2D donors. While these data do not address the reasons for the sex-biased risk of T2D, our findings suggest b cell gene expression changes in T2D are not fully shared between the sexes. These data provide a useful resource to support future studies on how medication, disease progression, age, or body mass index contribute to the sex difference in b cell gene expression in T2D. Sex-based analysis of human b cell gene expression data will also clarify the mechanisms underlying our finding that b cells from female donors with T2D maintain higher insulin production than male donors with T2D. In mice, we generated a large RNAseq dataset using islets isolated from 20-week-old males and females. We used an ITT to show that male and female mice have similar insulin sensitivity at this age. We acknowledge that the ITT may not be sensitive enough to pick up small differences, so future studies will need to use a hyperinsulinemiceuglycemic clamp to further compare insulin sensitivity between the sexes. Despite this potential limitation, our unbiased analysis of gene expression in islets from males and females revealed sex differences in genes associated with the UPR under normal physiological conditions. This differential gene expression was significant, as female islets were more resilient to phenotypes caused by ER stress and UPR activation than male islets, showed sex-specific transcriptional and proteomic changes in this context, and had a greater ability to maintain glucosestimulated insulin production and secretion during ER stress. Collectively, these data suggest that in rodents, b cells from females are more resilient to ER stress. Considering the well-established links between ER stress and T2D [90,106e108], our data suggests a model in which female b cells have a greater ability to maintain glucose-stimulated insulin secretion in T2D because they are more resilient to ER stress and UPR activation. While future studies are needed to test this working model, and to assess the relative contribution of sex differences in b cells to the sex-biased risk of T2D, our findings highlight the importance of including both sexes in islet and b cell studies. Including both sexes in our analysis of b cell gene expression in human ND and T2D allowed us to uncover genes that were differentially regulated in T2D in each sex. Because many of these genes may have been missed if the scRNAseq data was not analyzed by sex, our findings advance knowledge of b cell changes in T2D by identifying additional genes that are differentially regulated in this context. This knowledge adds to a growing number of studies that identify sex differences in b cell gene expression during aging in humans [12], and in mice fed either a normal [4,11] or a high fat diet [11]. Further, given that our RNAseq on islets from male and female mice with similar insulin sensitivity identifies genes and biological pathways that align with previous studies on sex differences in murine b cell gene expression [4,11], our data suggests that sex differences in islet and b cell gene expression cannot be explained solely by a male-female difference in peripheral insulin resistance. Instead, there is likely a basal sex difference in b cell gene expression that forms the foundation for sex-specific transcriptional responses to perturbations such as ER stress and T2D. By generating large islet gene expression datasets from male and female mice with similar peripheral insulin sensitivity and from islets subjected to pharmacological induction of ER stress, our studies provide a foundation of knowledge for future studies aimed at studying the causes and consequences of sex differences in islet ER stress responses and b cell function following UPR activation. This will provide deeper mechanistic insight into the sex-specific phenotypic effects reported in animal models of b cell dysfunction [35e39,109e112] and the sex-biased risk of diseases such as T2D that are associated with b cell dysfunction [12,22,113,114]. Beyond gene expression, our sex-based analysis of mouse islets allowed us to uncover male-female differences in ER stress-associated phenotypes (e.g. protein synthesis repression, cell death). While previous studies identify a sex difference in b cell loss in diabetic mouse models [37,39,115], and show that estrogen plays a protective role via estrogen receptor a (ERa) against ER stress to preserve b cell mass and prevent apoptosis in cell lines, mouse models, and human islets [39,115,116], we extend prior findings by showing that differences in ER stressinduced cell death were present in the context of similar insulin sensitivity between the sexes. This suggests sex differences in ER stressassociated phenotypes do not solely depend on male-female differences in peripheral insulin sensitivity. Indeed, islets isolated from males and females with similar insulin sensitivity also show a sex difference in protein synthesis repression, a classical ER stress-associated phenotype [86]. While estrogen affects insulin biosynthesis via ERa [117], future studies will need to determine whether estrogen also allows female islets to restore protein synthesis to basal levels faster than male islets following ER stress. We currently lack this knowledge, as most studies on UPR-mediated recovery from protein translation repression use singleand mixed-sex animal groups, or cultured cells [118e123]. Assessing whether the recovery of protein synthesis contributes to reduced cell death in female islets following ER stress will also be an important task for future studies, as prior work suggests the inability to recover from protein synthesis repression increases ER-stress induced apoptosis [118]. Ideally, this type of study would also monitor the activity of pathways known to regulate protein synthesis repression during ER stress. For example, while we did not detect any changes in levels of phosphorylated eIF2a (also known as Eif2s1), which is known to mediate UPR-induced protein synthesis repression [86], our chosen timepoints did not overlap with the rapid changes in phospho-eIF2a following ER stress published in other studies [124,125]. A more detailed time course will therefore be necessary to assess p-eIF2a levels during ER stress in both sexes, and to test a role for phospho-eIF2a in mediating differences in protein synthesis repression. More work will also be needed to determine whether males and females differ in b cell replication [126], another ER stress-related phenotype.
Ultimately, a better understanding of sex differences in ER stressassociated phenotypes in b cells will provide a mechanistic explanation for the strongly male-biased onset of diabetes-like phenotypes in mouse models of b cell ER stress (e.g. Akita, KINGS, Munich mice) [37,38,109]. Given the known relationship between ER stress, b cell death, and T2D, studies on the male-female difference in b cell ER stress-associated phenotypes may also advance our understanding of the male-biased risk of developing T2D in some population groups.
A further benefit of additional studies on the sex difference in b cell ER stress responses will be to identify mechanisms that support b cell insulin production. In rodents, we found that female islets maintained high glucose-stimulated insulin secretion and increased insulin content following ER stress, whereas male islets showed significant repression of high glucose-stimulated insulin secretion under the same conditions. In humans, while a study using a mixed-sex group of T2D donors shows b cells experience ER stress associated with b cell dysfunction [63], we found that changes to b cell insulin secretion in T2D were not the same between the sexes. Specifically, the magnitude of the reduction in insulin release by b cells from female donors with T2D was smaller than in b cells from male donors with T2D. Together with our data from rodents, this suggests female b cells maintain enhanced insulin production and/or secretion in multiple contexts, and the increased b cell function cannot be solely attributed to a sex difference in peripheral insulin sensitivity.
Clues into potential ways that female b cells maintain improved insulin production and secretion emerge from our examination of the transcriptional response to ER stress in mice of each sex. Our data shows that Tg treatment induces gene expression changes characteristic of ER stress [127], and revealed similar biological pathways that were upregulated in T2D donors. Furthermore, we identified significant differences between male and female islets in the transcriptional response to ER stress over time. One notable finding was that a greater number of b cell identity genes were downregulated between 6-and 12-hour Tg treatments in females, but not in males. Because most studies on the relationship between b cell identity and function used a mixed-sex pool of islets and b cells [79,128,129], more studies will be needed to test whether there are sex-specific changes to b cell identity during ER stress, and to determine the functional consequences of this sex-specific effect.
Overall, our data demonstrates sex differences in b cell function in multiple contexts. One potential explanation for these differences is the sex-specific regulation of b cell ER stress responses and function.
Indeed, sex differences in ER stress and protein markers of apoptosis were observed in mouse kidney cells [130], suggesting that studying b cell ER stress may provide insight into this difference in other cell types. Alternatively, it is possible that there are sex differences in b cell number that account for the male-female differences in b cell function that we observe. For example, considering two published studies indicate males have fewer b cells [6,37], the burden of maintaining glucose homeostasis may fall on a smaller number of cells in males, leading to higher susceptibility to ER stress. Future studies will need to address sex differences in b cell number relative to pancreas size and body size to test this possibility. Ultimately, a better understanding of changes to b cell gene expression and function in males and females will suggest effective ways to reverse disease-associated changes to this important cell type in each sex, improving equity in health outcomes [131].

Conclusions
Our study reports significant sex differences in islet and b cell gene expression and stress responses in both humans and mice. These differences likely contribute to sex differences in b cell resilience, allowing female b cells to show a greater ability to maintain glucosestimulated insulin production and secretion across multiple contexts. This knowledge forms a foundation for future studies aimed at understanding how sex differences in b cell function affect physiology and the pathophysiology of diseases such as T2D.