The functional mass of insulin-secreting pancreatic β-cells expands to maintain glucose homeostasis in the face of nutrient excess, in part via replication of existing β-cells. Type 2 diabetes appears when these compensatory mechanisms fail. Nutrients including glucose and fatty acids are important contributors to the β-cell compensatory response, but their underlying mechanisms of action remain poorly understood. We investigated the transcriptional mechanisms of β-cell proliferation in response to fatty acids. Isolated rat islets were exposed to 16.7 mmol/L glucose with or without 0.5 mmol/L oleate (C18:1) or palmitate (C16:0) for 48 h. The islet transcriptome was assessed by single-cell RNA sequencing. β-Cell proliferation was measured by flow cytometry. Unsupervised clustering of pooled β-cells identified different subclusters, including proliferating β-cells. β-Cell proliferation increased in response to oleate but not palmitate. Both fatty acids enhanced the expression of genes involved in energy metabolism and mitochondrial activity. Comparison of proliferating versus nonproliferating β-cells and pseudotime ordering suggested the involvement of reactive oxygen species (ROS) and peroxiredoxin signaling. Accordingly, N-acetyl cysteine and the peroxiredoxin inhibitor conoidin A both blocked oleate-induced β-cell proliferation. Our study reveals a key role for ROS signaling through peroxiredoxin activation in oleate-induced β-cell proliferation.
Both type 1 and type 2 diabetes are characterized by a loss of functional β-cells (1), and increasing β-cell mass is considered a promising therapeutic approach (2). In type 2 diabetes, a commonly accepted view is that insulin resistance drives a compensatory increase in both insulin secretion from individual β-cells and β-cell numbers (3), in part via replication of existing β-cells (4). This view is supported by the observations that factors secreted from peripheral tissues in response to insulin resistance, such as serpin B1 (5), promote β-cell proliferation. However, the notion that insulin resistance is the cause of compensatory β-cell mass expansion has been challenged (6). In humans, insulin hypersecretion can be observed before insulin resistance in obesity (7), and in mice, short-term exposure to high-fat diet leads to enhanced β-cell proliferation before an apparent increase in insulin resistance (8,9). This suggests a direct effect of nutrients in the promotion of β-cell replication. Accordingly, short-term glucose infusion in mice leads to a significant increase in β-cell proliferation (10), and we previously showed that a combination of glucose and fatty acids (FAs) increases β-cell proliferation and mass in infused rats as well as in rat and human islets ex vivo (11,12).
In β-cells, chronic exposure to elevated levels of glucose and FAs generally leads to β-cell dysfunction, apoptosis, and dedifferentiation, a condition known as glucolipotoxicity (13). However, FA species can affect β-cells differently, with saturated FAs such as palmitate having deleterious effects and unsaturated FAs such as oleate being protective (14).
Recently, we demonstrated that in the presence of elevated glucose concentrations, oleate but not palmitate promotes β-cell proliferation (15). To further decipher the underlying mechanisms, we analyzed the islet transcriptome in response to either FA by single-cell RNA sequencing (scRNA-seq). Specifically, the aims of the study were 1) to ascertain the differential effects of oleate and palmitate on the major islet cell types and 2) to characterize the transcriptome of proliferating β-cells in response to these FAs at single-cell resolution.
Research Design and Methods
Islet Isolation and Culture; Measurement of β-Cell Proliferation, Reactive Oxygen Species Levels, and Gene Expression; and Immunostaining of Islet Sections
All procedures were approved by the Institutional Committee for the Protection of Animals at the Centre Hospitalier de l’Université de Montréal. Male Wistar rats (Charles River, Saint-Constant, QC, Canada) weighing 250–300 g (∼2 months old) were group housed (two animals per cage) under controlled temperature on a 12-h light-dark cycle with free access to water and standard laboratory chow. Islets were isolated from 2-month-old male Wistar rats by collagenase digestion of the pancreas as described in Supplementary Materials. After isolation, islets were recovered for 24 h in RPMI 1640 with 10% (vol/vol) FBS with 11.1 mmol/L glucose. For treatment, pools of 200 islets were cultured in RPMI 1640 with FBS for 6–48 h in the presence of glucose (2.8 or 16.7 mmol/L) with or without palmitate or oleate (0.5 mmol/L) or vehicle (0.1 mmol/L BSA + 50% ethanol [vol/vol]) as indicated in the figure legends. FAs were dissolved in ethanol before 1-h complexation with BSA at a 5:1 molar ratio (FA to BSA). Islets were treated with N-acetylcysteine (NAC), conoidin A, 10,058-F4/1-RH, or harmine as indicated in the figure legends. Media were replaced daily, and 5-ethynyl-2′-deoxyuridine (EdU) (10 μmol/L) was added to the culture medium. Measurement of β-cell proliferation and reactive oxygen species (ROS) levels by flow cytometry, gene expression by quantitative PCR, and immunostaining of islet sections were performed as described in Supplementary Materials.
Statistical Analysis (excluding scRNA-seq data)
Statistical analyses were performed using Prism 9 (GraphPad Software, Inc., San Diego, CA) using one- or two-way ANOVA followed by Tukey, Dunnett, or Sidak post hoc analysis for multiple comparisons as indicated in the figure legends. Results are presented as mean ± SEM. P < 0.05 was considered significant.
Before scRNA-seq library generation, islets were dissociated (Supplementary Materials), washed once using PBS plus BSA (1%) solution, and resuspended with MACS MicroBeads (Miltenyi Biotec, Waltham, MA). Dead cells were removed using MS Columns (Miltenyi Biotec) according to manufacturer instructions. Live cells were then counted using a Countess II automated cell counter (Thermo Fisher Scientific). Single-cell suspensions with <10% dead cells were kept for library generation. Single-cell libraries were generated using the Chromium Single-Cell 3′ Library and Gel Bead Kit (version 2; 10× Genomics, Pleasanton, CA). In brief, to reach a target cell number of 6,000 cells per sample, 10,500 cells per sample were loaded onto a channel of the 10× Genomics chip to produce Gel Bead-in-Emulsions. This underwent reverse transcription to barcode RNA before cleanup and cDNA amplification followed by enzymatic fragmentation and 5′ adaptor and sample index attachment. Library quantification and quality control was performed on a 2100 Bioanalyzer (Agilent, Santa Clara, CA). Libraries were sequenced on the NovaSeq 6000 (Illumina, San Diego, CA) with 150-bp paired-end sequencing. Further scRNA-seq analysis including data processing, clustering, differentially expressed gene (DEG) analysis, and trajectory inference are described in Supplementary Material.
Data and Resource Availability
scRNA-seq data generated during the study are available in the National Center for Biotechnology Information Gene Expression Omnibus repository (https://www.ncbi.nlm.nih.gov/geo/; accession no. GSE193857).
scRNA-seq Identifies a Subpopulation of Proliferative β-Cells
We previously showed that infusion of a triglyceride emulsion enriched in oleate in rats increases β-cell proliferation only in the presence of elevated glucose concentrations (12). Therefore, to investigate β-cell proliferation in islets ex vivo, isolated rat islets were cultured for 48 h in the presence of 16.7 mmol/L glucose (vehicle; n = 4) with or without 0.5 mmol/L palmitate (C16:0; n = 3) or oleate (C18:1; n = 4) (Fig. 1A). After scRNA-seq, unbiased graph-based clustering of 52,570 pooled cells from all three culture conditions identified the four main endocrine cell types based on expression of insulin (Ins2; n = 39,910), glucagon (Gcg; n = 7,821), somatostatin (Sst; n = 1,419), and pancreatic polypeptide (PP) (Ppy; n = 1,157), corresponding to β, α, δ, and PP cells, respectively, as well as nonendocrine cells (n = 2,263) (Fig. 1B and C and Supplementary Table 2). The number of cells in each population (Supplementary Table 2), as well as uniform manifold approximations and projections (UMAPs) (Supplementary Fig. 1A), was comparable between conditions and biological replicates. Endocrine cell types clustered together irrespective of the treatment condition.
Refined clustering of the Ins2+ cells identified five β-cell subpopulations: β1 (n = 28,411), β2 (n = 638), β3 (n = 5,811), β4 (n = 3,596), and β5 (n = 1,454) (Fig. 1D). The percentages of cells in each cluster across biological replicates and treatment conditions (Fig. 1E) as well as UMAP projections (Supplementary Fig. 1B) were comparable, except for the β2 cluster, where we observed a significant increase in the presence of oleate and, to a lesser extent palmitate, compared with the vehicle condition. Except for the β4 cluster, where only 10 genes (including Ins1 and Ins2) involved in hormone activity were enriched (Fig. 1F and Supplementary Tables 3 and 4), the other clusters displayed many uniquely enriched genes (log2 fold change ≥0.25: β1 227, β2 1,063, β3 611, and β5 662) (Supplementary Table 3). DEG and pathway enrichment analyses from the top 50 cluster markers for the β1 subpopulation suggested a role in hormone secretion (e.g., Ptprn2 and Vamp2) (Fig. 1F and Supplementary Tables 3 and 4). Genes involved in cell division were highly enriched in the β2 subpopulation, as shown by expression of the proliferation markers Mki67, Cdk1, and Pcna (Fig. 1F and G and Supplementary Tables 3 and 4). The β3 cluster was characterized by enriched expression of genes involved in the unfolded protein response (UPR) and endoplasmic reticulum (ER) stress, including Hspa1b, Ddit3, and Tmbim6 (Fig. 1F and Supplementary Tables 3 and 4). Finally, the β5 cluster was enriched in genes implicated in proinsulin biosynthesis and secretory vesicle formation as shown by the expression of Scg2, Pcsk2, Chga, and Cpe (Fig. 1F and Supplementary Tables 3 and 4).
To directly confirm the proliferative effect of oleate, we assessed β-cell proliferation by flow cytometry using EdU on the same biological replicates used for scRNA-seq. In line with the transcriptomic analysis, oleate stimulated β-cell proliferation, whereas the effect of palmitate was not significant (Fig. 1H and I).
Oleate and Palmitate Differentially Affect the β-Cell Transcriptome
We next compared the transcriptomes of individual β-cell clusters between oleate- or palmitate- and vehicle-treated islets and performed pathway enrichment analysis. For the β1 cluster, oleate treatment resulted in 117 DEGs, whereas palmitate treatment produced 155 DEGs. Approximately 45% of the DEGs were similar between both conditions (Fig. 2A and F). Both oleate and palmitate upregulated pathways controlling metabolic processes, including glycolysis (e.g., Aldoa, Gapdh, Pgk1, and Eno1) and oxidative phosphorylation (e.g., Ndufa5, Cox5b, and Atp5b) (Fig. 2A and Supplementary Tables 5 and 6).
In the proliferative β2 cluster, comparison of oleate- or palmitate- with vehicle-treated islets revealed only 35 and 24 DEGs, respectively (Fig. 2B and G). Except for downregulation of ribosomal protein subunits (e.g., Rps7, Rps3a, and Rps24) and their associated pathways in oleate-treated islets, no major differences were observed between conditions (Fig. 2B and Supplementary Tables 5 and 6).
The β3 cluster showed clear differences between palmitate and oleate treatments. Oleate-treated islets presented a total of 61 DEGs (Fig. 2C and H). Pathway analysis showed decreased expression of β-cell differentiation and maturity markers (e.g., Neurod1, Slc2a2, and Pdx1) and increased expression of antiapoptotic genes (e.g., Tmbim6) (Fig. 2C and Supplementary Tables 5 and 6). Palmitate-treated islets displayed 243 DEGs (Fig. 2C and H), including downregulated β-cell markers (e.g., NeuroD1, Pdx1, and Slc2a2) and genes involved in insulin secretion (e.g., Ptprn2, Gnas, and Rgs16) (Fig. 2C and Supplementary Tables 5 and 6). In addition, UPR, ER stress, oxidative stress, apoptotic pathways (e.g., Ddit3, Atf4, Herpud1, and Sod2), and translational processes (e.g., Eif1ad, Rpl37, and Rps28) were upregulated in response to palmitate but not oleate (Fig. 2C and Supplementary Tables 5 and 6).
In the β4 cluster, oleate- and palmitate-treated islets displayed 253 and 564 DEGs, respectively (Fig. 2D and I). No downregulated pathways were enriched in either treatment condition. Similar to the β1 cluster, upregulated pathways after exposure to FAs included cellular respiration (e.g., Cox4i2, Ndufs3, Cycs, or Atp5b) and antioxidant response (e.g., Sod2 or Prdx1) (Fig. 2D and Supplementary Tables 5 and 6). However, transcripts involved in translational processes (e.g., Eif1ad, Rpl37, Eif3d, and Rpl41) and ER stress–related pathways (e.g., Ddit3, Ppp1r15a, and Dnajb11) were also enriched in palmitate-treated islets (Fig. 2D and Supplementary Tables 5 and 6).
In the β5 cluster, oleate exposure produced 123 DEGs (Fig. 2E and J), including downregulation of β-cell differentiation and maturity markers (e.g., Neurod1, Pdx1, Slc2a2, or Nkx6–1) and upregulation of pathways related to cellular respiration and antioxidant defense (e.g., Cox6a1, Ndufs7, Sod2, or Gpx1) (Fig. 2E and Supplementary Tables 5 and 6). On the other hand, palmitate-treated islets displayed 389 DEGs (Fig. 2E and J). Similar to the β3 and β4 clusters, palmitate treatment increased expression of genes involved in oxidative phosphorylation (e.g., Ndufs4, Cycs, or Atp5d), oxidative stress (e.g., Sod2, Cycs, Gpx1, or Prdx1), translational processes (e.g., Rpl41 or Rps20), and UPR/ER stress (e.g., Ppia, Calr, Ddit3, Ppp1r15a, or Dnajb11) (Fig. 2E and Supplementary Tables 5 and 6). Because MAST analysis to detect DEGs (16) may lead to false-positive results (17), we also aggregated RNA counts in a pseudobulk fashion for each subpopulation of β-cells and then calculated DEGs using DESeq2 (18). We found that DEGs identified using both methods were highly correlated (Supplementary Fig. 2A and B).
Oleate Does Not Promote Non–β-Cell Proliferation
Refined clustering of pooled non–β-cells identified three subpopulations of Gcg-expressing cells (α1–3) and a population of Gcg/Ppy-coexpressing cells (Gcg+Ppy+), as well as δ, PP, and nonendocrine cell populations (Fig. 3A and B). Reclustering also identified a population of doublets, characterized by the coexpression of Ins2 and/or Gcg, Ppy, and Sst, as well as a higher number of genes and total RNA count (undefined cluster in Fig. 3B and C). Although we cannot exclude that the Gcg/Ppy double-positive cells also represent doublets, the existence of islet cells coexpressing glucagon and PP has been described previously in rats (19). No difference in the percentage of cells in each cluster across biological replicates and treatment conditions was observed (Fig. 3D).
Pathway enrichment analysis revealed that the α1, α2, and α3 clusters were characterized by elevated expression of the complement system, hormone secretion and protein translation, and ER stress–related pathways, respectively (Fig. 3E and Supplementary Tables 3 and 4). In the α1 and α2 clusters, oleate increased protein translation and secretion-related pathways, whereas palmitate upregulated secretion-related pathways (Supplementary Tables 5 and 6). In the α3 cluster, both FAs increased the expression of genes involved in protein translation and ER stress pathways (Supplementary Tables 5 and 6). In Gcg+Ppy+ cells, both FAs increased secretion-related pathways, and oleate also upregulated protein translation processes (Supplementary Tables 5 and 6). In PP cells, both FAs upregulated protein translation processes (Supplementary Tables 5 and 6). δ-Cells exhibited increased oxidative phosphorylation–related pathways (Fig. 3E and Supplementary Tables 3 and 4), and oleate increased protein translation processes, whereas palmitate increased ER stress–related pathways (Supplementary Tables 5 and 6). No downregulated pathways were observed in any population, and we did not detect any cluster of proliferating non–β-cells.
To directly confirm the lack of proliferative effect of FAs, we assessed α-cell proliferation by flow cytometry on the same biological replicates used for scRNA-seq. Although some proliferating α-cells were detected, the percentage was similar in the three conditions (Fig. 3F). Thus, oleate-induced cell proliferation and regulation of associated pathways seem unique to β-cells in rat islets.
Oxidative Phosphorylation and Oxidoreductase Activity Pathways Are Upregulated in Proliferating β-Cells
Because no major transcriptional differences were detected in the proliferating β2 cluster in response to oleate or palmitate compared with vehicle (Fig. 2B), we compared the β2 cluster with the nonproliferating β1 and β3–5 clusters (Fig. 4A, Supplementary Fig. 3, and Supplementary Table 7) irrespective of treatment condition and performed pathway enrichment analysis. This approach enabled improved visualization of gene sets organized into networks. Comparison of β2 with β3 and β5 showed downregulation of hormone and insulin response, glycosylation, cell secretion, UPR/ER stress, and lysosomal-related pathways (Supplementary Fig. 3A and C). Comparison of β2 with all nonproliferating clusters revealed that most upregulated pathways were involved in protein catabolism and cell division–related processes, as expected (Fig. 4A and Supplementary Fig. 3). Furthermore, upregulation of glucose metabolism, ATP synthesis, oxidative phosphorylation, and oxidoreductase activity (β2 vs. β1 and β3–5) (Fig. 4A and Supplementary Fig. 3) was also observed. Among these pathways, the main upregulated genes included the subunits of the complexes of the electron transport chain (e.g., Ndufa12, Cyc1, and Atp5g2), involved in oxidative phosphorylation and ATP synthesis, and the antioxidant family members thioredoxin (e.g., Txn1 and Txnl1) and peroxiredoxin (Prdx1, Prdx2, and Prdx4), involved in oxidoreductase activity (Fig. 4B).
Oleate Preferentially Directs β-Cells Toward Proliferation While Amplifying Oxidative Phosphorylation and Antioxidant Pathways and Diminishing β-Cell Maturation
Growing evidence suggests that β-cells are less differentiated during cell-cycle activation (20). Therefore, we used CytoTRACE (Supplementary Materials) to predict the differentiation state of cells from our scRNA-seq data set. We observed that the β4 and β5 clusters were more differentiated, whereas the proliferative β2 (including part of β1 and β3 clusters) were less differentiated (Fig. 5A). Interestingly, both palmitate and oleate treatment increased the immaturity score relative to vehicle (Fig. 5B), suggesting that both FAs promote dedifferentiation of β-cells in agreement with the DEG analyses shown in Fig. 2.
We then asked whether a transition state could be identified between the more differentiated β5 cluster and the less differentiated β2 cluster. To test this, we performed trajectory inference analysis (Supplementary Materials) with a root starting at the β5 cluster. Pseudotime ordering of β-cells from the most differentiated cluster revealed three lineages (Fig. 5C). These trajectories branched within the UPR/ER stress–related β3 subpopulation, leading to the secretory β1 (lineage 1), the more differentiated insulin-expressing β4 (lineage 2), and the proliferative β2 (lineage 3) subpopulations. Although no lineage was unique to a particular condition, comparing the mean cell weight of each lineage across the three conditions showed that oleate favored progression toward the proliferative path (lineage 3 had significant greater weights for oleate), that vehicle-treated islets differentiated preferentially through lineage 1, and that palmitate-treated islet cells differentiated preferentially through lineage 2 (Fig. 5D).
Next, we identified genes whose expression patterns along the path to proliferation (lineage 3) were differentially modulated between the oleate (Fig. 5E) or palmitate (Supplementary Fig. 4A) and vehicle conditions. For each gene and for each experimental condition, smoothers were fitted using a negative binomial generalized additive model. Significant differences between smoothers were identified by comparing smooths in a factor-smooth interaction model. Compared with vehicle, expression of several genes was significantly modulated in the palmitate and oleate groups (Supplementary Table 8) and involved in UPR/ER stress–related pathway (Supplementary Table 9).
Because visually interpreting the differential modulation of gene expression along a pseudotime lineage between conditions is challenging with heatmaps (Fig. 5E and Supplementary Fig. 4A), the differences between pairs of smoothers were visualized in Fig. 5F and Supplementary Fig. 4B. In lineage 3, genes involved in the electron transport chain, such as Ndufa12 and Atp5b, as well as the antioxidants Sod2, Prdx1, and Prdx2, were upregulated, whereas β-cell markers, such as Ins2, Neurod1, and Pdx1, were reduced in response to both oleate and palmitate compared with vehicle (Fig. 5F and Supplementary Fig. 4B). Atf4, associated with UPR/ER stress, was upregulated at the beginning of the pseudotime and downregulated at the end in both FA-exposed groups (Fig. 5F and Supplementary Fig. 4B).
ROS Generation and Peroxiredoxin Activity Are Necessary for Oleate-Induced β-Cell Proliferation
The potential role of ROS and antioxidant signaling in β-cell proliferation in response to oleate suggested by scRNA-seq was experimentally tested in isolated rat islets. We focused on the peroxiredoxin family because Prdx1, Prdx2, Prdx3, and Prdx4 were significantly upregulated in the proliferative β2 cluster as well as along the pathway to proliferation (lineage 3) in response to oleate (Supplementary Table 3 and Figs. 4 and 5). First, using the ROS sensor CellROX Green, which preferentially marks nuclear and mitochondrial ROS, we observed that oleate, but not palmitate, significantly increased ROS levels in islets exposed to FAs for 24 h (Fig. 6A). Next, we showed that a 48-h NAC treatment dose dependently reduced β-cell proliferation in response to oleate but not vehicle (Fig. 6B). Finally, the specific inhibitor of peroxiredoxins, conoidin A (21), dose dependently decreased β-cell proliferation in response to oleate but not vehicle (Fig. 6C). To confirm the selectivity of conoidin A, we tested whether it altered the proliferative response to the DYRK1A inhibitor harmine (22). At the maximal concentration used in Fig. 6C, conoidin A reduced the response to oleate but not to harmine (Fig. 6D).
MYC Is Regulated by Oleate and Is Necessary for Oleate-Induced β-Cell Proliferation
The proto-oncogene MYC plays a key role in β-cell proliferation (23) and is a potential effector of ROS signaling (24). Intriguingly, we identified a small group of Myc-expressing cells within the β3 subpopulation (Fig. 7A), along the path toward proliferation (Fig. 5), and MYC target genes (e.g., Cdk1, Cdc20, and Ccna2) were upregulated in proliferating β-cells (Fig. 1 and Supplementary Table 3). Consistent with this transcriptomic profile, immunohistochemical analysis of MYC+ β-cells showed a time-dependent increase in response to oleate but not palmitate (Fig. 7B and C). The small-molecule inhibitor of the MYC/MAX interaction 1-RH (25) decreased the β-cell proliferative response to both harmine and oleate (Fig. 7D). Accordingly, oleate increased expression of the MYC target gene Cdk1, but not Myc, and NAC blocked this effect (Fig. 7E).
The aims of this study were to ascertain the differential effects and transcriptional response to individual FAs on β-cell proliferation and to characterize the transcriptome of proliferating β-cells in response to FAs at single-cell resolution. Palmitate and oleate were selected as representative unsaturated and monounsaturated species, respectively, because they are the two most abundant FAs in human plasma (26), and our previous studies demonstrated that oleate increases β-cell proliferation in rat islets (15). scRNA-seq enabled us to capture several β-cell clusters, including proliferating β-cells, and non–β-cell clusters. Selective proliferation of β-cells was confirmed by the absence of proliferating non–β-cell clusters. In β-cells, both FAs increased expression of markers of energy metabolism and decreased markers of β-cell maturation. Notable differences between the two FAs included a marked increase in cellular stress in response to palmitate. Comparing the transcriptomic profile of proliferative and nonproliferative β-cells at the gene and trajectory levels highlighted a key role for ROS and peroxiredoxin signaling along the path to proliferation. Accordingly, ROS scavenging or peroxiredoxin inhibition blocked oleate-induced β-cell proliferation, a process that required MYC activity.
β-Cell functional heterogeneity is well established, and single-cell transcriptional profiling studies have reinforced this notion by identifying multiple β-cell subtypes with unique transcriptional signatures (reviewed by Mawla and Huising ). In this study, independent of the culture condition, we captured five transcriptionally distinct β-cell clusters characterized by high expression of maturity and/or function (β1, β4, and β5), proliferative (β2), and UPR/ER stress (β3) markers. Although β-cell clusters distinguished by expression of ER and oxidative stress as well as maturity markers were identified previously in human β-cells (28,29), not surprisingly, transcriptional signatures and β-cell clustering differed significantly in our study, likely because of species, culture, and methodological differences as well as the limitations of the scRNA-seq technology (27). Therefore, the biological significance of the β1 and β3–5 clusters remains to be established.
Differential effects of individual FAs on the β-cell have been investigated in the context of glucolipotoxicity (reviewed by Lytrivi et al. ). However, β-cells also adapt to glucolipotoxic stress by upregulating genes involved in the synthesis and storage of neutral lipids, an effect that is more pronounced after exposure to unsaturated versus saturated FAs (30). In general, saturated FAs (e.g., palmitate) have deleterious effects on β-cell function and survival (31), whereas unsaturated FAs (e.g., oleate) are protective (14). The advent of unbiased transcriptomic technologies has enabled examination of the effects of FAs on islet function at the level of global changes in gene expression. Microarray or bulk RNA-seq approaches of human islets as well as rodent insulin-secreting cell lines have shown that exposure to palmitate alone or in the presence of high glucose leads to a gene expression profile similar to that of islets from donors with type 2 diabetes, characterized by an increase in UPR/ER stress, protein degradation, immune surface receptor, autophagy, mRNA splicing regulation, and nuclear transport processes (32–36). To our knowledge, our study is the first to compare transcriptomic profiles of islets exposed to individual FAs at the single-cell level. In line with previous transcriptomic analyses of whole islets, genes involved in UPR/ER stress were upregulated in the β3–5 subpopulations in response to palmitate, as well as translational processes and oxidative stress (37–39), confirming the deleterious effects of saturated FAs. Interestingly, not all β-cell subpopulations were affected (e.g., β1 and β2); however, whether the heterogeneity described here is a snapshot of a dynamic flux or a stable cell state remains to be explored. On the other hand, cellular stress responses were not augmented in response to oleate, in agreement with previous studies (14). Instead, both oleate and palmitate increased expression of genes involved in glycolysis and oxidative phosphorylation in several β-cell clusters (β1, β4, and β5). In line with these findings, increased expression of genes involved in oxidative phosphorylation was observed in human islets exposed to oleate or palmitate (40). Intriguingly, our results resemble the transcriptional response of the human β-cell in diabetes (41) and may reflect an attempt by the β-cell, albeit failed in the case of palmitate, to meet the elevated secretory demand.
Although no β-cell cluster was unique to an individual culture condition and the relative proportion of β1 and β3–5 was independent of the specific FA treatment, we observed a higher proportion of proliferative β-cells (β2) in oleate-treated islets. Furthermore, we report distinct maturity states across the different clusters, especially along the path toward proliferation, where proliferative β-cells were less mature. These data are consistent with other studies where cycling β-cells have shown immature gene expression profiles (20,42). Interestingly, both oleate and palmitate decreased expression of β-cell markers and the overall maturity score. Although the effect of palmitate is in line with previous studies (43–45), β-cell dedifferentiation after chronic exposure to glucose and oleate has not, to our knowledge, been described previously.
By comparing the transcriptome of proliferative (β2) with that of nonproliferative β-cells, we found that besides the expected upregulation of pathways involved in cell division processes, proliferative β-cells showed upregulated genes and pathways related to oxidative phosphorylation and antioxidant activity. Importantly, ROS and peroxiredoxins were found to be essential for oleate-induced β-cell proliferation. Nutrients are important inducers of ROS (e.g., O2− and H2O2) (46) in the β-cell, and although excessive ROS production is associated with dysfunction (47), a moderate increase in ROS levels is required for glucose-induced insulin secretion (48) and promotes β-cell neogenesis in zebrafish (49) and postnatal β-cell proliferation in mice (50). Therefore, the increase in peroxiredoxin levels documented in this study may be important in limiting ROS but also in functioning as a signal mediator by oxidizing target thiols in a redox relay (46). Interestingly, the nuclear factor erythroid 2–related factor (Nrf2) transcription factor, involved in the maintenance of β-cell mass during metabolic stress (51), drives antioxidant gene expression and promotes β-cell survival and proliferation (52). Because the Keap1/Nrf2 pathway is activated by ROS (53), Nrf2 could link ROS to peroxiredoxins and β-cell proliferation in response to oleate.
Consistent with the important role of MYC in rat and human β-cell proliferation (54), MYC inhibition decreased β-cell proliferation in response to oleate. Myc was expressed in a small group of cells along the path toward proliferation, and MYC target genes were upregulated in proliferating β-cells. In agreement with these findings, nutrients increase Myc expression in islets in vivo and in vitro (55). Upstream of MYC, ROS promotes Myc expression in islets (24) and ERK-dependent phosphorylation in melanoma cells (56). We showed that oleate increased the percentage of MYC-expressing β-cells and that ROS signaling was important for MYC target gene expression in response to oleate, revealing a potential mechanistic link between oleate-induced ROS and the β-cell proliferative response.
In obese individuals, elevated plasma glucose levels arising from insulin resistance lead to increased functional β-cell mass that maintains normoglycemia in the prediabetic state. Mechanistically, several studies suggest that changes in glucose levels are directly sensed by the β-cell (57,58). Similarly, we speculate that FAs, whose plasma levels are also increased in obesity (59), contribute to β-cell adaptation to insulin resistance by potentiating the effects of glucose. Importantly, however, only certain FAs, such as oleate, improve β-cell function and survival (14) and, as this study suggests, promote β-cell proliferation.
In summary, our findings suggest that ROS and peroxiredoxin signaling via MYC are required for oleate-induced β-cell proliferation in rat islets, providing important information on the molecular mechanisms of β-cell adaptation to nutrient excess.
This article contains supplementary material online at https://doi.org/10.2337/figshare.21215594.
Acknowledgments. The authors thank G. Fergusson and M. Éthier from the Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM) Rodent Metabolic Phenotyping Core for assistance with islet isolations, D. Gauchat and P. St-Onge from the CRCHUM Flow Cytometry Core for assistance with measurements of β-cell proliferation, and V. Barrès and L. Meunier from the CRCHUM Pathophysiology Core for their help with immunostaining.
Funding. This study was supported by the National Institutes of Health (R01-DK-58096), the Canadian Institutes of Health Research (MOP 77686), and the Quebec Cardiometabolic Health, Diabetes and Obesity Research Network.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. A.V. was responsible for conceptualization, supervision, methodology, investigation, formal analysis, and writing of the original draft. J.G. was responsible for conceptualization, methodology, investigation, writing, and editing. A.F.-M. was responsible for methodology, investigation, formal analysis, and writing. Z.E.A. was responsible for investigation and formal analysis. A.-L.C. was responsible for methodology. R.S. was responsible for conceptualization and validation. V.P. was responsible for conceptualization, validation, writing, editing, funding acquisition, and project administration. V.P. 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.
Prior Presentation. Parts of this study were presented at the 81st Scientific Sessions of the American Diabetes Association, 25–29 June 2021.