**2. Results**

#### *2.1. Identification of Islet Cell Type Clusters by sc-RNAseq*

Previous studies from our group demonstrated that a one-week period of HFD (60% kcal from fat) feeding compared to a low-fat diet (LFD, 10% kcal from fat) in mice led to a significant increase in mRNA translation, mTOR pathway activation, and cellular proliferation in pancreatic islets [10,11]. However, these prior studies do not provide the granularity to assess which of the specific endocrine cellular populations account for these responses. To gain more insight into the molecular pathways and the individual changes happening at the single-cell level, we performed sc-RNAseq from a collection of dissociated islets (180–200 islets/mouse) isolated from a total of seven male C57BL/6J mice fed for one week with either a HFD (60% kcal from fat, n = 4) or a control LFD (10% kcal from fat, n = 3). Each mouse sample was processed separately. Our analysis of sc-RNAseq allowed the annotation of islet cell types into different clusters based on the expression of key identifying genes, depicted in the Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) plots in Figure 1a and Supplementary Figure S1. We identified 26 cell clusters representing 9 distinct cell types based on the gene expression of canonical markers and, as expected, the major populations identified were β cells (Ins1<sup>+</sup>, Ins2+, Mafa+), α cells (Gcg<sup>+</sup>, Ttr+, Irx2+), and δ cells (Sst<sup>+</sup>, Ghsr+, Rbp4+), with smaller populations of other cell types, including PP cells (Ppy+), ductal (Krt19<sup>+</sup>, Hnf1b+), endothelial cells (Plvap<sup>+</sup>, Esm1+), immune cells as macrophages (Adgre1<sup>+</sup>, Lyz2+), T cells (Trbc2<sup>+</sup>, Cd3g<sup>+</sup>, Cd4+), and B cells (Cd19<sup>+</sup>, Igkc<sup>+</sup>, Ighm+) (Figure 1a). Notably, no differences were observed in the percentage of cells identified in any given cluster in HFD-fed compared to control LFD-fed mice (Figure 1b–f). Furthermore, we found that the percentage of total α, β, and δ cells or other types of cells identified with sc-RNAseq does not change substantially within replicates and between the LFD and HFD groups (Supplementary Table S1). These results demonstrate the reproducibility of our data at a single-cell level.

**Figure 1.** Identification of islet cell type clusters by single-cell RNA sequencing. Single cells were obtained from dissociated islets from male *C57BL*/*6J* mice fed for one week with either a high-fat diet (HFD, 60% kcal from fat, *n* = 4) or a control low-fat diet (LFD, 10% kcal from fat, *n* = 3) and used to perform single-cell RNA sequencing. (**a**) Annotation of islet cell types into different clusters based on the expression of key identifying genes, depicted in the Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) plots of merged sc-RNAseq profiles from LFD and HFD mice; (**b**) UMAP of single-cell RNA sequencing profiles from islets of individual mice fed a HFD or a LFD, as indicated; In (**c–f**), the percentage of β cells (**c**), α cells (**d**), δ cells (**e**), and other cell types (**f**) identified per cluster relative to the total number of cells sequenced is shown. Data are mean ± SEM, *n* = 3 for LFD and *n*= 4 for HFD. The number in the *parentheses* above each bar indicates the average number of cells per cluster.

#### *2.2. Single-Cell RNA Sequencing Analysis Reveals Greatest Gene Expression Changes in Minor* β *Cell Clusters Following Short-Term HFD Feeding*

Next, we examined the gene expression profiles in β cells from mice fed either a HFD or LFD for one week. UMAP analysis identified a total of 11 distinct β-cell clusters (β1-β11, see Figures 1a and 2a). Based on the proportion of cells per cluster (Figure 1c), we identified three major clusters of β cells (β1–β3) and eight minor clusters (β4–β11). Di fferential gene expression between HFD and LFD were interrogated, and statistical significance was determined by using edgeR on the integrated single-cell data obtained by the R package Seurat (see Section 4). It is notable that the major clusters of β cells (β1–β3) showed minimal change in gene expression patterns (Supplementary Figure S2), whereas the greatest changes were observed in the minor clusters (most notably β5, β7, β8, β10, β11) (Figure 2b). These findings sugges<sup>t</sup> that minor β-cell clusters drive the earliest responses to HFD, and further emphasize how bulk RNA sequencing might miss these sentinel changes. In these minor clusters, the most notable gene expression changes reflect upon hormone secretion and intracellular inflammatory pathways. Clusters β5, β7, β8, β10, and β11 demonstrated significant *decreases* in genes that promote insulin secretion (*Atf6, Meg3, Herpud1*), β cell survival (*Sox4, Tnfaip3*), calcium signaling (*Robo2*), and activation of immune response (*Nfkb1, Il1r1, Cxcl10, Ifngr2*) (Figure 2b), with increases in genes that block insulin release (*Mt2*) and promote apoptosis (*Mif*) [12–17]. Furthermore, we observed an increased expression of *Mafa*, a gene that encodes a transactivator of the *Ins1*/*2* genes, in this group of clusters following HFD feeding.

**Figure 2.** Identification of di fferentially expressed genes of the minor β-cell clusters. B-cell clusters were identified from dissociated islets from male *C57BL*/*6J* mice fed for one week with either a high-fat diet (HFD, 60% kcal from fat) or a control low-fat diet (LFD, 10% kcal from fat). (**a**) Representative UMAP plot of β-cell clusters identified by single-cell RNA sequencing; (**b**) heatmaps of the minor β-cell clusters of genes significantly di fferentially expressed (*p* < 0.05) in the β-cell clusters β5, β7, β8, β10, and β11; genes are ordered from most positive to most negative fold-change.

Among the β-cell clusters that showed minimal gene expression changes between HFD and LFD, four of them (β1, β2, β6, and β9) showed downregulation of genes for the endoplasmic reticulum stress response (*Chac1*) and activation (*Nfkbia, Cxcl10*) of inflammatory processes (Supplementary Figure S2) [18,19]. Clusters β3 and β4 showed an increase in expression of genes associated with insulin secretion (*Dbp*) and a complementary decrease in expression of genes encoding a suppressor of insulin signaling (*Pdea5*) and inflammatory pathways (*Nfkbia, Tnfaip3*) (Supplementary Figure S2) [20,21].

Collectively, these gene expression changes provide a picture of the response of β cells to HFD feeding—namely, that specific, larger clusters of β cells demonstrate augmented insulin secretory capacity and reduction of ER stress, whereas smaller clusters exhibit a reduced insulin secretion capacity and survival. It remains unclear if these findings signify an early response of small clusters, which will be later reflected in larger clusters as the impact of HFD is prolonged, or if they reflect what will be a persistent, and perhaps competing, heterogenous response among β-cell subtypes whose function on-balance determines T2D outcomes. In this regard, our prior studies of HFD feeding in male mice [22] were suggestive of recurring patterns of β-cell loss, wherein it was suggested that di fferent subpopulations of β cells might have shown evolving susceptibility to death or dedi fferentiation as HFD feeding progressed.

Using the same diet that we have used in the present work, other studies from our group have demonstrated that although one week of HFD feeding results in only minor impairments of glucose tolerance, longer-term HFD feeding impairs both glucose tolerance and insulin secretion [10,11,22]. Therefore, further studies would be necessary to know if β-cell dysfunction in the smaller clusters drives hyperglycemia and impairs insulin secretion during obesity. Several studies have also demonstrated that proinflammatory signals, cellular stress, and genetic components contribute to T2D development [23–26]. Here, we found that despite the HFD feeding, β-cell clusters overall tend to decrease the expression of proinflammatory genes, suggesting that cellular stress and β-cell dysfunction are the initial triggers that propagate the inflammatory response in the HFD feeding model. However, more studies will be necessary to clarify if the events of cellular stress and β-cell dysfunction identified in the smaller clusters of β cells propagate cellular stress in other larger clusters.

#### *2.3. Gene Pathway Analysis Reveals Molecular Responses Related to Inflammation*/*Immunity and Oxidative Stress in* β *Cells Following HFD Feeding*

Whereas the preceding studies focused on specific islet-cell gene expression changes between HFD- and LFD-fed mice, they do not provide an unbiased context for how these alterations a ffect functional signaling pathways in the cell. Therefore, we performed gene ontology gene set enrichment analysis (GO GSEA) to evaluate the signaling pathways specifically altered by HFD feeding in di fferent β-cell clusters. We identified three general pathway responses exemplified by clusters β1, β4, and β7 (Figure 3). β-cell clusters with minimal gene expression changes in response to HFD had an overrepresentation of gene pathways in the response to cellular metabolism, endoplasmic reticulum (ER) stress, and oxidation–reduction process (Figure 3). A second pattern, seen largely among clusters β3 and β4 (Figure 3), showed an overrepresentation of genes associated with cell di fferentiation/development, immune response, and the negative regulation of apoptosis. Finally, a third group, exemplified by cluster β7 and exhibited the greatest gene expression changes in response to a HFD, showed changes in gene pathways that modulate immune/inflammatory response and cellular stress (Figure 3). Collectively, these pathway analyses are suggestive of β-cell adaptations to HFD feeding in which inflammation/immune cascades are a ffected and in which changes to cellular redox and metabolite utilization prevail. Our findings are consistent with the observations in the literature that free fatty acids present in HFDs impose oxidative stress [6] and enhance signaling pathways linked to cytokines and inflammation [27].

**Figure 3.** Altered signaling pathways of the minor β-cell clusters. Gene ontology gene set enrichment analysis (GO GSEA) was performed after single-cell RNA sequencing to evaluate the signaling pathways specifically altered by high-fat-diet feeding vs. low-fat-diet feeding in different β-cell clusters. Shown are the GO GSEA analyses of β1, β4, and β7 cell clusters.

#### Short-Term HFD Feeding Reveals Heterogeneity in α Cell Responses

Islet α cells produce glucagon, a major insulin counterregulatory hormone. The dysregulated hypersecretion of glucagon contributes to dysglycemia in obesity and T2D [28]. Bulk RNA-seq studies from long-term HFD-fed mice previously demonstrated that α cells display remarkably minor changes in transcriptome profile [29], but such studies might belie greater changes in gene expression patterns in specific subsets of α cells. In this study, sc-RNAseq identified minor distinct clusters of α cells (designated α12– α16, Figure 4a), the populations of which were not altered upon HFD feeding of mice for one week (Figure 1d). Di fferential gene expression analysis of these clusters was then performed. The cluster containing the largest pool of α cells ( α12) did not show statistically significant changes in gene expression; however, clusters α13, α14, α15, and α16 showed multiple di fferentially expressed genes in HFD-fed mice compared to LFD-fed mice (Figure 4b). Notably, we observed that HFD feeding led to a decrease in genes associated with inflammation (*Il1r, Cxcl1*, *and Nfkbia*) in all four α-cell clusters (Figure 4b). Only cluster α13 showed changes in hormone expression, with an increase in Ppy after HFD feeding, indicating that, in general, short-term HFD does not lead to the misexpression of hormone-encoding genes. Other general observations include an increase in the expression of genes involved in cell survival and proliferation (*Upk3a*) and cellular stress (*Hspa1a* and *Hspa1b*) in clusters α14 and α16 [30,31], the enhanced expression of genes related to endocrine progenitors (*Neurog3*), and an oxidative stress/ER stress (*Ero1b*) decrease in cluster α15 [32,33]. Collectively, these data indicate that, like β cells, the early gene expression changes in α cells (a) are dependent on the specific cluster and thereby exhibit cluster-dependent heterogeneity, and (b) might be missed by bulk sequencing approaches since a major cluster of α cells exhibits no significant changes.

**Figure 4.** Transcriptome profile in α-cell clusters identified by single-cell RNA sequencing. α-cell clusters were identified following single-cell RNA sequencing from male *C57BL*/*6J* mice fed for one week with either a high-fat diet (HFD, 60% kcal from fat) or a control low-fat diet (LFD, 10% kcal from fat). (**a**) Representative UMAP plot of α-cell clusters; (**b**) heatmaps depicting genes di fferentially regulated (*p* < 0.05) in α-cell clusters; genes are ordered from most positive to most negative fold-change.
