O304 treatment prevents islet gene expression signature changes induced by HFD
We have previously demonstrated that the AMPK activator O304 improves blood glucose homeostasis in both human T2D subjects as well as in high-fat diet induced obese and diabetic mouse models. In the present study, we have now analysed the in vivo effects of O304 treatment on the genome-wide transcriptional program of the pancreatic islets. To characterize gene expression signatures and chromatin marks of β-cells in the context of elevated glucose levels induced by obesity, and the potential positive effects of O304-mediated remission of dysglycemia in this context, we performed RNA-sequencing and Chromatin ImmunoPrecipitation sequencing (ChIP-seq) of islets isolated from 20-week-old mice on regular diet (RD), high-fat diet (HFD) and HFD formulated with O304 (HFD-O304). Diets were introduced at 10 weeks of age. Fasting blood glucose and plasma insulin levels were similar between the groups at start of the diet (Supplementary Fig. S1a). In line with our previous data8, HFD-O304 mice had significantly lower glucose and insulin levels than HFD mice after 9 weeks of the treatment period (Supplementary Fig. S1b). Food intake was highest for RD mice (4.2 g/m/day) and lowest for HFD mice (2.5 g/m/day), whereas food intake was increased in HFD-O304 mice compared to HFD fed mice (3.2 g/m/day, Supplementary Fig. S1c). A blunted first-phase insulin release is the first detectable defect in β-cell function in T2D aetiology and can be assessed by arginine stimulated insulin secretion tests, which provides an estimate of the β-cell secretory capacity independent of potential increased peripheral glucose uptake/improved insulin. Arginine stimulated insulin secretion was improved in HFD-O304 mice as compared to non-treated HFD mice (Supplementary Fig. S1d), confirming previous data on improved first-phase insulin secretory capacity of the β-cells in O304 treated diet-induced obese mice8.
RNA-seq and ChIP-seq analysis of islets isolated from mice following 10 weeks of treatment demonstrated that 1621 genes were differentially regulated in islets of HFD mice compared with that of RD mice (918 upregulated, 703 downregulated: FDR < 0.01) (Fig. 1a, left, Supplementary Table S1). In contrast, islets isolated from HFD-O304 mice presented a transcriptional profile similar to islets of RD-mice, with only 657 differentially expressed genes between islets of these two groups of mice (227 upregulated, 430 downregulated) (Fig. 1a, right, Supplementary Table S1). Of the 1621 differentially expressed genes in HFD mice, 1495 corresponds to genes that are not differentially expressed between islets of HFD-O304 mice and RD controls (Fig. 1b and Supplementary Fig. S2a). This implies that O304 prevented, directly or indirectly, the dysregulation of the majority of genes whose expression is altered in this HFD provoked pre-diabetic context. In fact, O304 preserved the expression of > 90% of the differentially expressed genes in islets of HFD mice (Fig. 1b). Accordingly, both principal component analysis (PCA) (Supplementary Fig. S2b) and hierarchical clustering (Supplementary Fig. S2c) group HFD-O304 islet samples closer to islets from RD controls than to islets of HFD mice. Taken together, these findings provide evidence that O304 treatment prevents HFD induced islet transcriptional changes, thus largely preserving transcriptional profiles similar to islets of RD mice.
K-means clustering analysis of differentially expressed genes revealed 4 major clusters, out of a total of 5, where clusters A and B contain genes primarily upregulated in islets of HFD mice, and clusters C and D contain genes primarily downregulated in islets from HFD mice (Fig. 1c). Enrichment analysis of gene ontology (GO) terms and KEGG pathways revealed profound functional differences between the expression profiles of the three experimental groups (Supplementary Fig. S3 and Supplementary Table S1). The most enriched GO- and KEGG-terms for cluster A (containing genes that were upregulated in HFD islets) are related to endoplasmatic reticulum (ER) function and protein processing, including response to ER stress (Supplementary Fig. S3, Supplementary Table S1). This finding corroborates our previous data on ex vivo cultured islets where O304 treatment prevents increased expression of genes related to unfolded protein response under high-glucose conditions8. In the setting of obesity-provoked insulin resistance, β-cell proliferation is a compensatory response to the increased demand of insulin. Consistently, upregulation of genes in cluster B, which are mainly related to cell cycle and proliferation (Supplementary Fig. S3, Supplementary Table S1), were observed in islets of HFD mice, but not in islets of HFD-O304 mice. β-Cell proliferation has also been suggested to be tightly coupled to a loss of mature β-cell identity21, and AMPK has been implicated in maintenance of a mature β-cell phenotype via inhibition of mTORC1 signalling22. Still, while β-cell proliferative markers were not induced in islets of HFD-O304 mice (Supplementary Fig. S4a), we find no clear evidence for altered mTORC1 signalling (Supplementary Fig. S4b). Taken together these findings show that O304 treatment prevents up-regulation of genes associated with metabolic β-cell stress. However, given that O304 averts HFD induced insulin resistance8 (Supplementary Fig. S1), the effect of O304 on β-cell stress is likely partly indirect.
Genes in Cluster C, which were most abundantly expressed in islets of HFD-O304 mice and least expressed in HFD mice, were strongly enriched for terms/pathways related to regulation of hormonal secretion and insulin secretion in particular (Supplementary Fig. S3, Supplementary Table S1), as well as genes related to AMPK- and glucagon-signalling (Supplementary Table S1). Compared to islets from RD mice, O304 treatment preserved or even increased the expression of these genes in mice fed HFD (Supplementary Fig. S3, Supplementary Table S1). These findings suggest that, in the context of HFD, O304 treatment enhances the insulin secretory capacity of β-cells by increasing the abundance of transcripts related to these processes. Cluster D was enriched in genes largely related to neuronal function, which may reflect that β-cells share several phenotypic traits with neurons that are dependent on similar transcriptional programs for their function23. Last, cluster E genes are mainly associated with amino acid metabolism and other catabolic processes (Supplementary Fig. S3), likely reflecting that AMPK activation inhibits protein synthesis and stimulates catabolic processes24.
Taken together, these data demonstrate that O304 treatment largely prevented the transcriptional changes observed in islets of HFD mice. Although the net action of O304 treatment on β-cells likely occurs indirectly via amelioration of insulin resistance and dysglycemia as well as through a combination of direct downstream targets, it significantly impacts on several key β-cell processes affected in T2D aetiology, and in particular O304 treatment prevented gene expression changes associated with insulin signalling and hormonal secretion.
O304 treatment mitigates gene expression signatures associated with β-cell stress
To further understand the impact of O304 treatment on glucose homeostasis and β-cell function in the context of HFD and insulin resistance, we explored our transcriptional dataset to investigate the effects on specific genes relevant for β-cell function and/or β-cell identity, and whose altered expression has been implicated in β-cell stress. As HFD mice are not overt diabetic but exhibit a pre-diabetic glucose intolerant phenotype, islets of HFD-mice were expected to display subtle expression changes, primarily of early markers of β-cell stress. Preserved expression levels of such markers would imply a maintained β-cell phenotype, and the functional annotation of differentially expressed genes (Supplementary Fig. S3, Supplementary Table S2) in our data indeed suggest that this may be the case.
Aldehyde dehydrogenase 1 isoform A3 (ALDH1A3) is a proposed marker for metabolically stressed β-cells25,26,27,28,29 and in agreement with these findings, Aldh1a3 expression was significantly increased in islets from HFD-mice compared to RD controls (Fig. 2a). Notably, Aldh1a3 expression was reduced in islets of HFD-O304 mice compared with both HFD and RD islets (Fig. 2a). Increased Aldh1a3 expression in β-cells has been linked to reduced expression of key transcription factors associated with maintenance of β-cell identity and function27. No major differences in gene expression were observed for a set of mature β-cell identity genes although the expression profile of HFD-O304 islets were more similar to RD islets than to HFD islets22,30 (Supplementary Fig. S5a). However, O304 treatment not only prevented the reduced expression of NeuroD1 and Isl1 (Fig. 2b,c) observed in HFD islets but also enhanced the expression of these genes compared with that observed in RD islets (Fig. 2b,c). In addition, a similar trend was observed for MafA, the expression of which was slightly decreased in HFD islets (Fig. 2d) as compared to RD and HFD-O304 islets. FoxO1 that regulates NeuroD1 and MafA was slightly upregulated in HFD-O304 islets while O304 treatment had no effect on Pdx1 or Nkx6.1 expression (Fig. 2e–g). The limited transcriptional changes observed for only a subset of β-cell identity genes suggest that increased Aldh1a3 expression is not directly linked to a reprogrammed β-cell state but rather appear to be a general marker of early metabolic β-cell stress.
Urocortin-3, Ucn3, is a marker for functionally mature β-cells that showed reduced expression under metabolic stress conditions31,32. Ucn3 expression tended to be decreased in HFD islets as compared with that of RD islets and O304 significantly increased Ucn3 expression compared with that of both RD and HFD islets (Fig. 2h). Taken together, these findings indicate that preservation of β-cell function in O304 treated HFD mice involves both maintenance of mature β-cell identity and amelioration of β-cell stress.
Reprogramming due to metabolic stress has also been associated with increased expression of “disallowed genes”, i.e. genes normally repressed in β-cells33,34,35,36. Although we do not observe major changes in the expression of such genes, the transcriptional profile of this gene set also tends to be more similar between RD islets and HFD-O304 islets (Supplementary Fig. S5b). This finding suggests that the expression of disallowed genes in dysfunctional β-cells is limited at very early stages of T2D but gradually increases in response to sustained hyperglycemia and progression of T2D.
O304 treatment preserves the expression of insulin secretory associated genes in HFD mice
HFD mice displayed highly increased levels of fasting plasma insulin (Supplementary Fig. S1a), and islet mRNA levels of Ins1 tended to be elevated while the abundance of Ins2 transcripts were significantly increased in HFD compared to RD islets (Fig. 2i,j). In HFD-O304 mice, Ins2 transcripts still tended to be increased compared to RD islets whereas Ins1 mRNA levels were significantly reduced as compared to HFD mice (Supplementary Fig. S1a and Fig. 2i), likely reflecting a decreased demand for insulin production and secretion as a consequence of improved insulin sensitivity. The lower metabolic load on β-cells of HFD-O304 mice would in turn contribute to β-cell rest and preserved β-cell function. Interestingly, the differentially expressed genes in our data were enriched for GO-terms associated with response to Glucose Stimulus and Insulin Secretion (GSIS) (Supplementary Fig. S3, Supplementary Table S1). Hierarchical clustering of samples based on genes related to insulin secretion (99 genes from 3 gene sets) segregates HFD islets from RD and HFD-O304 islets, which further underscores the preservation of gene expression profiles related to insulin secretion in HFD-O304 islets as compared to lean controls (Supplementary Fig. S6).
Under conditions of hyperglycemia and perturbed GSIS, the glucose transporter Slc2a2 is markedly downregulated, which correlates with an impaired insulin secretory response in both mice and rats37,38,39,40. Consistently, Slc2a2 expression was downregulated in HFD islets as compared to RD islets, and this expression was completely restored in islets of HFD-O304 mice (Fig. 2k). We observed a similar expression pattern for Trpm5, another glucose-responsive gene also associated with functional GSIS41,42. Alike Slc2a2, HFD resulted in downregulation of Trpm5 expression in islets (Fig. 2l) suggesting that reduced levels of Trpm5 is also an early indicative of β-cell stress and disturbed glucose homeostasis. However, in HFD-O304 islets, Trpm5 levels were increased compared with both HFD and RD islets (Fig. 2l). These data demonstrate that, in pre-diabetic HFD mice, O304 treatment largely normalized islet gene expression signatures associated with β-cell stress and impaired glucose response. Taken together, these findings provide evidence of preserved β-cell health status in HFD mice treated with O304 that in turn might prevent long-term functional deterioration of β-cells due to overnutrition.
O304 treatment induces remobilisation of chromatin marks in islets of HFD mice
In order to further elucidate how O304 impacts on gene regulation and epigenetic marks, we characterized the distribution of active (H3K27Ac) and repressive (H3K27Me3) chromatin marks by ChIP-seq in islets from RD controls, HFD, and HFD-O304 mice. In RD islets we identified 16,314 H3K27Ac and 15,517 H3K27Me3 binding regions which present a highly non-overlapping distribution as expected (Fig. 3a and Supplementary Fig. S7a). We also observed that active and repressive chromatin marks correlated positively or negatively, respectively, with gene expression, further validating the quality of our data (Supplementary Fig. S7b,c).
A large fraction of the called peaks for both active (H3K27Ac) and repressed (H3K27Me3) chromatin were common for RD, HFD, and HFD-O304 islets. Importantly, a substantial number of peaks were unique for each condition or only present in two of them (Fig. 3b–d). Interestingly, the abundance of H3K27Ac binding sites was higher in HFD and HFD-O304 islets, as compared to RD islets (21,674; 20,305; and 16,314 respectively). In HFD islets, binding of H3K27Ac to 7428 new genomic regions, that were not present in RD islets, was observed together with the depletion of this mark from 2068 sites as compared with that of RD islets (Fig. 3b,d). O304 treatment prevented H3K27Ac binding in 4321 out of these 7428 new sites and also preserved its binding to 567 of the 2068 sites lost in HFD islets (Fig. 3b,d). These results provide evidence that O304 prevented ~ 58% of the redistribution of active chromatin marks observed in HFD islets, as well as preserved ~ 27% of H3K27Ac regions lost in HFD islets (Fig. 3d).
In contrast to H3K27Ac, the number of H3K27Me3-enriched regions detected in islets from lean controls and HFD mice were similar, while these regions were more abundant in HFD-O304 islets (15,517; 15,438; and 22,049 respectively). Still, the distribution of these regions was considerably different, where only 9412 of the peaks were shared between the three groups (Fig. 3c,d). In HFD islets, 5377 new H3K27Me3-enriched regions were detected as compared to RD, but > 55% (3034) of these did not carry H3K27Me3 in HFD-O304 islets (Fig. 3d). In addition, 45% (2469 out of the 5456) of regions depleted for H3K27Me3 in HFD islets as compared to RD islets were preserved in islets of HFD-O304 mice (Fig. 3d). Taken together, these findings suggest that O304 treatment also prevented the redistribution of a large fraction of repressive chromatin marks induced by HFD in islets.
Interestingly, most of the active and repressed regions that were preserved by O304 treatment were located in intergenic regions, not in promoters (< 1 kb from TSS), suggesting that at least some of the gene expression changes induced in HFD mice and preserved in HFD-O304 mice may depend on regulatory input from distal enhancer elements (Supplementary Fig. S7). These data provide evidence that remobilisation of active and repressive chromatin marks contributed to the observed gene expression changes associated with HFD, and that O304 treatment preserved a significant fraction of the epigenetic signature characteristic of lean RD islets.
Increased Aldh1a3 gene expression is accompanied by changes in histone modifications
As stated above, Aldh1a3 has in recent years emerged as a potential biomarker for dysfunctional β-cells27, and Aldh1a3 expression is significantly increased in our RNA-seq data in HFD islets, but not in treated HFD-O304 islets. In addition, a significant decrease in the percentage of ALDH1A3-positive β-cell area in HFD-O304 mice (~ 2.5%) with respect to that of HFD mice (~ 19%) (Fig. 4b,c) was observed. Notably, in isolated islets cultured ex vivo Aldh1a3 expression was significantly increased at high glucose concentrations (22 mM), as compared with islets cultured at 11 mM glucose concentration, whereas Aldh1a3 expression in islets cultured at 22 mM glucose in the presence of 5 μM O304 was significantly reduced even compared with that of islets cultured at 11 mM glucose (Fig. 4a). Thus, O304 prevented an increase in Aldh1a3 expression both in vivo in islets of HFD-O304 mice and ex vivo in isolated islets cultured at high glucose levels. Together these data provide evidence that O304 ameliorates β-cell stress both directly and indirectly, i.e. via reduced metabolic burden.
To investigate further how these changes in gene and protein levels correlated with regulatory changes, we investigated H3K27Ac and H3K27Me3 occupancy at the Aldh1a3 locus in our ChIP-seq data. Notably, the increased expression of Aldh1a3 in HFD-islets was accompanied by enrichment of H3K27Ac in the promoter region as well as two distant regions (~ 70 kb and ~ 114 kb) upstream of the gene (Fig. 4d). None of these binding sites were present in islets of RD mice, and in HFD-O304 islets both the promoter region and the most distant element were devoid of H3K27Ac-enrichment. In addition, in islets from RD controls, HFD or HFD-O304 mice, all of these regions were bound by H3K27Me3, likely reflecting a heterogenous β-cell population with only a subset (~ 19%, Fig. 4b,c) of the β-cells of HFD mice expressing ALDH1A3. Most importantly, these findings also provide evidence that Aldh1a3 expression is actively repressed via H3K27Me3 in healthy islets and, as a consequence of the increased metabolic pressure induced by HFD, expression of Aldh1a3 is activated upon remobilisation of H3K27Ac to the gene promoter and to two distal upstream elements that may represent potential regulatory regions. Taken together, these data demonstrate that O304 treatment induced remobilisation of chromatin marks in pre-diabetic islets, which thus likely contributes to preservation of their transcriptional profile.
O304 treatment reverts gene expression changes associated with β-cell stress and impaired insulin secretion
To test whether O304 treatment could not only prevent, but also revert the islet gene expression changes associated with diet-induced obesity, mice were fed HFD for 9 weeks and then switched to HFD formulated with O304 (0.8 mg/g) for another 9 weeks (from here on referred to as HFD-switch mice) (Fig. 5a). Vice versa, mice that had been fed HFD-O304 for 9 weeks were switched to HFD without O304 (from here on referred to as O304-switch mice). Control mice were fed RD throughout the whole time period. Mice fed HFD had a significantly decreased food intake (> 30% reduction) as compared to both mice fed RD or HFD-0304, but food intake increased after the switch to HFD-O304 (Supplementary Fig. S8). Reciprocally, food intake of HFD-O304 mice was reduced after the switch to HFD (Supplementary Fig. S8).
In HFD-switch mice the transition from HFD to HFD-O304 completely reverted the hyperinsulinemia, dysglycemia, and insulin resistance established by the 9 weeks HFD feeding (Fig. 5b,c). Notably, although weight gain was halted by switching mice from HFD to HFD-O304, neither body weight, lean body mass, nor fat mass decreased after the switch from HFD only to HFD-O304 (Fig. 5d,e), suggesting that the positive metabolic effects exerted by O304 treatment are not secondary to weight loss. In O304-switch mice, before the switch (T = 9 weeks), fasting plasma insulin were slightly decreased and glucose levels similar compared with that of RD mice (Fig. 5a–c). The switch from HFD-O304 to HFD resulted in an increased weight and fat mass gain as well as a marked increase in insulin and glucose levels (Fig. 5b–e). Together, these data demonstrate that O304 treatment not only prevents dysglycemia in HFD mice but also potently reverses dysfunctional glucose homeostasis in pre-diabetic HFD mice, and that this is uncoupled from changes in body mass.
To test if the improved metabolic profile of HFD-switch mice is also accompanied by significant improvements in islet gene expression profiles, we isolated total RNA from pancreatic islets at the time of the switch (Fig. 5a, T = 9 weeks) as well as after 9 weeks of intervention (Fig. 5a, T = 18) and analysed the expression of key islet genes associated with β-cell stress and insulin secretion. Importantly, HFD-switch mice presented with an improved gene expression profile at T = 18 weeks for key islet genes associated with β-cell stress and insulin secretion (Fig. 5f), indicating that O304 treatment can revert gene expression changes associated with HFD. The expression of the β-cell stress marker Aldh1a3 was reduced to minimal levels in HFD-switch islets after 9 weeks of treatment, while both Slc2a2 as well as Trpm5 expression was restored, and in the latter case even increased as compared to controls (Fig. 5f). Together these data provide evidence that O304 treatment potently can reduce β-cell stress and improve glucose sensing and secretory capacity also after an initial 9-week period of HFD feeding. The expression of β-cell identity gene MafA was restored in HFD-switch islets at T = 18 weeks and Isl1 expression was significantly increased as compared to control islets (Fig. 5f). The trend was similar for the mitochondrial biogenesis regulator Pgc1a (Fig. 5f). In addition, in line with previous ex vivo data on islets cultured in high glucose8, at T = 18 weeks the expression of the unfolded protein response genes Bip and Txnip were reduced, providing further support that O304 treatment reverted β-cell stress in HFD-switch mice. In O304-switch islets, the opposite change in expression profiles was observed with increased expression of Aldh1a3, as well as a decreased expression of Slc2a2, Trpm5, and MafA at T = 18 weeks, in comparison to the timepoint of the switch, i.e. T = 9 weeks (Fig. 5f). We did not detect any major change in the expression levels of the β-cell maturity marker Ucn3 nor in that of the disallowed gene Ldha at T = 18 weeks for HFD-switch islets. Taken together, these data demonstrate that O304 treatment not only potently prevents impaired glucose homeostasis and islet transcriptional changes associated with high-caloric food intake in HFD mice, but that O304 treatment also can restore metabolic function and associated islet gene expression changes in HFD mice with established obesity and dysglycemia.
Read more here: Source link