This study shows that palmitate exposure alters mRNA expression genome-wide in human islets in parallel with impaired insulin secretion, a defect often seen in T2D patients. Several genes with altered expression in palmitate-treated human islets also exhibited differential expression in islets from patients with T2D. We also demonstrate for the first time that the genome-wide DNA methylation pattern in human islets was affected by palmitate treatment. Several genomic regions had significantly higher global DNA methylation levels in the palmitate-treated islets compared with control islets, although these differences were generally small. This may be the result of the relatively short treatment (48 h) and that DNA methylation changes of a larger magnitude may require longer exposure to hyperlipidemia, a condition seen in many T2D patients. Also, since T2D is known to be a polygenic disease, it is possible that a combination of several modest changes in DNA methylation might have a combined larger effect which together could contribute to the pathogenesis of the disease. In support of this hypothesis, previous studies have shown relatively modest differences of DNA methylation in non-cancerous tissues and cell types, ranging from 0.13% to 11% [9, 49, 50]. However, even an absolute change of only a few percent units can represent a large difference in relative terms, as evident by the findings in our study where the fold change of DNA methylation between the treatment groups (palmitate treatment/control treatment) ranged from 0.54 to 1.84. This is in line with data from a recent study, where we found differential DNA methylation of 3,116 CpG sites in human pancreatic islets from subjects with T2D compared with non-diabetic controls with a fold change ranging from 0.58 to 1.61 when dividing the degree of methylation in diabetics with that in controls .
We also identified many genes with a difference in mRNA expression and a corresponding change in DNA methylation. This could suggest that altered DNA methylation influences the expression of the corresponding genes. Indeed, we have previously shown that increased DNA methylation reduces the transcriptional activity in functional in vitro studies [5, 8]. Interestingly, here we find decreased expression in parallel with increased DNA methylation of several candidate genes for T2D, such as TCF7L2 and GLIS3, in palmitate-treated human islets, suggesting that lipid-induced epigenetic modifications may affect the risk for diabetes. The fact that many of the up-regulated genes have corresponding increased DNA methylation could be due to the location of these CpG sites in the gene body. Indeed, DNA methylation of the gene body has been demonstrated to have a positive effect on gene expression . The gene regions with differential gene expression but without any change in DNA methylation could be targets for other forms of transcriptional regulation, such as histone modifications and/or altered activation by transcription factors. Also, genetic and epigenetic variation may interact to affect gene expression and subsequently contribute to the development of complex metabolic disease, such as obesity and T2D. Indeed, it has previously been shown that SNPs that introduce or remove a CpG site, so called CpG-SNPs, can influence the expression of target genes by interfering with certain proteins . Moreover, we recently showed that approximately 50% of SNPs associated with T2D are CpG-SNPs, which affect the degree of DNA methylation in the SNP site as well as gene expression and alternative splicing events in human pancreatic islets . It has been hypothesized that since DNA methylation can affect the regulation of splicing, CpG-SNPs can possibly affect alternative splicing events .
There is an increased risk for obesity and T2D among children with obese and/or diabetic parents [55, 56]. Additionally, rodent studies demonstrate that an altered intrauterine environment gives rise to epigenetic changes, which later in life can predispose the offspring to impaired metabolism and T2D [57–59]. These data suggest that epigenetic modifications contribute to the pathogenesis of T2D. Based on the results from our study, we speculate that early exposure to palmitate may affect the epigenetic patterns of genes which are known to affect the risk of T2D. This may increase the risk of disease later in life. However, we cannot exclude that epigenetic changes seen in patients with T2D are secondary to the disease [4, 5, 48, 60, 61].
Our human insulin secretion data are in concordance with previous rodent studies, where palmitate treatment was found to lower glucose-stimulated insulin secretion in rodent pancreatic islets [17, 18]. A tight coupling of glycolysis to mitochondrial respiration and ATP production is essential for proper beta-cell function and glucose-stimulated insulin secretion. Palmitate treatment of human islets resulted in altered expression of individual metabolic genes as well as of genes in metabolic pathways such as glycolysis/gluconeogenesis, pyruvate metabolism and biosynthesis of unsaturated fatty acids. Additionally, several down-regulated genes in the enriched metabolic pathways encode proteins which are part of the respiratory chain, for example, NDUFA4, NDUFB5, NDUFS1, NDUFS2, SDHA and UQCRB. Decreased expression of these genes may contribute to decreased oxidative phosphorylation and subsequently decreased ATP production and insulin secretion in islets exposed to lipotoxicity. Indeed, our previous study showed that decreased expression of genes involved in oxidative phosphorylation results in impaired insulin secretion .
While some studies have found decreased beta-cell number in T2D islets, others do not find an altered cell composition in diabetic islets [10, 63–65]. In the present study, palmitate had no significant effect on apoptosis in human islets and it is hence unlikely that the beta-cell number is significantly decreased. As the majority of cell types in human islets have important effects on whole body glucose homeostasis , it is physiologically warranted to study both whole human islets and cell-lines representing the individual cell types in the pancreatic islets.
Additionally, the insulin signaling pathway was significantly enriched when performing a pathway analysis on all the significant expression data, including both up- and down-regulated genes. Interestingly, this pathway was also enriched when performing a pathway analysis on the differentially methylated genes. Previous studies have shown that insulin signaling contributes in the regulation of beta-cell mass and apoptosis as well as insulin synthesis and secretion  and here we show that this pathway is affected by palmitate treatment in human islets. This in turn could potentially affect insulin secretion in these islets. PPARGC1A (encoding PGC1α) is a part of the insulin signaling pathway and its expression was reduced in human islets exposed to palmitate. We have previously shown that PPARGC1A expression is decreased in islets from T2D patients compared to non-diabetics, and PPARGC1A expression correlated positively with insulin secretion in human islets . PPARGC1A encodes a transcriptional co-activator of mitochondrial genes involved in oxidative phosphorylation and silencing of PPARGC1A in human islets results in decreased insulin secretion . Furthermore, SCD (encoding for stearoyl-CoA desaturase (delta-9-desaturase)) was up-regulated in human islets due to palmitate treatment. SCD is a component of the biosynthesis of unsaturated fatty acids pathway, which was enriched in the KEGG pathway analysis. Stearoyl-CoA desaturase catalyzes the conversion of saturated fatty acids to unsaturated fatty acids, and it has been shown to protect rodent and human beta-cells from palmitate-induced ER stress and apoptosis [67, 68]. Our result is in accordance with these previous studies and could provide an explanation for the absence of an increase in apoptosis in the palmitate-treated human islets.
Furthermore, the “one carbon pool by folate” pathway was enriched in the KEGG pathway analysis using both mRNA expression data and DNA methylation data. Altered expression of genes in this pathway may affect the amount of methyl donors, for example, S-adenosyl methionine in the islets exposed to palmitate and, thereby, contribute to differential DNA methylation. SHMT2 and MTHFD2 were both up-regulated due to palmitate exposure. The enzymes encoded by these genes are involved in the folate cycle which is linked to the methionine cycle, which in turn controls the amount of S-adenosyl methionine .
Importantly, our study demonstrates that palmitate directly affects the expression of genes that also show differential expression in islets from diabetic donors . Additionally, some of our in vitro findings were validated in a cohort of islets from donors with a large spread in BMI (17.6 to 40.1 kg/m2) suggesting that lipid-induced changes seen in vitro correspond to those in vivo. While some previous studies have examined the impact of lipotoxicity on the expression of a limited number of candidate genes in human islets in vitro[24–26], the present study is to our knowledge the first to perform a genome-wide analysis of gene expression in lipotoxic-treated human islets of more than five donors [28, 29].
It is debated whether lipotoxicity can occur in the absence of high glucose levels, a phenomenon known as glucolipotoxicity. However, previous in vivo studies in humans have shown that prolonged exposure (24 to 48 h) to free fatty acids, in the absence of elevated glucose levels, perturbs islet function . Moreover, a recent study showed that the lipotoxic effect of palmitate occurs even at low concentrations of glucose in intact human islets . Our findings provide further evidence that palmitate-induced lipotoxicity under normal glucose conditions results in extensive transcriptional changes and impaired insulin secretion in human islets. It is important, however, to note that our study only examined the effects of palmitate on human islets, and it is known that different fatty acids can have divergent, and even opposite, effects on cell function. Also, the in vivo fatty acid composition in plasma contains several different fatty acids  where palmitate is one of the most abundant saturated fatty acids. We can therefore in our study not rule out that other types of fatty acids have additional effects on human islets. However, our study provides evidence for palmitate-induced changes on gene expression, DNA methylation and insulin secretion which might be of relevance to phenotypes seen in obese individuals and T2D patients. Finally, as our previous studies have shown that the genome-wide methods used in the present study are robust and reproducible, we did not technically validate the array results in the present study [8, 72–74].