In this issue, Cnop et al. (1) report on using RNA-sequencing methodology and the response of the whole transcriptome of human islets to 48-h exposure to saturated free fatty acid (FFA) palmitate. They demonstrated that palmitate altered the expression of 1,325 genes and shifted alternate splicing of 3,525 transcripts. This follows on from a similar study by the same group on the effects on human islets of 48-h exposure to the proinflammatory cytokines interleukin-1β and interferon-γ in which 3,065 (16%) of transcripts were modified and, again, alternate splicing of transcripts was commonly seen (2). However, islet β-cell failure causing diabetes, particularly type 2 diabetes, develops over years and not 48 h. So how do these technically remarkable in vitro experiments on human islets help us to understand islet β-cell failure?
Islet β-cell failure causing diabetes is due to impaired β-cell function and/or loss of β-cell mass. The pathways to islet β-cell failure, however, are very heterogeneous, requiring interaction between extrinsic stressors, such as cytokines or nutrient oversupply, and genetic or acquired islet susceptibility factors (3,4) (Fig. 1). In some circumstances, islet susceptibility may predominant such that extrinsic stressors are less important (e.g., some forms of monogenic diabetes ); while at the other end of the spectrum, severe extrinsic insults may be enough to cause failure of quite robust β-cells (e.g., aggressive autoimmune attack of type 1 diabetes or as a consequence of morbid obesity) (Fig. 1). Very often though, it will be the interaction of extrinsic and intrinsic factors that will result in the β-cell failure. Of note, the separation of type 1 from type 2 diabetes is becoming more blurred with overweight and insulin resistance possibly underlying some of the increasing prevalence of type 1 diabetes (4,6). Complex interactions among various external islet stressor factors and islet susceptibility factors are likely to underlie much of the heterogeneity in the diabetes phenotypes that we see (Fig. 1).
The above-mentioned in vitro studies are really looking at the effects of severe insults from the outside, that of toxic saturated FFA or cytokines, on normal human islets (Fig. 1). They are not designed to investigate how various islet susceptibility factors interact with these extrinsic stresses. The experimental conditions are very different from those that occur in vivo. For example, palmitate is not the only FFA present in the circulation in vivo, but occurs within a mix of other FFAs. Importantly, monounsaturates, such as oleate, are well known to attenuate palmitate toxicity (7). The experimental conditions used, however, will result in islet damage, allowing analysis of the mechanisms involved, many of which will be important in islet β-cell failure in diabetes. The RNA-sequencing method of analysis used here confirms potential mechanisms already suspected, but also is a powerful tool of discovery for previously unsuspected processes by which islets may fail, with the new hypotheses generated requiring confirmation in other systems.
Of note, this RNA-sequencing methodology identified the expression of 15,200 and 18,463 genes in human islets in the first and second of these studies, respectively (1,2). This has increased, by more than twofold, the number of genes known to be expressed in human β-cells. Importantly, 25/41 of type 1 candidate genes and 59/69 type 2 candidate genes were found expressed in human islets, consistent with genetic factors being involved in islet susceptibility. It also has become apparent how prone the human islet transcriptome is to alternative splicing in response to extrinsic stresses. Both cytokines and palmitate altered the expression of splicing factors, but the pattern of alternate splice variants in response to the two stresses was very different. These studies show, however, that altered transcript splicing is likely to be an important mechanism by which extrinsic stresses alter β-cell function and survival.
There was surprisingly little overlap between the gene expression modifications induced by cytokines compared with palmitate, with only 10% of those upregulated and 19% of those downregulated by cytokines being modified by the FFA (1). The cytokines in particular affected genes of the inflammatory system, including the innate immune response, as well as the genes involved in apoptosis (2). The most upregulated genes were those of cytokines and chemokines (2). The authors comment on this being evidence of dialogue between the islet β-cells and the invading immune cells in type 1 diabetes pathogenesis, with the β-cells being actively involved in the destructive inflammation rather than being passive victims (2).
The palmitate human islet experiment elegantly strengthens and builds on much of what is already known about how saturated fatty acids impair islet function and damage them (1). Genes that define the islet β-cell phenotype are inhibited by palmitate, such as the β-cell transcription factors PDX1, PAX4, MAFA, MAFB, and NEUROD1 and genes of mitochondrial metabolism involved in ATP production, the latter being so pivotal to the nutrient-secretion coupling in the β-cell (1). These changes, as well as the observed modifications in the expression of hormone receptors and ion channels, are reflective of islet dedifferentiation as one overriding mechanism by which elevated FFAs cause islet dysfunction. Of particular interest was the finding that palmitate reduced the expression of the transcription factor GATA6. GATA6 mutations are known to cause pancreatic agenesis and neonatal diabetes and can contribute to adult-onset diabetes (8). In additional experiments, GATA6 downregulation by small interfering RNA increased the susceptibility of islet β-cells to palmitate-induced apoptosis (1).
Palmitate upregulated genes related to fatty acid metabolism, much of which may be an adaptive response to limit their toxicity (1). Its effects to induce the unfolded protein response also were clearly evident, confirming previous studies implicating saturated fatty acids in causing β-cell endoplasmic reticulum stress, which can be both adaptive or harmful depending on its degree and type of activation (1,9–11). Palmitate also altered the expression of several cytokines, chemokines, and apoptosis-related genes, showing how lipotoxic and inflammatory pathways of islet injury may interact (1).
Dysfunction of protein trafficking, the ubiquitin-proteasome pathway, and autophagy have recently been considered to be important components of how lipotoxicity causes β-cell failure in type 2 diabetes (12–14). The proteasome pathway and autophagy have roles in the safe disposal of abnormal proteins and damaged organelles. Both were shown to be inhibited by palmitate in the human islet transcriptome experiment (1).
Interestingly, the expression of thioredoxin-interacting protein (TXNIP), which is well known to be upregulated by hyperglycemia and to be a contributor to β-cell apoptosis in hyperglycemic conditions, was downregulated by palmitate in the isolated human islets (1,15). This highlights the point that glucotoxicity, lipotoxicity, and their interaction—glucolipotoxicity—will affect islet function and survival in different ways (11,15). RNA-sequencing experiments in human islets exposed to 48 h of hyperglycemia and the combination of hyperglycemia and elevated fatty acids also would be informative.
These in vitro islet experiments have looked at the transcriptome at just one point in time, 48 h after exposure to stressors (1,2). They miss very acute and late changes that occur in the in vivo setting. Of importance though is their relevance to human diabetes. The authors compared the gene changes induced in vitro by palmitate with those previously reported in islets isolated from donor subjects with diabetes and/or HbA1c levels of ≥6% (16). Islets from donor subjects with HbA1c <6% were used as control subjects (16). They found 7–16% of the islet genes with altered expression from the type 2 diabetic and hyperglycemic subjects were also altered in normal islets exposed in vitro to palmitate (1,16).
See accompanying article, p. 1978.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.