Expression Profiling of Palmitate- and Oleate-Regulated Genes Provides Novel Insights Into the Effects of Chronic Lipid Exposure on Pancreatic β-Cell Function

Culture media, TRIzol, and the Superscript Choice system were from Gibco BRL (Gaithersburg, MD). Falcon plasticware was obtained from Becton Dickinson (Franklin Lakes, NJ). The molecular biological reagents used (and their sources) were: RNeasy mini kits (Qiagen, Melbourne, Australia), Wizard PCR preps (Promega, Madison, WI), High Yield RNA transcript labeling kits (Enzo Biochem, New York), Faststart DNA Master kit and SYBR-Green II (Roche Diagnostics, Castle Hill, Australia). U74A murine genome microarrays and T2 test arrays were purchased from Affymetrix (Santa Clara, CA). Kits for radioimmunoassay of rat insulin were obtained from Linco Research (St. Louis, MO). Nonesterified FA 1C kits for assaying free FAs were from Wako (Osaka, Japan). Diazoxide, 98% FA-free BSA, palmitate, oleate, and all other biochemicals and specialized reagents were from Sigma Aldrich (St. Louis, MO).

MIN6 cells were passaged in 75-cm² flasks with 20 ml Dulbecco’s modified Eagle’s medium (DMEM) containing 25 mmol/l glucose, 24 mmol/l NaHCO₃, 10 mmol/l HEPES, 10% (vol/vol) FCS, 50 IU/ml penicillin, and 50 μg/ml streptomycin. Cells were seeded at 3 × 10⁵ per well in 0.5 ml DMEM in a 24-well dish for secretory experiments, and they were seeded at 5 × 10⁶ per 25-cm flask for transcript profiling or RT-PCR. At 48 h before the experiment (24 h after seeding), the medium was replaced with DMEM (as above but with 6 mmol/l glucose) and supplemented with either BSA alone or BSA coupled to palmitate or oleate. In some experiments, diazoxide (0.1 mmol/l) was added to maintain cellular insulin contents during culture. For FA coupling, 18.4% BSA was dissolved in DMEM (25 mmol/l glucose) by gentle agitation at room temperature for 3 h. Palmitate or oleate (8 mmol/l) was then added as Na⁺ salts, and the mixture was agitated overnight at 37°C. The pH was then adjusted to 7.4, and then, after sterile filtering, FA concentrations were verified using a commercial kit, and aliquots were stored at −20°C. Similar couplings were made using glucose-free modified Krebs-Ringer bicarbonate (KRB) buffer containing 5 mmol/l NaHCO₃, 1 mmol/l CaCl₂, 0.5% (wt/vol) BSA, and 10 mmol/l HEPES (pH 7.4) instead of DMEM. This procedure generated BSA-coupled FA in a molar ratio of 3:1 (generally, 0.4 mmol/l to 0.92% BSA, final).

Cultured cells were washed once (twice for diazoxide-treated cells) in modified KRB buffer (see above) containing 2.8 mmol/l glucose, and then they were preincubated for a further 30 min in 0.5 ml of the same medium at 37°C. This buffer was then replaced with 0.5 ml of prewarmed KRB containing other additions as indicated, and it was incubated for a further 60 min at 37°C. An aliquot was then removed for analysis of insulin content by radioimmunoassay. The cell monolayers were washed twice in PBS and then extracted for measurement of total insulin content by lysis in 0.5 ml H₂O per well, followed by sonication.

RNA (25–40 μg) was isolated using RNeasy columns, and total and double-stranded cDNA was synthesized using the Superscript Choice system with an high-performance liquid chromatography–purified T7-(dT)24 primer. This was converted to biotin-labeled, double-stranded cRNA using a high yield RNA transcript labeling kit according to the manufacturer’s instructions. After purification on RNeasy columns, the cRNA was fragmented at 94°C for 35 min in 100 mmol/l potassium acetate, 30 mmol/l magnesium acetate, and 40 mmol/l Tris-acetate, pH 8.1. After quality verification using Test2 arrays, the fragmented cRNA was hybridized to oligonucleotide microarrays, according to the Affymetrix instructions, in 1 mol/l NaCl, 20 mmol/l EDTA, 0.01% Tween, 100 mmol/l 2-[N-morphalino]ethanesulfonic acid (MES), pH 6.6, for 16 h at 45°C. Mouse genome U74A version 2 arrays were used, except for the first palmitiate experiment, which was conducted with the original version U74A arrays, now known to contain some faulty probe sets. The arrays were then washed on an Affymetrix fluidics station and stained with streptavidin-phycoerythrin according to the manufacturer’s instructions. Probe sets were visualized on a Hewlett-Packard GeneArray scanner and then quantified for intensity and comparison between experimental groups using Affymetrix Genechip software. This raw data were further analyzed using Phenzomix, a data management tool developed in-house for use with these microarrays. Based on FileMaker Pro 5.3, this tool facilitated cluster formation based on both inputs from the arrays (mRNA intensity, fold change, and present/absent calls) and known functionality. For the current study, we assembled clusters of palmitate- or oleate-regulated genes satisfying the following criteria: 1) genes were called as “changed” by the Genechip software and showed alterations of at least 1.9-fold by lipid treatment, 2) those called “increased” or “marginally increased” were also described as “present” in the experimental group, and 3) those called “decreased” or “marginally decreased” were also “present” in the control group. Gene clusters satisfying these conditions in two independent experiments for each lipid were saved and combined in multiple-comparison statements, using Phenzomix to define the smaller dataset of genes coregulated in both experiments. These genes were classified using Swiss Prot and TrEMBL databases.

Total RNA was prepared using RNeasy or Trizol, and 0.2 μg per reaction was converted to cDNA using a Superscript cDNA kit according to the manufacturer’s instructions. Real-time PCR was undertaken on a LightCycler (Roche) with a commercial kit containing 10 mmol/l MgCl₂ and 1 μl cDNA in a PCR of 45–55 cycles (annealing temperature of 55°C for all primers). Standards for each transcript were prepared using appropriate primers in a conventional PCR and purified using Wizard PCR preps. For experimental analyses, PCR products were quantified fluorometrically using SYBR-Green II. Concentrations were then calculated from standard curves as copies per microliter. These PCR products of known concentration for each primer pair were then used in the corresponding LightCycler PCR to provide standards for these reactions. β-Actin expression was calculated in parallel as a control (this was not altered by lipid pretreatment). The following primers were used (forward and reverse): AAT CCT GTG GCA TCC ATG AAA C and CGC AGC TCA GTA ACA GTC CG (β-actin); TGG CAA GGA AGG TGA CAA GCA C and CGG TCC CAT TTT ATT TCA GAG C (calcyclin); AAC CTG TAT GGG ATT TCG GGG and TCC AGA GGG TCA AAG CAA ACC (fructose 1,6-bisphosphatase [FBPase]); TCC TTT GTC GGC AAT CCC TAC and CCC TCC TTG ACT TTC TCT TCA TTC C (phosphatidic acid phosphohydrolase 2 [PAP2]); and TGG ATG AAC GGG AGC AGA TG and GGT TTT GTT GGA GTC AGC CTT CTC (SNAP25).

Unless otherwise indicated, results are expressed as the means ± SE, with the number of observations in brackets. Statistical significance was determined with Students’ t test.

As a prerequisite to analysis of gene expression, it was first essential to establish that lipid-treated MIN6 cells showed secretory defects similar to those seen in other β-cell models. To preclude potentially confounding influences due to chronically elevated glucose, cells were cultured at a substimulatory (6 mmol/l) glucose concentration during the 48-h period of lipid treatment. As shown in Fig. 1A, prior exposure to either palmitate or oleate dramatically reduced the subsequent secretory response to high glucose as compared with control cells. However, consistent with earlier islet studies (7,10,17), an accompanying loss of cellular insulin content caused by the chronic FA treatment (Fig. 1A, insert) may have contributed to the ablated glucose response. Recalculating the results as a percentage of cell content revealed that basal secretion was actually increased two- or fourfold, respectively, after exposure to oleate or palmitate (Fig. 1B). In addition, the fold stimulation due to 16.7 mmol/l glucose was reduced from 14 times basal in control cells to four- to fivefold in cells pretreated with either lipid. The increased basal secretion and diminished fold increase to high glucose are comparable to numerous studies using lipid-pretreated islets or β-cell lines (7,11,12,17–20). However, our results additionally suggested that oleate and palmitate were not entirely equivalent in their effects. This is more apparent in Fig. 1C, in which palmitate is shown to augment insulin secretion slightly over a range of glucose concentrations. In contrast, oleate pretreatment tended to flatten the glucose dose-dependency by both elevating responses at low glucose and inhibiting those at high glucose.

We next undertook transcript profiling, using microarrays to characterize alterations in β-cell gene expression caused by 48-h exposure to palmitate and oleate. As with the secretory experiments, a FA-to-BSA ratio of 3:1 was used, which was previously shown as half-maximal for alterations in expression of some β-cell genes (12). This corresponds to a circulating concentration of ∼2 mmol/l, which is within the low physiologically elevated range. Of the 12,500 genes or expressed sequence tags represented on the microarray, ∼4,700 were present in the control MIN6 cells. The expression of 40 genes was increased ≥1.9-fold in common by both palmitate and oleate; 66 were augmented solely by palmitate, but only 4 were augmented by oleate and not by palmitate. Of the genes decreased by ≥1.9-fold by lipid pretreatment, only 2 were coregulated by both palmitate and oleate, with 18 responding solely to palmitate and 12 solely to oleate. These altered genes are listed in Table 1 according to function. The changes in expression of four genes not previously known to be regulated by FAs were also examined using RT-PCR (Fig. 2). Comparing these results with those in Table 1, it is apparent that the two methodologies gave highly equivalent values for expression of all of the chosen genes: FBPase2, calcyclin, PAP2, and SNAP25.

The largest group of genes regulated by FAs encoded metabolic enzymes. Of the five genes relating specifically to lipid metabolism, all except fatty acyl elongase are involved in β-oxidation. Some genes implicated in steroid metabolism were also increased. Those grouped under amino acid metabolism encoded either transporters or enzymes controlling polyamine metabolism. Many genes relating to carbohydrate metabolism were also altered, generally in a manner that might be expected to disrupt the tight coupling between glycolysis and oxidative phosphorylation, which is a hallmark of β-cell metabolism (21–23). The biggest change was a decrease in the glucose transporter GLUT2, which was much more pronounced with palmitate than oleate. Also decreased were mitochondrial glycerol 3-phosphate dehydrogenase (mGPDH), which forms part of the glycerol phosphate shuttle, and pyruvate carboxylase (PC), which controls a critical step in the anaplerotic pathway (21–23). Consistent with general belief (22,23), FBPase and lactate dehydrogenase (LDH) were absent in untreated β-cells. However, the FAs induced expression of both muscle (type 2) and liver (type 1) forms of FBPase, and palmitate also upregulated LDH. Increases in FBPase, which reverses an early step of glycolysis, could augment metabolic flux through branch pathways such as the hexosamine biosynthesis pathway (HBP). Indeed, expression of another enzyme in the HBP, glucosamine-phosphate n-acetyl transferase (GPAT), was also increased by palmitate.

There were also alterations in 15 genes controlling transcription; 4 of these genes were decreased by palmitate and are probable transcription factors. One of these, Nkx6.2, is closely related Nkx6.1, a transcription factor specific to differentiated β-cells (24). Conversely, of the 11 transcriptional genes showing increased expression due to palmitate, as many as 7 potentially act as repressors, including ATF3 (25), Id1 (26), mTGIF (27), and EGR-binding protein 1 (28). CCAAT/enhancer binding protein-β (C/EBPβ) can either inhibit or stimulate transcription (29,30). Finally, the upregulated CREM family member is probably the inducible cAMP early repressor (ICER), the sole member documented in β-cells (31). Of these potential repressors, only Id-1 was co-induced by oleate. Both lipids upregulated the transcriptional factors C/EBPδ and c-fos. They also increased expression of two genes that would be expected to stimulate cell growth: the IGF-1 receptor and amphiregulin, a close relative of β-cellulin, which binds to the EGF family of receptors. Oleate and palmitate exerted conflicting effects on genes controlling the cell cycle, with upregulation of p21 and p15 by palmitate expected to favor growth arrest, as opposed to downregulation of p57 by oleate, which would facilitate cell cycle progression.

An unexpected category of regulated genes was that mediating inflammatory responses. Most of these genes were absent in control cells and markedly upregulated by both lipids. Three genes (MCP-1, stromal cell-derived factor 1, and GRO1 oncogene) encode chemokines, whereas pentraxins are cell surface proteins potentially implicated in proinflammatory responses. Lipocalin-2 is a secreted factor promoting leukocyte apoptosis (32). Serum amyloid A3 is a secreted lipoprotein whose induction is indicative of the acute phase response, or activation of the innate immune system (33). Interestingly, other features of this response include upregulation of some antiproteolytic enzymes and cell matrix proteins (33), which also occurred in β-cells exposed to palmitate.

Many genes encoding signal transduction enzymes were also lipid-regulated, including categories involved in protein phosphorylation and G-protein function. The observed upregulation of phospholipase D1 (PLD1) and PAP is of potential interest because these enzymes catalyze the two-step formation of diacylglycerol from precursor phospholipids. This implies increased activity of the diacylglycerol/protein kinase C (PKC) pathway in lipid-treated cells. There was also a (generally) increased expression of genes for Ca²⁺ and phospholipid-binding proteins. One of these, calcyclin, whose expression was most sensitive to palmitate exposure, has been directly implicated in the control of insulin secretion (34,35). SNAP25, an essential component of the exocytotic machinery (36), was also increased in the lipid-treated cells. Increased expression of calcyclin and SNAP25 suggests that the distal exocytotic pathway might be upregulated in cells pretreated with palmitate. Such a possibility would also be consistent with enhancement of genes relating to protein processing and export, such as chaperone proteins, disulphide isomerases, and an endoplasmic reticulum translocon protein. Indeed, there was also a general upregulation of genes for secreted products and hormones. The increased expression of islet hormones, such as pancreatic polypeptide and glucagon, which are not normally found in β-cells, might point to a relative loss of cellular differentiation induced by FA exposure.

To address the prediction that lipid pretreatment would lead to a sensitization of distal secretory pathways, functional responses to a variety of nonnutrient secretagogues were next examined. As shown in Fig. 3, pretreatment with palmitate for 48 h markedly enhanced secretion due to agents acting at the level of activation of PKC (the phorbol ester 12-O-tetradecanoylphorbol-13-acetate [TPA]), the cAMP-dependent protein kinase (forskolin), and Ca²⁺-dependent secretion elicited with a depolarizing concentration of KCl. In cells correspondingly pretreated with oleate, only TPA-induced secretion was augmented to an extent equivalent to that seen with chronic palmitate treatment, although the forskolin response was very slightly sensitized relative to control. Ca²⁺-dependent secretion was not sensitized at all by oleate pretreatment. These observations would be consistent with the finding that whereas oleate and palmitate both elevated PLD1, expression of SNAP25 and calcyclin were more markedly upregulated by palmitate.

Palmitate-pretreated cells appeared to show a much larger enhancement in their subsequent responses to the pharmacological agents (Fig. 3) than was seen with glucose (Fig. 1), possibly because of a negative effect on carbohydrate metabolism, which might partially counteract the upregulation of distal secretory processes. We therefore used palmitate as both an acute and chronic stimulus to determine whether a physiological, but noncarbohydrate, secretagogue was capable of bypassing such a proximal defect. In cells not preexposed to FAs, acute addition of palmitate doubled insulin secretion at low glucose and markedly potentiated the response to 16.7 mmol/l glucose (Fig. 4A). In cells chronically exposed to palmitate, the responses to either high or low glucose alone were slightly increased, consistent with the findings of Fig. 1. In contrast, there was a marked sensitization to palmitate, most obvious at low glucose (nearly 10 times the palmitate response in control cells), but also observed with the combined stimuli of 16.7 mmol/l glucose and palmitate. To exclude that the accompanying depletion of insulin stores might somehow contribute to this sensitization, the studies were repeated using MIN6 cells cultured during the pretreatment period in the presence of diazoxide, a specific inhibitor of insulin secretion. Although under these conditions there was minimal depletion of insulin content due to palmitate-pretreatment (Fig. 4B, insert), a marked sensitization was still observed.

We have made use of the highly differentiated and glucose-responsive cell line MIN6 (16) to define β-cell genes regulated by palmitate and oleate. These cells possess the advantage over primary islets of being a homogeneous cellular population and have been previously used in a similar approach to define genes sensitive to glucose (15). The suitability of MIN6 cells for these studies is underscored by our findings that when cultured in the presence of elevated FAs, they display secretory alterations broadly similar to those observed using isolated islets (7,10,16,19) and other β-cell lines (11,12,18,20). However, our detailed analyses point to additional secretory changes that have not been previously described, probably because of difficulties in obtaining sufficient pancreatic islets to conduct comparable studies and because of the poor glucose-responsiveness of many other β-cell lines. Thus, we have shown that oleate and palmitate, though both enhancing basal secretion, are far from equivalent in other respects. Oleate pretreatment inhibited secretion due to maximal glucose concentrations and, with the exception of TPA, barely sensitized the cells to stimuli acting on distal secretory processes. In contrast, palmitate-treated cells were sensitized to all secretagogues tested, including glucose and palmitate, as well as the pharmacological activators. However, the sensitization to glucose was less apparent than that seen with other secretagogues. This is potentially explained by an additional effect of lipid pretreatment to disrupt the coupling between glucose metabolism and proximal signaling pathways, which might partially counter the upregulation of distal secretory processes. In the case of oleate, a similar disruption of glucose signaling might be more obvious because upregulation of the distal pathways is less apparent.

Should this modest augmentation of glucose-induced insulin secretion in palmitate-treated cells be viewed as a sensitization or, alternatively, as a desensitization if the maximal response, is calculated as the diminished fold increase over an elevated baseline? In our view, this question is no longer especially relevant, since it is based on a rather artificial protocol in which β-cells are pretreated with FAs and then challenged with glucose in the absence of FAs. In contrast, our studies suggest that under physiological conditions of constant exposure to FAs, the degree of resultant sensitization to the combined stimuli would far outweigh any small effect seen on the response to glucose alone. Such a conclusion would be consistent with the findings of one earlier study, in which the effect of a combined challenge of glucose plus FA was enhanced in lipid-pretreated islets (17). However, the experimental design of that study did not allow for a full appreciation of the multiple influences at work. On the other hand, oleate-pretreated INS-1 cells did not show an enhanced acute response to a subsequent challenge with oleate (20), but this might represent a difference in the potency of palmitate versus oleate in the sensitization process. Moreover, our findings are supported by in vivo evidence that secretory responses to nonglucose stimuli were upregulated in fat-fed mice and in several animal models of diabetes (37–39). However, it is uncertain whether these enhanced responses, and hence the palmitate-induced sensitization described here, are ultimately beneficial in terms of compensatory insulin secretion, or whether they are implicated in a pathophysiological overstimulation of the β-cell, and hence progression to diabetes. In the latter context, it needs to be stressed that there was a depletion of intracellular insulin content in MIN6 cells pretreated with either oleate or palmitate. This is consistent with previous studies using islets, and the knowledge that lipids are unable to stimulate transcriptional or translational pathways for insulin synthesis to compensate for chronically stimulated secretion (7,10,17).

The further utility of the MIN6 cell model for transcript profiling is witnessed by the fact that four genes, GLUT2, carnitine palmitoyl transferase 1 (CPT1), c-fos, and PC, previously known to be lipid-regulated in β-cells, were also detected in our screen (10,11,13,40). Similarly, expression of the β-cell–specific transcription factor pancreatic duodenal homeobox-1 (PDX-1), known to be downregulated in islets by palmitate treatment (10), was also inhibited in the MIN6 cells by 1.4 ± 0.1-fold. Although this decrease was too small to be included in Table 1, it is likely to be functionally significant. Many other genes known to be altered during β-cell dysfunction were also shown here to be regulated by palmitate (see below). However, we failed to detect changes in some other β-cell genes previously reported to be lipid-regulated, including: acetyl CoA carboxylase (12) and nur-77 (13) (not present on the microarrays), glucokinase (10), inducible NO synthase (9), and uncoupling protein 2 (14), as well as insulin (10). The latter was expressed so highly in MIN6 cells that it saturated even the “mismatch” probe sets on the array and was therefore paradoxically called as absent. Glucokinase is downregulated by only 25% by palmitate in islets (15), which was probably too a small a difference to be detected here. Uncoupling protein 2 was well expressed in MIN6 cells, but its levels were unaltered by lipid pretreatment (not shown), in contrast to a previous study using INS-1 cells (14). The discrepancy is possibly due to differences in treatment times (18 vs. 48 h here) and the fact that we used a much lower effective FA concentration (molar FA-to-BSA ratio of 3:1, in contrast to 6:1 in the previous study). We were also unable to demonstrate upregulation of NO synthase by lipid, but this has not been previously shown using homogenous β-cells. Macrophages present in the islet might be important for induction of this gene in situ.

Our results provide novel insights into the effects of chronic FA exposure on β-cell function in terms of the distal sensitization (described here) as well as the enhancement of basal secretion and disruption of glucose-dependent stimulus-secretion coupling that have been previously reported. The increased expression of SNAP25, calcyclin, and PLD in palmitate-treated cells is consistent with the observed sensitization to distally acting stimuli, especially those activating Ca²⁺-dependent secretion. SNAP25 is essential for targeting insulin secretory granules to the sites of exocytosis, and inhibition of its function blocks insulin secretion (36). Similarly, calcyclin is also required for Ca²⁺-dependent insulin secretion, and, crucially, its overexpression is sufficient for upregulation of the secretory response (34,35). Enhanced PLD signaling should also be stimulatory, either by activation of PKC or at the level of vesicular trafficking. Most importantly, the relative capacities of palmitate versus oleate to alter these genes is consistent with their observed effects on distal secretory pathways. Thus, oleate pretreatment only sensitized the MIN6 cells to PKC-dependent secretion, in keeping with the accompanying increases in PLD1 expression, relatively modest effects on calcyclin, and no effect at all on SNAP25. However, causative relationships between these alterations in gene expression and secretory function will need to be substantiated in future experiments.

Upregulation of SNAP25, calcyclin, and PLD1 might also contribute to the enhancement of basal secretion that accompanies lipid pretreatment of β-cells. However, the observed decrease in the expression of short chain L-3-hydroxyacyl CoA dehydrogenase (SCHAD) is also of particular interest because an inactivating mutation in that enzyme has very recently been causally associated with basal insulin hypersecretion (41). On the other hand, downregulation of GLUT2, PC, and mGPDH might be expected to contribute specifically to the decoupling of glucose metabolism from proximal signaling pathways (21–23). Moreover, the unexpected induction of two forms of FBPase, by both palmitate and oleate, might be particularly relevant to the loss of glucose sensitivity because these enzymes could divert glycolytic flux into alternative metabolic pathways. Indeed, upregulation of one of these branch pathways, the HBP, is sufficient to inhibit glucose-stimulated insulin release (42). Other evidence implicates the HBP in the β-cell apoptosis that accompanies prolonged exposure to high glucose concentrations (43). It is noteworthy that another enzyme in the HBP pathway, GPAT, was also upregulated in palmitate-pretreated MIN6 cells. Interpreting these earlier observations in the light of our new findings, it is conceivable that augmented flux through the HBP could contribute both to the secretory defects and the apoptotic response that accompany chronic β-cell exposure to lipids.

Importantly, our results also demonstrate that palmitate pretreatment is sufficient to alter β-cell gene expression in a similar fashion to that occurring in models of type 2 diabetes. The coregulated genes include GLUT2 (3,44), mGPDH (44,45), LDH (44), PC (45), SNAP25 (46,47), p21 (48), ATF3 (25), ICER (49), and C/EBPβ (50). The fact that highly abundant β-cell genes like GLUT2, PC, and mGPDH were downregulated after lipid treatment, whereas poorly expressed ones like LDH and FBPase (as well as those for hormones other than insulin) were induced, is suggestive of a loss of β-cell differentiation. Upregulation of C/EBPβ might be important in this context (50,51) because this also occurs with supraphysiological glucose levels in vitro and is causally associated with decreased expression of PDX-1 (30). Indeed, a primary role for C/EBPβ in de-differentiating β-cells has been postulated as an underlying cause of type 2 diabetes (51). Our results would be consistent with this hypothesis, but they suggest that other transcriptional regulators are also involved. ATF3 is induced in β-cells as a result of oxidative stress, and its ectopic overexpression leads to abnormal islet development (25). Likewise, induction of ICER correlates with transcriptional repression of the insulin gene in vitro (31), and expression of ICER is increased in the islets of diabetic GK rats relative to nondiabetic controls (49).

Another major finding was the upregulation by both lipids of proinflammatory genes encoding chemokines and mediators of the acute phase response. Although the latter is known to be triggered by cellular stress, its induction in β-cells by the physiologically relevant FA concentrations used here is surprising. On the other hand, type 2 diabetes is associated with increases in circulating markers of the acute phase response, and a causative role of a hypersensitive innate immune system in the progression of the disease has been hypothesized (52). According to this hypothesis, however, β-cells are passive victims of this response and not, as our results might now suggest, themselves playing a more active role. Whether this, along with chemokine release, could form part of self-destructive loop will obviously require further investigation. However, the fact that binding elements for C/EBPβ are found in the promoters of many acute-phase genes (29) suggests, at the very least, that β-cell dysfunction and the acute phase response share a common etiology.

In conclusion, we have demonstrated that palmitate and oleate induce profound, but differing, alterations in β-cell gene expression. Our results suggest novel explanations, such as induction of FBPase and upregulation of the HBP, for the secretory defects that also occur in lipid-treated islets. Moreover, palmitate pretreatment was sufficient to reproduce many of the alterations in gene expression previously documented in models of β-cell dysfunction. This is in marked contrast to those genes regulated by glucose, as assessed in a comparable study (25). However, palmitate pretreatment also increased expression of genes controlling distal secretory processes and was shown functionally to sensitize β-cells to secretory stimuli. Our study highlights the contribution of regulated gene expression to the phenotypic alterations induced in β-cells after FA exposure, but our results additionally demonstrate that these phenotypic alterations are multiple and complex. The balance between them is likely to be influenced by genetic factors. Allelic variation in some of the lipid-regulated genes that we have documented here might therefore predispose to type 2 diabetes.

This work was supported by the National Health and Medical Research Council of Australia and a “High Risk” grant from the Garvan Institute.

We thank Andrew Sutherland for help with the array hybridization and Dr. Naras Lapsys and Megan Lim-Fraser for technical advice. We also thank Drs. Christopher Ormandy, Lee Carpenter, and Christopher Mitchell for critically reading the manuscript. The MIN6 cells were a generous gift from Dr. Jun-ichi Miyazaki.