FoxO1 Deacetylation Decreases Fatty Acid Oxidation in β-Cells and Sustains Insulin Secretion in Diabetes

Abstract

Pancreatic β-cell dysfunction contributes to onset and progression of type 2 diabetes. In this state β-cells become metabolically inflexible, losing the ability to select between carbohydrates and lipids as substrates for mitochondrial oxidation. These changes lead to β-cell dedifferentiation. We have proposed that FoxO proteins are activated through deacetylation-dependent nuclear translocation to forestall the progression of these abnormalities. However, how deacetylated FoxO exert their actions remains unclear. To address this question, we analyzed islet function in mice homozygous for knock-in alleles encoding deacetylated FoxO1 (6KR). Islets expressing 6KR mutant FoxO1 have enhanced insulin secretion in vivo and ex vivo and decreased fatty acid oxidation ex vivo. Remarkably, the gene expression signature associated with FoxO1 deacetylation differs from wild type by only ∼2% of the >4000 genes regulated in response to re-feeding. But this narrow swath includes key genes required for β-cell identity, lipid metabolism, and mitochondrial fatty acid and solute transport. The data support the notion that deacetylated FoxO1 protects β-cell function by limiting mitochondrial lipid utilization and raise the possibility that inhibition of fatty acid oxidation in β-cells is beneficial to diabetes treatment.

Introduction

Defects in pancreatic β-cell function and mass are a critical factor in the onset of type 2 diabetes (1). They include a variety of cellular pathologies that culminate in the three main pathophysiological abnormalities that characterize diabetic endocrine pancreatic function: decreased insulin release in response to glucose, excessive glucagon production, and reduced number of insulin-producing β-cells (2, 3). Although the contribution of each abnormality to the impairment of endocrine pancreatic function is likely heterogeneous in different individuals and variable according to stage of disease and type of treatment, a strong case can be made that reversal of these abnormalities is necessary to move from the current therapeutic approaches to ones that can be considered disease-modifying (4).

We have proposed a theory of β-cell dysfunction that attempts to reconcile diverse etiologies under a single mechanism and incorporates the role of insulin resistance in the process (4). We have shown that FoxO, a family of transcription factors with important roles in peripheral insulin signaling (5), also integrate various aspects of β-cell biology. In the developing pancreas, FoxO1 promotes β-cell maturation, i.e. the acquisition of proper nutrient-sensing abilities (6). In the adult pancreas, FoxO1 is activated through transient elevations of glucose or lipids (7–9) to modulate the metabolic flexibility of β-cells (10). A key event in regulating FoxO activity in β-cells is the acetylation of specific lysine residues. When fully acetylated, FoxO1 is primarily cytoplasmic, whereas when fully deacetylated it lingers in the nucleus and translocates to the cytoplasm more slowly in response to hormones (7). Deacetylation has a complex effect on FoxO1 function: on the one hand, it increases its transcriptional activity, but on the other it accelerates its degradation. The net outcome is that metabolic conditions associated with increased FoxO1 deacetylation, such as a persistent hyperglycemic state ultimately decrease FoxO1 (7), paving the way for β-cell dedifferentiation (11).

We previously showed that mice homozygous for a mutant Foxo1 allele encoding a constitutively deacetylated protein (6KR) develop metabolic abnormalities that can be accounted for by increased nuclear residence of the mutant protein, leading to mild hyperglycemia (12) and increased atherosclerosis (13). Despite the FoxOs central role in the progression of β-cell dysfunction, the effect of FoxO1 deacetylation on β-cell function has not been explored in detail and remains somewhat conjectural. Thus, in this work we sought to test the hypothesis that FoxO1 deacetylation benefits β-cell function and to identify the underlying mechanism. To this end, we investigated in vivo and ex vivo properties of β-cells bearing the 6KR FoxO1 mutant.

Experimental Procedures

Animals

We have described generation, characterization, and genotyping protocols for this mouse strain in a previous publication (12). Unlike in the original publication, mice used for these studies were maintained on a C57BL/6 background. All mice were fed normal chow and maintained on a 12-h light/dark cycle (lights on at 7 a.m.). All experiments were performed in 12–20-week-old male mice unless specified otherwise in figure legends (10). Animals were fasted from 8 to 48 h and re-fed for 4 h before killing. Animals were always killed at the same time of day to avoid the effects of the circadian clock on protein acetylation. The Columbia University Institutional Animal Care and Utilization Committee approved all experiments.

RNA Profiling

We used GeneChip Mouse Exon arrays (Affymetrix) and performed data analysis with Partek Genomics Suite (Partek, Inc.) and Ingenuity Pathway Analysis (Ingenuity System, Inc.). We present four types of Ingenuity analyses: pathways, regulators, diseases, and cellular functions. For each analysis, we report the top hits by p value. For the individual gene analyses shown in this study, we used a threshold of an adjusted p < 0.05 and >1.3-fold change to declare significance. We used n = 4 WT, 3 KO per group. Each array was performed with pooled islets from three mice per genotype.

Metabolic Analyses

We performed glucose tolerance tests in overnight-fasted mice by intraperitoneal injection of glucose (2 g/kg) (14). We performed glucose-stimulated and arginine-induced insulin secretion tests as described (15). We measured glucagon by radioimmunoassay and insulin by ELISA (Millipore).

Immunoblot, Immunohistochemistry, and Morphometry

Western blots used the primary antibodies directed against the following proteins: Foxo1, Foxo3, Foxo4, α-tubulin from Cell Signaling (#2880, 2497, 9472, and 2144). We fixed and processed tissues for immunohistochemistry as described (16). We applied perfused fixation and antigen retrieval for nuclear transcription factor detection (Nacalai USA) (11). We used anti-Foxo1 (Cell Signaling #2880) and anti-insulin and anti-glucagon from Dako (#A056401-2, A056501-2) for primary antibodies and FITC-, Cy3-, and Alexa-conjugated donkey as secondary antibodies (Jackson ImmunoResearch Laboratories, and Molecular Probes) (16). Antibodies to Pcsk1, Pcsk2, Glut-2, and Pdx-1 were from EMD Millipore (#AB10553, AB15610, 07-1402, 061384). For β-cell morphometry, 4 pancreatic sections from 4 mice of each genotype were sampled 150 μm apart and used for a two-tailed paired Student's t test analysis. Total pancreas area was captured at 100×, and immunostaining area captured at 200× (11). The ratio of immunoreactive area/pancreas area was converted from pixels to mm² and plotted as mm².

Mitochondrial Function

We used the XF24-3 respirometer (Seahorse Bioscience) with 24-well plates designed to confine islets to measure oxygen consumption rates. We used the F₁F₀ ATP synthase inhibitor oligomycin to assess uncoupling, FCCP² to estimate maximum respiration, and rotenone to measure non-mitochondrial respiration (10).

Glucose and Lipid Metabolism

After overnight recovery and 2-h starvation, 20 islets were preincubated in KRBH for 45 min at 2.8 mm glucose followed by incubation in KRBH containing 0.5 μCi of d-[5-³H]glucose (16 Ci/mmol) and 1μCi/ml d-[U-¹⁴C]glucose (250 mCi/mmol) at 2.8, 8.3, and 16.7 mm glucose. Incubation was terminated by the addition of citrate/NaOH buffer (400 mm, pH 4.9) containing antimycin-A (10 m), rotenone (10 m), and KCN (5 mm). Glucose oxidation was measured by the generation of KO-trapped ¹⁴CO₂ after 60 min, and glucose utilization was determined by measuring ³H₂O (17). Lipid oxidation was measured as described with minor modifications. Briefly, islets were cultured overnight in RPMI 1640 and then incubated in RPMI 1640 with 10% FBS and 2 mm glucose for 2 h. Thereafter, batches of 50 islets were preincubated in KRBH with 2.8 mm glucose for 45 min and transferred to KRBH containing 0.25% BSA, 0.1 mm palmitate, and 0.2 μCi/ml [9,10(n)-³H]palmitate (74 kBq/ml) at 2.8 and 16.7 mm glucose for 2 h. The supernatant was collected to separate ³H₂O from radioactive fatty acids, and fatty acid oxidation was calculated by measuring ³H₂O produced (18).

RNA Measurements

We used standard techniques for mRNA isolation and quantitative PCR. PCR primer sequences have been published (11). In experiments with purified islets, samples were incubated in either 5 mm or 20 mm glucose for 24 h before mRNA measurements. In some instances islets were transduced with adenovirus expressing GFP, FoxO1ADA, or FoxO16KR for 24 h (7) and incubated in 11 mm glucose before isolating mRNA.

Statistical Analyses

Sample sizes were estimated from the expected effect size based on previous experiments. No randomization or blinding was used. We present data as the means ± S.E. We used two-tailed Student's t test, one-way ANOVA or two-way ANOVA for data analysis and the customary threshold of p < 0.05 to declare statistically significant differences.

Results

Metabolic Effects of Constitutive FoxO1 Deacetylation

We analyzed FoxO1 acetylation in islets from fasted or re-fed mice and found that, similar to other tissues (12), FoxO1 acetylation peaks 4 h after feeding and reaches its nadir after a prolonged, 48-h fast (Fig. 1A). We confirmed that 6KR mice have mild fasting and fed hyperglycemia, associated with increased plasma insulin levels (Fig. 1, B and C) (12). The latter likely results from increased insulin secretion, as indicated by a commensurate increase in C-peptide levels (Fig. 1D). Fasting glucagon levels, on the other hand, did not change (Fig. 1E). We measured the acute insulin response to glucose and found a brisker response in 6KR mice, consistent with the possibility that β-cells are hyperresponsive to glucose (Fig. 1F). The relatively modest response in WT controls can be attributed to the C57Bl/6 background of the animals (19).

FIGURE 1.

Metabolic features of FoxO1–6KR mice. A, FoxO1 deacetylation detected by immunoblotting with anti-acetyl-FoxO1 antibodies in islets purified from mice fasted or re-fed for the indicated periods of time. B, blood glucose levels in fasted and re-fed 12-week-old male mice (n = >6 for each genotype). Glucose levels in 12-week-old male mice (n = >6 for each genotype). p < 0.05 by ANOVA. *, p < 0.05, ***, p < 0.001. C–E, insulin (B), c-peptide 2 (C), and glucagon levels (D) in the same mice. p < 0.05 by ANOVA. F, insulin release in response to an intraperitoneal glucose challenge (n = >6 for each genotype). p < 0.05 by ANOVA.

Because 6KR mice have complex metabolic abnormalities (20), we cannot conclude from these experiments that the increase in insulin secretion results from intrinsic effects of the mutant FoxO1 in β-cells, as opposed to representing a response to insulin resistance. To answer this question, we isolated islets and measured their response to glucose. Consistent with the in vivo data, we found a near doubling of insulin secretion at 11.2 mm glucose and a trend toward increased release at 16.8 mm glucose (Fig. 2A). The changes in insulin secretion cannot be explained by changes in β- or α-cell mass, as both were comparable to WT mice (1.02 ± 0.09 versus 1.14 ± 0.08 mg for β-cell mass and 0.09 ± 0.03 versus 0.12 ± 0.02 mg for α-cell mass in WT and KR, respectively) as was islet insulin content (16.2 ± 1.14 versus 17.0 ± 0.77 arbitrary units in WT and KR, respectively).

FIGURE 2.

Insulin secretion, glucose, and lipid metabolism in isolated islets. A, insulin release from isolated islets. B, [³H]glucose oxidation. C, [¹⁴C]glucose utilization. D, palmitate oxidation (n = 3 per group). Each experiment was performed with pooled islets from five mice per genotype. Each data point consists of six-nine replicates. Glc, glucose. All data are presented as means ± S.E. **, p < 0.01 by Student's t test.

RNA Sequencing Analysis

To understand how the gain of FoxO1 function affects the islet response to physiologic changes in nutrient availability, we analyzed transcriptomes from islets obtained after animals had been fasted for 24 h or re-fed for 4 h. We compared islet RNA expression profiles according to feeding state (fasted versus re-fed) and genotype (WT versus 6KR). When analyzed irrespective of genotype, the fasting to re-feeding transition was characterized by changes to 4119 genes with an adjusted p < 0.05. Ingenuity analysis indicated that protein translation and degradation as well as mitochondrial function were key pathways affected by (or affecting) the transition. Key regulators of the process were the Maturity Onset Diabetes of Youth-associated gene HNF4α, Rictor, and Perk (Eif2ak3) (Table 1). Next we analyzed individual genes. We detected the most profound decreases in Trb3, glutamate transporter Slc1a4, Myc, and transcription factors Cebp-β and Akna; in contrast Txnip, Isl1, and Pax6 showed the most robust increases (Table 2). The strong induction of Txnip, the most highly glucose-responsive gene in human islets (21), provides an important quality control for the robustness of the data. ATF transcription factors were coordinately decreased, consistent with decreased mRNA translation initiation after feeding (22). Three gene families were conspicuous by their extensive changes: 9 of the 22 members of the Rho guanine nucleotide exchange factors (Arhgefs) decreased, whereas 29 of the 41 subunits of the mitochondrial NADH dehydrogenase (complex I) and 20 of the 53 members of the ribosomal protein L family (Rpl) increased. These changes are consistent with increased intracellular vesicle trafficking, ATP production, and protein synthesis. Notably, there were coordinate decreases of genes required for lipid synthesis and utilization, including all C/ebp isoforms, Pparγ, and Cpt1α (Table 2).

TABLE 1.

Summary of Ingenuity analysis of transcriptome data

We selected four parameters from the analysis: pathways, regulators, diseases, and cellular functions. For each parameter, we report the top statistically significant categories. The right column shows absolute numbers and percentages of changed genes in each pathway or the activation state of each regulator, as predicted by the Z-score. A Z-score >2 predicts activation; <−2 predicts inhibition.

Fast vs. feed	p	% Changes/predicted state
Pathways
EIF2 signaling	2.15 × 10−17	97/185 (0.524)
Protein ubiquitination pathway	6.24 × 10−16	119/255
Mitochondrial dysfunction	1.66 × 10−13	85/172 (0.494)
Regulation of eIF4 and p70S6K signaling	3.54 × 10−13	75/146 (0.514)
Oxidative phosphorylation	4.14 × 10−11	58/110 (0.527)
Regulators
HNF4A	2.66 × 10−42
RICTOR	3.79 × 10−35	Inhibited
TP53	1.17 × 10−33
CD 437	5.39 × 10−19	Inhibited
EIF2AK3	3.56 × 10−18	Inhibited
Diseases
Cancer	3.47 × 10−33 − 1.84 × 10−4
Infectious disease	1.02 × 10−28 − 4.00 × 10−7
Cell functions
Gene expression	3.08 × 10−44 − 8.58 × 10−5
Cellular growth and proliferation	3.80 × 10−35 − 1.46 × 10−4
Cell death and survival	2.85 × 10−25 − 1.76 × 10−4

WT fast vs. feed	p	% Changes/predicted state
Pathways
Unfolded protein response	3.08 × 10−8	12/54 (0.222)
STAT3 pathway	6.83 × 10−6	11/73 (0.151)
Regulators
EIF2AK3	3 1.85 × 10−25
ATF4	4.51 × 10−22	Inhibited
Tosedostat	2.04 × 10−16	Inhibited
Tunicamycin	3.28 × 10−15	Inhibited
TGFB1	2.92 × 10−13	Inhibited
Diseases
Endocrine system disorders	1.54 × 10−8 − 2.62 × 10−4
Cell functions
Cellular growth and proliferation	1.93 × 10−14 − 4.72 × 10−4
Cell death and survival	8.11 × 10−11 − 5.10 × 10−4

KR fast vs. feed	p	% Changes/predicted state
Pathways
Unfolded protein response	1.92 × 10−6	9/54 (0.167)
Axonal guidance signaling	1.72 × 10−4	22/433 (0.051)
Adipogenesis	4.02 × 10−4	10/127 (0.079)
Regulators
EIF2AK3 3.41 × 10−25	3.41 × 10−25	Inhibited
ATF4 5.93 × 10−21	5.93 × 10−21	Inhibited
Tosedostat	5.39 × 10−15	Inhibited
Tunicamycin	4.49 × 10−14	Inhibited
Diseases
TGFB1	1.09 × 10−12
Cancer	2.27 × 10−9 − 1.34 × 10−3
Gastrointestinal disease	7.72 × 10−8 − 8.57 × 10−4
Cell functions
Cell death and survival	1.79 × 10−10 − 1.34 × 10−3
Cell growth and proliferation	6.76 × 10−9 − 1.15 × 10−3
Cellular development	3.10 × 10−8 − 1.19 × 10−3

Fasted WT vs. KR	p	% Changes/predicted state
Pathways
Hepatic fibrosis/stellate cell activation	2.63 × 10−9	25/197 (0.127)
B cell development	6.75 × 10−7	9/34 (0.265)
Regulators
TNF	4.00 × 10−27
IFNG	1.16 × 10−24
TGFB1	2.29 × 10−24
Lipopolysaccharide	4.08 × 10−20
Diseases
Metabolic disease	2.53 × 10−18 − 1.28 × 10−5
Cell functions
Cell death and survival	1.40 × 10−19 − 3.84 × 10−5

Re-fed KR vs. WT	p	% Changes/predicted state
Pathways
FXR/RXR activation	2.42 × 10−4	11/127 (0.087)
Adipogenesis	3 × 10−3	9/124 (0.073)
Regulators
HNF1A	2.48 × 10−11	Activated
TGFB1	7.26 × 10−11
Tp53	4.47 × 10−9
H2O2	5.24 × 10−9
Tretinoin	5.34 × 10−9
Diseases
Metabolic disease	3.73 × 10−13 − 5.73 × 10−4
Endocrine system disorders	1.01 × 10−12 − 5.73 × 10−4
Cell functions
Cellular movement	9.79 × 10−8 − 6.54 × 10−4
Lipid metabolism	1.01 × 10−7 − 6.50 × 10−4

TABLE 2.

Selected candidate genes from RNA sequencing

Shown is a partial list of differentially expressed genes, as identified in RNA sequencing experiments. The numbers show the mean FPK value for each gene. Red-shaded values indicate genes with greater expression in fasted vs. re-fed or fasted WT vs. fasted 6KR or re-fed KR vs. re-fed WT. Note that the order of WT and KR is reversed in the last set.

When we analyzed the data by genotype, comparing the transition from fasting to re-feeding in WT and 6KR islets, we found that among pathways the unfolded protein response was preserved, as were top regulators. The latter included members of the UPR and Tgf1 signaling pathways, both of which have been implicated in the islet response to glucose (22, 23). In 6KR islets we observed changes in lipid metabolic pathways (axonal guidance signaling and adipogenesis). The endocrine system was among the top disease pathways identified, whereas cellular functions were related to cell growth, death, and survival (Table 1).

Next we compared gene expression in fasted WT versus 6KR islets. Although the top pathways identified by Ingenuity analysis were not related to β-cell-specific functions, top regulators included Tnfα, Tgf1β, and γ-interferon. These data indicate that FoxO1 is likely required for the response to these cytokines, which are not themselves expressed in β-cells but are known to activate inflammatory and apoptotic pathways in this cell type (24). Moreover, metabolic diseases topped the list of disease pathways (Table 1). Of the >4000 genes affected by the fasting/re-feeding transition, only 80 (∼2%) showed significantly different expression between WT and 6KR, but they included increased expression of a remarkable collection of genes important for β-cell function, including Chrebp, MafA, Pdx1, Glut2, G6pc, Gpr119, cannabinoid receptor Cnr1, and folate receptor Folr1. Notably, expression of insulin-1 and -2 increased by 60–80% even though it did not achieve statistical significance, possibly due to the high number of sequences detected. In contrast, the fatty acid receptor Cd36 was decreased, consistent with decreased free fatty acid utilization (Table 2). These data appear to indicate that deacetylation allows FoxO1 to selectively target critical genes for the maintenance of β-cell identity.

In the comparison between re-fed WT versus 6KR islets, Ingenuity analysis specifically identified Fxr/Rxr and adipogenesis as the top differentially modulated pathways (Table 1). We were gratified to see that the top regulator identified in re-fed 6KR islets was the MODY gene Hnf1α, which we have previously shown to be among key FoxO targets required for metabolic flexibility (10). Consistent with the observation that hydrogen peroxide activates FoxO1 nuclear translocation in β-cells (as deacetylation would) (7, 25), this chemical was also among the top regulators in 6KR islets. Forty-one differentially expressed genes met the significance criteria of an adjusted p < 0.05. Among them, the most remarkable was insulin-1, which increased ∼5-fold in 6KR (Table 2). Unlike the fasted state, none of the other β-cell-specific or -enriched transcripts showed significant departures from the means. Because FoxO1 is inactive in the fed state (7), this result should not surprise.

To validate the RNA sequencing analysis, we selected aldolase B (aldoB) as a gene that was strongly induced by re-feeding in both genotypes. In liver, AldoB transcription is induced by carbohydrates, possibly through Chrebp (26). We isolated islets and performed batch incubations at 5 and 20 mm glucose before measuring AldoB mRNA. We found a 25-fold induction in WT islets and a nearly 100-fold induction in 6KR islets (Fig. 3A). To determine whether this was due to gain of function of FoxO1, we measured AldoB levels in WT islets incubated in 11 mm glucose after transduction with two FoxO1 mutants, ADA and 6KR, both of which confer gain-of-function (7). In both instances, FoxO1 overexpression was sufficient to induce AldoB in the absence of additional contributions from glucose (Fig. 3B). These data confirm that 6KR acts as a gain-of-function mutant in β-cells and validate the RNA sequencing findings.

FIGURE 3.

Validation of Aldolase B as a differentially regulated gene. A, qPCR analysis of mRNA encoding AldoB at the indicated concentrations of glucose. Each experiment was performed with pooled islets from three mice. Each data point consists of three replicates. Data are expressed as -fold-increase over the values at 5 mm glucose. Pancreatic islets were isolated from 10-week-old control and 6KR mice. B, AldoA and AldoB levels after transduction of WT islets with adenovirus encoding GFP, phosphorylation-defective (ADA), or deacetylated FoxO1 (6KR). Islets were incubated at 11 mm glucose in RPMI. Data are presented as means ± S.E. (n = 9–12 per genotype). ***, p < 0.001 by one-factor ANOVA.

Decreased Lipid Oxidation in 6KR Islets

Can the changes in gene expression explain the increased insulin response to glucose of 6KR islets? To answer this question, we assessed islet metabolism. Glucose oxidation to CO₂ was normal, as was the generation of pyruvate from glucose (i.e. glucose utilization), indicating that there are no gross changes to glucose metabolism (Fig. 2, B and C). Next we measured [¹⁴C]palmitate oxidation. We found no difference in islets incubated in 2.8 mm glucose but a nearly 50% decrease in 6KR islets incubated in 16.8 mm glucose compared with WT (Fig. 2D). These data are consistent with prior observations that loss of FoxO function in islets is associated with increased fatty acid oxidation (10).

Given the decrease in lipid oxidation, we sought to determine whether 6KR islets had changes in mitochondrial function (10). We found no differences in mitochondrial area as well number of mitochondria by electron microscopy (Fig. 4A). Next, we tested mitochondrial function in purified islets by measuring cellular respiration (oxygen consumption rates (OCR)). Under basal conditions, ATP turnover (oligomycin-sensitive respiration), proton leak (oligomycin-insensitive respiration), and maximal respiration (FCCP-induced) were identical between WT and 6KR animals (Fig. 4B). Raising glucose concentrations from 3 to 20 mm doubled ATP turnover in 6KR islets and had a smaller, but not statistically different effect in WT islets. The addition of the electron transport chain inhibitor rotenone suppressed OCR in both groups to the same extent, indicating that non-mitochondrial respiration was unaffected (Fig. 4B). Measurements of citrate synthase activity as a readout of mitochondrial function showed no significant difference in the presence of basal or elevated glucose levels (Fig. 4C). We profiled expression of mitochondrial genes using low density cDNA arrays. We found a striking, coordinate decrease of transcripts encoding members of the solute transporter Slc25 family (Table 3). Specifically, we found decreases of the carnitine/acylcarnitine carrier Slc25a20 and the glutamate carriers Slc25a12 and Slc25a22 as well as nucleotide and phosphate transporters (27). Of note, both Slc25a20 and Slc25a22 are significantly increased in β-cells exposed to hyperglycemia or hyperlipidemia (28), and Slc25a20 inhibition is associated with improved insulin secretion (29). The fact that they are decreased in 6KR islets raises the interesting possibility that reducing substrate flux into the mitochondrial inner core protects against mitochondrial dysfunction. Consistent with the observed decrease in free fatty acid oxidation, we saw decreased expression of Cpt1β, the enzyme responsible for ferrying long chain free fatty acids across the outer mitochondrial membrane, a rate-limiting step in fatty acid oxidation. Additional translocases were also decreased (Table 3). In sum, these data are consistent with the possibility that FoxO1 deacetylation decreases mitochondrial permeability to a variety of molecules and solutes and may, therefore, explain the decreased free fatty acid oxidation.

FIGURE 4.

Mitochondrial function in isolated islets. A, mitochondrial morphometry in pancreatic islets of WT and FoxO1–6KR mice (n = 3 per group). B, oxygen consumption rates measured in isolated islets from WT and FoxO1–6KR mice. Glc, glucose. Oligomycin inhibits ATP synthase, FCCP is an uncoupler, and rotenone inhibits complex I (n = 6 per group). Each experiment was performed with pooled islets from three mice per genotype. Each data point consists of 3 replicates. C, citrate synthase activity calculated at different glucose concentrations (n = 3 per genotype). All data are presented as the means ± S.E.

TABLE 3.

Mitochondrial low-density array

List of differentially expressed mitochondrial genes, as identified by expression profiling of islets with a low density array. The right column indicates -fold change compared to WT. Only values meeting the p < 0.05 are presented.

Gene	-Fold change from WT
Cpt2	2.0936
Hspd1	2.1376
Rhot1	2.3554
Timm8a1	2.3554
Hsp90ab1	2.5071
RTC	2.5071
PPC	2.5955
Aifm2	−3.0398
Aip	−2.998
Bcl2	−3.061
Cdkn2a	−2.1347
Cox10	−4.039
Cox18	−4.4816
Cpt1b	−2.2408
Dnajc19	−3.4678
Mfn1	−2.521
Rhot2	−2.0763
Sfn	−5.0771
Slc25a1	−2.0907
Slc25a12	−2.0195
Slc25a14	−2.5919
Slc25a15	−2.3038
Slc25a19	−6.3379
Slc25a2	−3.2355
Slc25a20	−2.6647
Slc25a22	−4.3893
Slc25a23	−2.2879
Slc25a24	−6.5614
Slc25a25	−3.3729
Slc25a30	−8.1908
Slc25a31	−4.6396
Slc25a37	−2.4351
Sod2	−2.3198
Taz	−3.3496
Timm44	−2.1199
Timm50	−3.8745
Tomm40	−8.1908
Tspo	−2.521

Discussion

We undertook the present study to probe in greater mechanistic detail the role of FoxO transcription factors in the pathogenesis of β-cell dysfunction and their potential as a direct or indirect target for the development of new approaches to treat this condition. The key conclusion of this work is that a moderate FoxO1 gain-of-function, as conferred by mutations that mimic its deacetylation (7), has a selective effect on a narrow swath of β-cell-specific genes and improves β-cell performance in the context of diabetes, possibly by decreasing fatty acid oxidation.

The present work strengthens the connection between FoxO and β-cell function by filling conjectural gaps. Specifically, most of the supportive evidence for a role of FoxO1 in β-cell dysfunction thus far derives from loss-of-function studies (4). Indeed, we previously studied transgenic mice carrying a constitutively nuclear, phosphorylation-defective mutant in β-cells, but there are three important new aspects to the present work. First, at the time of our prior experiments, our understanding of the function of FoxO1 in β-cells was not nearly as advanced as it is today, and most relevant parameters were simply not analyzed. Second, the phosphorylation-defective mutant analyzed in those studies has a strong effect on β-cell proliferation that likely masked some of the subtler effects on β-cell differentiation and metabolism (30). And finally, the deacetylated mutant used in the present studies is expressed as a knock-in, preserving both transcriptional and some of the post-translational regulation of FoxO1 function through phosphorylation (31).

The gene expression signature of WT islets transitioning from fasting to re-feeding appears to involve modulation of protein translation, degradation, mitochondrial complex I function, and Rho/Gef activity (specific examples are described below). The latter might be involved in the cytoskeletal remodeling associated with docking and fusion of secretory granules that occurs during second-phase insulin release (32). A novel finding of our study is the robust induction of two transcription factors, Pax6 and Isl1, with re-feeding. The former might be involved in the activation of glucagon production (33), whereas the latter has been less studied for its role in terminally differentiated endocrine cells. Likewise, we detected a strong inhibition of the neutral amino acid transporter Slc1a4 (Asct1), which has been implicated in the control of nutrient availability in the CNS through glia (34). What role if any it may have in islets is unclear, but its cognate isoform Asct2 has been implicated in the anaplerotic response by increasing glutamate transport and has been shown to be regulated by Atf4 and Myc (35). Interestingly, the substantial (∼90%) fall of Asct1 levels in response to feeding is associated with strong inhibition of both Myc and Atf4 in islets. If Asct1 acts as a glutamate transporter, a disputed notion onto itself (27), it could participate in the anaplerotic pathway in β-cells.

These responses are preserved in 6KR β-cells. Only ∼2% of the >4000 genes affected by re-feeding show preferential patterns of expression in 6KR islets, but this narrow swath provides stronger “β-cell identity”: key β-cell-specific factors, including insulin-1 and 2, Pdx1, MafA, and Glut2 are overrepresented in islets of 6KR mice. Of interest is the increase in Gpr119 levels, an orphan G-protein-coupled receptor whose activation is pursued for diabetes therapeutic purposes (36). These data provide the key, heretofore missing evidence that FoxO1 deacetylation can actively “protect” β-cells. Consistent with this notion, 6KR mice maintain better in vivo and ex vivo insulin secretion despite the hyperglycemia caused by increased hepatic glucose production (12).

In addition to maintaining robust expression of β-cell identity factors, FoxO1 appears to regulate the balance of glucose versus lipid utilization in islets as it does in liver (37). The key finding of the present study is the decrease of lipid utilization by purified islets in the presence of elevated glucose. These findings can be explained by several concurrent features of the gene expression analysis: (i) the strong inhibition of Pparγ and C/ebps with feeding; (ii) Cd36 is decreased, as are Cpt1α and -β; (iii) Chrebp is increased and so is its target gene AldoB, which would funnel more glucose toward oxidation; (iv) AldoB is normally inhibited by Pparα activation when lipid oxidation is favored over glucose oxidation (38); (v) the decrease of the Slc25 family and specifically the carnitine/acylcarnitine carrier is consistent with the observation that these carriers are increased in cells exposed to glucose and lipotoxicity (28) and that inhibition of carnitine translocase enhances insulin secretion (29). Thus, the ability of 6KR to tune down expression of solute carriers might be part of a protective function. Interestingly, the mitochondrial gene expression signature of FoxO1 in β-cells appears to differ substantially from that in liver (39). Whether this depends on different levels of deacetylation in the two organs is an interesting question that will have to be addressed in subsequent studies. These data are entirely consistent with previous work, demonstrating that loss of FoxO function is associated with de-repression of the Pparα pathway, with attendant increases in lipid oxidation that might affect mitochondrial function (10).

In conclusion, we provide critical experimental support for the hypothesis that deacetylation is indeed a molecular mechanism underlying the ability of FoxO1 to protect β-cells, and we reveal a new dimension to its molecular targets. These data provide an opportunity to evaluate new targets for therapeutic intervention against β-cell failure.

Author Contributions

J. Y. K.-M. designed and conducted the experiments, analyzed the data, and wrote the article. Y. J. R. K., J. F., and S. Z. conducted the experiments and analyzed the data. A. S. B. generated the transgenic mice. M. P. and D. A. designed the experiments, analyzed the data, and wrote the manuscript.