The Mammalian INDY Homolog Is Induced by CREB in a Rat Model of Type 2 Diabetes

Reduced expression of the INDY (I’m not dead yet) gene regulates life span by mechanisms that share important similarities with caloric restriction—the most reliable intervention to prolong life span over a wide range of species (1–3)—in Drosophila melanogaster (4) and Caenorhabditis elegans (5). For example, INDY-reduced long-lived flies show decreased whole-body fat stores and expression of insulin-like proteins (6). The INDY gene product is a cation-independent, electroneutral tricarboxylate carrier (7–9) able to transport citrate across the plasma membrane as its preferred substrate and is mainly expressed in organs involved in energy homeostasis in flies (10).

In mammals, the gene product of the mammalian Indy (mIndy, SLC13A5) homolog, the sodium-coupled citrate transporter (NaCT) INDY, shares the highest sequence and functional similarity with D melanogaster INDY (8), and mIndy is predominantly expressed on the plasma membrane of hepatocytes (9,11,12). We recently showed that deleting mIndy in mice also mimics important aspects of caloric restriction (13), without reducing caloric intake, and protects from high-fat diet (HFD)– and aging-induced adiposity and insulin resistance. Therefore, upstream modulators of mIndy expression hold the intriguing potential to modify mammalian metabolic regulation by controlling di-/tricarboxylate uptake into hepatocytes (14); however, no such modulators have been identified.

Limited recent evidence suggests that mIndy is regulated on the transcriptional level through changes in hormonal and/or nutritional status, such as with lipid gavaging or long-term fasting (6,13,15). Moreover, previous studies have demonstrated that the hormone glucagon markedly contributes to the uptake of citrate from the plasma into the liver (16). It is thus plausible to hypothesize that glucagon is a hormonal regulator of mIndy transcription and/or activity, increasing the uptake of citrate by the liver and thereby affecting hepatic intermediary metabolism.

For fasting studies, animals were fed ad libitum on standard rodent chow leading up to the experiment. The night before the start of the experiment, food was removed in one group but was left in the control group. In the morning, rats were killed and livers were taken and kept snap-frozen in liquid nitrogen until analysis. For citrate uptake experiments, 10 µCi of [¹⁴C]-citrate was injected intraperitoneally 1 h before rats were killed. Uptake of citrate was normalized to protein. For the type 2 diabetic (T2D) rat model, rats were then given a 175 mg/kg dose of nicotinamide by intraperitoneal injection; after 15 min, rats were then dosed with 65 mg/kg of streptozotocin (STZ) (17).

Rats were given 4 days of a recovery period and randomized for blood glucose values before the first cAMP-responsive element–binding protein (CREB) or control antisense oligonucleotides (ASO) injection. Rats with overt diabetes (defined by a glucose value greater than 200 mg/dL) were excluded from the study. Body weight and food consumption were monitored weekly. The food consumed consisted of an HFD (26% carbohydrate, 59% fat, 15% protein calories), in which the major constitute is safflower oil (17), to mimic most closely the situation in a T2D patient. CREB ASO or control ASO was injected at 75 mg/kg semiweekly for 4 weeks. Liver tissue was harvested after an overnight fast and immediately snap-frozen after the animals were killed with a deep isoflurane narcosis. All procedures were approved by the Yale University School of Medicine Institutional Animal Care and Use Committee, the animal care and use committee of an accredited Pfizer vendor, or the Landesamt für Umwelt, Gesundheit und Verbraucherschutz (LUGV) Brandenburg.

Rat hepatocytes were isolated from healthy male Wistar rats as described previously (18). Hepatocytes were snap-frozen for subsequent RNA preparation or luciferase assays or were used to determine [¹⁴C]-citrate uptake.

Total RNA was isolated from rat hepatocytes, and 1–5 µg total RNA was reverse-transcribed into cDNA. Real-time PCR was carried out using the oligonucleotides used are listed in Table 1. The expression level was calculated as an n-fold induction of the gene of interest (int) in treated versus control cells with actin (act) as a reference gene according to the formula: fold induction = 2^(c-t)int/2^(c-t)act.

Immunoblots were performed as described (19).

Hepatocytes were treated with 0 or 10 nmol/L glucagon for 8 h, as described above. At the end of the stimulation period, they were washed twice with uptake buffer (25 mmol/L HEPES/NaOH [pH 7.5], 140 mmol/L NaCl, 5 mmol/L KCl, 1.8 mmol/L CaCl₂, 0.8 mmol/L MgSO₄, and 5 mmol/L glucose) and then incubated at room temperature in the same buffer containing 16 μmol/L [¹⁴C]-citrate for 15 min. Nonspecific transport and nonspecific binding were determined in the presence of a 1,000-fold excess of unlabeled citrate. Cells were washed twice with ice-cold uptake buffer, and cell-associated radioactivity was released by lysing cells in 0.3 mol/L NaOH (400 µL) in 1% (w/v) SDS and counted.

To determine gluconeogenesis, 1 µCi/mL [¹⁴C]-citrate was added to the culture, which was continued for 2 h in presence or absence of glucagon. Glucose in cell culture supernatants was separated from citrate by anion exchange chromatography on Dowex formiate.

Total RNAs from two different rats was used for 5′-rapid amplification of cDNA ends (RACE) using the 5′-RACE system 2.0 (Life Technologies, Eggenstein, Germany), as described previously (20).

Fragments containing 290 or 324 bp of the putative mIndy promoter were generated from rat genomic DNA by PCR using the primers listed in Table 1. Fragments were cloned in the right orientation into pGL3-basic (Promega, Mannheim, Germany). A mutation in the potential CRE-site in the 324 bp mIndy promotor fragment was introduced by PCR-based site-directed mutagenesis using the primers listed in Table 1.

Reporter gene assays with transfected primary rat hepatocyte were performed as described previously (21).

Small pieces of fresh liver tissue from three rats underwent cross-linking in 1% formaldehyde for 15 min, followed by quenching with 1/20 volume of 2.5 mol/L glycine solution, and one wash with 1× PBS. Nuclear extracts were prepared by homogenizing in 20 mmol/L HEPES, 0.25 mol/L sucrose, 3 mmol/L MgCl₂, 0.2% NP-40, 3 mmol/L β-mercaptoethanol; complete protease inhibitor tablet from Roche, Mannheim, Germany. Chromatin fragmentation was performed by sonication 50 mmol/L HEPES, 1% SDS, and 10 mmol/L EDTA, using a Bioruptor (Diagenode, Seraing, Belgium) for 20 cycles of 30 s at the highest level. Proteins were immunoprecipitated in 50 mmol/L HEPES/NaOH at pH 7.5, 155 mmol/L NaCl, 1.1% Triton X-100, 0.11% Na-deoxycholate, and complete protease inhibitor tablet, using an anti-CREB antibody (sc-186 Lot# F0412; Santa Cruz Biotechnologies). Crosslinking was reversed overnight at 65°C and DNA isolated using phenol/chloroform/isoamyl alcohol. For chromatin immunoprecipitation (ChIP) quantitative (q)PCR, enrichment was measured using SYBR Green PCR Mastermix (Roche) and the ABI 7300 instrument (Applied Biosystems, Life Technologies). Analysis of ChIP samples and the corresponding inputs was performed by the standard curve method and normalization to a nontarget control region of the insulin gene. Primer sequences used for qPCR analyses are given in Table 1.

The cDNA coding for FLAG-tagged human mIndy was amplified from human hepatocyte cDNA and cloned into the pcDNA3. Clonal HepG2 cells stably expressing human Indy were generated by single-cell cloning and selected with 500 units/mL G-418. To determine lipid synthesis in these cells, they were cultured with 10 nmol/L insulin and 100 μmol/L [¹⁴C]-citrate for 8 h. Radiolabeled sterols and fatty acids were extracted by basic and acid hexane extraction, respectively.

A two-tailed Student t test was used to test differences between two groups. One-way ANOVA analysis was performed for multiple comparisons. Values are presented as mean ± SE. P < 0.05 was considered statistically significant.

Cultured hepatocytes were stimulated with 10 nmol/L glucagon or 10 nmol/L insulin as a negative control. Hepatocytes were harvested after 2 h. Although stimulation with insulin neither induced nor repressed mIndy mRNA in hepatocytes, glucagon induced mIndy mRNA about 2.5-fold (Fig. 1A). Phosphoenolpyruvate-carboxykinase (PCK) mRNA, which served as a positive control, was also not affected by insulin but was strongly induced by glucagon (Fig. 1B). At the same time, glucokinase, as a prototypical gene that is induced in response to insulin, was induced about twofold (Fig. 1C). However, due to the culture conditions chosen, the induction of glucokinase was comparatively small. To exclude that the lack of insulin-dependent regulation of mIndy expression was attributable to the culture conditions that might obscure smaller insulin effects, the experiment was repeated under conditions that allowed for a much larger induction of glucokinase by insulin (i.e., hepatocytes were starved for insulin for 5.5 h before stimulation with 10 nmol/L insulin). Although glucokinase was strongly induced by insulin under these conditions (33.5- ± 8.1-fold, P < 0.01), insulin still did not affect mIndy expression (0.9- ± 0.2-fold, data not shown).

Induction of rat mIndy mRNA by glucagon was time- and dose-dependent. With 10 nmol/L glucagon, mIndy mRNA was slightly induced as early as 1 h after the stimulus was added (Fig. 1D). The induction was maximal between 2 and 4 h and then gradually declined. Notably, the response was biphasic, with a reduction of mIndy mRNA below the starting level after 8 and 24 h of glucagon stimulation.

Expression of mIndy mRNA was induced by glucagon in the physiological concentration range between 1 and 10 nmol/L, and induction reached a maximum between and 10 nmol/L (Fig. 1E) and had a tendency to decline toward 100 nmol/L. The half-maximal effective concentration of the response to glucagon was ∼2 nmol/L. This dose-response curve closely paralleled the dose-response curve for the prototypic fasting-induced gene PCK in the same experiments (not shown).

mIndy codes for the di-/tricarboxylate carrier mIndy (NaCT). The preferred substrate of this carrier is citrate. To further elucidate the functional relevance of the mIndy induction by glucagon, [¹⁴C]-citrate uptake was determined in control and glucagon-stimulated rat hepatocytes. At a concentration of 16 μmol/L citrate, which is close to the reported K_m value for the rat transporter, citrate uptake was increased about 1.5-fold by a prior induction with glucagon for 8 h (Fig. 2A). If hepatocytes were stimulated with glucagon for shorter periods (i.e., 4 h), no significant induction of transport activity was observed (data not shown). This is in accordance with previous observations concerning the induction of mRNA and activity of the prototypic glucagon target gene PCK. The mRNA induction reached a maximum after 2 h and then declined rapidly, whereas the activity lagged behind and remained elevated (22).

The principal signaling cascade downstream of the glucagon receptor is a Gs-dependent activation of adenylate cyclase and a subsequent activation of PKA due to increased cAMP concentrations. Therefore, whether PKA-activating agents could mimic the glucagon effect in elevating mIndy mRNA was tested next. In accordance with expectations, the PKA-activator 6-Bnz-cAMP significantly increased the rat mIndy mRNA although to a somewhat lesser extent than 10 nmol/L glucagon, which served as positive control (Fig. 2B).

5′-RACE experiments were performed to identify the transcription start site with gene-specific primers for reverse transcription in exon 4 and nested PCR primers that were located in the fourth and third exon (Fig. 3A). A possible transcription start site that is located ∼16 nucleotides upstream of the 5′-end of the published mRNA (39 bp upstream of the start ATG) was found in five of seven clones that contained sequences continuing into the first exon (Fig. 3B). One clone each was found that terminated at the published 5′-end (23 bp upstream of the start ATG) and one that ended ∼110 bp upstream. The region upstream of the most frequent start point contains a putative CAAT-box motif. The region upstream of the most frequent start point was tentatively assigned as a promoter region. A reporter gene construct was generated by cloning a stretch of 324 bp upstream of the start ATG of the rat mIndy gene, including the putative transcription start site, in front of luciferase in the vector pGL3-basic (Fig. 3B). The reporter gene expression under the control of this rat mIndy promoter fragment in rat hepatocytes was significantly activated by glucagon and the PKA activator 6-Bnz-cAMP (Fig. 3C).

Reporter gene activity was enhanced by glucagon in a time- and dose-dependent manner (Fig. 3D and E). The dose-response curve of the induction of the reporter gene activity closely reflected the dose-response curve for the induction of the endogenous rat mIndy mRNA. Similar to what was observed with the endogenous mRNA, reporter gene activity increased rapidly after the addition of glucagon. In contrast with the endogenous mRNA, after reaching a maximum at 4 h, luciferase reporter gene activity did not rapidly decline after but remained elevated over the subsequent 20 h.

A nuclear target of PKA-dependent regulation of promoter activity is the protein CREB. This transcription factor binds to a cis-acting element called cAMP-responsive element (CRE). In silico analysis of the 324 bp rat mIndy promoter fragment revealed a potential CRE close to the 5′-end (Fig. 3B). Therefore, a second construct was generated that was shortened by 34 nucleotides and hence was missing the CRE. In contrast to the full-length construct, no glucagon-dependent induction of the reporter gene activity was observed in hepatocytes transfected with the truncated construct (Fig. 3C, last panel). Truncation, in addition to removing the CRE, might have resulted in a loss of other functionally important sequence elements in the promoter. Therefore, an additional reporter gene construct was generated that retained the full length of the original active 324 bp construct in which, however, the CRE was functionally inactivated by a C to G mutation within the core consensus site (Fig. 3B). Similar to what was observed with the truncated construct, mutation of the CRE core sequence largely abolished the glucagon-dependent induction of the reporter gene under control of the mutated rat mIndy promoter (Fig. 3C, middle panel).

To elucidate the interaction between CREB as a transcriptional regulator and its putative target gene mIndy in vivo, we used ChIP in rat livers. Known binding sites of CREB near the PCK and signal transducer and activator of transcription-3 (Stat3) genes served as positive controls (23), whereas a region downstream of the putative CREB-binding site within the mIndy gene served as negative control. We found that CREB occupied the mIndy promoter at the proposed site nearly as strongly as it did near the PCK and Stat3 gene (Fig. 4). No enrichment was observed in the control region of mIndy.

Glucagon is the major hormone controlling hepatic gluconeogenesis and, thus, hepatic glucose output through the induction of CREB during an initial phase of fasting (24, 25). Therefore, hepatic mIndy expression was assessed after an overnight fast in rats. In this setting, mIndy expression was induced threefold (Fig. 5A). To establish a functional effect, citrate uptake was determined in fed versus fasted rats after an intraperitoneal injection of [¹⁴C]-citrate. Overnight-fasted rats integrated about 30% more citrate into the liver than fed rats (Fig. 5B). Cytoplasmic citrate can be converted by ATP-citrate lyase to acetyl-CoA and oxaloacetate, the latter opening into gluconeogenesis. Thus, we studied whether glucagon treatment in primary rat hepatocytes increased the incorporation of [¹⁴C]-labeled citrate into glucose. Interestingly, glucagon significantly enhanced the formation of glucose from citrate (Fig. 5C).

CREB is constitutively activated in insulin-deficient diabetes. To establish the pathophysiological significance of our findings, an animal model of STZ-induced, HFD-fed diabetes was used, mimicking most closely the situation in T2D patients, who are characterized not only by a relative lack of insulin but also by hyperglucagonemia and subsequent activation of CREB (26). These rats were compared with rats fed a pure HFD, without STZ treatment and STZ-treated, HFD-fed rats in which CREB was depleted by CREB-ASO. Biochemical serum parameters of mice are given in Table 2. The T2D rat model with control ASO and the T2D rat model with CREB-ASO both developed marked hyperglucagonemia. In accordance with the assumption that CREB is constitutively activated in T2D, the phospho-CREB-to-CREB ratio was significantly elevated in STZ-treated versus HFD controls (Fig. 5D). CREB-ASO reduced CREB protein expression by 50% compared with control-ASO (P < 0.01, data not shown), and CREB mRNA expression was reduced by 74 ± 3% compared with the T2D rat model treated with control-ASO (P < 0.01; Fig. 5E). In this setting, PCK was reduced by 30 ± 3% (P < 0.05) in the CREB-ASO–treated rats (data not shown). mIndy expression was induced by the treatment of STZ compared with the solely HFD-fed rats with control-ASO. This induction was completely prevented by the depletion of CREB (Fig. 5F). Thus, deficiency of CREB was able to prevent the induction of mIndy-induced by hyperglucagonemia. To further test the consequences of increased mIndy expression for hepatic lipid metabolism, an mIndy-overexpressing HepG2 cell line was generated. mIndy expression was fourfold higher in this cell line than in untransfected HepG2 cells. Fatty acid and sterol synthesis in this setting was induced approximately twofold in the mIndy-overexpressing cells (Fig. 5G)

Our study is the first to identify glucagon as a transcriptional regulator of the rat mIndy gene and to describe a promoter sequence of mIndy that is located upstream of the most frequent transcription start site, which was determined by 5′-RACE (Fig. 3). Evidence is provided for rat mIndy being a CREB-dependent glucagon target gene, because treatment of primary cultured rat hepatocytes with glucagon (Fig. 1) induced mIndy mRNA, which codes for the plasma membrane citrate transporter mIndy (NaCT). As a consequence, citrate uptake into glucagon-stimulated rat hepatocytes was increased (Fig. 2A). The induction of mIndy was cAMP-dependent, as shown by the mIndy induction in response to the cell-permeable cAMP analog 6-Bnz-cAMP (Fig. 2B). cAMP activated PKA, which can in turn phosphorylate and activate the transcription factor CREB. In accordance with the assumption that CREB phosphorylation and direct activation of the rat mIndy promoter was the mechanism underlying the induction of mIndy mRNA, a cAMP-responsive element consensus sequence was identified by in silico analysis of the rat mIndy promoter 5′ upstream of the putative transcription start site (Fig. 3B). A reporter gene construct containing this site was activated by glucagon or 6-Bnz-cAMP (Fig. 3C), whereas truncation as well as site-directed mutagenesis of the consensus sequence of this site resulted in an almost complete elimination of the glucagon-dependent promoter activation (Fig. 3C). Finally, ChIP studies showed a strong occupation of this mIndy promoter site by CREB, pinpointing CREB as the transcription factor that mediates the induction of mIndy by glucagon.

Comparable with the induction of the prototypic glucagon target gene in hepatocytes, the gluconeogenic key enzyme PCK, mIndy mRNA levels were elevated only transiently, with a maximum between 2 and 4 h, despite the continued presence of glucagon in our cell culture assays. After prolonged stimulation with glucagon, mIndy mRNA levels decreased below starting levels (Fig. 1D). This might be due to rapid degradation of glucagon by hepatocytes in culture (27,28). One alternative possible explanation might be a pattern that is known from PCK activation by glucagon: Glucagon not only induces mRNA of PCK but also activates its degradation. Possibly, a similar mechanism mediates the effect on mIndy mRNA (29).

Glucagon is the major hormone controlling hepatic gluconeogenesis, and thus, hepatic glucose output through the induction of CREB during an initial phase of fasting (24,25). We show that mIndy expression was induced after an overnight fast in rats. Accordingly, our data show that the uptake of citrate into the liver was increased in fasted rats, suggesting that the induction of mIndy leads to an increased uptake of citrate into the liver. Citrate can be converted into oxaloacetate as a carbon source for gluconeogenesis. Accordingly, our data show that glucagon increased uptake of citrate and the incorporation of citrate into glucose in primary rat hepatocytes in vitro. Previous studies have shown that the knockout of Indy in mice reduced hepatic glucose production. Taken together, these data suggest that glucagon-induced mIndy expression may contribute to the regulation of gluconeogenesis in the fasted liver by supplying carbon sources as building blocks for the generation of glucose.

We have previously shown that mIndy expression was reduced in 36-h starved mice. This seemingly discrepant finding might be explained by the fact that glucagon effects on gluconeogenesis are attenuated during late fasting (25), when the CREB-regulated transcription coactivator 2 (CRTC2, TORC2) is downregulated after an initial activation. This explanation is supported by our data showing that glucagon induced mIndy initially, followed by a decrease in expression in primary hepatocytes. Clearly, more studies are needed to decipher the time course and mechanism of this finding.

To test if our findings also have pathophysiological implications, we studied a rat model of T2D with or without knockdown of CREB by ASOs. The model was chosen because T2D is characterized not only by an absolute or relative lack of insulin but also by failure to repress glucagon secretion. In this setting, CREB has been shown to be constitutively activated (26). In line with the hypothesis, CREB-ASO resulted in reduced CREB expression, PCK expression, and reduced mIndy mRNA levels by more than 50% compared with the control-ASO–treated diabetic rats. Our present and previous data (12,13) show that increased expression of mIndy enhances the uptake of citrate into the cytoplasm, whereas reducing mIndy expression reduced cellular import of citrate. Citrate is a central metabolite, in both cytosolic and in mitochondrial metabolism, by connecting carbohydrate catabolism and lipogenesis. Citrate is the main carbon source of fatty acid and cholesterol synthesis and acts as an allosteric activator of acetyl-CoA carboxylase. Moreover, cytosolic citrate furnishes NADPH generation via malic enzyme for lipogenesis (30). Overexpression of mIndy in HepG2 cells resulted in an increase in intracellular fatty acid and sterol synthesis. In line with this, enhancing the activity of citrate transport by Indy in HepG2 cells has been shown to enhance citrate-induced lipid synthesis (14). It is tempting to speculate that in conditions with increased CREB activity, such as in T2D, increased mIndy expression could contribute to ectopic lipid accumulation and aggravate nonalcoholic fatty liver disease, which is associated with T2D (31). The increase in glucagon-induced hepatic glucose production from citrate, as observed in our studies, might further contribute to this association. In line with this notion, knockdown of CREB (17) in rats as well as mIndy in mice (13) has been shown to reduce diet-induced hepatic lipid accumulation and hepatic glucose production. Reducing mIndy activity in the setting of T2D might therefore be an interesting target to control excessive hepatic lipid accumulation and hepatic glucose production.

In summary, we have shown that mIndy is a CREB-dependent target gene of glucagon in rats. Our work describes a proximal promoter sequence of mIndy that is located upstream of the most frequent transcription start site, which was determined by 5′-RACE. Induction of mIndy was observed in an early phase of fasting and in a rat model of T2D. In these conditions, mIndy might be involved in the control of hepatic glucose and lipid metabolism. Reducing mIndy activity in the setting of T2D might therefore be an interesting target to control excessive hepatic lipid accumulation and hepatic glucose production.

Acknowledgments. The authors thank ISIS Pharmaceuticals for kindly providing the CREB-ASO and Patrick Schmerler (Institute of Pharmacology, Charité University Medicine Berlin) for providing rat liver tissue for ChIP studies. The contribution of Lucretia Hernandez-Valenciano, who performed part of the initial screening experiments and helped to construct a reporter gene vector as part of her bachelor's thesis, is gratefully acknowledged.

Funding. This work was funded in part by project grants from the German Research Foundation (Emmy Noether grant SCHU 2546/1-1) and the Einstein Foundation Berlin (grant A-2010-83) to M.S.; the U.S. Public Health Service (R01-DK-40936, R24-DK-085638, P30-DK-45735) to G.I.S; and the German Diabetes Association, the German Research Foundation (DFG, BI1292/4-1), and the Fritz-Thyssen Stiftung (Az 10.12.2.140) to A.L.B.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. F.N.-R. researched data and supervised most of the research work. S.L., M.K., J.H., R.J.P., D.M.E., D.P., D.M.W., S.B., C.v.L., and A.T. researched data. A.P.-N.-R. generated the HepG2-mIndy cell line. G.P.P. and A.L.B. researched data, supervised most of the research work, planned the study, and wrote the manuscript. M.S. researched data and reviewed the manuscript. A.F.H.P and G.I.S. reviewed the manuscript. A.L.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.