Growing attention has been focused on the roles of the proximal tubules (PTs) of the kidney in glucose metabolism, including the mechanism of regulation of gluconeogenesis. In this study, we found that PT-specific insulin receptor substrate 1/2 double-knockout mice, established by using the newly generated sodium–glucose cotransporter 2 (SGLT2)-Cre transgenic mice, exhibited impaired insulin signaling and upregulated gluconeogenic gene expression and renal gluconeogenesis, resulting in systemic insulin resistance. In contrast, in streptozotocin-treated mice, although insulin action was impaired in the PTs, the gluconeogenic gene expression was unexpectedly downregulated in the renal cortex, which was restored by administration of an SGLT1/2 inhibitor. In the HK-2 cells, the gluconeogenic gene expression was suppressed by insulin, accompanied by phosphorylation and inactivation of forkhead box transcription factor 1 (FoxO1). In contrast, glucose deacetylated peroxisome proliferator–activated receptor γ coactivator 1-α (PGC1α), a coactivator of FoxO1, via sirtuin 1, suppressing the gluconeogenic gene expression, which was reversed by inhibition of glucose reabsorption. These data suggest that both insulin signaling and glucose reabsorption suppress the gluconeogenic gene expression by inactivation of FoxO1 and PGC1α, respectively, providing insight into novel mechanisms underlying the regulation of gluconeogenesis in the PTs.
The kidney plays a pivotal role in systemic glucose metabolism by regulation of glucose reabsorption, glycolysis, and gluconeogenesis. Gluconeogenesis occurs exclusively in the liver and the kidney (1). In the absorptive state, the kidney accounts for only 10% of the systemic gluconeogenesis, whereas in the prolonged fasting state, the rate rises to as much as 40% (2). Among precursors of gluconeogenesis, glutamine is used as the major glucogenic amino acid in the kidney, whereas alanine is used as the major amino acid in the liver (3,4).
Gluconeogenic enzymes, including phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase), are expressed mainly in the proximal tubules (PTs) of the kidney and in the liver (5,6). In the process of gluconeogenesis in the PTs, glucose diffuses outward through glucose transporter 2 (GLUT2) in the basolateral membrane (7,8). On the contrary, GLUT2 plays an important role in glucose uptake in the liver, and sodium–glucose cotransporters (SGLTs) in the luminal membrane are involved instead in inward efflux of glucose in the PTs (9,10).
It is well known that gluconeogenic gene expression in the liver is mainly regulated by insulin, especially via suppression of forkhead box transcription factor 1 (FoxO1). FoxO1 is a major transcription factor that binds to the promoter regions of PEPCK and G6Pase to induce the gene expression (11–13). Insulin signaling activated after feeding, however, promotes translocation of FoxO1 to the cytoplasm, consequently suppressing gluconeogenic gene expression via the insulin receptor substrates (IRSs) and Akt (14,15). We have previously reported enhanced hepatic gluconeogenic gene expression in liver-specific IRS1/2 double-knockout mice, which was reversed by inactivation of FoxO1 (16).
Similarly, it was shown using PT-specific insulin receptor (IR) knockout mice (17) that insulin also downregulated gluconeogenic gene expression in the PTs of the kidney. Rich expression of the IRs is seen in the renal cortex, including in the PTs (17,18), and it is assumed that the IRs on the basolateral side sense plasma insulin and play an important role in intracellular signaling (19). However, the role of insulin in the regulation of gluconeogenesis in the kidney is still under debate, because suppression as well as elevation of gluconeogenic gene expression has been reported from experiments in rodents in which insulin secretion was suppressed by treatment with streptozotocin (STZ) (19,20). It is thus suggested that other potential mechanisms could exist.
Another molecule to regulate gluconeogenic gene expression in the liver is sirtuin 1 (Sirt1), known as an NAD-dependent deacetylase, by deacetylating peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α), which plays a pivotal role as a coactivator of FoxO1 in the transcription of gluconeogenic genes (21,22). In the field of nephrology, various roles of Sirt1 in the kidney have been reported (23–26), and, especially in the context of the responses to nutrition, its importance seems to be highlighted in the presence of relatively low glucose concentrations, both in vitro and in vivo (27). Involvement of Sirt1 in gluconeogenesis in the PTs, however, remains to be clarified.
In this study, we focused on insulin signaling using a PT-specific knockout mouse model, as well as on glucose reabsorption by SGLTs, and explored the mechanisms underlying the regulation of gluconeogenic gene expression in the PTs of the kidney.
Research Design and Methods
Generation of the Transgenic Mice
To generate SGLT2-cre transgenic (Tg) mice with a construct according to a previous report (28), we obtained a genomic fragment of Slc5a2 (encoding SGLT2) including the promoter region, exon 1, intron 1, and the first part of exon 2 from the RP24–178K1 BAC clone (BACPAC Resources) with the BAC Subcloning Kit (Gene Bridges). It was inserted upstream of the Cre recombinase DNA that was flanked by rabbit β-globin and a polyadenylation sequence, as previously described (29), following deletion of the start codon using the KOD -Plus- Mutagenesis Kit (TOYOBO). The linearized construct was injected into the pronuclei of fertilized C57BL/6J embryos, and two-cell stage embryos were transferred into the oviduct of a pseudopregnant mouse. After confirmation of germline transmission, the Tg mice were crossed with ROSA26-LacZ mice (003474; The Jackson Laboratory) to analyze the expression of Cre recombinase. Then, the Tg mice were crossed with IRS1/2-floxed mice (16) for generating mice with PT-specific knockout of the genes.
C57BL/6J and ob/ob mice were purchased from CLEA Japan. C57BL/6NCr Slc and Akita/Slc mice were purchased from SLC Japan. All mice were housed under a 12-h light/12-h dark cycle and had free access to sterile water and regular chow, CE-2 (CLEA Japan). In the experiments conducted in the fasting and fed states, the mice were denied access to pellet food or given access to pellet food after 24-h fasting in individual cages. The animal care and experimental procedures were approved by the Animal Care Committee of The University of Tokyo.
STZ (Sigma-Aldrich) was diluted in citrate sodium buffer and administered intraperitoneally at the dose of 180 mg/kg body weight (BW) twice at an interval of 4 days. Mice were used for the experiments 10 days after the treatments. Phlorizin (PHZ; Sigma-Aldrich) was dissolved in a solution containing 10% ethanol, 15% DMSO, and 75% saline and injected subcutaneously.
The insulin tolerance test, pyruvate tolerance test, glutamine tolerance test, and alanine tolerance test were performed after the mice had been denied access to food for 6 h, unless otherwise indicated. Mice were injected intraperitoneally with insulin (0.75 units/kg: Humulin R; Eli Lilly Japan), pyruvate (2.0 g/kg; Wako), glutamine (2.0 g/kg; Biowest), or alanine (2.0 g/kg; Wako) (16,30). Blood glucose levels were measured using the Glutest sensor (Sanwa Kagaku Kenkyusho) at the indicated time points. The plasma insulin levels were measured using an ELISA kit (Morinaga).
β-Galactosidase staining was performed using the β-Gal Staining Set (Roche), according to the manufacturer’s instructions. For macroscopic analysis, tissues were fixed in 4% paraformaldehyde before the staining, and images were acquired with Optio WG-2 (RICOH). For microscopic analysis, tissues were first embedded in optimal cutting temperature compound (Sakura Finetek Japan), frozen on dry ice, and then carved out for fixation with 2% glutaraldehyde/PBS at room temperature. After β-galactosidase staining, the specimens were re-embedded in paraffin and sectioned for periodic acid Schiff (PAS) staining, and images were acquired with BZ-X710 (Keyence).
Frozen tissues were cut into 15-μm–thick sections and mounted onto a PEN-Membrane (Leica Microsystems). These sections were stained with PAS and excised using the Laser Microdissection DM6000B (Leica Microsystems).
RNA Preparation and Quantitative PCR
The RNeasy Mini Kit (Qiagen) was used to prepare total RNA from the mouse tissues and cultured cells, and the RNeasy Micro Mini Kit (Qiagen) was used to prepare total RNA from the microdissected tissues. Reverse-transcription reaction was carried out with a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) after treatment with DNase (Promega). Quantitative PCR analyses were performed using ABI Prism 7900, with Power SYBR Green PCR Master Mix (Applied Biosystems) (16). The relative expression levels after normalization to the expression level of cyclophilin were compared.
The primers used for the RT-PCR were as follows: mouse PEPCK forward, 5′-CCCCTTGTCTATGAAGCCCTCA-3′ reverse, 5′-GCCCTTGTGTTCTGCAGCAG-3′; mouse G6Pase forward, 5′-TTACCAAGACTCCCAGGACTG-3′ and reverse, 5′-GAGCTGTTGCTGTAGTAGTCG-3′; mouse Podocin forward, 5′-TGAAAGAGTAATTATATTCCGACTGG-3′ and reverse, 5′-TGATAGGTGTCCAGACAGGGTAA-3′; mouse Megalin forward, 5′-AGGCCACCAGTTCACTTGCT-3′ and reverse, 5′-AGGACACGCCCATTCTCTTG-3′; mouse aquaporin 2 forward, 5′-TAGCCCTGCTCTCTCCATTG-3′ and reverse, 5′-GAGCAGCCGGTGAAATAGAT-3′; mouse IRS1 forward, 5′-CTATGCCAGCATCAGCTTCC-3′ and reverse, 5′-TTGCTGAGGTCATTTAGGTCTTC-3′; mouse IRS2 forward, 5′-TCCAGGCACTGGAGCTTT-3′ and reverse, 5′-GGCTGGTAGCGCTTCACT-3′; mouse Hk1 forward, 5′-CGGAATGGGGAGCCTTTGG-3′ and reverse, 5′-GCCTTCCTTATCCGTTTCAATGG-3′; mouse Hk2 forward, 5′-TGATCGCCTGCTTATTCACGG-3′ and reverse, 5′-AACCGCCTAGAAATCTCCAGA-3′; mouse Gck forward, 5′- GTGGCCACAATGATCTCCTGC-3′ and reverse, 5′- TCGGCGACAGAGGGTCGAAGGC-3′; mouse SGLT1 forward, 5′-TCTGTAGTGGCAAGGGGAAG-3′ and reverse, 5′-ACAGGGCTTCTGTGTCTTGG-3′; mouse SGLT2 forward, 5′- TATTGGTGCAGCGATCAGG-3′ and reverse, 5′- CCCAGCTTTGATGTGAGTCAG-3′; mouse GLUT1 forward, 5′-CGTGCTTATGGGTTTCTCCAAA-3′ and reverse, 5′-GACACCTCCCCCACATACATG-3′; mouse GLUT2 forward, 5′-TTTGCAGTAGGCGGAATGG-3′ and reverse, 5′-GCCAACATGGCTTTGATCCTT-3′; mouse GLUT4 forward, 5′-CAACTGGACCTGTAACTTCATCGT-3′ and reverse, 5′-ACGGCAAATAGAAGGAAGACGTA-3′; mouse cyclophilin forward, 5′-GAGCTGTTTGCAGACAAAGTTC-3′ and reverse, 5′-CCCTGGCACATGAATCCTGG-3′; human PEPCK forward, 5′-TGAGCTGTGTCAGCCTGATCAC-3′ and reverse, 5′-ACCGTCTTGCTTTCGACCTG-3′; human cyclophilin forward, 5′- ATGCTGGACCCAACACAAAT-3′ and reverse, 5′-TCTTTCACTTTGCCAAACACC-3′.
Immunoprecipitation and Western Blot Analysis
To prepare the lysates, mouse tissues and cells were homogenized in buffer A (25 mmol/L Tris-HCl [pH 7.4], 10 mmol/L sodium orthovanadate, 10 mmol/L sodium pyrophosphate, 100 mmol/L sodium fluoride, 10 mmol/L EDTA, 10 mmol/L EGTA, and protease inhibitor). For immunoprecipitation (IP) of IR, IRS1, IRS2, and PGC1α, 4 mg lysates was incubated with specific antibodies, respectively, overnight at 4°C. Then, protein G-Sepharose was added, followed by incubation for 1 h at 4°C. After washing three times with buffer A, the target proteins were eluted with sample loading buffer. The lysates were resolved on SDS-PAGE and transferred to polyvinylidene difluoride membranes using the Trans-Blot Turbo Transfer System (Bio-Rad) (16).
Anti-IRS1 (IP and immunoblotting [IB]: 06–248), anti-IRS2 (IB: MABS15), and anti-Sirt1 antibodies (05–1243) were purchased from Merck Millipore. Anti-IRS2 (IP: 3089), anti-FoxO1 (9454), anti-phosphorylated (p-)FoxO1 (Ser256) (9461), anti-Akt (9272), anti–p-Akt (Ser473) (9271), and anti-PGC1α antibodies (IB: 2178) were purchased from Cell Signaling Technology. β-Actin (sc-1616), insulin Rβ (IP and IB: sc-711), p-Tyr (pY99) (sc-7020), and anti-PGC1 antibodies (IP: sc-13067) were purchased from Santa Cruz Biotechnology.
Urine Electrolyte Analysis
Mice were housed in metabolic cages (Shinano Seisakusho) for 24 h to collect the urine samples for analysis of the sodium, potassium, chloride levels, and osmolality using PVA-EX II (A&T) and an osmometer (model 3320; Advanced Instruments, Inc.), respectively.
Human kidney PT (HK-2) cells were purchased from Lonza and maintained at 37°C in a humidified 5% CO2 incubator. The cells were grown in DMEM/Hams F12 medium (Gibco) supplemented with 10% FBS, 5% antibiotics, and 2.5 mmol/L glutamine and subcultured to 80% confluence using 0.05% trypsin-EDTA (Gibco).
In experiments, HK-2 cells were cultured in DMEM/F12 (Biowest) with low or high glucose concentrations (5 or 30 mmol/L, respectively), with or without 50 mmol/L PHZ overnight. The cells were harvested after stimulation with 100 μmol/L 8CPT-cAMP (Abcam) and 100 nmol/L dexamethasone (Wako), with or without 100 nmol/L insulin or supplemented with or without 1 mmol/L glutamine, for 4 h (31,32). The NADH/NAD+ ratio in the HK-2 cells and the glucose concentrations in the medium were measured using the NAD+/NADH Quantification Colorimetric Kit (BioVision) and the Glucose Assay Kit (Cell Biolabs, Inc.), respectively.
Small interfering RNAs targeting murine FoxO1, Sirt1, and PGC1α were purchased from Ambion (Silencer Select Predesigned siRNA: ASO224JQ, ASO224JW, and ASO224JR, respectively; Thermo Fisher Scientific) and transfected into cultured cells using Lipofectamine RNAiMax Reagent (Life Technologies) according to the manufacturer’s instructions.
Data are expressed as means ± SEM, and statistical significance was set at P < 0.05 (one asterisk) or P < 0.01 (two asterisks). Differences between two groups were assessed by unpaired two-tailed t tests, unless otherwise indicated, whereas those among three or more groups were assessed by one-way ANOVA with post hoc Tukey honest significant difference in EZ-R (33).
Insulin Signaling and Its Related Gene Expression in the Renal Cortex in the Fasting and Fed States
First, we investigated insulin signaling in the renal cortex in the fed state in wild-type mice. As the serum insulin levels as well as blood glucose levels became higher (Fig. 1A), tyrosine phosphorylation of the IRSs, as well as phosphorylation of downstream molecules, including Akt and FoxO1, was enhanced (Fig. 1B). Feeding and treatment with an excess dose of insulin enhanced phosphorylation of the insulin signal cascade to an equivalent degree (Fig. 1B). The expression of the gluconeogenic genes, including PEPCK and G6Pase, was downregulated in the renal cortex (Fig. 1C). Similar changes were seen in the liver, as previously reported by us (Supplementary Fig. 1A and B) (16). These data suggest that the expression levels of the gluconeogenic gene expression could be inversely associated with insulin signaling in the renal cortex in the fed state.
Expression of the IRSs and Gluconeogenic Enzymes in the PTs
To explore the insulin actions in the kidney, we used laser microdissection (LMD) to investigate the precise distribution of the insulin signaling-related genes in the kidney, which was poorly understood, even though the IRs had been reported to be abundantly expressed in the renal cortex (17–19). We identified and isolated the glomeruli, PTs, and distal tubules (DTs) morphologically in PAS-stained sections prepared from wild-type mice (Fig. 2A), and successfully confirmed rich expression of the distribution markers, namely, of podocin in the glomeruli, megalin in the PTs, and aquaporin 2 in the DTs (Fig. 2B) (34). Among the insulin signal-related genes IRS1 and IRS2, as well as IR, were expressed in all three segments at expression levels that were roughly equivalent to those found in the liver. Gluconeogenic enzymes, SGLTs and GLUT2 among the GLUTs, were abundantly expressed in the PTs (Fig. 2C).
Insulin Mediates Gluconeogenesis in the PTs
These data prompted us to investigate the roles of insulin signaling in the regulation of gluconeogenesis in the PTs using genetically modified mouse models. We generated mice with IRS1/2 knockout specifically in the PTs, because we intended to explore the roles of insulin signaling by focusing on IRS1 and IRS2 as key molecules of the insulin-signaling cascade. First, we generated SGLT2-cre Tg mice to express Cre recombinase in the PTs, with the construct prepared according to a previous report (Supplementary Fig. 2A) (28). The SGLT2-cre mice appeared to show efficient expression of the Cre recombinase in the PTs, whereas no expression was detected in any of the other tissues (Supplementary Fig. 2B and C). We then crossed the Tg mice with IRS1/2-floxed mice, which yielded PT-specific IRS1/2 double-knockout (SIRS1/2DKO) mice. Both the IRS1 and IRS2 mRNA levels were significantly lower in the PTs, but not in the glomeruli or the DTs, as isolated by LMD, of the SIRS1/2DKO mice (Fig. 3A). In the renal cortex, at the protein level, expression of IRS1 and IRS2 was also reduced (Fig. 3B). Besides, insulin-mediated tyrosine phosphorylation of the IRS1 and IRS2 proteins and, consequently, phosphorylation of downstream molecules Akt and FoxO1 were markedly attenuated (Fig. 3B), suggesting the existence of an insulin signaling defect in the PTs of the SIRS1/2DKO mice.
The SIRS1/2DKO mice showed scarce differences in the BW (Fig. 3C), blood glucose levels (Fig. 3D), urine electrolytes (Supplementary Fig. 3A), and histological findings (Supplementary Fig. 3B). Although these mice failed to exhibit glucose intolerance (Supplementary Fig. 3C), they showed elevated blood glucose levels after administration of glutamine (Fig. 3E) and pyruvate (Fig. 3F), but not after administration of alanine, a major substrate in the liver (Fig. 3G). The SIRS1/2DKO mice also showed systemic insulin resistance (Fig. 3H) and consistently elevated gluconeogenic gene expression in the PTs (Fig. 3I). These phenotypes were not seen in the SGLT2-cre Tg mice (Supplementary Fig. 2E–G) or PT-specific IRS1 or IRS2 single-knockout mice (Supplementary Fig. 4). These data suggest that insulin signaling, mediated by both IRS1 and IRS2, suppresses gluconeogenesis in the PTs and affects systemic glucose metabolism.
SGLT Inhibition Modulates Gluconeogenic Gene Expression
Next, we investigated the gluconeogenic gene expression in other mouse models: ob/ob mice, a model of hyperglycemia and hyperinsulinemia (Fig. 4A and B), and mice treated with STZ, a model of hyperglycemia and hypoinsulinemia (Fig. 4C and D). In the ob/ob mice, gluconeogenic gene expression was downregulated in the renal cortex in the fed state, although they were upregulated rather than downregulated in the liver (Fig. 4B). Moreover, we analyzed the latter model just 10 days after the treatment to eliminate secondary effects and found that, contrary to the findings in the SIRS1/2DKO mice, the gluconeogenic gene expression was downregulated in the renal cortex in the fed state, although they were upregulated in the liver (Fig. 4D). These data suggest the existence of mechanisms other than insulin signaling that regulate the gluconeogenic gene expression profile unique to the PTs.
We then focused on glucose, another potential regulator of gluconeogenesis, whose uptake is mediated by the SGLTs in the PTs, and analyzed the primary impact of SGLT inhibition by a single administration of an SGLT1/2 inhibitor, PHZ. The administration to the STZ-treated mice reduced the blood glucose levels to almost those found in the control mice, without elevating the serum insulin levels (Fig. 4E). Interestingly, treatment with PHZ restored the gluconeogenic gene expression to levels similar to those seen in the untreated control mice (Fig. 4F). Similar results were observed in Akita mice, a model of chronic hypoinsulinemia and hyperglycemia (Supplementary Fig. 5).
We further explored the impact of SGLT1/2 inhibition in the SIRS1/2DKO mice, a model of suppressed insulin actions in the PTs and normoglycemia. A single administration of PHZ lowered the blood glucose and plasma insulin levels in a similar manner in both the SIRS1/2DKO mice and floxed mice (Fig. 4G). As expected, the enhanced gluconeogenic gene expression in the SIRS1/2DKO mice was further enhanced by the PHZ treatment (Fig. 4H). These data suggest that the gluconeogenic gene expression in the PTs can be modulated not only by insulin signaling, but also by the reabsorbed glucose via the SGLTs, and also that the impact of the latter could be larger than that of the former under certain conditions.
Insulin and Glucose Suppress Gluconeogenic Gene Expression via FoxO1 and Sirt1/PGC1α
We then explored the molecular mechanisms underlying the regulation of the gluconeogenic gene expression by insulin and glucose in vitro. In HK-2 cells, a cell line derived from human PT cells, insulin promoted Akt phosphorylation (Fig. 5A) and suppressed PEPCK expression (Fig. 5B). Knocking down of FoxO1 mimicked the suppressed expression of PEPCK seen under the condition of insulin stimulation (Fig. 5A and B).
We also found that high glucose levels suppressed PEPCK expression in the HK-2 cells, which was reversed by treatment with PHZ (Fig. 5C). Moreover, the glucose levels were elevated in the medium supplemented with glutamine under a low-glucose condition, but not a high-glucose condition, suggesting that the gluconeogenesis in the PTs is glucose dependent (Fig. 5D). We focused on Sirt1, because it has been reported to promote gluconeogenesis in the liver by deacetylation of PGC1α under the condition of a reduced NADH/NAD+ ratio (22). In the HK-2 cells, the NADH/NAD+ ratio was elevated in the presence of high glucose levels, which was reversed by PHZ treatment (Fig. 5E). Although high glucose levels did not alter the expression levels of Sirt1, they enhanced acetylation of PGC1α, suggesting suppression of the activity of Sirt1 as a deacetylase, which was abrogated by PHZ treatment (Fig. 5F). Consistent with these data, acetylation of PGC1α was enhanced in the renal cortex of the STZ-treated mice, which was abrogated by PHZ treatment (Fig. 5G). Knockdown of either Sirt1 or PGC1α replicated the suppressed expression of PEPCK seen under high glucose conditions (Fig. 5H and I). These data suggest that gluconeogenic gene expression was regulated by not only insulin but also glucose via suppression of the activities of FoxO1 and Sirt1/PGC1α, respectively.
Although the kidney has been known to play an important role in systemic glucose metabolism, the mechanism of regulation of gluconeogenesis in the PTs remains poorly understood. In this study, we suggest novel mechanisms unique to the PTs of the kidney: gluconeogenesis could be collaboratively regulated by both insulin signaling and the reabsorbed glucose in the PTs. In the fasting state, suppressed insulin signaling transduced by the IR/IRSs expressed on the basolateral side increases FoxO1 activity. In parallel, decreased glucose reabsorption via SGLT2 on the luminal side downregulates the NADH/NAD+ ratio, activating Sirt1 and PGC1α. Therefore, both pathways result in enhanced gluconeogenesis. Inversely, in the fed state, gluconeogenic gene expression is suppressed by both activated insulin signaling and increased glucose reabsorption (Fig. 6).
Just as in the liver, insulin signaling in the PTs is activated in the fed state: the IRSs and Akt are phosphorylated, followed by suppressed gluconeogenic gene expression. Moreover, our findings in the microdissected tissues indicate that the expression levels of IRS1 and IRS2 in the PTs are similar to those in the liver, suggesting that they could play important roles in glucose homeostasis. To explore their roles in the PTs, we successfully generated SGLT2-cre Tg mice and then PT-specific IRS1/2 DKO mice. Indeed, residual expression of the IRSs and phosphorylation of Akt were detected by Western blotting in the renal cortex, but this finding could be attributable to inclusion of cells other than the PT cells, as was argued for the case of the γ-glutamyl transferase–driven PT-specific IR knockout mice (17). Still, gluconeogenic phenotypes were clearly observed in the SIRS1/2DKO mice, suggesting that reduced expression, even if not amounting to complete knockout of the IRSs, is sufficient to disrupt physiological glucose metabolism in the PTs.
The knockout mice showed renal and systemic insulin resistance and enhanced renal gluconeogenesis, phenotypes that were compatible with the phenotypes of γ-glutamyl transferase–driven PT-specific IR knockout mice (17), suggesting that insulin signaling through IRS1 and IRS2 in the PTs plays important roles in systemic glucose homeostasis. These phenotypes are not seen in PT-specific IRS1 or IRS2 single-knockout mice, suggesting that each of the IRSs can compensate for the lack of the other in the PTs. This redundancy is not seen in insulin-mediated sodium reabsorption in the PTs, in which IRS2 plays a pivotal role (35,36), or in the regulation of glucose and lipid metabolism in the liver, in which IRS1 and IRS2 show functional differences (16), although FoxO1 is shared as the key downstream molecule of insulin signaling.
In another model of impaired insulin action in the PTs in the STZ-treated mice, we analyzed gluconeogenic gene expression relatively soon after the treatment and found that it was suppressed in the fed state, in clear contrast to the case in the liver, in which it was enhanced in the fed state. Similar results were obtained also in Akita mice, a model of chronic impaired insulin action. Existence of mechanisms other than insulin signaling was suggested, and we focused on the differences in the blood glucose levels among the three models: SIRS1/2DKO mice showed almost normoglycemia, whereas the STZ-treated mice and Akita mice showed hyperglycemia. We further hypothesized that the gene expression could be affected by the glucose reabsorbed via the SGLTs in the luminal membrane in parallel with primary urine glucose. Acute glucose administration and SGLT1/2 inhibition modulated the gene expression and activity of the gluconeogenic enzymes, as reported previously (5,37,38). Together with the results of the experiments involving a single PHZ treatment of the SIRS1/2DKO mice and those involving HK2 cells treated with PHZ, these data show the importance of glucose reabsorption in the regulation of gluconeogenic gene expression in the PTs, sharing the downstream molecules Sirt1 and PGC1α in response to the inward efflux of glucose with the liver.
Such dual regulation of gluconeogenesis could have developed because the PTs are a unique tissue that releases glucose into the general circulation via two energy-consuming mechanisms, namely gluconeogenesis and glucose reabsorption. PTs are different from the liver in that they reabsorb glucose to prevent external glucose loss in the urine. Theoretically, to release a single molecule of glucose into the bloodstream, glucose reabsorption mediated by SGLT1 and SGLT2 requires only two out of three and one out of three molecules of ATP (39), respectively, whereas gluconeogenesis requires six molecules of ATP. It is reasonable that in the fed state, which is associated with a high risk of external glucose loss, ATP-consuming gluconeogenesis is suppressed, not only by insulin signaling, but also by glucose reabsorption. It is not until the fasting state is reached that the inability to reabsorb sufficient glucose, together with inactivated insulin signaling, promotes the ATP-consuming gluconeogenesis. In this context, the PTs are in clear contrast with the liver, which is equipped with only a single mechanism of glucose release, namely gluconeogenesis, which seems to be regulated mainly by insulin.
Another difference between PTs and the liver lies in how to handle ammonia produced in breakdown of glucogenic amino acids for gluconeogenesis. In the PTs, production of ammonia is important in maintenance of acid–base homeostasis, and excessive ammonia is discharged directly into urine (40,41), whereas in the liver, ammonia is converted to urea by consuming additional ATP. That could account for the reason why the PTs prefer glutamine, which contains an amino group in addition to an amide group and produces two molecules of ammonia, and the liver prefers alanine, which lacks an amino group and produces a single molecule of ammonia. Moreover, the PTs could have acquired such preference to cope with a prolonged fasting, which is characterized not only by lowered plasma glucose levels but also acidosis because of enhanced ketogenesis, because glutamine can be used for efficient production of ammonia to maintain the acid–base balance, as well as gluconeogenesis. In this sense, the PTs could have played much larger roles in survival through the maintenance of homeostasis in the prehistoric era when the humans were forced to cope with severe fasting.
In clinical practice, SGLT2 inhibitors are drawing attention as a new class of antidiabetic agents (42). Our data, however, show that SGLT inhibition actually enhances gluconeogenic gene expression in the PTs via interfering with the physiological glucose-sensing mechanism of the PTs and generating a discrepancy between the primary urine glucose and reabsorbed glucose levels. Recently, we reported that administration of an SGLT2 inhibitor enhanced the gluconeogenic gene expression in the liver (43). Precursors of gluconeogenesis are known to be increased in diabetes (32) and possibly further increased by SGLT2 inhibition via hypercatabolism in the skeletal muscle (38). Therefore, these progluconeogenic aspects of SGLT2 inhibitors, possibly contributing to prevention of hypoglycemia, should be kept in mind when using the drug in clinical practice.
Moreover, our data also suggest potential mechanisms underlying hypoglycemia in chronic renal failure (CRF), which is a clinically important issue. Hyperinsulinemia is known to be caused in CRF, probably because of lowered renal insulin clearance, which could suppress the gluconeogenic gene expression in the PTs. Moreover, Sirt1 was reported to be inactivated in renal diseases (44), which could also suppress the expression. ATP-consuming gluconeogenesis might be further impaired by inability of the hypoxic PTs in CRF to produce ATP (45), altogether resulting in decreased insulin requirement and elevated risk of hypoglycemia in CRF.
Taken together, these findings suggest that gluconeogenesis is dually regulated in the kidney by both the reabsorbed glucose via the SGLTs in the luminal membrane and insulin signaling transduced from the basolateral membrane in the PTs. Our findings provide a novel insight into the unique mechanisms underlying the regulation of gluconeogenesis in the PTs.
Acknowledgments. The authors thank Ayumi Nagano, Eishin Hirata, Ayumi Ohuchi, Yuka Kobayashi, Yuko Kanto, Ritsuko Hoshino, Yoshiko Ito, and Naoki Ishikawa, all of whom are at The University of Tokyo, for excellent technical assistance and assistance with the animal care and especially Katsuyoshi Kumagai, formerly at The University of Tokyo and currently at Tokyo Medical University, for generating the Tg mice. The authors also thank Prof. Ryuichi Nishinakamura (Kumamoto University) for advice on designing the SGLT2-cre Tg mice and Drs. Yu Ishimoto and Motonobu Nakamura (The University of Tokyo) for advice on isolation of segments of the kidney.
Funding. This work was supported by a grant for Translational Systems Biology and Medicine Initiative, Creation of Innovation Centers for Advanced Interdisciplinary Research Areas Program of the Ministry of Education, Culture, Sports, Science and Technology of Japan (MEXT), and a MEXT Grant-in-Aid for Scientific Research (B) (15H04847 to N.K.).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. M.S. developed the hypothesis, designed and performed the experiments, analyzed the data, and wrote the manuscript. T.S. developed the hypothesis, designed the experiments, generated and validated the Tg mice, analyzed the data, and wrote the manuscript. N.K. developed the hypothesis, designed the experiments, analyzed the data, and wrote the manuscript. Y.S. validated the Tg mice. I.T., T.Ku., R.I., G.S., M.G., K.U., M.N., and T.J. designed the experiments. T.Ka. developed the hypothesis, designed the experiments, analyzed the data, and reviewed and edited the manuscript. T.Ka. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.