MicroRNA-451 Negatively Regulates Hepatic Glucose Production and Glucose Homeostasis by Targeting Glycerol Kinase–Mediated Gluconeogenesis

Type 2 diabetes (T2D) is characterized by insulin resistance and abnormally elevated hepatic glucose production primarily from sustained gluconeogenesis (1–3), but the underlying mechanisms are poorly understood. Hepatic gluconeogenesis is induced by glucagon and glucocorticoids and inhibited by insulin through the transcriptional regulation of glucose-6-phosphatase (G6Pase) and PEPCK, two rate-limiting gluconeogenic enzymes (3,4). CREB, FOXO1, and peroxisome proliferator–activated receptor γ coactivator 1α (PGC-1α) are major transcription factors and transcriptional coactivators that mediate hormonal regulation of hepatic gluconeogenic enzyme expression (3,5–7). The expression or activity of gluconeogenic enzymes and transcriptional factors/coactivators are increased in the livers of humans with diabetes and diabetic rodents (8,9); knockdown or inhibition of these molecules significantly improves hyperglycemia in diabetic mouse models (10–12). Thus, inhibition of the gluconeogenic pathway in the liver could be a potential strategy for combating diabetic hyperglycemia.

MicroRNAs (miRNAs) are small endogenous noncoding RNAs that regulate gene expression through translational repression or degradation of target mRNAs. Recent studies of liver samples from diabetic animal models and humans with diabetes suggest that dysregulation of hepatic miRNAs may lead to T2D. Several miRNAs contribute to insulin resistance in vivo (13–17), but very few are involved in gluconeogenesis and T2D (17–19). miRNA-451 (miR-451) is dysregulated in multiple types of cancers and contributes to tumorigenesis, tumor progression, and metastasis by targeting various molecules (20–22). Recent studies have indicated that miR-451 may be involved in metabolic disorders. miR-451 is decreased in liver tissues of patients with nonalcoholic steatohepatitis and the corresponding mouse model. It negatively regulates fatty acid–induced inflammation through the LKB1/AMPK/AKT pathway by targeting CAB39, a component of the LKB1-STRAD-CAB39 complex (23). Microarray analysis has shown that miR-451 levels are elevated in the livers of diabetic and obese animal models (13,24). We also found that hepatic miR-451 is elevated in high-fat diet (HFD)–induced diabetic mice and db/db mice by real-time PCR (S. Zhuo, unpublished observation). However, whether miR-451 contributes to diabetes is unclear. We speculate that miR-451 might be involved in the regulation of hepatic glucose metabolism and T2D because miR-451 has been reported to be upregulated by glucose in glioma cells, and some target molecules and signaling pathways that mediate the effects of miR-451 on cancer cell growth and migration, such as 14-3-3 proteins, CAB39, and LKB1/AMPK and PI3K/AKT (25–28), are critical regulators of hepatic glucose metabolism and therapeutic targets of T2D (29–34).

In this study, we show that miR-451 is elevated in livers of HFD-induced diabetic mice and db/db mice and identified glycerol kinase (Gyk) as a target of miR-451. Through adenovirus-mediated gain- and loss-of-function studies, we demonstrate that miR-451 negatively regulates hepatic gluconeogenesis and glucose homeostasis. We further explored the signaling pathways involved in the regulation of hepatic gluconeogenesis by miR-451 and Gyk and examined the regulation of hepatic miR-451 and/or Gyk in vitro and in HFD-fed mice. Finally, we checked the effect of overexpression of miR-451 or knockdown of Gyk on glucose metabolism in the diabetic mouse models.

Male C57BL/6 and db/db mice were obtained from Shanghai Laboratory Animal Co. (Shanghai, China) and the Model Animal Research Center of Nanjing University (Nanjing, China), respectively. Eight-week-old mice were fed an HFD (D12492; Research Diets) or control chow (D12450J; Research Diets) for the same period. All animals were killed under an anesthetic condition, serum was stored at −80°C, and livers were collected and snap-frozen in liquid nitrogen for further analysis. Animal experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of the Institute for Nutritional Sciences, Chinese Academy of Sciences.

Recombinant adenoviruses expressing miR-451 pre-miRNA (Ad-451) or control scrambled short hairpin RNA (shRNA), miR-451 sponges (Anti-451), Gyk (Ad-Gyk), or green fluorescent protein (Ad-GFP) were prepared by using the AdEasy Adenoviral Vector System (Agilent) according to the manufacturer’s instructions and the literature (35). Recombinant adenoviruses expressing Gyk shRNA and control scrambled shRNA were generated by using the BLOCK-iT Adenoviral RNAi Expression System (Thermo Fisher Scientific). Eight- to 10-week-old C57BL/6 mice, HFD-induced diabetic mice, or db/db mice were injected with adenoviruses through the tail vein at 1 × 10⁹ plaque-forming units in 0.2 mL of PBS. All experiments were carried out within 1 week after injection.

For glucose tolerance test (GTT), pyruvate tolerance test (PTT), and glycerol tolerance test (GlyTT), mice fasted for 6 h were intraperitoneally injected with glucose, pyruvate, or glycerol, respectively, at 2 g/kg body weight. For the insulin tolerance test (ITT), mice fasted for 4 h were injected with insulin intraperitoneally at 0.75 or 1 unit/kg body weight. Blood glucose concentrations were measured before and after injection with a glucometer (FreeStyle; Abbott).

Serum insulin and glycerol levels were determined by an ELISA kit (Millipore) and an enzymatic glycerol assay kit (Applygen, Beijing, China), respectively. Liver glycogen content was measured as previously described (36).

Total RNA was extracted from tissues or cells using TRIzol reagent (Invitrogen) and treated with RNase-free DNase to degrade contaminating genomic DNA. cDNA was synthesized with 2 μg of total RNA using Moloney murine leukemia virus reverse transcriptase and oligo (dT) primers; 500 ng of total RNA was reverse-transcribed by using an miR-451–specific stem-loop primer (5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGAACTCAGT-3′). Quantitative real-time PCR (qRT-PCR) was performed by using an ABI Prism 7900 sequence detection system (Applied Biosystems) with SYBR Green PCR Master Mix (Applied Biosystems). Transcriptional levels for mRNA and miR-451 were normalized to 36B4 and U6, respectively. The relative expression of mRNA or miR-451 was calculated using the 2^-ΔΔCT method. Primer sequences used for qRT-PCR are listed in the Supplementary Table 1.

Western blotting was carried out according to standard protocols. Target proteins were detected with SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific), quantified with Adobe Photoshop CS5, and normalized to total protein or β-actin. Primary antibodies used for Western blotting are listed in Supplementary Table 2.

Gyk luciferase reporter construct (Gyk wild type [WT]) was constructed by inserting a Gyk mRNA 3′ untranslated region (UTR) fragment containing the miR-451 binding site into a pRL-TK vector (Promega). Luciferase reporter constructs containing the mutated miR-451 binding site were generated by mutation of the miR-451 binding site from 5′-AACGGTT-3′ to 5′-AACGCTT-3′ (Mut1 Gyk) or 5′-AAGCGTT-3′ (Mut2 Gyk). Human embryonic kidney (HEK) 293T cells were transfected with 20 ng of luciferase reporter construct together with 20 nmol of miR-451 mimics or negative control oligos using Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific). A reporter plasmid encoding firefly luciferase was cotransfected for normalization purposes. Cells were collected 24 h after transfection and assayed using the Dual-Luciferase Reporter Assay System (Promega).

Primary hepatocytes were prepared by collagenase perfusion as described previously (37). Hepatocytes infected with Ad-GFP or Ad-Gyk for 24 h were cultured in serum-free DMEM overnight and stimulated with 100 nmol/L glucagon for 2 h to examine the expression of gluconeogenic genes with qRT-PCR. Hepatocytes cultured in serum-free DMEM overnight were treated with 100 nmol/L insulin, 10 nmol/L dexamethasone, or various concentrations of glucose for 4 h to examine the expression of miR-451.

Glucose production was measured as described previously with slight modification (38). Briefly, mouse primary hepatocytes infected with Ad-Gyk or Ad-GFP for 24 h were washed, incubated for 2 h in glucose-, L-glutamine–, phenol red–, sodium pyruvate–, sodium bicarbonate–free DMEM (Sigma) supplemented with or without gluconeogenic substrates (200 μmol/L glycerol only or 10 mmol/L sodium lactate and 2 mmol/L sodium pyruvate). Glucose in the medium was measured using a glucose oxidase kit (Sigma) and normalized with total cellular protein. Glucose production through gluconeogenesis was calculated by subtracting glucose production without gluconeogenic substrates from glucose production with gluconeogenic substrates.

All data are expressed as mean ± SEM. Unpaired two-tailed Student t test was used to assess significance between control and treated groups. P < 0.05 was considered statistically significant.

We examined the expression of miR-451 in liver tissues of two mouse models of diabetes: HFD-fed mice and db/db mice. Mice fed an HFD for 12 weeks were obese and had hyperglycemia (Fig. 1A and B). Hepatic miR-451 expression was significantly increased in HFD-fed mice compared with control chow–fed mice (Fig. 1C). Similarly, hepatic miR-451 levels were higher in db/db mice than in WT mice (Fig. 1D). These data show that hepatic expression of miR-451 is significantly increased in dietary and genetic mouse models of diabetes.

To determine whether hepatic miR-451 is involved in glucose homeostasis, we overexpressed miR-451 (Ad-451) in liver tissues of C57BL/6 mice through adenovirus infection (Fig. 2A). miR-451 overexpression had no effect on body weight and food intake (Supplementary Fig. 1A and B). Both fed and fasting blood glucose levels were significantly lower in Ad-451–treated mice compared with control adenovirus (Ad-NC)–treated mice (Fig. 2B). Serum insulin levels and insulin sensitivity were not different between these mice (Fig. 2C and D). GTT showed that exogenous glucose was cleared faster in Ad-451–treated mice than in Ad-NC–treated mice (Fig. 2E). We further checked hepatic glycogen metabolism and gluconeogenesis, two main contributors to blood glucose homeostasis, in Ad-451– and Ad-NC–infected mice and found hepatic glycogen content to be comparable (Fig. 2F). PTT showed that de novo hepatic glucose production in Ad-451 mice was lower than in control mice (Fig. 2G), indicating that hepatic overexpression of miR-451 inhibited gluconeogenesis. These results demonstrate that elevation of miR-451 in liver leads to hypoglycemia by inhibiting hepatic gluconeogenesis.

We further inhibited the function of hepatic miR-451 with miR-451 sponges expressed by adenoviruses (Anti-451). The blood glucose levels in Anti-451–treated mice slightly increased under the fed condition compared with control adenovirus–treated mice (Anti-null) (Fig. 2H). GTT and PTT showed that inhibition of miR-451 in mice impaired glucose tolerance and enhanced hepatic gluconeogenesis, respectively (Fig. 2I and J). Inhibition of miR-451 had no effect on serum insulin levels, hepatic glycogen content, body weight, and food intake (Fig. 2K and L and Supplementary Fig. 1C and D). These results support the finding that hepatic miR-451 plays an important role in glucose homeostasis through regulating hepatic gluconeogenesis.

To address the mechanisms underlying gluconeogenesis regulation by miR-451, we examined the effect of miR-451 on hepatic gluconeogenic gene expression by adenovirus-mediated overexpression of miR-451 or miR-451 sponges. Overexpression of miR-451 in mice significantly inhibited hepatic G6Pase and PEPCK expression (Fig. 3A) consistent with the inhibitory effect of miR-451 on hepatic glucose production (Fig. 2G). Conversely, inhibition of miR-451 function with miR-451 sponges in mice increased the mRNA levels of G6pase and PEPCK in liver tissues (Fig. 3B). These data reveal that miR-451 regulates hepatic gluconeogenesis by affecting the transcription of gluconeogenic genes.

CREB and FOXO1 are two transcription factors that control hepatic G6Pase and PEPCK expression, and PGC-1α is a coactivator of FOXO1. We found that overexpression of miR-451 in mice had no effect on hepatic CREB phosphorylation and PGC-1α expression (Fig. 3A and C) but significantly enhanced hepatic FOXO1 phosphorylation (Fig. 3D, left panel). Besides the reduction of G6pase and PEPCK expression, IGF binding protein 1 (IGFBP1), an additional downstream molecule of FOXO1, was decreased in the livers of miR-451–overexpressed mice (Fig. 3A). In contrast, inhibition of miR-451 in mice through miR-451 sponges greatly increased the hepatic expression of IGFBP1 (Fig. 3B) and decreased phosphorylation of FOXO1 (Fig. 3D, right panel). Similarly, no changes of CREB phosphorylation and PGC-1α expression were observed in the livers of miR-451–inhibited mice (Fig. 3B and C). These results indicate that hepatic miR-451 negatively regulates gluconeogenic gene expression through FOXO1. In the fed state, FOXO1 is mainly phosphorylated by AKT, a key molecule in the insulin signaling pathway. We found that miR-451 overexpression increased the phosphorylation of AKT and its downstream target glycogen synthase kinase 3β (GSK3β), but had no effect on phosphorylation of insulin receptor β (IRβ) in mouse liver (Fig. 3D, left panel). Further studies showed that inhibition of miR-451 in mouse liver suppressed the phosphorylation of AKT and GSK3β and did not affect IRβ phosphorylation (Fig. 3D, right panel). Collectively, these results indicate that miR-451 regulates the transcription of gluconeogenic genes through the AKT-FOXO1 signaling pathway.

We used bioinformatics tools (TargetScan, MiRanda, and DIANA microT-CDS) to predict the targets of miR-451. Among the putative targets, Cab39, 14-3-3ζ, and Gyk are all involved in glucose metabolism (26,29,30,39). Although Cab39 and 14-3-3ζ have been reported to mediate miR-451 functions other than gluconeogenesis (25–27,40), the relationship between miR-451 and Gyk is unknown. We found that adenovirus-mediated overexpression of miR-451 in mice had no effect on protein levels of Cab39 and 14-3-3ζ in liver tissues (data not shown), indicating that hepatic Cab39 and 14-3-3ζ were not targets of miR-451. Sequence alignment showed that a 3′ UTR fragment of Gyk mRNA is complementary to the seed region of miR-451 and exhibits high conservation among human, mouse, rat, chimpanzee, and rhesus (Fig. 4A). To investigate whether Gyk expression can be regulated by miR-451, pRL promoter–based Gyk 3′UTR reporter was cotransfected with miR-451 mimics or negative control oligos to HEK293T cells. miR-451 mimics significantly inhibited luciferase activity of WT Gyk 3′UTR reporter but has no effect on mutated Gyk 3′UTR reporters (Fig. 4B). Correspondingly, protein levels of Gyk were decreased by miR-451 mimics in HEK293 cells and the murine hepatic cell line AML-12 (Fig. 4C). In addition, adenovirus-mediated overexpression or inhibition of miR-451 in mouse liver showed that miR-451 negatively regulated hepatic Gyk expression at protein levels but had no effect on Gyk mRNA levels (Fig. 4D and E). These results indicate that miR-451 directly inhibits Gyk protein expression at the translational level.

Gyk is a rate-limiting enzyme that converts glycerol to glycerol 3-phosphate, which is the first step on the glycerol gluconeogenesis axis (39,41,42). After we identified Gyk as a target of miR-451 and found that miR-451 regulated the hepatic AKT-FOXO1-PEPCK/G6Pase pathway, we investigated whether Gyk could regulate both the glycerol gluconeogenesis axis and the AKT-FOXO1-PEPCK/G6Pase pathway. We found that overexpression of Gyk (Ad-Gyk) in mouse liver tissues (Fig. 5A and G) reduced blood glycerol levels and increased blood glucose levels in a random fed state (Fig. 5B and C). Both GlyTT and PTT showed that de novo hepatic glucose production was higher in Ad-Gyk mice than in Ad-GFP mice (Fig. 5D and E), indicating that Gyk enhanced gluconeogenesis by promoting glycerol metabolism and using both glycerol and pyruvate as precursors. Compared with Ad-GFP–infected mice, Ad-Gyk–infected mice had higher mRNA levels of G6Pase, PEPCK, and IGFBP1 (Fig. 5F), lower levels of phosphorylated AKT and FOXO1 (Fig. 5G), and comparable mRNA levels of PGC-1α and phosphorylated CREB (Fig. 5F and G), indicating that hepatic Gyk promotes gluconeogenesis through both glycerol and AKT-FOXO1 gluconeogenic pathways. Gyk overexpression in mice had no effect on body weight and food intake (Supplementary Fig. 1E and F). In contrast with results of Gyk overexpression, knockdown of Gyk in mice with adenoviruses expressing Gyk shRNA enhanced phosphorylation of AKT and FOXO1 in liver tissues (Fig. 5H), inhibited G6Pase and PEPCK expression (Fig. 6I), and reduced glucose production by using pyruvate as a precursor (Fig. 5I). These results indicate that Gyk regulates gluconeogenesis through both the glycerol gluconeogenic axis and the AKT-FOXO1-PEPCK/G6Pase pathway. In mouse primary hepatocytes, overexpression of Gyk promoted glucose production by using glycerol or lactate and pyruvate as precursors (Fig. 5J), inhibited AKT and FOXO1 phosphorylation (Fig. 5K), and enhanced glucagon-induced G6Pase and PEPCK expression (Fig. 5L), further supporting that Gyk promotes hepatic gluconeogenesis through the AKT-FOXO1-PEPCK/G6Pase pathway. Altogether, these results indicate that Gyk regulates gluconeogenesis not only by promoting glycerol metabolism in the glycerol gluconeogenesis axis but also through modulating the AKT-FOXO1-PEPCK/G6Pase pathway.

To verify that miR-451 regulates gluconeogenesis through Gyk in vivo, we first examined the effect of miR-451 on glycerol metabolism and glycerol gluconeogenesis in mice. We found that blood glycerol levels were upregulated in miR-451 overexpressed mice and downregulated when miR-451 function was inhibited by miR-451 sponges (Fig. 6A). Overexpression of hepatic miR-451 in mice impaired glycerol gluconeogenesis (Fig. 6B). Because miR-451 directly regulates the expression of Gyk, which is involved in glycerol metabolism and glycerol gluconeogenesis, these results indicate that miR-451 inhibits hepatic glycerol metabolism and glycerol gluconeogenesis through Gyk. We next checked whether elevation of Gyk in mice could reverse the inhibitory effect of miR-451 on hepatic gluconeogenesis through the AKT-FOXO1-PEPCK/G6Pase pathway. We observed that overexpression of Gyk with miR-451 in mice totally reversed miR-451 elevation–induced hypoglycemia, inhibition of glucose production, and gluconeogenic gene expression as well as AKT and FOXO1 phosphorylation (Fig. 6C–F), indicating that miR-451 inhibits hepatic gluconeogenesis and the AKT-FOXO1 gluconeogenic pathway by targeting Gyk. We further checked whether knockdown of Gyk in mice could reverse miR-451 inhibition–induced upregulation of gluconeogenesis and found that knockdown of Gyk by adenovirus-mediated RNA interference in mice could decrease blood glucose levels and gluconeogenic gene expression in normal mice (Fig. 6G–I). These results show that the decreased expression of Gyk mimics the effect of increased miR-451 on glucose metabolism (Figs. 2B and 3A). Furthermore, the upregulation of blood glucose level and gluconeogenic gene expression by miR-451 inhibition could be reversed by Gyk RNA interference (Fig. 6H and I). Taken together, the data demonstrate that miR-451 regulates hepatic gluconeogenesis and glucose homeostasis through targeting Gyk-mediated glycerol and AKT-FOXO1 gluconeogenesis.

Because hepatic miR-451 negatively regulates gluconeogenesis and blood glucose levels in normal C57BL/6 mice through Gyk, forced expression of miR-451 in liver under diabetic conditions may be helpful in reducing blood glucose level. We checked this possibility by adenovirus-mediated overexpression of miR-451 in diabetic mouse models. Mice fed an HFD for 12 weeks showed a higher fasting blood glucose level, impaired glucose tolerance, enhanced gluconeogenesis, and decreased insulin sensitivity compared with normal chow-fed mice (Fig. 7A–C). The results of GTT, PTT, and ITT showed that infection of HFD-induced diabetic mice with control adenoviruses had no significant effect on blood glucose levels, glucose tolerance, glucose production, and insulin sensitivity (Fig. 7A–C). However, overexpression of miR-451 in HFD-fed mice significantly improved glucose tolerance, decreased gluconeogenic capacity, improved insulin sensitivity, decreased blood glucose and insulin levels as well as the expression of Gyk and gluconeogenic genes (G6Pase and PEPCK), and increased the phosphorylation of their upstream molecules AKT and FOXO1 in liver tissues (Fig. 7A–G). These results demonstrate that the elevation of miR-451 in HFD-induced diabetic mice could alleviate hyperglycemia and inhibit gluconeogenesis and the AKT-FOXO1 gluconeogenic pathway. We further found that knockdown of Gyk in HFD-fed mice significantly reduced blood glucose levels (Fig. 7H) and serum insulin levels (Fig. 7I), and improved glucose intolerance (Fig. 7J). These results indicate that Gyk plays an important role in HFD-induced diabetic syndrome. Taken together, these results from HFD-induced diabetic mice indicate that elevation of miR-451 could inhibit gluconeogenesis and alleviate hyperglycemia through Gyk.

We also examined the effect of miR-451 on glucose metabolism in db/db mice. Compared with WT control mice treated with Ad-NC, Ad-NC–treated db/db mice had hyperglycemia, enhanced gluconeogenesis, impaired glucose tolerance, and insulin sensitivity (Fig. 7M–O). Consistent with the effect of miR-451 on glucose metabolism in HFD-induced diabetic mice, overexpression of miR-451 in db/db mice significantly decreased blood glucose levels, increased glucose tolerance, inhibited gluconeogenic capacity, and decreased serum insulin levels (Fig. 7K–N and P). Forced expression of miR-451 in both HFD-induced diabetic mice and db/db mice had no significant effect on food intake and body weight (data not shown). Taken together, these results demonstrate that elevation of hepatic miR-451 in diabetic mice decreases blood glucose levels by inhibiting Gyk-mediated hepatic gluconeogenesis.

Because hepatic miR-451 is elevated in diabetic mice, which have hyperglycemia and hyperinsulinemia, we examined whether glucose and insulin could regulate hepatic miR-451 expression. We found that glucose and insulin significantly induced miR-451 expression in mouse primary hepatocytes, but dexamethasone had no effect on miR-451 expression (Fig. 8A and B). These results indicate that hepatic miR-451 may be upregulated by glucose and insulin in T2D and elevation of hepatic miR-451 may be helpful in lowering blood glucose levels. To check this possibility, we first examined the levels of hepatic miR-451 and Gyk in mice fed an HFD for various periods. We found that both miR-451 and Gyk mRNA levels increased during obesity progression. However, the protein levels of hepatic Gyk did not continue to rise after 1 month (Fig. 8C), indicating that Gyk expression was inhibited at the posttranscriptional level possibly by miR-451. We then examined the effect of miR-451 inhibition on glucose metabolism in db/db mice. We found that inhibition of miR-451 with miR-451 sponges tended to exacerbate hyperglycemia and significantly enhanced hepatic gluconeogenic gene expression in these mice (Supplementary Fig. 2). These results support that the increase of miR-451 in diabetic mice contributes to attenuating hyperglycemia.

In this study, we show that hepatic miR-451 is markedly increased in liver tissues of diabetic animal models and identified Gyk as a target of miR-451. We found that miR-451 regulated hepatic gluconeogenesis, glucose homeostasis, and the AKT-FOXO1-PEPCK/G6Pase pathway, which was conversely regulated and reversed by Gyk. miR-451 was upregulated by glucose and insulin in hepatocytes. Overexpression of miR-451 or knockdown of Gyk in diabetic mice significantly alleviates hyperglycemia and hyperinsulinemia. The findings indicate that miR-451 and Gyk play an important role in glucose homeostasis, and manipulating these molecules could provide novel opportunities for treating diabetes.

Liver gluconeogenesis plays a critical role in the maintenance of glucose homoeostasis (43,44). We have identified Gyk as a target of miR-451 both in vitro and in vivo and demonstrate that miR-451 negatively regulates hepatic gluconeogenesis and blood glucose levels through Gyk at both pre- and posttranslational levels. We demonstrate that miR-451 inhibits hepatic gluconeogenesis by downregulating Gyk, which limits glycerol-to-gluconeogenic flux. Furthermore, we found that miR-451 negatively regulates hepatic gluconeogenic gene (PEPCK and G6Pase) expression through the Gyk-AKT-FOXO1 signaling pathway.

Madiraju et al. (38) reported that mitochondrial glycerophosphate dehydrogenase (mGPD), a downstream enzyme of Gyk in the glycerol gluconeogenesis axis, regulates hepatic lactate and glycerol gluconeogenesis. Metformin suppresses gluconeogenesis by inhibiting mGPD (38). Glycerol normally is considered a minor gluconeogenic substrate (45), but the current results and those of Madiraju et al. suggest that enzymes in the glycerol gluconeogenesis axis, such as Gyk and mGPD, play important roles in regulating hepatic gluconeogenesis. Although mGPD regulates gluconeogenesis through modulating glycerol and lactate metabolism but has no effect on gluconeogenic enzyme expression (38), the current study demonstrates that Gyk regulates hepatic gluconeogenesis (using glycerol or other precursors) through affecting glycerol metabolism and the AKT-FOXO1-PEPCK/G6Pase pathway. Therefore, Gyk may regulate hepatic gluconeogenesis independent of mGPD and play a more extensive role than mGPD in gluconeogenesis and glucose homeostasis. To our knowledge, this study demonstrates for the first time that Gyk can regulate AKT-FOXO1–mediated PEPCK and G6Pase expression and plays an important role in glucose homeostasis. The study shows that miR-451 and Gyk regulated blood glucose levels, hepatic glucose production, the glycerol gluconeogenesis axis, and the AKT-FOXO1-PEPCK/G6Pase pathway in an opposite manner. Overexpression of Gyk in mice could reverse miR-451–induced phosphorylation of AKT and FOXO1, inhibition of gluconeogenic gene expression, and decrease blood glucose levels; knockdown of Gyk in mice reversed miR-451 inhibition-induced elevation of blood glucose level and enhancement of gluconeogenic gene expression. According to these results, we propose that Gyk, a target of miR-451, regulates hepatic gluconeogenesis through both the glycerol gluconeogenesis axis and the AKT-FOXO1-PEPCK/G6Pase pathway and mediates the function of miR-451 in gluconeogenesis and glucose homeostasis. The mechanisms underlying the inhibition of AKT phosphorylation by Gyk require further investigation. We also found that hepatic Gyk is upregulated in an HFD-induced diabetic mouse model. Knockdown of Gyk could significantly alleviate hyperglycemia and hyperinsulinemia and improve glucose tolerance. These results indicate that Gyk is a potential therapeutic target against diabetes.

miR-451 has been reported to be regulated by glucose availability in glioblastoma cells (26). Low glucose inhibits miR-451 transcription by phosphorylating AMPK, which in turn phosphorylates and inactivates OCT1, a direct transcription factor controlling miR-451 expression (46). We show that in addition to glucose, insulin could also induce miR-451 expression in hepatocytes. These results indicate that upregulation of hepatic miR-451 by glucose and insulin may be helpful for lowering blood glucose level in T2D. The molecular mechanisms underlying the regulation of hepatic miR-451 by glucose and insulin need further investigation. The current study shows that hepatic miR-451 expression is significantly upregulated in two diabetic animal models, which is consistent with previous microarray data from obese and diabetic mice (13,24). In HFD-fed mice, although hepatic miR-451 and Gyk mRNA kept increasing during the progression of obesity, the protein level of Gyk only increased in the first month and then remained stable. These results indicate that elevation of miR-451 in the livers of diabetic mice may be helpful in inhibiting Gyk expression and gluconeogenesis.

Because hepatic glucose output derived from glycerol and other glucogenic precursors is increased in T2D (45) and Gyk is upregulated in liver tissues of diabetic mice, we hypothesized that elevation of hepatic miR-451 in diabetes may be helpful in maintaining glucose homeostasis. As predicted, we found that overexpression of miR-451 in HFD-induced diabetic mice and db/db mice significantly inhibited hepatic Gyk expression and gluconeogenesis, alleviated hyperglycemia and hyperinsulinemia, and improved glucose tolerance and insulin sensitivity (Fig. 7). In contrast, inhibition of miR-451 in db/db mice exacerbated hyperglycemia and enhanced hepatic gluconeogenesis (Supplementary Fig. 2). These results support that miR-451 is a pharmacological target in the treatment of T2D.

In summary, this study demonstrates that miR-451 and Gyk are critical regulators for the hepatic glycerol gluconeogenesis axis, the AKT-FOXO1-PEPCK/G6Pase pathway, and glucose homeostasis and suggests miR-451 and Gyk as potential therapeutic targets for the control of hyperglycemia in diabetes. We have identified Gyk as the target of miR-451 and propose that miR-451 regulates hepatic glucose production through targeting the Gyk-mediated glycerol gluconeogenesis axis and AKT-FOXO1 gluconeogenic pathway (Fig. 8D).

Funding. This study was supported by grants from The Science and Technology Service Network Initiative Project, Chinese Academy of Sciences (KFJ-EW-STS-099, KFJ-EW-STS-031); The National Basic Research Program of China (973 Program) (2011CB504002); The Science and Technology Commission of Shanghai Municipality (13JC1404003); and the China Postdoctoral Science Foundation (2015M571609).

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

Author Contributions. S.Z. and Y.Le contributed to the experimental design and writing of the manuscript. M.Y., Y.Z., X.C., F.Z., N.L., P.Y., T.Z., H.M., S.W., and S.C. researched data. Y.Li provided the HFD-fed mouse liver samples and contributed to the discussion. Y.Le 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.