p38 Mitogen-activated Protein Kinase Plays a Stimulatory Role in Hepatic Gluconeogenesis*

Hepatic gluconeogenesis is essential for maintaining blood glucose levels during fasting and is the major contributor to postprandial and fasting hyperglycemia in diabetes. Gluconeogenesis is a classic cAMP/protein kinase A-dependent process initiated by glucagon, which is elevated in the blood during fasting and in diabetes. In this study, we have shown that p38 mitogen-activated protein kinase (p38) was activated in liver by fasting and in primary hepatocytes by glucagon or forskolin. Fasting plasma glucose levels were reduced upon blockade of p38 with either a chemical inhibitor or small interference RNA in mice. In examining the mechanism, inhibition of p38 suppressed gluconeogenesis in liver, along with expression of key gluconeogenic genes, including phosphoenolpyruvate carboxykinase and glucose-6-phosphatase. Peroxisome proliferator-activated receptor γ coactivator 1α and cAMP-response element-binding protein have been shown to be important mediators of hepatic gluconeogenesis. We have shown that inhibition of p38 prevented transcription of the PPARγ coactivator 1α gene as well as phosphorylation of cAMP-response element-binding protein. Together, our results from in vitro and in vivo studies define a model in which cAMP-dependent activation of genes involved in gluconeogenesis is dependent upon the p38 pathway, thus adding a new player to our evolving understanding of this physiology.

Gluconeogenesis in the liver is the process of de novo synthesis of glucose from non-hexose carbohydrate precursors such as lactate, pyruvate, alanine, and glycerol. This process plays a key role in maintaining blood glucose concentrations within a very narrow range during fasting (1,2). Gluconeogenesis is largely controlled by the balance between insulin and glucagon. In the fed state, plasma insulin levels are increased while glucagon levels are decreased (2)(3)(4). Insulin propels glucose uptake into peripheral tissues such as muscle and adipose tissue while serving as a brake to stop hepatic gluconeogenesis (3,(5)(6)(7)(8)(9). In contrast, in the fasted state insulin levels are diminished while glucagon levels are increased, leading to elevation of gluconeogenesis. In diabetes, the ability of insulin to suppress gluconeogenesis is lost due to absolute insulin deficiency in type I diabetes or is severely compromised in type II dia-betes due to insulin resistance and relative insufficiency of insulin production. As a result, hepatic gluconeogenesis becomes the major contributor to postprandial and fasting hyperglycemia in both forms of diabetes (see Refs. 3

and 4 for review).
Gluconeogenesis is a classic cAMP/protein kinase A (PKA) 3 -dependent process initiated by glucagon (see Ref. 9 for review). Although the molecular mechanism(s) that are responsible for the elevation of hepatic gluconeogenesis in diabetes have been extensively studied, much remains unresolved (reviewed in Ref. 4). It is generally understood that the increase of gluconeogenesis in diabetes is due to the unrestrained expression and activity of gluconeogenic enzymes (see Refs. 3 and 4 for reviews). Normally, insulin can inhibit gluconeogenesis through multiple mechanisms. First, in the pancreas insulin inhibits glucagon secretion from ␣-cells so as to promptly eliminate the dominant stimulator of gluconeogenesis (reviewed in Ref. 4). Second, in the liver insulin blocks the glucagon signaling mechanism by activating a cAMP phosphodiesterase. Third, insulin can directly suppress the transcription of key gluconeogenic genes, including phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase). Specifically, insulin blocks the recruitment of the transcriptional coactivators PGC-1␣ and CREB-binding protein to the promoters of the PEPCK and G6Pase genes (10). As a result, the transcription of gluconeogenic genes is strongly inhibited by insulin. Because insulin action is deficient in diabetes, all these suppressive mechanisms are lost, and as a consequence the stimulatory machinery of gluconeogenesis is activated. These include the cAMP/PKA-dependent expression of the PGC-1␣ gene, in addition to PEPCK and G6Pase (11,12). However, the cascade of signaling events downstream of PKA, including the possible role of other kinases, has not been explored.
We and others have previously shown that p38 is a downstream effector of PKA (13)(14)(15)(16)(17). More recently, we identified p38 as necessary for the control of energy balance by mediating PKA-dependent transcription of brown fat thermogenic genes, such as uncoupling protein 1 (UCP1) and PGC-1␣ (13,16,17). In this study, we have extended these observations in adipocytes to another metabolically important cell type, hepatocytes. Our results identify p38 as a component in the signaling mechanism for cAMP-dependent expression of gluconeogenic genes and hepatic gluconeogenesis.
Preparation of Primary Hepatocytes and Viral Infections-Primary hepatocytes were prepared as previously described (19 , 0.1 mM EGTA, 5.5 mM glucose, and 20 mM HEPES). The viability of hepatocytes was examined by trypan blue exclusion and was typically Ͼ95%. Hepatocytes were inoculated into collagen-coated 6-well plates (5 ϫ 10 5 /well) in Williams' medium. Cells were incubated for 24 h before any experimentation and were subsequently treated with 10 M forskolin or 100 nM glucagon in the presence or absence of 5 M SB as noted. For adenoviral infection, 100 active particles/cell were used. The viruses were prepared in 2% fetal bovine serum medium by standard techniques (18,20). At 36 h post-infection, levels of p38 in hepatocytes were detected by immunoblotting.
Immunoblotting-Tissue or cell lysates were prepared by homogenization and sonication, followed by addition of 2ϫ Laemmli sample buffer. Aliquots (5-10 g of protein/well) were resolved using mini Tris-glycine gels (4 -20%) (Invitrogen) and transferred onto nitrocellulose membranes. Levels of phosphorylated and total p38 or CREB were detected with a 1:1,000 dilution of each specific antiserum, followed by a 1:10,000 dilution of goat anti-rabbit immunoglobulin G conjugated with alkaline phosphatase (RPN5783; Amersham Biosciences). Fluorescent bands were visualized with a Typhoon PhosphorImager (Amersham Biosciences).
Transient Transfection and Promoter Assays-Hepa1c1c7 hepatoma cells were purchased from ATCC and maintained in minimum essential medium-␣ (Invitrogen). Cells were transiently transfected with Lipofectamine 2000 (Invitrogen) as noted and harvested to assay PGC-1␣ and PEPCK promoter activities 36 h after transfection with chloramphenicol acetyltransferase and luciferase assays, respectively. Results from these assays were normalized to the internal control, ␤-galactosidase. Glucagon (100 nM) was added for the last 6 h of the transfection to stimulate cAMP production. Chloramphenicol acetyltransferase, luciferase, and ␤-galactosidase assays were performed as described previously (21) or in manuals from manufacturers.
RNA Isolation, Semiquantitative RT-PCR, Taqman Real-time RT-PCR, and Northern Blotting-Total RNA from livers were prepared using RNA purification kits from Qiagen. RT-PCR reactions were performed according to manuals from the manufacturer (Applied Biosystems). The Taqman assays for measuring PEPCK and PGC-1␣ transcripts by real-time RT-PCR were purchased from Applied Biosystems (Mm00440636_m1; Ms0017330 -4). For Northern blot analysis, RNA was denatured, fractionated through 1.2% agarose gels, and transferred onto Biotrans nylon membranes (ICN) (22). Radiolabeled probes were prepared by random primer extension (Prime-It RmT; Stratagene) in the presence of [␣-32 P]dCTP to a specific activity of Ͼ2 ϫ 10 9 cpm/g of DNA. Blots were hybridized and washed, all as previously described (23,24).
Animal Experiments-The role of p38 in hepatic gluconeogenesis in vivo was examined in mouse models with or without diabetes mellitus. Male C57Bl/6J mice (8 weeks of age) were used in the experiments. Mice were fasted for 24 h as indicated, whereas those of the control groups were fed ad libitum. SB was administered via intraperitoneal injection twice during the fast at 12.5 mg/kg of body weight. The first dose was given at the beginning of the fast, and the second followed 12 h later. The dosage and regimen of SB administration were established in previous reports (16,25). A model of Type I diabetes was induced by administration of streptozotocin for 5 days at 50 mg/kg of body weight/day (26). For all experiments, plasma glucose levels were measured with a Hemocue glucose meter. Livers were collected for protein and RNA analyses. Levels of p38 and CREB phosphorylation and expression of PEPCK, G6Pase, and PGC-1␣ genes in the liver were assessed.
Measurements of Hepatic Glucose Production-Male C57BL/6J mice (8 weeks of age) were used in the experiments. The protocol used has been previously described (27). Briefly, mice were fasted for 24 h. Under anesthesia with pentobarbital (intraperitoneal injection, 30 mg/kg of body weight), a fine infusion tube was inserted into an external jugular vein through an intravenous puncture. Animals were allowed to recover for 30 min before any experiments. SB was administered twice via intraperitoneal injection at 12.5 mg/kg. One dose was administered at the beginning of the fasting, while another was 12 h later. A bolus D-[3-3 H]glucose (10 Ci; Amersham Biosciences) was administered, followed by continuous infusion at the rate of 0.1 Ci/min for 2 h. Blood samples (20 l) were collected at 0-, 60-, 90-, and 120-min time points. Plasma glucose levels were measured with a Hemocue glucose meter.
To determine plasma levels of [3-3 H]glucose, plasma was deproteinized with 10% trichloroacetic acid, dried to remove 3 H 2 O, resuspended in water, and counted by a scintillation counter. Hepatic glucose production (HGP) was calculated as follows: HGP ϭ R a Ϫ R i (27). R a is infusion rate of D-[3-3 H]glucose (DPM/plasma glucose-specific activity (DPM/ mol of glucose))/kg of body weight, and R i is exogenous glucose infusion rate.
Statistical Analysis-All the data were analyzed with either Student's t-test or two-way analysis of variance as indicated for each experiment.

Blockade of p38 Reduces Fasting Plasma Glucose Levels and
Expression of Gluconeogenic Enzymes-Because we recently showed that p38 is activated by cAMP/PKA in white and brown adipocytes and is necessary for transcription of the genes for the thermogenic uncoupling protein 1 (UCP1) and PGC-1␣ (13,16,17), we postulated that p38 might be similarly activated in the liver by glucagon during fasting to control the expression of gluconeogenic genes. To test this hypothesis, mice were pretreated with either saline or the p38 inhib-itor, SB, as described previously (13) and subjected to a 24-h fast. Control groups of mice fed ad libitum were similarly treated. As shown in Fig. 1A, phosphorylation of p38 was elevated by fasting and completely blocked by SB. As expected, fasting reduced plasma glucose levels significantly (p Ͻ0.05) (Fig. 1B), while treatment with SB caused further reduction (p Ͻ0.01) (Fig. 1B). SB had no effect on glucose levels in fed mice. As anticipated, expression of the key gluconeogenic genes, PEPCK, G6Pase, and PGC-1␣, was induced in the liver by fasting. However, this response was largely suppressed by treatment with SB (Fig. 2). Together, these results suggested a stimulatory role for p38 in hepatic gluconeogenesis.
Inhibition of p38 Reduces HGP-To directly examine the role of p38 in HGP, D-[3-3 H]glucose was infused into mice through an external jugular vein as detailed under "Materials and Methods" in the presence or absence of SB. As shown in Fig. 3, HGP was reduced 49 Ϯ 5.9% by SB treatment. These results support a role for p38 in gluconeogenesis in the liver.
Silencing of the p38␣ Gene in Liver Suppresses Expression of Key Gluconeogenic Genes and Reduces Fasting Plasma Glucose Levels-To confirm the role of p38 in hepatic gluconeogenesis by an independent method, we used siRNA against p38␣ encoded by adenoviruses to silence the expression of the p38␣ gene. (Because p38␣ is the predominant isoform in hepatocytes (Ref. 28 and data not shown), it was targeted in these experiments.) As shown in Fig. 4A, following 24 h of fasting, levels of total p38 proteins in the liver were reduced Ͼ80 Ϯ 4.4% by siRNA, whereas scrambled siRNA had no effect. When the p38␣ gene was silenced, transcripts of PEPCK and PGC-1␣ genes were decreased by 75 Ϯ 5.3 and 70 Ϯ 6.2%, respectively (Fig. 4, B and C), while plasma glucose levels were reduced by 21 Ϯ 8.7% (Fig. 4D). Scrambled siRNA had no effect. Liver aspartate aminotransferase was in the normal range (data not shown), indicating that liver function was not significantly damaged by adenoviral infection. Together, these results confirm a stimulatory role for p38 in expression of central gluconeogenic genes in liver.

Studies in Primary Hepatocytes Recapitulate the Results in Vivo-
To determine whether the regulation of gluconeogenesis by p38 occurs directly in hepatocytes, cells were freshly isolated from mouse liver and treated with glucagon in the presence or absence of either SB or siRNA against p38␣. As shown in Fig. 5A, infection of hepatocytes with adenoviral siRNA-p38␣ reduced p38 protein levels by more than 80% in comparison to untreated cells. The stimulation of PEPCK, G6Pase, and PGC-1␣ genes by glucagon was similarly suppressed by either SB or siRNA-p38␣ (Fig. 5B). Scrambled siRNA had no effect. Together, these results from primary hepatocytes support a stimulatory role of p38 in expression of gluconeogenic genes, consistent with observations from the in vivo studies.
p38 Plays a Stimulatory Role in PGC-1␣ Promoter Activation-To further study the role of p38 in transcription of the PGC-1␣ gene, a potent amplifier of hepatic gluconeogenesis (11), we examined the effect of p38 on PGC-1␣ transcription in a hepatoma cell line (hepa1c1c 7). The part of the PGC-1␣ promoter tested contains a functional consensus cAMP-response element (10,16). As shown in Fig. 6, the promoter was stimulated by glucagon but suppressed by either the catalytic    PGC-1␣ Phosphorylation by p38 Is Involved in Transcription of the PEPCK Gene-Phosphorylation of PGC-1␣ by p38 is required for its coactivator function in transcription of several genes (16,29,30). To determine whether this phosphorylation is also involved in tran-scription of the PEPCK gene in hepatocytes, the PEPCK promoter was introduced into hepa1c1c7 cells and stimulated by glucagon together with either the wild-type or a phosphorylation-deficient mutant of PGC-1␣ (29). As shown in Fig. 7, the PEPCK promoter was stimulated by glucagon, and overexpression of the wild-type PGC-1␣ further enhanced this stimulation. However, overexpression of the PGC-1␣ mutant actually suppressed glucagon stimulation of the promoter. These results are consistent with the idea that phosphorylation of PGC-1␣ by p38 is one of the events required for transcription of the PEPCK gene.
p38-dependent Phosphorylation of CREB-CREB has been recently identified as an important cAMP/PKA-dependent regulator of gluconeogenic gene expression (10,11). p38 appears to be a downstream effector of PKA, as we have shown in previous studies in adipocytes (13,16,17) and this study in hepatocytes. In addition, p38 can indirectly activate CREB (31)(32)(33). Therefore, we postulated that p38 might also be   Silencing of the p38␣ gene in liver by siRNA suppresses expression of key gluconeogenic genes and reduces fasting plasma glucose. As described under "Materials and Methods," adenoviruses encoding siRNAs (10 11 pfu/ mouse) against either p38␣ or a scrambled sequence (Scram) were administered to mice via tail vein injection. Five days after injection, mice were fasted for 24 h. Subsequently, livers were collected for detection of p38␣ protein by immunoblotting (A). Transcripts of PEPCK (B) and PGC-1␣ (C) genes in the liver were quantified with Taqman real-time RT-PCR. Fasting plasma glucose levels were measured with a glucose meter (D). Results represent two independent experiments, each with 5 mice/group. involved in activation of CREB in hepatocytes. To test this hypothesis, mice were either fed or fasted in the presence or absence of SB. As shown in Fig. 8A, CREB phosphorylation was elevated by fasting in the liver and suppressed by blockade of p38, indicating that p38 is a required component for CREB activation. To more directly study the role of p38 in activation of CREB, isolated hepatocytes were stimulated by glucagon or forskolin in the presence or absence of SB or siRNA-p38␣. As shown in Fig. 8B, phosphorylation of both p38 and CREB was stimulated by glucagon or forskolin; this stimulation was blocked by inhibition of p38. Together, these results support the idea that CREB is downstream of p38 in the control of gluconeogenesis. Interestingly, even though CREB is a known classic substrate for PKA, our results would indicate that there is a requirement at some level beyond PKA for p38.

Acute Inhibition of p38 Reduces Expression of Hepatic Gluconeogenic Genes and Plasma Glucose Levels in Streptozotocin-induced
Diabetes-One of the characteristic features of diabetes in animal models and in human is persistent elevation in hepatic gluconeogenesis, and this situation is further aggravated during fasting (see Refs. 34 and 35 for review). Therefore, we examined the role of p38 in expression of gluconeogenic genes under fasting conditions in mouse models of type I diabetes induced by streptozotocin. As shown in Fig. 9A, whereas fasting alone decreased plasma glucose levels by 28 Ϯ 5.8% (n ϭ 8, p Ͻ 0.05) they were further reduced to   . Acute blockade of p38 inhibits expression of gluconeogenic genes in liver of mouse models with type I diabetes. As detailed under "Materials and Methods," type I diabetes was induced by streptozotocin in mice. Diabetic mice were either fed ad libitum or fasted for 24 h in the presence or absence of SB treatment. A, plasma glucose levels were measured. B, livers were collected for detection of p38 (p38-T) and phospho-p38 (p38-P) by immunoblotting. C, transcripts of PEPCK, G6Pase, and ␤-actin genes in the liver were measured by semiquantitative RT-PCR. Results represent means Ϯ S.D. of two independent experiments, with 4 mice/group. *, p Ͻ 0.05 between ϪSB and ϩSB. **, p Ͻ 0.01 between ϪSB and ϩSB. DECEMBER 30, 2005 • VOLUME 280 • NUMBER 52 45 Ϯ 6.4% by blockade of p38 (n ϭ 8, p Ͻ 0.01). Phosphorylation of p38 and expression of PEPCK and G6Pase genes in the liver were all elevated in fasted diabetic mice and blocked by inhibition of p38 (Fig.  9, B and C).

DISCUSSION
Because of deficiency of insulin function, hepatic gluconeogenesis becomes unrestrained in diabetes, resulting in hyperglycemia. In addition, the function of gluconeogenic stimulators is enhanced in diabetes, further elevating blood glucose levels. We have identified p38 as a mediator of the cAMP-dependent gluconeogenic response by regulating the expression of the rate-limiting enzyme PEPCK and G6Pase genes, as well as transcription factors that control them. There had been observations in the literature variously associating p38 activation in hepatocytes with cAMP-generating hormones and transcription of the PEPCK gene (36 -38). However, until now the role of p38 in regulating gluconeogenic genes per se as part of the physiologic response to fasting had not been specifically addressed.
In considering potential targets of p38 in the liver, the transcriptional coactivator PGC-1␣ became an obvious candidate for the following reasons. It is recognized as an important player in the response from cAMP to gluconeogenesis (10,11), and we also recently demonstrated in brown adipocytes that both phosphorylation and transcription of PGC-1␣ are direct targets of p38 (16). In hepatocytes we found that, as in adipocytes, p38 activity was activated downstream of PKA in regulating expression of the PGC-1␣ gene. It is interesting that, although these findings suggest that regulation of PGC-1␣ gene transcription by p38 may be conserved in several cell types, the specific transcription factor targets of p38 differ between tissues. For example, in adipocytes the cAMP-dependent activation of p38 leads to phosphorylation of ATF-2, which utilizes the CRE in the PGC-1␣ gene to drive transcription (16), whereas in the liver the mediator is CREB. Some studies in hepatocytes have also shown that ATF-2 may be involved in the control of PEPCK promoter activity in cell lines (37,38). However we could find no evidence of ATF-2 activation in the liver during fasting (data not shown), and Herzig et al. (10) previously concluded that CREB was required for expression of PGC-1␣ and gluconeogenic genes in the liver. Although the role for ATF-2 was not directly examined in that study, it is also very unlikely that overexpression of a dominant-negative form of CREB (A-CREB) could have inhibited ATF-2 (39). In our studies, phosphorylation of CREB in the liver and isolated hepatocytes was elevated by fasting and blocked by inhibition of p38. It is still unclear at this point what specific steps lead to CREB phosphorylation from p38 and why p38 is required for CREB activation in this tissue when CREB is a known classic substrate for PKA.
In summary, we have observed that p38 plays a stimulatory role in hepatic gluconeogenesis. The activator of p38 function is most likely glucagon, which is elevated significantly in the blood during fasting and in diabetes. However, free fatty acids and cytokines such as interleukin 6 can also be potential activators of p38 (40 -43) because they are all reported to be elevated in the circulation at various stages of fasting and/or in diabetes (44 -48). A better understanding of the signaling pathways that can activate p38 in the liver should provide new insights into the unrestrained gluconeogenesis in diabetes. In addition to the stimulatory role in gluconeogenesis described here, p38 is better known for its role in a variety of inflammatory diseases (see Ref. 49 for review). Increasingly, diabetes is viewed as possessing an inflammatory basis (50 -53). Although it may seem that inhibition of p38 would be an appealing avenue for the management of diabetes, only by deciphering this pathway in greater detail may we be in a position to identify other unique targets for intervention.