PDK4 Deficiency Suppresses Hepatic Glucagon Signaling by Decreasing cAMP Levels

The liver is an important organ for maintenance of glucose homeostasis throughout the starve-feed cycle. Hepatic gluconeogenesis is tightly regulated by the interwoven actions of insulin and glucagon, the hormones mainly affected by nutritional status. In the fasted state, glucagon is responsible for the increase in cAMP level and protein kinase A (PKA) activity (1). PKA, in turn, activates the transcription factor CREB by phosphorylation, which promotes expression of key regulatory gluconeogenic genes, including peroxisome proliferator–activated receptor γ coactivator 1α (Pgc-1α), Pepck, and glucose-6-phosphatase (G6pase) (2,3). Because increased glucagon secretion plays a pivotal role in the pathophysiology of type 2 diabetes (4), this signaling cascade is of considerable therapeutic interest.

Pyruvate dehydrogenase complex (PDC) plays a central role in cellular energy metabolism by catalyzing the conversion of pyruvate to acetyl-CoA via oxidative decarboxylation (5,6). This process is critically important in the maintenance of hepatic glucose homeostasis, as evidenced by changes in its activity according to the starve-feed cycle (7–9). In the fed state, for example, activated PDC increases glucose oxidation for the provision of acetyl-CoA for the citric acid cycle, fatty acid synthesis, and cholesterol synthesis. In the fasted state, however, PDC is inactivated by pyruvate dehydrogenase kinase (PDK)–induced phosphorylation, which prevents the conversion of the glucogenic compounds alanine, lactate, and pyruvate to acetyl-CoA, a nonglucogenic compound.

Four isoenzymes of PDKs (PDK1-4) are expressed in mammalian tissues. Among these, PDK2 and PDK4 are highly expressed in the liver, and especially PDK4 is known to be increased in diabetic conditions as well as in the fasting condition when glucagon action is augmented (10,11). Thus, several efforts have been made to unveil whether the hyperglycemia of pathologic conditions can be alleviated by modulating PDK4 activity. Indeed, PDK-knockout (KO) mice exhibit lower fasting blood glucose levels, which is generally believed to be due to reduced availability of precursors for hepatic gluconeogenesis (12,13). Likewise, the well-known PDK inhibitor dichloroacetate (DCA) is believed to lower blood glucose by the same mechanism (14).

To the best of our knowledge, however, the possibility that PDK4 deficiency or inhibition might also reduce expression of gluconeogenic enzymes has not been properly evaluated. Here we show evidence that besides its effect on limitation of substrate for gluconeogenesis previously observed in global PDK4-KO mice, liver-specific inhibition of PDK4 is sufficient to modulate glucagon-simulated hepatic gluconeogenesis by decreasing cellular cAMP level and, therefore, gluconeogenic gene expression.

Primary hepatocytes were prepared as described previously (15). For virus infection experiments, cells were infected with adenoviruses for 2–3 days in 10% complete 199 media and then treated with glucagon (G2044; Sigma-Aldrich) or 8-bromo (Br)-cAMP (B5386; Sigma-Aldrich). For DCA (347795; Sigma-Aldrich) experiments, cells were incubated with different doses of DCA in complete medium for 16 h and then treated with glucagon.

C57BL/6J male mice, db/db mice and control db/+ mice (8–10 weeks old) were acclimated in our animal facility for 1 week before experiments were initiated. To compare mRNA and protein levels, mice were fasted for 16 h and refed for 4 h. For DCA experiments with diet-induced obese mice, 4-week-old male mice were fed a high-fat–high-sucrose (HFHS) diet (D12231; Research Diets). DCA (300 mg/kg) and vehicle were delivered by intraperitoneal injection every day for the indicated number of days, and blood glucose levels were measured after 6-h fasting at 10-day intervals during the 50-day experiment. For liver-specific PDK4 overexpression or knockdown experiments, the indicated adenoviruses were injected via the tail vein. Body composition was measured with a minispec plus Live Mice Analyzer (LF50; Bruker). Livers were harvested after the mice were fasted for 16 h. All experiments were approved by the Deagu-Gyeongbuk Medical Innovation Foundation Institutional Animal Care and Use Committee (DGMIF-10630802-00).

After indicated adenovirus (target templates sequence: short hairpin green fluorescent protein [shGFP], 5′-GCATCAAGGTGAACTTCAAGA-3′; shPdk4, 5′-GGAAGGAATCAAAGCACTTTA-3′) (16) or 7-day DCA treatment, 16-h fasted mice were intraperitoneally injected with 2 g/kg of sodium pyruvate (P2256; Sigma-Aldrich). For glucose tolerance test, mice fasted for 16 h were intraperitoneally injected with 1.5 g/kg of d-(+)-glucose (G7528; Sigma-Aldrich). Tail vein blood was collected at 0, 15, 30, 60, and 120 min for the measurement of blood glucose.

Liver triglyceride was measured with a kit according to the manufacturer’s instructions (K622-100; BioVision), and plasma glucagon was measured with the Wako Glucagon ELISA Kit (297-57101; Wako).

G6Pase (−231/+57) promoter (provided by S.-H.K.) was prepared as previously described (17). Adenoviral G6Pase-luc (1.5 × 10⁹/plaque-forming units [pfu]) was delivered by i.v. injection to C57BL/6J male mice for 4–7 days. Mice fasted for 15 h were injected with 30 mg/mL firefly D-luciferin intraperitoneally 5 min before imaging. Mice were anesthetized with 2% isoflurane gas and imaged with an IVIS bioluminescence imaging system (Xenogen).

Transfections were conducted in AML-12 cells with Lipofectamine 2000 (52887; Invitrogen) with Plus Reagent (10964; Invitrogen), according to the manufacturer’s instructions. Reporter plasmids encoding the human G6Pase promoter (−1,227/+57) and rat Pepck promoter (−2,000/+73) were used as previously described (18). Wild-type CREB vector (15221; Addgene) and mutant CREB vector (15222; Addgene) were purchased from Addgene.

Mouse liver tissue and primary hepatocyte RNA were isolated using QIAzol (79306; Qiagen) and cDNA was synthesized from 2 μg total RNA using oligo dT primer (K1622; Thermo Scientific). Quantitative real-time PCR was performed on a Viia7 instrument (Applied Biosystems) using SYBR green reagent (4367659; ABI). Mouse primer sequences for real-time PCR are reported in Supplementary Table 1.

Western blot assay was performed using antibodies specific for PDK4 and PDC (provided by R.A.H.), phosphorylated (p)–PDH-E1α ser293 (AP1062; Calbiochem), p-CREB (9198; Cell Signaling), p-PKA substrate (5045; Cell Signaling), p-AMPK (2532; Cell Signaling), total CREB (9197; Cell Signaling), Hsp90 (4874; Cell Signaling), p-PDE4B (provided by M.H. Rider, Université catholique de Louvain and de Duve Institute, Brussels, Belgium), and total PDE4B (sc-25812; Santa Cruz Biotechnology).

Primary mouse hepatocytes were fixed on collagen type I–coated cover slips (GG-22-collagen; neuVitro) with 4% paraformaldehyde for 30 min. Cells were washed with filtered PBS, followed by incubation for 15 min with 0.1% Triton X-100 containing permeable solution. After permeabilization, cells were incubated overnight at 4°C with anti–p-CREB at a dilution of 1:400. Cells were washed with filtered PBS three times and then incubated for 2 h with secondary antibody (Alexa Fluor 488 or 568). The cells were further washed three times with PBS before the cover slips were mounted with mounting solution containing DAPI.

For glucose production assay, hepatocyte growth medium was replaced with Krebs-Ringer bicarbonate (KRB) buffer including 0.5% BSA, 10 mmol/L sodium lactate, and 1 mmol/L sodium pyruvate, with or without glucagon. The culture medium was collected after incubation of the plates for 4–6 h. Glucose formed by the cells was determined with a hexokinase/glucose-6-phosphate dehydrogenase (H4502 and G7877; Sigma-Aldrich) assay and normalized to total protein concentrations and incubation times.

cAMP was determined in primary hepatocytes with a kit from ABI (T1500) and in mouse livers with an ELISA kit (ab65355).

To establish steady-state labeling, cells were incubated for 2 h with 200 μmol/L BSA-conjugated U-[¹³C]palmitic acid (605573; Sigma-Aldrich) or 1 mmol/L U-[¹³C]sodium pyruvate (490717; Sigma). Before collection, cells were treated with glucagon for 15 min.

The liquid chromatography–tandem mass spectrometry analysis was performed as described previously (19). Briefly, analytes were separated on a Mastro C18 (3 μm particle size, length 150 mm, and inner diameter 2.0 mm) column by gradient elution using an HPLC Nexera instrument coupled with an LCMS-8060 mass spectrometer (Shimadzu, Japan).

Compared with the refed state, the expression of the mRNA for Pdk4 was greatly increased in the liver in the fasted state, along with increased expression of the gluconeogenic genes Pgc-1α, G6pase, and Pepck (Fig. 1A). Accordingly, the PDK4 protein level was higher in the livers of fasted mice (Fig. 1B), and compared with db/+ mice, hepatic Pdk4 mRNA and its protein were highly induced in diabetic db/db mice where gluconeogenesis is aberrantly reinforced (Supplementary Fig. 1A and B). We therefore, we checked whether cAMP and glucagon, which are augmented in the fasting condition and diabetes, affect PDK4 expression in primary mouse hepatocytes. As expected, gluconeogenic gene expression was significantly increased by the glucagon and 8-Br-cAMP challenge (Fig. 1C). Pdk4 mRNA expression was also markedly increased without increases in the mRNAs for the other PDK isoenzymes (Fig. 1C). The protein level of PDK4 and p-PDHE1α were increased by 8-Br-cAMP, suggesting that hepatic PDK4 expression is positively regulated by cAMP stimulation (Fig. 1D).

Next, hepatocytes were pretreated with PKA inhibitor H89, followed by cAMP stimulation. H89 markedly suppressed levels of p-PKA substrates, p-CREB , and PDK4 protein, suggesting that the hepatic PDK4 level is tightly coupled with the gluconeogenic signaling pathway (Fig. 1D). To further confirm PKA-CREB signaling is responsible for PDK4 transcription, we transfected wild-type and mutant CREB, which is mutated at serine 133 to alanine for cytosolic retention, and examined PDK4 promoter activity (Fig. 1E). Unlike wild-type CREB, mutant CREB failed to increase PDK4 transcription (Fig. 1E). This finding is also concordant with a recent report that direct binding of p-CREB to the PDK4 promoter is responsible for cAMP/PKA induction of PDK4 transcription in stromal fibroblasts (20).

To gain better insight into the role of PDK4 in hepatic gluconeogenesis, adenoviral-mediated shRNA against PDK4 (shPDK4) was injected into HFHS-fed mice. Successful knockdown of PDK4 in the liver was confirmed by a substantial decrease in Pdk4 mRNA level, whereas the mRNAs for other PDK isoenzymes remained unchanged (Fig. 2A). The amount of PDK4 protein in the liver was markedly decreased in shPDK4 mice, whereas PDK4 protein in the adipose tissue was preserved, suggesting specific targeting of the liver (Fig. 2B). Surprisingly, the expression levels of the mRNAs for Pgc-1α, G6pase, and Pepck were decreased by liver-specific PDK4 knockdown (Fig. 2A). Furthermore, luciferase activity driven by the G6Pase promoter was decreased (Fig. 2C). The pyruvate tolerance test confirmed that hepatic gluconeogenesis was attenuated in PDK4-knockdown mice (Fig. 2D). Glucose tolerance test results were not altered by PDK4 deficiency (Fig. 2E), ruling out the possibility that improved glucose tolerance contributed to reduced glucose levels in the pyruvate tolerance test. Furthermore, food intake and the lean-to-fat mass ratio were not altered by hepatic PDK4 deficiency (Supplementary Fig. 2A).

To delineate the underlying molecular mechanism by which PDK4 deficiency inhibits hepatic gluconeogenesis, primary mouse hepatocytes were infected with shPDK4. PDK4-deficient cells showed decreased glucose production (Fig. 3A) and blunted gluconeogenic gene expression in response to glucagon (Fig. 3B). The amount of p-CREB was diminished by PDK4 knockdown (Fig. 3C). Moreover, unlike control (shGFP) hepatocytes, p-CREB did not merge with DAPI staining in shPDK4 hepatocytes under glucagon stimulation, indicating that PDK4 deficiency blocked p-CREB entrance into the nucleus (Fig. 3D). Because PKA and CREB are regulated by cAMP, we measured cAMP levels and found them decreased in PDK4-deficient primary mouse hepatocytes with or without glucagon stimulation (Fig. 3E). PDK4 deficiency also decreased ATP levels and increased p-AMPK levels (Fig. 3F). Given that hepatic ATP is primarily synthesized by fatty acid oxidation (FAO), followed by oxidation of acetyl-CoA in the citric acid cycle (21), we asked whether FAO is altered by the amount of PDK4. U-[¹³C]palmitate-driven acetyl-CoA [M+2] and citrate [M+2] enrichment, which reflect FAO flux (Fig. 3G), were significantly decreased by PDK4 deficiency in the cultured primary hepatocyte (Fig. 3G). Taken together, these findings suggest that inhibition of PDK4 in the liver is sufficient to inhibit hepatic gluconeogenesis by decreasing cAMP levels and increasing AMPK phosphorylation.

A recent study reported that AMPK activation results in the activation of PDE4B by phosphorylation (22). To examine whether this mechanism should be considered in our working model, we measured Ser304 p-PDE4B by Western blot analysis. p-PDE4B and p-AMPK were positively correlated, suggesting activation of PDE4B by phosphorylation may be responsible for the decrease in cAMP levels in PDK4-deficient hepatocytes (Fig. 3F). Furthermore, when hepatocytes were challenged with 8-Br-cAMP rather than glucagon, the reduction in p-CREB signaling and the reduction in hepatic glucose production by PDK4 deficiency were cancelled (Fig. 3H and I). This finding suggests that the effect of hepatic PDK4 ablation on gluconeogenesis relies on regulation of cAMP levels.

To examine whether inhibition of hepatic gluconeogenesis could be observed by pharmacologic inhibition of PDK, HFHS-fed mice were treated with 300 mg/kg DCA by daily intraperitoneal injection for 50 days. DCA-treated mice showed significantly decreased 6 h-fasted blood glucose level (Fig. 4A). Of note, short-term (7-day) DCA administration was sufficient to decrease blood glucose (Fig. 4A and D) without affecting body weight and body composition (Supplementary Fig. 2B). We therefore conducted the rest of the experiments after the administration of DCA for 7 days. DCA downregulated mRNAs for Pgc-1α and Pepck but, paradoxically, not G6pase (Fig. 4B). Concordant with the findings with shPDK4-injected mice, DCA-treated mice showed marked attenuation of luciferase activity driven by the G6Pase promoter (Fig. 4C). Pyruvate tolerance test analysis showed that gluconeogenesis was markedly suppressed by DCA (Fig. 4D). Unlike shPDK4 mice, DCA treatment also improved glucose tolerance (Fig. 4E), presumably by increased PDH-mediated glucose oxidation in the muscle (12), which is consistent with previous findings noted in global PDK2/4 double-KO and PDK4-KO mice (10,23). The beneficial result on pyruvate tolerance by DCA may therefore reflect the combined effects of inhibition of gluconeogenesis in the liver and increased glucose oxidation in muscle.

In primary mouse hepatocytes, DCA inhibited glucose production in a dose-dependent manner (Fig. 5A). Increased gluconeogenic gene expression in response to glucagon was attenuated by DCA (Fig. 5B). This correlated with decreased levels of p-CREB and the phosphorylation of PKA substrates (Fig. 5C) and perturbed nuclear trans-localization of p-CREB (Fig. 5D). Furthermore, glucagon-stimulated production of cAMP was significantly abrogated by DCA in a dose-dependent manner (Fig. 5E). In line with the increased p-AMPK level in response to PDK4 deficiency, DCA treatment increased the p-AMPK level in a dose-dependent fashion (Fig. 5F). Furthermore, DCA also increased the p-PDE4B level concordantly with p-AMPK (Fig. 5F). We then determined the effects of the PDE4B inhibitor rolipram on DCA-treated hepatocytes. Inhibition of PDE4B not only increased cAMP levels but also restored downregulated PKA-CREB signaling of hepatic glucose production by DCA (Supplementary Fig. 3A–C).

Given that PDK4 expression was increased in db/db mice (Supplementary Fig. 1A), we generated adenoviruses for overexpression of PDK4 (Ad-PDK4) and a control mock adenovirus (Ad-Mock) to gain further insight into the pathologic role of PDK4 in hepatic gluconeogenesis. Successful hepatic PDK4 overexpression was achieved by tail vein injection of the adenovirus, as evidenced by increased Pdk4 mRNA (Fig. 6A). These mice showed significant upregulation of gluconeogenic gene expression (Fig. 6B). Importantly, luciferase expression driven by the G6Pase promoter was significantly upregulated in Ad-PDK4–injected mice relative to control mice (Fig. 6C). p-PDHE1α–to–total PDHE1α and CREB ratios were increased by PDK4 overexpression as well as phosphorylation of PKA substrates (Fig. 6D). These hepatic PDK4-overexpressed mice manifested no appreciable difference compared with control mice in body weight, body composition, or food intake (Supplementary Fig. 2C).

PDK4 overexpression in primary mouse hepatocytes also resulted in a significantly higher rate of glucose production (Fig. 7A) and gluconeogenic gene expression (Fig. 7B). Because FAO flux is decreased by PDK4 deficiency, we checked the carnitine palmitoyltransferase-1 (Cpt-1) mRNA level, which was significantly increased by PDK4 overexpression (Fig. 7B). In addition, phosphorylation of PKA substrates and p-CREB levels were increased (Fig. 7C). Ad-PDK4 hepatocytes showed increased p-CREB nuclear localization as analyzed by immunofluorescence staining (Fig. 7D). These findings were a consequence of the increased cAMP level by PDK4 (Fig. 7E). Intriguingly, PDK4-induced cAMP increment and subsequent glucose production were also observed without glucagon stimulation, suggesting that PDK4 overexpression overrides the need for hormonal stimuli (Fig. 7A and E). To further address the pathologic role of facilitated cAMP-PKA-CREB signaling, we compared this signaling pathway in db/+ and db/db mice. Livers from db/db mice showed augmented PKA-CREB signaling (Supplementary Fig. 1B) and contained higher cAMP concentrations (Supplementary Fig. 1C), wherein PDK4 levels were increased (Supplementary Fig. 1A). To further address the role of hepatic PDK4, we knocked down PDK4 in db/db mice (Supplementary Fig. 1D and E). PDK4 deficiency in the liver decreased the cAMP level in the liver without affecting the plasma glucagon level per se. Taken together with Figs. 3 and 5, the findings suggest that the glucagon antagonizing effect of PDK4 deficiency is achieved by acceleration of cAMP degradation without modulation of plasma glucagon level.

Because Cpt-1 mRNA expression was increased by PDK4 (Fig. 7B), and FAO flux was decreased by PDK4 deficiency in the cultured primary hepatocytes (Fig. 3G), we hypothesized that the effect of PDK4 on gluconeogenesis might be dependent on its ability to modulate FAO. Acetyl-CoA [M+2] and citrate [M+2] from U-[¹³C]palmitate were enriched by PDK4 overexpression in the cultured primary hepatocyte, confirming that increasing PDK4 abundance increases the FAO rate (Fig. 8A). The increase in FAO flux caused by PDK4 overexpression was normalized in the presence of the CPT-1 inhibitor etomoxir (Fig. 8A). Indeed, the glucose production rate correlated with FAO; that is, it was blocked by etomoxir (Fig. 8B). The PDK4-induced increase in ATP was cancelled by etomoxir, suggesting that increased ATP production in PDK4-overexpressed hepatocytes was driven by increased FAO (Fig. 8C). As expected, etomoxir completely eliminated the effect of PDK4 overexpression on the cAMP increment (Fig. 8D). These effects were also observed in control hepatocytes, suggesting that FAO-driven energy production is required for hepatic gluconeogenesis (Fig. 8B–D). We further investigated the causal relationship between FAO flux and the cAMP level by detecting changes in p-AMPK and p-PDE4B levels after CPT-1 inhibition. Inhibition of FAO restored p-AMPK and p-PDE4B levels as a result of the decreased ATP level and the corresponding increase in the AMP-to-ATP ratio (Fig. 8E). Collectively, these findings suggest that the effect of PDK4 on potentiation of cAMP-PKA-CREB signaling and inhibition of AMPK phosphorylation is largely dependent upon its ability to increase FAO flux and ATP synthesis during gluconeogenesis.

In the current study, we identified a novel and unexpected role of PDK4 in hepatic gluconeogenesis. Our results show that hepatic PDK4 expression is enhanced by glucagon or cAMP treatment of hepatocytes and that overexpression of PDK4 stimulates the cAMP-PKA-CREB pathway and, therefore, glucose production. Conversely, pharmacologic inhibition of PDK4 in vitro and in vivo reduces the expression of gluconeogenic genes and, therefore, hepatic glucose production. Importantly, we demonstrate that the effect of PDK4 on hepatic gluconeogenesis is largely dependent on modulating the cytosolic cAMP concentration. These changes were tightly coupled with the rate of FAO. Given that FAO coupled to the citric acid cycle is the primary source of ATP for hepatic gluconeogenesis, it can be concluded that hepatic PDK4 plays a critical role in fuel selection during gluconeogenesis, which controls the overall rate of the gluconeogenic pathway, including cAMP levels, PKA-CREB signaling, and gluconeogenic enzyme expression (Supplementary Fig. 4; graphical summary). Interestingly, inhibition of FAO by hepatic PDK4 deficiency did not affect the liver triglyceride level (Supplementary Fig. 2). In addition to the decrease in Cpt-1 expression, mRNA expression of lipogenesis-related genes, such as Pparγ, Scd1, and Acc was simultaneously downregulated in PDK4-deficient hepatocytes (data not shown).

The role of PDK4 in diabetes or insulin resistance was previously discussed in the context of its role in shutting down the pyruvate dehydrogenase complex to conserve the gluconeogenic substrates alanine, pyruvate, and lactate (6,24). Remarkable upregulation of PDK4 in skeletal muscle in response to starvation or diabetes increases the delivery of these gluconeogenic substrates to the liver. Here, however, we found that PDK4 also affects the rate of gluconeogenesis by modulating the enzymatic capacity for gluconeogenesis and FAO.

A dramatic decrease in the phosphorylated form of CREB in response to PDK4 deficiency was the most surprising and important observations of this study. The significance of CREB in the regulation of hepatic gluconeogenesis is well established (3). Liver-specific deficiency of p-CREB reduces blood glucose levels, an apparent consequence of reduced expression of gluconeogenic genes (2). Our results show that besides the known effects on the expression of gluconeogenic genes, inhibition of PKA-CREB signaling reduced PDK4 protein levels in hepatocytes (Fig. 1D). Because p-CREB has been shown to bind and activate the PDK4 promoter in fibroblasts (20), this finding is presumably due to reduced transcription of the PDK4 gene by p-CREB deficiency. Conversely, in tumor cells, PDK4 was reported to physically bind to CREB and maintain CREB stability (25). However, no effect of PDK4 deficiency on the amount of CREB was found in this study with normal cells, and mutant CREB did not increase promoter activity of PDK4 in the hepatocytes, suggesting a critical role of p-CREB in PDK4 gene transcription (Fig. 1E).

Glucagon stimulates hepatic gluconeogenesis by increasing cAMP levels. Metformin antagonizes glucagon signaling by inducing the accumulation of AMP, which lowers cAMP independently of AMPK by direct inhibition of adenylyl cyclase (26) and dependently on AMPK by activation of PDE4B (22).We show here that PDK4 deficiency likewise antagonizes glucagon signaling by inducing the accumulation of AMP, which lowers cAMP. Although direct inhibition of adenylyl cyclase by AMP may be involved, we document activation of PDE4B.

Altered production of ATP by FAO in response to the level of PDK4 expression is responsible for changes in AMP levels, given that FAO and glucose oxidation reciprocally regulate each another (27) and that PDK is an important negative regulator of glucose oxidation. This is further supported by the recent finding that acetamiprid, the insecticide, suppresses testosterone synthesis by decreasing testicular ATP level, and in turn, cAMP level and StAR expression, which is downstream of CREB (28). A manipulation that restored ATP synthesis also recovered cAMP level (28). This finding, in accordance with our finding, suggests that the cAMP level can be tightly coupled with the rate of ATP synthesis, and in turn, affect the PKA-CREB pathway.

One of the unexpected findings was the relationship between PDK4 and AMPK. Others have noted that intensification of the cAMP-PKA pathway by cAMP treatment increases ATP production and thereby decreases AMPK phosphorylation in Leydig cells (29). The 8-Br-cAMP dose dependently increased cellular ATP levels, and p-AMPK was reciprocally decreased in this model (29). This finding strongly suggests that the gluconeogenic and steroidogenic pathways, both of which require activation of PKA-CREB signaling, positively affect ATP synthesis and thereby negatively affect phosphorylation of AMPK.

We show preclinical evidence that PDK4 is an important therapeutic target of diabetes. Although novel small molecules targeting PDK2 are under development (30), no drug that specifically targets PDK4 is available.

Acknowledgments. The authors thank Dr. Xiaocheng C. Dong (Indiana University School of Medicine, Indianapolis, IN) for providing shGFP- and shPDK4-expressing adenovirus and Dr. Mark H. Rider (Université catholique de Louvain and de Duve Institute, Brussels, Belgium) for providing the Ser304 p-PDE4B antibody.

Funding. This research was supported by the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare (MOHW), Republic of Korea (HI16C1501); Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2017R1A2B3006406 and NRF-2016R1D1A1B03935408); and the Bio & Medical Technology Development Program of NRF and funded by the Korean government (Ministry of Science, ICT and Future Planning and MOHW) (NRF-2016M3A9B6902872).

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

Author Contributions. B.-Y.P., J.-H.J., Y.G., R.A.H., and I.-K.L. generated the hypothesis, designed the experiments, and wrote the manuscript. B.-Y.P., Y.G., H.J.H., J.-E.K., E.K.Y., W.H.K., and Y.H.J. performed the experiments. N.-H.J., S.-H.K., B.-G.K., L.H., K.-G.P., R.A.H., and I.-K.L. analyzed and discussed the data. I.-K.L. 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.