Fasting-Induced Protein Phosphatase 1 Regulatory Subunit Contributes to Postprandial Blood Glucose Homeostasis via Regulation of Hepatic Glycogenesis

Descriptions of the detailed protocols used in many assays of this study are provided in the Supplementary Data.

All animal protocols were approved by the Institutional Animal Care and Use Committee of the Institute for Nutritional Sciences, Chinese Academy of Sciences. Male C57BL/6 J mice were from Shanghai Laboratory Animal Co. Ltd.

The full-length mouse cDNA of PPP1R3C, PPP1R3G, and PP1c was cloned from mouse liver by RT-PCR and confirmed by DNA sequencing. PPP1R3C, PPP1R3G, and PPP1R3G with deletion of putative glycogen binding domain PPP1R3G(ΔGB) (deletion of the amino acid residues 225 to 247) were subcloned into the mammalian expression vector pRc/CMV with a Flag epitope tag added to the N-termini. PP1c was cloned into the mammalian expression vector pCS2+MT with six Myc tags at the N terminus. PPP1R3G was also cloned into the pGEX-4T1 vector to generate glutathione-S-transferase (GST)-fused protein. PP1c was also cloned into the pET-30a vector to generate His-fused protein. Cell culture and transfection were as previously described (18).

The antibodies were purchased as follows: monoclonal anti-Flag antibody from Sigma-Aldrich (St. Louis, MO); antibodies against Myc, tubulin from Santa Cruz Biotechnology (Santa Cruz, CA); antibodies against PP1c and GS from Cell Signaling Technology (Boston, MA); antibody against phosphorylated GS at Ser641 from Abcam (Boston, MA); Alexa Fluor 488 donkey anti-mouse IgG, Alexa Fluor 546 goat anti-mouse IgG, and Hoechst 33342 from Molecular Probes (Eugene, OR). The polyclonal PPP1R3G antibody was generated by mouse immunization with a GST-fused PPP1R3G protein. Immunoblotting and coimmunoprecipitation assays were previously described (18).

Isolation of primary hepatocytes and adenovirus infection was previously described (19). Flag-tagged PPP1R3G adenovirus was generated using the AdEasy Adenoviral Vector System (Stratagene, La Jolla, CA) according to the manufacturer’s instructions. Adenoviruses with either scrambled sequence or short hairpin (sh)RNA sequence specific for mouse PPP1R3G were generated using the BLOCK-iT Adenoviral RNAi Expression System (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions.

Cell glycogen content measurement was performed as previously described (20). Liver glycogen content measurement was described previously (21).

Mice were fasted overnight and injected intraperitoneally with glucose (2 g/kg body wt) containing a trace amount of D-[3-³H]glucose (Perkin Elmer Life and Analytical Sciences, Waltham, MA; specific activity 740 GBq/mmol) at 0.5 μCi/g body wt. One hour later, the mice were killed and liver glycogen was isolated using an ethanol-based method (21). The glycogen precipitate was dissolved in H₂O, and an aliquot was used in scintillation counting.

GS activity assay was performed as previously described (14,22). GP activity assay was previously described (14,23).

All data were analyzed by two-tailed Student t test.

To gain a global view of genes regulated by fasting in the liver, we examined genes of the whole mouse genome by mRNA microarray (Supplementary Table 1). Unexpectedly, PPP1R3G, a predicted glycogen-targeting regulatory subunit of PP1 (13), was found to be highly upregulated by fasting. By RT-PCR analysis, PPP1R3G was expressed in many mouse tissues with a relatively high level in liver, brain, lung, white adipose, and adrenal gland (Supplementary Fig. 1A). As confirmed by real-time RT-PCR, PPP1R3G mRNA level was increased to ∼12-fold and 35-fold in the liver after fasting for 6 and 12 h, respectively (Fig. 1A). As a positive control, FGF21, which was reported to be highly induced by fasting (24), was also elevated in our experiment (Fig. 1A). The protein level of PPP1R3G was upregulated consistently by fasting in the mouse liver (Fig. 1B) using an antibody generated in our laboratory (Supplementary Fig. 1B and C). However, none of the other three glycogen-related tissues, including brain, white adipose, and skeletal muscle, had an induction of PPP1R3G on fasting at the mRNA and protein levels (Fig. 1C, Supplementary Fig. 1D). PPP1R3G mRNA level was quickly reduced after refeeding, reaching the prefasting level after 2-h feeding (Fig. 1D). PPP1R3G protein was also significantly downregulated after refeeding (Fig. 1E). Collectively, these data demonstrate that PPP1R3G is a cyclic gene that changes along with the fasting–feeding cycle in the mouse liver.

To identify the molecular mechanism that underlies the regulation of PPP1R3G expression, we analyzed the effects of three major signals/hormones—glucagon, dexamethasone, and insulin—that have been shown by others to regulate other fasting-related gene expression (25). In primary mouse hepatocytes, treatment of glucagon or dexamethasone alone could not significantly elevate PPP1R3G at both protein and mRNA levels (Fig. 1F and G). However, when the cells were treated with these two factors together, PPP1R3G was significantly elevated at both protein and mRNA levels (Fig. 1F and G). We next analyzed the activity of a putative PPP1R3G promoter that contained a 2-kb fragment in the 5′ region upstream of PPP1R3G coding region. In concert with the change at the protein and mRNA levels, the PPP1R3G promoter activity was also significantly induced by dexamethasone and glucagon (Fig. 1I). We found that PPP1R3G mRNA level and PPP1R3G promoter activity were both reduced by insulin (Fig. 1H and J), indicating that insulin signaling likely plays a role to turn off PPP1R3G expression during refeeding.

To confirm whether PPP1R3G is indeed a glycogen-related regulatory subunit of PP1, we first investigated whether PPP1R3G and PP1c were colocalized in the cell. When the catalytic subunit of PP1 (the α isoform of PP1c, the same for other experiments used in Fig. 2) was expressed alone, it was localized in the cytoplasm and nuclei (Fig. 2A), consistent with a previous report (26). However, when coexpressed with PPP1R3G, the cytoplasmic localization of PP1c in the cytoplasm was significantly increased with a profound colocalization with PPP1R3G (Fig. 2A). Endogenous PP1c and endogenous PPP1R3G also had a profound cytoplasmic colocalization in primary hepatocytes isolated from overnight-fasted mouse (Fig. 2B).

We used GST pull-down and coimmunoprecipitation assays to analyze the interaction between the two proteins. We found that the purified PPP1R3G GST fusion protein could pull down PP1c in vitro (Fig. 2C). At the in vivo level, these two proteins could also interact with each other in a coimmunoprecipitation assay when they were coexpressed in HEK293T cells (Fig. 2D). As a positive control, a known PP1 regulatory subunit, PPP1R3C (27), could also interact with PP1c in the assay (Fig. 2D). Next, we explored whether the interaction is direct. We found that the purified His-tagged PP1c protein was able to interact with the purified PPP1R3G GST fusion protein in vitro (Fig. 2E). At the animal level, the interaction of endogenous PPP1R3G with endogenous PP1c in the liver of fasted mice was confirmed (Fig. 2F). In addition, the β isoform of PP1c was also able to colocalize with PPP1R3G (Supplementary Fig. 2A) and interact with PPP1R3G (Supplementary Fig. 2B and C).

To determine whether PPP1R3G is localized in the glycogen-enriched pellet (GEP) fraction within the cell, we fractionated PPP1R3G-transfected HepG2 cells by ultracentrifugation to obtain cytosolic and GEP fractions from post-nuclear supernatant (28). PPP1R3G could be found in both post-nuclear supernatant and GEP fractions (Fig. 2G), and deletion of the predicted glycogen-binding motif led to complete loss of GEP localization of PPP1R3G (Fig. 2G). Furthermore, PPP1R3G could be detected in the supernatant fraction after GEP fractions were treated with α-amylase and sedimentation, further indicating glycogen association of PPP1R3G (Supplementary Fig. 2D).

We next investigated the functional role of PPP1R3G in glycogen synthesis. We constructed an adenovirus that could overexpress PPP1R3G in mouse hepatocytes (Fig. 3A). In vivo staining of glycogen in the primary hepatocytes and HepG2 cells revealed that the cells with high expression of PPP1R3G were associated with an increase of glycogen content (Supplementary Fig. 3A and B). Direct measurement of glycogen also demonstrated that PPP1R3G dose-dependently increased the glycogen content in primary hepatocytes (Fig. 3B). However, when the predicted glycogen binding motif of PPP1R3G was deleted, the glycogen-stimulating effect was completely abrogated (Supplementary Fig. 3B and C).

Although overexpression of PPP1R3G itself was able to elevate glycogen content in the absence of glucose in the cultured cells (Fig. 3C), PPP1R3G-induced elevation of glycogen synthesis was more effective with increasing the level of glucose than the cells without PPP1R3G overexpression (Fig. 3C), indicating that PPP1R3G-regulated glycogenesis is dependent on available glucose substrate. On the other hand, it has been proposed that glycogen-targeting subunits of PP1 are involved in insulin regulation of glycogen synthesis (29,30). We found that insulin could increase glycogenesis in primary hepatocytes with or without PPP1R3G overexpression (Fig. 3D). Meanwhile, forskolin could reduce glycogen content in the presence of overexpressed PPP1R3G (Fig. 3E). To further confirm that PPP1R3G is able to elevate glycogenesis, we used small interfering RNA strategy to silence the expression of endogenous PPP1R3G. Two of three shRNA sequences could significantly downregulate PPP1R3G expression at both the mRNA level (Fig. 3F) and protein level (Fig. 3G). Consistently, these two shRNA sequences reduced glycogen content in primary hepatocytes (Fig. 3H). We also analyzed glucose dependence of PPP1R3G shRNA (#962) on glycogenesis and found that PPP1R3G knockdown could reduce glycogen content at various glucose concentrations (Fig. 3I). Collectively, our data indicate that PPP1R3G is a regulatory subunit of PP1 to facilitate stimulation of glycogen synthesis in hepatocytes.

We next analyzed the in vivo function of PPP1R3G on the regulation of glycogen synthesis and blood glucose homeostasis. We first used recombinant adenovirus to deliver PPP1R3G-expressing plasmid to the mouse liver. Previous studies have shown that intravenous administration of adenoviral vectors in mice almost exclusively targets the transgene to the liver (31). C57BL/6 J mice were injected with the control or PPP1R3G-expressing adenovirus. Evaluation of aminotransferases in the serum indicates that the PPP1R3G-expressing adenovirus did not cause apparent functional damage to the liver (Supplementary Fig. 4A).

By immunoblotting assay and real-time RT-PCR, we confirmed that the PPP1R3G-expressing adenovirus was able to drive abundant expression of PPP1R3G in the mouse liver (Fig. 4A and B). The PPP1R3G mRNA level was elevated by the adenovirus to ∼12-fold (Fig. 4B), comparable to the level induced by fasting (Fig. 1A). Consistent with the observation with cultured hepatocytes, the liver glycogen content of mice infected with PPP1R3G-expressing adenovirus was significantly elevated, reaching to approximately threefold higher than in control animals (Fig. 4C). We next performed glucose tolerance tests with the animals. PPP1R3G overexpression led to a decrease in fasting blood glucose level and an increase in glucose clearance rate (Fig. 4D), with the area under curve (AUC) of glucose tolerance test reducing by ∼20% (Fig. 4E). However, insulin tolerance test revealed no difference between the two groups of mice (Fig. 4F), indicating that insulin sensitivity is not altered by PPP1R3G overexpression in vivo. Collectively, these data suggest that overexpressed PPP1R3G could increase clearance of blood glucose likely by increasing conversion of blood glucose into liver glycogen.

To support our hypothesis that fasting-induced PPP1R3G is involved in postprandial blood glucose homeostasis, we injected the mice with recombinant adenovirus that contains shRNA specific for PPP1R3G, and the adenovirus did not cause apparent functional damage to the liver (Supplementary Fig. 4). When PPP1R3G was silenced in the liver, the fasting-induced PPP1R3G expression was largely abrogated (Fig. 5A). After 12-h fasting, the mice infected with PPP1R3G shRNA adenovirus had less hepatic glycogen content than the control mice (Fig. 5B). Refeeding for 1 h increased the hepatic glycogen content, and such increase was significantly attenuated when PPP1R3G was downregulated (Fig. 5B). Most important, the postprandial blood glucose clearance rate was altered when PPP1R3G expression was silenced. The glucose tolerance test revealed that the glucose clearance rate was significantly decreased in PPP1R3G-downregulated mice in comparison with the control animals (Fig. 5C), with AUC increasing by ∼50% (Fig. 5D). However, insulin sensitivity did not seem to be affected by downregulation of PPP1R3G (Fig. 5E).

To provide further evidence that PPP1R3G is directly involved in postprandial hepatic glycogenesis, we analyzed the amount of newly synthesized glycogen in the mouse by using ³H-labeled glucose as a tracer. Mice injected with control or PPP1R3G-shRNA adenovirus were fasted for 12 h, followed by intraperitoneal injection of glucose containing a trace amount of ³H -glucose. The mice were killed in 1 h, and the isolated liver glycogen was subjected to radioactivity measurement. As shown in Fig. 5F, the amount of newly synthesized glycogen in the liver was reduced to ∼50% by PPP1R3G knockdown. These data collectively indicate that PPP1R3G is directly involved in postprandial regulation of hepatic glycogenesis.

Because GS and GP are two key enzymes modulated by PP1-mediated dephosphorylation, we investigated whether PPP1R3G could influence the activities of these two enzymes. When PPP1R3G was overexpressed by adenovirus in the liver in the fed state, GS activity was markedly elevated (Fig. 6A), whereas GP activity was not affected (Fig. 6B). Because our results indicate that PPP1R3G mainly acts on GS instead of GP to regulate glycogenesis in the liver, we next focused on analyzing the effect of PPP1R3G on GS activity during the fasting–feeding cycle (Fig. 6C). In the fasting–feeding cycle, GS activity deceased ∼50% after fasting for 6 h and slightly elevated to the unfasted level in 24 h. However, GS activity was robustly increased on refeeding for 1 to 2 h and subsequently declined to a very low level in 12 h. This phenomenon is consistent with previous reports (32–34). We next investigated how PPP1R3G knockdown could affect GS activity during the fasting–feeding cycle (Fig. 6C). We found that the major effect of PPP1R3G knockdown is to reduce GS activity around the fasting–refeeding transition. During constant feeding and at early fasting or refeeding for 12 h, PPP1R3G knockdown only slightly reduced GS activity. However, the GS activity was decreased to >40% in the liver at fasting for 24 h and at refeeding for 1 to 2 h when PPP1R3G was silenced. Accordingly, we found that the phosphorylation level of GS at Ser641 in mouse livers was very high during fasting and robustly reduced by refeeding for 1 to 2 h (Supplementary Fig. 5), consistent with the observation that GS activity was robustly stimulated by refeeding (Fig. 6C). Furthermore, GS phosphorylation at Ser641 was elevated by PPP1R3G knockdown during refeeding (Supplementary Fig. 5), consistent with the finding that GS activity was reduced by PPP1R3G knockdown (Fig. 6C). Collectively, these data indicate that PPP1R3G mainly functions during the period of fasting–refeeding transition to regulate GS activity.

It is noteworthy that PPP1R3G is not the only glycogen-targeting regulatory subunit of PP1 to regulate hepatic glycogenesis. In addition to PPP1R3G, other glycogen-targeting regulatory subunits of PP1, such as PPP1R3B, 3C, 3D, and 3E, are also expressed in the liver (13,27,35,36). In the fed state, G_L/PPP1R3B accounts for ∼60% of GS phosphatase activity, and the PPP1R3C, 3D, and 3E account for the remaining activity (13,37). To investigate how these glycogen-targeting regulatory subunits orchestrate to control hepatic glycogenesis, we analyzed the expression patterns of PPP1R3B, 3C, 3D, 3E, and 3G at different times during fasting and refeeding (Fig. 6D). The mRNA of PPP1R3A and PPP1R3F was hardly detectible in the liver in our experiment (data not shown). Fasting for 24 h could reduce the mRNA levels of PPP1R3B, 3C, and 3D, whereas refeeding for 1 h significantly elevated the mRNA levels of these three subunits. The mRNA level of PPP1R3E seemed to be unaffected by fasting and refeeding. In contrast, the expression of PPP1R3G was markedly stimulated by fasting and rapidly reduced by refeeding. Taken together, these data indicate that the expression pattern of PPP1R3G differs from other glycogen-targeting subunits during the fasting–feeding cycle.

So far there are seven glycogen regulatory subunits (G subunits) of PP1 in mammals, PPP1R3A to PPP1R3G (11,13). Phylogenetic tree analysis reveals that although all seven human subunits and their rodent orthologs possess known or putative PP1-interacting and glycogen-binding domains, none of the subunits shares more than 40% amino acid identity, suggesting that each subunit may serve a nonredundant function in mammals (13). In this study, we demonstrate that PPP1R3G is indeed a glycogen-targeting regulatory subunit of PP1. At the cellular level, PPP1R3G is associated with glycogen pellet, interacts with the catalytic subunit of PP1, and regulates GS activity. At the animal level, PPP1R3G is able to regulate glycogen synthesis in the liver and modulate blood glucose homeostasis. Fasting glucose level is reduced when PPP1R3G is overexpressed in the liver. On the other hand, hepatic silencing of PPP1R3G reduces postprandial elevation of GS activity and slows down postprandial clearance of blood glucose. Collectively, our data reveal for the first time that PPP1R3G is a functional regulatory subunit of PP1 and plays a role in regulating GS activity, hepatic glycogenesis, and postprandial blood glucose homeostasis.

One of the most intriguing findings of this study is that PPP1R3G is involved in the regulation of hepatic glycogenesis in a manner coupled to the fasting–feeding cycle and distinct from other G subunits, especially G_L/PPP1R3B, a major protein that regulates liver glycogen metabolism. It was originally reported in the early 1970s that liver GSP is inhibited by GPa (38), explaining why GS becomes inhibited while GP is activated. It was later found that glucose binds to GPa and promotes its dephosphorylation and inactivation, thereby terminating the inhibition on GSP by GPa so that glycogen can be resynthesized when blood glucose is high after a meal (8). In the mid-1980s it became clear that a form of PP1 was the major hepatic GSP in the fed state and that this enzyme was inhibited allosterically by GPa (39–41). The liver GSP in the fed state was later purified and shown to be a complex of GL/PPP1R3B with PP1 (42). The allosteric binding site for GPa was found to be located on the G_L/PPPR3B subunit, and not the PP1 catalytic subunit (42), and the GPa-binding site was located at the extreme C-terminus of G_L/PPPR3B (43,44). Mice that expressed G_L/PPP1R3B(Y284F) mutant that could not by inhibited by GPa consistently showed enhanced activation of hepatic GS and conversion of blood glucose into hepatic glycogen (16). This information has been exploited to develop small molecule inhibitors to disrupt the interaction between GPa and G_L/PPP1R3B to enhance the conversion of blood glucose to hepatic glycogen (45). It is noteworthy that inhibition by GPa is the unique feature of G_L/PPP1R3B not shared by any other G subunit, including PPP1R3G. We propose that the lack of a GPa binding site in PPP1R3G, and therefore presumably the lack of allosteric inhibition of the PP1-PPP1R3G complex by GPa, comprises a crucial difference from the PP1-PPP1R3B complex. Such a difference may explain why PPP1R3G is needed at the fasting–feeding transition. During starvation, when the glucagon/insulin ratio is high, GP would be expected to be largely in the active form. At the early stage of the fasting–feeding transition when GPa has not been inactivated, one would not want GPa to inhibit GSP activity; otherwise, GS activity would not be activated efficiently and the glucose could not be rapidly used to replenish hepatic glycogen after a meal. At the early stages of the fasting–feeding transition, the expression of PPP1R3G reaches its maximum while G_L/PPP1R3B expression is minimal (Fig. 6D). Thus, PP1-PPP1R3G complex may function as the major GSP at this time. In the normally fed state, the PP1-PPP1R3B complex may replace the PP1-PPP1R3G complex as the major GSP so that the important allosteric regulation by GPa can be introduced. Consistent with our model, it was found that G_L/PPP1R3B accounts for ∼60% of GSP activity in the fed state (13,37). Because of the functional and expressional difference between PPP1R3G and other G subunits, especially G_L/PPP1R3B in the liver, it is expected that the major physiologic mission of PPP1R3G is to ensure rapid activation of GS and rapid glycogen synthesis in the liver shortly after a meal, subsequently contributing to postprandial glucose clearance (Fig. 6E).

When feeding triggers stimulation of GS activity via different means, such as a rapid increase of blood glucose and insulin levels, insulin-mediated phosphorylation and inactivation of GSK-3, translocation of GS to the cellular periphery, conversion of intracellular glucose to G6P, and portal signals (2,5,10,11), the PPP1R3G-mediated GS activity would rapidly lead to hepatic glycogenesis and removal of postprandial blood glucose. In a simplistic way, and as judged by the increase of AUC in the glucose tolerance test in PPP1R3G knockdown mice (Fig. 5D), the reduction of postprandial newly synthesized liver glycogen (Fig. 5F), and the decrease of GS activity by these mice shortly after refeeding (Fig. 6C), it can be estimated that at least 50% of the postprandial hepatic glycogen synthesis and the reduction in blood glucose are mediated by a PPP1R3G-mediated mechanism. In humans, hepatic glycogenesis is reduced in diabetic patients, and genetic variations of genes involved in glycogen metabolism have been found in diabetic patients (46–48). In mice, although Suzuki et al. (14) reported that deletion of G_M/PPP1R3A had no obvious defects, other reports indicate that deletion of G_M/PPP1R3A leads to increased weight gain, obesity, glucose intolerance, and insulin-resistance (15,49). Hepatic expression of a C-terminus–truncated form of G_M/PPP1R3A in streptozotocin-induced diabetic rats can reverse hyperglycemia and hyperphagia (50). Mice that expressed the G_L/PPP1R3B(Y284F) mutant that could not be inhibited by GPa had an enhanced activation of hepatic GS activity and improved glucose tolerance (16). In addition, heterozygous deletion of PTG/PPP1R3C in mice led to glucose intolerance, hyperinsulinemia, and insulin resistance with aging (17). Therefore, the next challenge will be to determine whether alteration of PPP1R3G is associated with insulin resistance and type 2 diabetes and whether modulation of PPP1R3G can serve as a new strategy to improve glucose metabolism.

This work was supported by research grants from the Chinese Academy of Sciences (KSCX2-EW-R-08), the National Natural Science Foundation of China (30830037 and 81021002), and the Ministry of Science and Technology of China (2007CB947100) to Y.C. The work was also supported by the Ministry of Science and Technology of China (2010CB529506 to Y.P. and Z.W.) and the National Natural Science Foundation of China (30971660 to Y.P.).

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

X.L. performed the experiments, analyzed data, and wrote the article. Y.Z., X.R., X.J., L.Z., X.W., and Q.D. performed the experiments. W.L., Y.P., and Z.W. contributed the reagents, material, and analysis tools. Y.C. analyzed data, wrote the article, and conceived and designed the experiments.

The authors thank Drs. Peter J. Roach and Anna A. DePaoli-Roach at the Indiana University School of Medicine for valuable suggestions and discussion; Qiong Wang and Lei Jiang at the Institute for Nutritional Sciences (INS) for technical assistance with adenovirus purification; Shanshan Pang and Haiqing Tang at the INS for assistance with tail-vein injection; Ting Mao and Mengle Shao at the INS for assistance with hepatocyte isolation; and Dr. Yong Liu at the INS for providing recombinant GFP adenovirus.