Hepatic Expression of a Targeting Subunit of Protein Phosphatase-1 in Streptozotocin-diabetic Rats Reverses Hyperglycemia and Hyperphagia Despite Depressed Glucokinase Expression*

Glycogen-targeting subunits of protein phosphatase-1 (PP-1) are scaffolding proteins that facilitate the regulation of key enzymes of glycogen metabolism by PP-1. In the current study, we have tested the effects of hepatic expression of GMΔC, a truncated version of the muscle-targeting subunit isoform, in rats rendered insulin-deficient via injection of a single moderate dose of streptozotocin (STZ). Three key findings emerged. First, GMΔC expression in liver was sufficient to fully normalize blood glucose levels (from 335 ± 31 mg/dl prior to viral injection to 109 ± 28 mg/dl 6 days after injection) and liver glycogen content in STZ-injected rats. Second, this normalization occurred despite very low levels of liver glucokinase expression in the insulin-deficient STZ-injected rats. Finally, the hyperphagia induced by STZ injection was completely reversed by GMΔC expression in liver. In contrast to these findings with GMΔC, overexpression of another targeting subunit, GL, in STZ-injected rats caused a large increase in liver glycogen stores but only a transient decrease in food intake and blood glucose levels. The surprising demonstration of a glucose-lowering effect of GMΔC in the background of depressed hepatic glucokinase expression suggests that controlled stimulation of liver glycogen storage may be an effective mechanism for improving glucose homeostasis, even when normal pathways of glucose disposal are impaired.

In recent years, we and others have expressed the various targeting subunits in mammalian cells in culture or in liver of intact animals, leading to new insights into their relative metabolic potencies and regulatory properties (5). This has included work on a novel targeting subunit construct, G M ⌬C, derived by truncation of the unique 735-amino acid C-terminal domain of native G M (6). We have found that when expressed in hepatocytes, the targeting subunits stimulate glycogen synthesis in the rank order G L Ͼ PTG Ͼ G M ⌬C Ͼ G M (6,7). Surprisingly, G M ⌬C, but not G L or G M , ameliorates glucose intolerance when expressed in livers of rats fed on a high fat diet (8). This appears to be explained by our finding that liver cells with overexpressed G M ⌬C maintain full responsiveness to glycogenolytic signals such as forskolin and low glucose, unlike cells with overexpressed G L or PTG, which have impaired responsiveness to these agents, resulting in accumulation of large amounts of glycogen in the fasted state (6,8,9).
In light of these encouraging findings, the current study has extended our investigation of the properties of G M ⌬C by its expression in liver of rats with streptozotocin (STZ)-induced diabetes. Three surprising observations have been made as a result of this work. First, we find that G M ⌬C expression in liver is sufficient to lower blood glucose levels and raise liver glycogen levels to normal in STZ-injected diabetic rats. Second, these corrective effects on hepatic glycogen metabolism and glucose homeostasis occur despite very low levels of liver glucokinase expression in the insulin-deficient animals. Finally, we also find that hepatic G M ⌬C expression reduces food intake to normal levels in STZ-diabetic rats, which are otherwise hyperphagic.

MATERIALS AND METHODS
Recombinant Adenoviruses-A recombinant adenovirus containing the cDNA encoding a truncated G M glycogen-targeting subunit isoform from which 735 C-terminal amino acids were removed (G M ⌬C) is termed AdCMV-G M ⌬C; its preparation has been described previously (6). A recombinant virus containing the cDNA encoding the full-length hepatic targeting subunit G L (AdCMV-G L ) has also been described (6). Both the G M ⌬C and G L inserts include a C-terminal FLAG tag for ready identification of the transgene products by immunoblotting (6). As a control, some animals received a virus containing the Escherichia coli ␤-galactosidase gene, termed AdCMV-␤GAL (10). These viruses were amplified and purified for injection into animals using previously described procedures (11).
Animal Studies-Male Wistar rats (Charles River Laboratories) weighing 250 -300 g were housed on a 12-h light-dark cycle and were allowed free access to water and standard laboratory chow (65% carbohydrate, 4% fat, 24% protein; Harlan Tekland laboratory diet 9100). These animals were injected with a single moderate dose of streptozotocin (60 mg/kg; Sigma S-0130) intraperitoneally, followed by daily monitoring of blood glucose levels in the ad libitum fed state, using an automated blood glucose analyzer (Hemocue AB; Angelholm, Sweden). Only those animals in which blood glucose rose to levels greater than 250 mg/dl within 3 days of STZ injection were studied further. Five days after STZ injection, 0.5 ϫ 10 12 particles of AdCMV-G M ⌬C, AdCMV-G L , or AdCMV-␤GAL were administered via tail vein injection to rats anesthetized with intraperitoneal Nembutal (50 mg/kg of body weight; Abbott). As an additional control, some animals received no viral injection. After viral administration, animals were individually caged for daily monitoring of food intake, body weight, and blood glucose levels. Six days after viral administration, animals were sacrificed in the ad libitum fed state or following 18 h of fasting. A blood sample was taken, and liver and muscle samples were excised, rapidly clamp-frozen in liquid nitrogen, and stored at Ϫ70°C for further analysis. In a separate set of STZ-injected and virus-treated animals, urine volume and glucose concentration was monitored by four 12-h collections of urine from individual rats and measurement of urine glucose using a glucose oxidase-based assay (Sigma).
Immunoblot Analysis and Glycogen Measurements in Liver Samples-Powdered liver samples were homogenized in lysis buffer (50 mM Tris-HCl, pH 7.5, 5 mM EDTA, 1% Triton X-100, and proteinase inhibitors) using a Polytron homogenizer (model PT10 -35; VWR Inc.), followed by two rounds of freeze-thawing. Cell lysates were centrifuged at 3,000 ϫ g for 5 min, and total protein concentration was measured by the Bradford method (12). Glycogen was measured by extraction in 10% trichloroacetic acid, precipitation with methanol, and digestion of glycogen to free glucose by incubation with 0.4 mg/ml amyloglucosidase, as previously described (13).
For analysis of G M ⌬C protein levels, samples were centrifuged at 8,000 ϫ g for 2 min, and 2 mg of total proteins were incubated with 10 g of anti-FLAG antibody in 1 ml of buffer A (phosphate-buffered saline, 2% bovine serum albumin, 5 mM EDTA, and 100 M phenylmethylsulfonyl fluoride) at 4°C for 1 h. Then 20 l of Ezview Red anti-FLAG M2 affinity gel (catalog no. F-2426; Sigma) was added and incubated at 4°C for 1 h. Samples were centrifuged at 8,000 ϫ g and washed three times with buffer. The pellets were mixed with 50 l of SDS-running buffer and boiled for 5 min. Samples were resolved on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. The membranes were incubated in blocking buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% bovine serum albumin) for 1 h and treated overnight at 4°C with rabbit polyclonal serum specific for G M (1) (a generous gift of Dr. Anna A. Depaoli Roach, University of Indiana Medical Center) at a dilution of 1:1000. The membranes were washed and subsequently treated with horseradish peroxidase-labeled antirabbit IgG secondary antibody at 4°C for 2 h. The protein-antibody complexes were visualized using an enhanced chemiluminescence detection kit (PerkinElmer Life Sciences). Glucokinase protein levels were analyzed as described previously (6), using a rabbit polyclonal antiglucokinase antibody (14) at a dilution of 1:5000.
Statistical Analysis-Data are expressed as the mean Ϯ S.E. Statistical significance was determined by unpaired Student's t test using the statistics module of Microsoft Excel (Microsoft Excel X for Mac; Microsoft Corp., Redmond, WA). Statistical significance was assumed at p Ͻ 0.05.

RESULTS
Expression of G M ⌬C in Liver of STZ-treated Rats-Our prior work has shown that expression of G M ⌬C in liver of rats with diet-induced obesity and insulin resistance results in reversal of glucose intolerance, in concert with stimulation of liver glycogen storage (8). However, these studies did not address the utility of the targeting subunit to correct blood glucose levels in animals with frank diabetes. To investigate this point, we injected AdCMV-G M ⌬C or, as a control, AdCMV-␤GAL into Wistar rats that received a single moderate dose of STZ (60 mg/kg). Only animals in which blood glucose exceeded a level of 250 mg/dl within 3 days of STZ injection (ad libitum fed state) and with blood aspartyl-aminotransferase activity of less than 200 units/ml after viral injection (indicative of the absence of virus-induced liver damage) were used for further study. As a first step in our analysis, we sought to confirm expression of G M ⌬C specifically in animals injected with the AdCMV-GM⌬C virus. The cDNA insert contained in this virus includes a FLAG tag to allow specific identification of the virus-encoded protein.
As shown in Fig. 1, multiplex PCR analysis of liver RNA reveals an RNA-derived band of the expected size only in animals that received the AdCMV-GM⌬C virus and not in control animals (receiving either AdCMV-␤GAL or no virus). Similarly, a protein detectable with an anti-FLAG antibody was immunoprecipitated and immunoblotted only in liver samples from AdCMV-G M ⌬C-injected animals. It should be noted that we expect systemic administration of AdCMV-G M ⌬C to result in expression of G M ⌬C in a largely liver-specific fashion, based on prior analysis of a wide array of tissues in rats and mice infused with this and other recombinant adenoviruses (8,10,15,16).
Normalization of Blood Glucose Levels in STZ-treated Rats by Hepatic Expression of G M ⌬C-Blood glucose was measured daily for 5 days after STZ treatment and then for an additional 6 days after adenoviral injection. The glucose values for individual STZ-treated animals in the ad libitum fed state are shown immediately before and 6 days after injection of Ad-CMV-G M ⌬C ( Fig. 2A) or in a control group injected with Ad-CMV-␤GAL or left uninjected (Fig. 2B). All animals that received the AdCMV-G M ⌬C virus experienced a decline in blood glucose to the normal range, whereas this never occurred in AdCMV-␤GAL-injected or uninjected controls. A summation of this data is provided in Fig. 2C and shows that control STZinjected animals had no decline in average glucose values (381 Ϯ 11 versus 379 Ϯ 35 mg/dl prior to and 6 days after viral injection, respectively). In sharp contrast, AdCMV-G M ⌬C-injected animals experienced a decline in blood glucose from 335 Ϯ 31 mg/dl prior to viral injection to 109 Ϯ 28 mg/dl 6 days after injection, the latter value being indistinguishable from the average in normal controls (122 Ϯ 10 mg/dl). We also examined blood glucose levels in rats following an overnight fast (Fig. 2D). Fasting caused a lowering of blood glucose levels into the normal range in all three groups of rats. AdCMV-G M ⌬C-treated, STZ-injected rats had blood glucose levels that were slightly and significantly lower than STZ-injected controls, but these values were not different from those in uninjected controls (75 Ϯ 13, 89 Ϯ 8, and 84 Ϯ 13 mg/dl for the AdCMV-G M ⌬C-treated/STZ-injected, STZ-injected control, and uninjected control groups, respectively). Thus, AdCMV-G M ⌬C injection completely normalized blood glucose levels in STZdiabetic rats.
Effect of G M ⌬C Expression on Circulating Hormone and Metabolite Levels-Lowering of blood glucose levels by expression of genes that regulate carbohydrate metabolism in liver has the potential to perturb lipid homeostasis, as was evident in a prior study from our laboratory involving adenovirus-mediated expression of glucokinase (15). Table I presents a profile of several key blood hormones and metabolites in the various groups of rats from the current study. With regard to key metabolic regulatory hormones, STZ injection resulted in a 68% decrease in circulating insulin levels relative to uninjected controls, with a similar decrease occurring in STZ-treated animals that received the AdCMV-G M ⌬C virus. These decreases in circulating insulin were accompanied by a 40% increase in glucagon levels in both groups. Thus, both groups of STZ-treated rats experienced a fall in insulin/ glucagon ratio of ϳ75% (from 0.12 to 0.03) despite the fact that one group was hyperglycemic (AdCMV-␤GAL/uninjected controls) and the other normoglycemic (AdCMV-G M ⌬C-injected animals). Thus, lowering of blood glucose in the Ad-CMV-G M ⌬C-treated group was achieved by a mechanism independent of changes in insulin/glucagon ratio.
Table I also shows that there were no significant changes in circulating triglycerides, free fatty acids, ketones, or lactate in either group of STZ-injected animals (AdCMV-␤GAL/unin-jected or AdCMV-G M ⌬C-injected), relative to control rats that did not receive STZ. We presume that the absence of an increase in circulating lipids or ketones in the STZ-treated groups is attributable to our use of a single moderate dose of the drug that allows some residual insulin production. In sum, lowering of blood glucose by hepatic G M ⌬C expression does not perturb other indices of lipid or carbohydrate homeostasis measured in this study.
Normalization of Food Intake by Hepatic Expression of G M ⌬C Occurs by a Leptin-independent Mechanism-During routine monitoring of food consumption during these studies, we observed a potent effect of G M ⌬C expression on this variable. Many groups have previously reported increased food intake in response to STZ injection and type 1 diabetes, and consistent with this, our STZ-injected rats consumed 42% more food on a daily basis than uninjected controls (Fig. 3). Remarkably, injection of AdCMV-G M ⌬C into STZ-treated rats caused food consumption to return to normal (Fig. 3).
The hormone leptin plays a major role in control of food intake and feeding behavior. We therefore investigated whether the reduction in food intake in G M ⌬C-expressing animals was secondary to changes in circulating leptin levels. As shown in Fig. 4, STZ injection caused leptin levels to decrease by 82 and 70% in the AdCMV-␤GAL-treated/untreated and AdCMV-G M ⌬C-treated groups, respectively. On average, leptin levels were slightly but significantly higher in AdCMV-G M ⌬Cinjected animals than in the control STZ-treated rats (p Ͻ 0.01). It is unclear whether this slight increase in leptin in the former animals is of any functional significance. In fact, within a subset of several AdCMV-G M ⌬C (n ϭ 4) or control (n ϭ 3) STZ-treated-animals with no significant differences in their plasma leptin levels (0.70 Ϯ 0.15 versus 0.74 Ϯ 0.01 ng/ml in G M ⌬C-expressing versus control animals, respectively), food intake was still dramatically reduced in the G M ⌬C-expressing animals relative to controls (86 Ϯ 4.8 versus 126 Ϯ 2.5 mg/g/ day, respectively). Furthermore, work from another laboratory shows that STZ treatment does not influence leptin sensitivity in rats (17). In sum, whereas we consider it unlikely that leptin makes a significant contribution to the complete normalization of food intake in the STZ-injected, AdCMV-G M ⌬C-treated rats, some contribution of the hormone cannot be rigorously excluded at the present time.
Another possible explanation for the decline in food intake in AdCMV-G M ⌬C-treated rats could be that the lowering of blood glucose in these animals prevented spilling of calories in the form of glucose in the urine, thus avoiding the need for compensatory hyperphagia. To test this idea, we measured urine volume and glucose concentration in a separate group of STZinjected rats with and without G M ⌬C expression. Animals treated with STZ and AdCMV-G M ⌬C had blood glucose levels of 316 Ϯ 50 mg/dl immediately prior to and 172 Ϯ 67 mg/dl 6 days after viral injection (n ϭ 3). Control animals not injected with AdCMV-G M ⌬C had blood glucose levels of 320 Ϯ 54 and 317 Ϯ 44 mg/dl at the same time points (n ϭ 8). As measured in the 24-h period between the fifth and sixth day after Ad-CMV-G M ⌬C injection, total urine volume and urine glucose dropped from 78 to 27 ml and from 7.6 g of glucose/24 h to 0.9 g of glucose/24 h, respectively, in response to hepatic G M ⌬C expression. For comparison, animals that did not receive an injection of STZ had blood glucose levels of 120 mg/dl and no detectable glucose in their urine and produced 18 ml of urine/ day (n ϭ 7). The difference in glucose spilling in the untreated versus AdCMV-G M ⌬C-treated STZ-injected animals of 6.7 g of glucose/24 h (7.6 -0.9 g) is equivalent to a loss of 26 kcal/day. This is nearly exactly matched by the reduction in food intake of 25 kcal/day in G M ⌬C expressing STZ-injected rats versus controls. Thus, the reversal of caloric spilling in the urine is accounted for by a compensatory reduction in food intake in STZ-treated rats with hepatic expression of G M ⌬C.
Expression of G M ⌬C in Liver of STZ-treated Rats Restores Glycogen Levels to Normal-Hepatic glycogen metabolism is impaired in all forms of diabetes (18 -20). We therefore investigated the effect of G M ⌬C expression on liver glycogen stores in STZ-treated rats. As shown in Fig. 5, STZ treatment caused a 61% decrease in liver glycogen content relative to untreated controls despite significantly higher glucose levels in the STZtreated animals (379 Ϯ 35 versus 122 Ϯ 10 mg/dl). In contrast, the expression of G M ⌬C in liver of STZ-treated rats increased liver glycogen levels by 4-fold relative to STZ-treated controls and by 47% relative to untreated controls (p Ͻ 0.01).

Hepatic Overexpression of G L Fails to Mimic the Effects of G M ⌬C on Blood Glucose and Food
Intake-To address the potential role of hepatic glycogen repletion in mediating reduced food intake, we decided to express an alternate targeting subunit, G L , which in our prior studies was shown to cause strong increases in liver glycogen stores, even in fasted animals (8). As shown in Fig. 5, adenovirus-mediated expression of G L in liver of STZ-injected rats caused hepatic glycogen to increase to levels 8 times higher than in STZ-injected controls and double those in AdCMV-G M ⌬C-treated STZ-injected animals. Unlike G M ⌬C-expressing rats, in which blood glucose levels were lowered between the second and fourth day after viral injection and then remained low for the remainder of the experimental period (6 days), G L -expressing rats experienced a

FIG. 2. Lowering of blood glucose levels in STZ-treated animals by hepatic expression of G M ⌬C. A, individual STZ-injected animals prior to (closed diamonds) or 6 days after (open squares) infusion of AdCMV-G M ⌬C adenovirus. B, individual STZ-injected animals prior to (closed diamonds) or after infusion of AdCMV-␤GAL virus or no viral infusion (open squares;
a mixture of five animals that received the AdCMV-␤GAL control virus and three animals that received no virus). C, summary of data in A and B. Data represent the mean Ϯ S.E. for eight animals in each STZ-injected group and six animals in the "no STZ" group. *, blood glucose was lower in animals after treatment with AdCMV-G M ⌬C virus than before treatment, with p Ͻ 0.001. D, blood glucose levels in animals subjected to an overnight fast 5 days after virus injection. Data represent the mean Ϯ S.E. for AdCMV-G M ⌬C, STZ-injected rats (n ϭ 7); AdCMV-␤GAL or untreated, STZ-injected rats (n ϭ 10); and uninjected control rats (n ϭ 17). *, blood glucose was significantly lower in the AdCMV-G M ⌬C, STZ-injected rats than in the STZ-injected controls, with p ϭ 0.025.

TABLE I Blood metabolite and hormone levels
Animals received either a single bolus of 60 mg/kg STZ or no STZ. One group of STZ-injected rats were treated with AdCMV-␤GAL adenovirus or received no viral treatment (AdCMV-␤GAL/No virus). A separate group of STZ-injected rats received the AdCMV-G M ⌬C virus. Blood samples were taken 6 days after viral treatment and used for measurement of the indicated metabolites and hormones. Data represent the mean Ϯ S.E. for eight animals in each of the STZ-treated groups and six animals in the "no STZ" group. *** and *, significant differences relative to the no STZ group, with p Ͻ 0.001 and 0.05, respectively. FFA, free fatty acid. sudden, transient decline in blood glucose values at 3-5 days after viral injection but then returned to frank hyperglycemia by 6 days (blood glucose levels of 316 Ϯ 73 versus 379 Ϯ 35 mg/dl in G L -expressing STZ-injected rats compared with STZinjected controls, respectively) (Fig. 6A). Food intake was tightly correlated with blood glucose levels in both the G M ⌬C and G L -injected groups. Thus, at the time where either G M ⌬C or G L expression caused glucose to decline to normal levels, food intake also declined to the level observed in normal non-STZ-injected rats (Fig. 6B). However, as hyperglycemia returned in G L expressing rats, food intake climbed back toward the rates observed in STZ-injected controls, whereas food intake in normoglycemic G M ⌬C-treated STZ-injected animals was indistinguishable from non-STZ-injected controls (Fig. 6B). Thus, G L expression provides only a transient lowering of blood glucose and food intake in this model of STZ-induced diabetes, despite causing dramatic increases in hepatic glycogen storage.
These data argue that liver glycogen content per se is not a primary regulator of food intake.

G M ⌬C Expression in Liver of STZ-treated Rats Exerts Metabolic Effects without Affecting Glucokinase Expression-Insu-
lin-deficient states are known to be associated with a sharp decline in hepatic glucokinase expression (21)(22)(23), and reduction of glucokinase activity by liver-specific gene knock-out in mice results in impaired hepatic glycogen storage and glucose intolerance (24). These findings prompted us to investigate glucokinase expression in our model system. As shown in Fig.  7, STZ injection caused a sharp reduction in glucokinase mRNA and protein levels, as measured by reverse transcriptase-PCR and immunoblot analysis, respectively. Remarkably, glucokinase mRNA and protein levels were also very low in AdCMV-G M ⌬C-injected, STZ-treated rats.
Glucokinase-catalyzed phosphorylation of glucose is normally perceived as an important regulatory step in hepatic glucose balance. One possible mechanism by which G M ⌬C could have enhanced hepatic glucose disposal despite reduced glucokinase expression might be a compensatory increase in expression of other genes that are involved in hepatic glucose uptake and storage. To investigate this point, we measured the levels of GLUT-2 and glycogen synthase mRNA by semiquantitative multiplex reverse transcriptase-PCR. No significant differences in expression of these genes was noted between the AdCMV-G M ⌬C/STZ-injected, STZ-injected control, and uninjected groups (Fig. 8, A and B). Moreover, application of quantitative real time PCR methods for measurement of glucokinase and hexokinase I levels confirmed the sharp drop in expression of glucokinase in both the AdCMV-G M ⌬C-treated and control STZ-injected groups relative to uninjected controls but showed no change in expression of hexokinase I among these groups of animals (Fig. 8C). DISCUSSION Glycogen-targeting subunits of protein phosphatase-1 serve as important scaffolding proteins that juxtapose key enzymes of glycogen metabolism with the regulatory enzyme protein phosphatase-1. The various targeting subunit isoforms have FIG. 3. Expression of G M ⌬C in liver of STZ-injected rats normalizes food intake. The same animals as described in the legend to Fig. 2A were used for measurement of food intake. Animals received either a single bolus of 60 mg/kg streptozotocin (STZ-injected) or no streptozotocin injection (No STZ). One group of STZ-injected rats was treated with AdCMV-␤GAL adenovirus or received no viral treatment (AdCMV-␤GAL/no virus). A separate group of STZ-injected rats received the AdCMV-G M ⌬C virus. Food intake was measured during three successive 24-h periods, beginning 3 days day after viral treatment. Data represent the mean intake/24 h Ϯ S.E. for eight animals in each STZ-treated group and six in the "no STZ" group. *, STZ-injected rats had increased food intake relative to noninjected controls, with p Ͻ 0.001. #, AdCMV-G M ⌬C-treated, STZ-injected rats had reduced food intake relative to STZ-injected control rats, with p Ͻ 0.001.

FIG. 4. Circulating leptin levels in STZ-injected rats.
The animal treatment groups and their labeling are as described in the legend to Fig. 3. Plasma leptin levels were determined by radioimmunoassay 6 days after viral injection. Data represent the mean Ϯ S.E. for eight animals in each STZ-injected group and six animals in the "no STZ" group. *, both the STZ-injected, AdCMV-G M ⌬C and STZ-injected control groups had leptin levels lower than the no STZ group, with p Ͻ 0.02. #, leptin levels were higher in the STZ-injected, AdCMV-G M ⌬C-treated rats than in the STZ-injected control group, with p Ͻ 0.01. FIG. 5. Expression of G M ⌬C or G L in liver of STZ-injected rats stimulates hepatic glycogen synthesis to differing degrees. Animals received either a single bolus of 60 mg/kg streptozotocin (STZinjected) or no streptozotocin injection (No STZ; n ϭ 12). One group of STZ-injected rats was treated with AdCMV-␤GAL adenovirus or received no viral treatment (AdCMV-␤GAL/no virus; n ϭ 16). Separate groups of STZ-injected rats received either AdCMV-G M ⌬C (n ϭ 8) or AdCMV-G L (n ϭ 8) viruses. Animals were sacrificed for measurement of liver glycogen levels 6 days after viral injection in the ad libitum fed state. *, significant differences relative to STZ-injected controls, with p Ͻ 0.001. #, significant differences relative to the no STZ, no virus group, with p Ͻ 0.003. ϩ, a difference between the STZ-injected, G L expressing group and the STZ-injected, G M ⌬C-expressing group, with p ϭ 0.004. different potencies for activating glycogen deposition and also affect the way cells respond to regulatory signals for glycogen synthesis and breakdown (3,(5)(6)(7)(8)(9)25). Recently, we demonstrated reversal of glucose intolerance in rats fed a high fat diet by hepatic expression of G M ⌬C, a novel truncated version of the muscle isoform G M but, surprisingly, not the more glycogenic liver isoform, G L (8). Further work supported the idea that the unique efficacy of G M ⌬C in this experiment was a result of its intermediate glycogenic potency, coupled with retention of full responsiveness to glycogenolytic signals such as forskolin and low glucose in cells in which it is expressed (6). The latter property distinguished G M ⌬C from other targeting subunits (6) and proved essential in preventing glycogen stores from fil-ling up in the fasted state in high fat-fed rats, thereby allowing glycogen synthesis to be stimulated during a glucose challenge (8).
In the current study, we have extended our investigation of G M ⌬C to include its delivery to livers of animals with STZinduced diabetes. Three new findings have emerged from this work. First, we show that G M ⌬C expression in liver is sufficient to fully normalize blood glucose levels and glycogen storage in this model of insulin-deficient diabetes. Importantly, and in contrast to what we have observed with hepatic overexpression of glucokinase (15), other circulating metabolites, including free fatty acids and triglycerides, were not perturbed in these experiments. Second, this normalization occurred despite very FIG. 6. Differential effects of G M ⌬C and G L on blood glucose and food intake in STZ-injected rats. Animals were treated as described in the legend to Fig. 5. A, blood glucose levels were measured after STZ injection but prior to administration of the indicated recombinant adenoviruses (Start), at the point of first strong decrease in blood glucose levels following viral administration (Inflection point), and 6 days after viral injection (Final). B, food intake was measured in the same animals at the inflection and final time points described for A. Data represent the mean Ϯ S.E. for 4 G M ⌬C-expressing rats; 5 G L -expressing rats, 10 no STZ, no virus controls; and 12 STZ-injected controls (a mixture of AdCMV-␤GAL-injected and no virus). *, significant differences between the STZ-injected controls and STZ-injected, AdCMV-G L treated groups relative to other groups, with p Ͻ 0.006. low levels of liver glucokinase expression. Finally, G M ⌬C expression in liver curtailed food intake in STZ-injected rats, which were otherwise hyperphagic.
The normalization of blood glucose and liver glycogen by hepatic G M ⌬C expression is remarkable in that it occurred in the face of a 68% reduction in circulating insulin levels. Some insight into this outcome is derived from our prior work showing that stimulation of glycogen synthesis caused by overexpression of glycogen-targeting subunits in hepatocytes is glucosedependent but largely insulin-independent (7,25). Thus, our data are consistent with a model in which expression of G M ⌬C increases hepatic glucose disposal when the system is challenged by elevated glucose but allows normal emptying of glycogen stores and protection against hypoglycemia in fasting or other catabolic states. This interpretation fits both our prior study in insulin-resistant high fat-fed rats treated with Ad-CMV-G M ⌬C, which were normoglycemic in the fasting state and exhibited normal glucose tolerance (8), and the current study, in which blood glucose levels were normalized in STZinjected animals, with no evidence of hypoglycemia. Thus, our current and prior (8) studies taken together argue that appropriately regulated augmentation of hepatic glycogen storage is sufficient to reverse perturbations in glucose homeostasis caused by either impaired insulin action or impaired insulin secretion. Consistent with this conclusion, overexpression of another targeting subunit, G L , in liver of insulin-resistant (8) or STZ-injected diabetic rats (current study) resulted in unregulated glycogen overstorage in liver and a consequent failure to stably reverse glucose intolerance or hyperglycemia, respectively.
Glucokinase is generally thought of as a key regulatory step in hepatic glucose metabolism. Consistent with this idea, patients with maturity-onset diabetes of the young, type 2, have mutations in their glucokinase gene that affect both insulin secretion and hepatic glucose disposal (20). Furthermore, tissue-specific knock out of the glucokinase gene in liver of mice results in perturbed glucose homeostasis and impaired liver glycogen storage (24), whereas overexpression of the enzyme in the liver increases glycogen deposition and lowers blood glucose (15). It was therefore quite surprising that G M ⌬C expression in the liver was able to lower blood glucose levels and normalize liver glycogen content in STZ-injected rats in the face of a dramatic reduction in glucokinase expression. The mechanism by which G M ⌬C overcomes the decrease in glucokinase expression in insulin-deficient animals remains to be defined. One idea is that STZ injection and/or G M ⌬C expression could have caused a compensatory increase in other enzymes involved in hepatic glucose disposal, but such changes were not observed at the RNA level for the glycogen synthase, GLUT2, or hexokinase I genes. However, G M ⌬C might still have activated glycogen synthesis by a "pull mechanism," in which stimulation of glycogen synthase activity resulted in lowering of glucose 6-phosphate levels and consequent removal of product inhibition of hexokinase I activity. Alternatively, G M ⌬C expression may have diverted gluconeogenic precursors away from the glucose-6-phosphatase reaction and into the glycogen storage pathway, thereby contributing to lowering of glucose levels via a decrease in hepatic glucose production. Consistent with this model, we have recently employed 2 H 2 O NMR to demonstrate that a larger proportion of glycogen is synthesized from the level of gluconeogenic precursors in hepatocytes with overexpressed glycogen-targeting subunits than in cells with overexpressed glucokinase (6). Also consistent with this idea, expression of the PTG-targeting subunit in liver of normal rats resulted in reduced expression of the catalytic subunit of glucose-6-phosphatase (9). Whereas these ideas will require further investigation, our data clearly establish that normalization of blood glucose can be achieved by manipulation of steps distal to glucokinase, even when this important enzyme is present at very low levels.
Another possible mechanism contributing to lowering of blood glucose in animals with hepatic G M ⌬C expression is their decrease in food intake relative to controls. This is clearly not the only mechanism at work, however, based on a study by others of the effects of leptin infusion in STZ-induced diabetic rats (17). In these studies, infusion of leptin at a rate that allowed circulating levels of the hormone to be restored to those found in non-STZ-injected controls resulted in complete normalization of food intake. Hyperglycemia was partially ameliorated in these animals but clearly not normalized (glucose levels fell from 24.3 to 17.2 mM). In contrast, expression of G M C in STZ-injected animals not only normalized food intake but also lowered blood glucose levels completely to normal (from 18.6 to 6.1 mM).
There are several possible mechanisms by which G M ⌬C expression in liver could lead to normalization of food intake in FIG. 8. GLUT2, glycogen synthase and hexokinase I mRNA levels in liver. The animal treatment groups and their labeling are as described in the legend to Fig. 3. Animals were sacrificed for measurement of hepatic glycogen synthase (GS) and GLUT2 mRNA levels by multiplex reverse transcriptase-PCR. A representative set of samples is shown in A, and quantitative analysis of this data by densitometric scanning, normalized to the internal ␣-tubulin control, is shown in B. C, the levels of glucokinase (GK) and hexokinase I (HKI) mRNA in liver samples measured by quantitative real time PCR, using ␣-tubulin as the internal control. Data represent the mean Ϯ S.E. for six samples/group. *, significantly lower glucokinase expression in either group of STZ-injected rats relative to uninjected controls, confirming the data of Fig. 7.
STZ-injected animals. Three that were considered and tested experimentally in this study were the following: 1) G M ⌬C expression in liver affected leptin levels; 2) G M ⌬C expression reduced calorie spilling in the urine; and 3) liver glycogen repletion had a direct effect on food intake. STZ injection caused a large decrease in circulating leptin levels, probably secondary to the increased glucagon/insulin ratio and consequent reduction in fat mass in the diabetic animals. Expression of G M ⌬C in liver of STZ-injected animals resulted in a small but statistically significant rise in circulating leptin levels relative to animals that received the drug alone. Whereas we think it is unlikely that this small increase in leptin had a major impact on the reduced food intake in G M ⌬C-expressing, STZ-injected rats, for reasons set forth under "Results," the possibility cannot be rigorously excluded at present. G M ⌬C expression caused a clear reduction in excretion of glucose in the urine. This amelioration of caloric "spilling" was proportional to the reduction in food intake. This finding is consistent with the idea that G M ⌬C expression lowered food intake secondary to its effect on blood glucose. However, the factors that allowed G M ⌬C-expressing rats to sense the reduction in caloric spilling and translate this into a reduction in food intake remain to be defined, especially since these changes occurred at a time when hormones known to regulate food intake such as insulin (26) and leptin were present at low levels. It remains possible that levels of peptides that influence feeding behaviors such as glucagon-like peptide-1, CCK, or ghrelin could be altered by G M ⌬C expression, and this will require further investigation.
Finally, we considered the possibility that reduced food intake in STZ-injected, G M ⌬C-expressing rats might be related to changes in glucose disposal and storage in the liver. Fifty years ago, Mayer (27, 28) developed a "glucostatic theory" of feeding behavior in which changes in glucose utilization rates were proposed to regulate hunger and satiety. A "glycogenostatic" model in which hepatic glycogen stores play a central role in regulating food intake and energy balance has also been proposed (29 -32), but direct and consistent experimental support for this idea has not emerged. In the current study, expression of the targeting subunit G L in STZ-injected animals caused sharp but transient decreases in food intake and circulating glucose levels. Unlike G M ⌬C-expressing animals, in which both variables declined and then remained low, G L -overexpresssing animals experienced a return to high rates of food consumption in parallel with the return of hyperglycemia (Fig. 6). Because the G L -expressing animals had much higher liver glycogen content than the G M ⌬C animals at sacrifice (Fig. 5), these data argue against a primary role for liver glycogen stores as a regulator of food intake. However, it remains possible that a sudden increment in hepatic glycogen storage, such as induced upon initial overexpression of G L , could generate satiety signals. On balance, however, our current best understanding is that the main driver of reduced food intake in G M ⌬C expressing, STZ-injected rats is the lowering of blood glucose, which in turn results in reduced caloric spilling in the urine, leading to abrogation of compensatory hyperphagia.
We have now established that expression of G M ⌬C in liver not only reverses glucose intolerance in insulin-resistant rats but also normalizes blood glucose in STZ-induced diabetic animals. However, we recognize that whereas G M ⌬C is an effective reagent for treatment of diabetes in rodents, its application to human diabetes is hampered by the lack of safe and efficacious vectors for delivery of foreign genes to liver of diabetic patients. Nevertheless, the surprising demonstration of a glucose-lowering effect of G M ⌬C in the background of depressed hepatic glucokinase expression and insulin insufficiency may place new focus on drugs that activate liver glycogen storage as a means of controlling blood glucose.