Acute AKT signaling increases glucose phosphorylation and contribution to glycogen in hepatocytes Authors: Megan Stefkovich, Won Dong Lee, Talia Coopersmith, Joseph A Baur, Joshua D Rabinowitz, Paul M Titchenell Published: 2026 ABSTRACT SummaryInsulin’s regulation of hepatic glucose production and glycogen is critical for postprandial glucose disposal. AKT, a serine-threonine kinase and insulin signaling intermediate, regulates liver glucose metabolism through transcriptional and posttranslational mechanisms. However, current knowledge largely stems from genetic loss-of-function models, precluding observation of AKT’s non-transcriptional effects. To measure rapid changes to glucose and glycogen metabolism, isotope tracing using [U-13C]-glucose and [U-14C]-glucose was coupled with the AKT inhibitor MK-2206 in primary rat hepatocytes. MK-2206 treatment decreased AKT phosphorylation and glucose contribution to glucose 6-phosphate and uridine diphosphate glucose within minutes without affecting metabolite pool sizes or protein levels of glucokinase, glucose 6-phosphatase, or phosphoenolpyruvate carboxykinase. MK-2206 also decreased glucose contribution to glycogen, independent of glycogen breakdown or glycogen synthase phosphorylation. These results demonstrate that AKT acutely regulates glucose contribution to glycogen and upstream precursors, suggesting a transcription-independent mechanism that is proximal to glucose 6-phosphate generation for glycogen synthesis.Keywords: glycogen, glucose homeostasis, glucokinase, glucose 6-phosphate, glycolysis, gluconeogenesis Introduction Insulin regulation of hepatic glucose production (HGP) is critical for postprandial glucose homeostasis. Insulin inhibits HGP via two mechanisms, one involving direct action at the liver and the other working indirectly through controlling adipose tissue lipolysis.1,2,3 Infusing glucose and insulin into the portal vein of the liver drives rapid suppression of HGP and stimulates hepatic glucose clearance, allowing for net liver glucose uptake.1,2,4,5,6,7 Net glucose uptake following portal infusion of glucose and insulin is estimated to clear 30% of glucose administered,8,9,10,11 depending on the rate of the infusion. Thus, insulin’s direct effect on liver HGP and glucose uptake may play a significant role in blood glucose homeostasis. Studies performed in humans suggest that insulin’s direct effect on liver glucose metabolism is blunted in metabolic diseases. Patients with type 2 diabetes exhibit decreased splanchnic glucose uptake and liver glycogen in response to insulin,12,13,14 especially when glucose and insulin are delivered to the gut rather than peripherally.14 Hyperinsulinemic-euglycemic clamps performed in high-fat and fructose-fed (HFFD) dogs demonstrate that metabolic disease impairs insulin’s direct action on the liver. Regardless of a high insulin infusion concentration in the portal vein and adequate suppression of adipose tissue lipolysis, HFFD dogs showed decreased liver glucose uptake compared to normal chow controls.5 While an aberrant response to insulin impacts glucose homeostasis, the mechanism by which insulin rapidly suppresses liver glucose output and stimulates uptake is not completely understood. Hyperinsulinemic clamp studies have interrogated insulin’s relative control of glycogen breakdown and gluconeogenesis during its rapid suppression of HGP. Experiments in dogs suggest that insulin predominantly controls HGP by suppressing glycogenolysis.15,16 Two similar studies indicate an increase in glycolysis and decrease in gluconeogenesis also contribute,7,17 but gluconeogenic control depends on insulin’s anti-lipolytic effects.7 Other studies rebut this, finding that a low dose of insulin, even when delivered via the portal vein, does not impact gluconeogenesis despite suppression of glycogenolysis. This was established using a 3-3H-glucose tracer method or by measuring 14C-phosphoenolpyruvate contribution to glucose 6-phosphate (G6P) and plasma glucose.15,16 In humans, one study showed that glycogenolysis is more sensitive to insulin but that gluconeogenesis decreases as well.18 While additional studies would provide clarity as to whether insulin’s inhibition of glycogenolysis accounts for its control of HGP, evidence from humans indicate glycogenolysis supplies approximately half of glucose produced following overnight fasting.19,20,21,22,23 Thus, glycogenolysis makes a considerable contribution to HGP under physiologically relevant conditions. It is well-established that insulin controls glucose metabolism through direct action at the liver by signaling to the serine/threonine kinase AKT.24 AKT is a major effector of insulin signaling, required for insulin-stimulated glycogen storage25 and insulin’s suppression of HGP.24,25 The prevailing model for how AKT regulates HGP is based on a transcriptional mechanism involving the suppression of gluconeogenic genes Pck1 and G6pc,26 which encode phosphoenolpyruvate carboxykinase (PEPCK) and glucose 6-phosphatase (G6Pase), respectively. However, the slow degradation of PEPCK and G6Pase is insufficient to explain how insulin shifts hepatic glucose balance from output to uptake within minutes.4,7 Furthermore, Pck1 and G6pc expression correlates poorly with blood glucose control and decreases well after insulin achieves suppression of HGP during a clamp.17,24,27 AKT may rapidly suppress HGP by preventing glycogen breakdown. Insulin signaling via AKT opposes the action of cAMP-dependent protein kinase A (PKA).28 PKA phosphorylates glycogen synthase (GS),29 glycogen phosphorylase kinase,30 and glycogen synthase kinase 3β (GSK-3β),31 with the net effect of increasing glycogenolysis.32,33 Mechanistically, AKT blocks PKA signaling by activating phosphodiesterase 3B (PDE3B), which degrades cyclic AMP (cAMP) and thus decreases PKA activity.28 AKT can also decrease glycogenolysis by phosphorylating a regulatory subunit of protein phosphatase 1 (PP1), protein phosphatase 1 regulatory subunit 3G (PPP1R3G).34 This activates PP1, eliciting an activating dephosphorylation of GS and inhibitory dephosphorylation of glycogen phosphorylase, increasing glycogen levels. Both of these mechanisms converge on control of GS phosphorylation. However, the requirement of the AKT-regulated phosphosites on PDE3B and PPP1R3G for insulin’s activation of glycogen synthase and inhibition of glycogen phosphorylase remains to be tested. A third mechanism by which AKT controls GS phosphorylation is by inhibiting GS’s major upstream regulator GSK-3β.35 GSK-3β phosphorylates GS to block insulin-stimulated GS activity.36 However, a GSK-3β phosphomutant refractory to AKT signaling exhibits normal insulin-stimulated glycogen synthesis,37 despite decreased insulin-stimulated GS activity.36 This suggests that there are GS-independent mechanisms for acute AKT regulation of glycogen in hepatocytes. One explanation for how insulin stimulates liver glucose uptake and suppresses HGP is substrate push. Substrate push dictates that a higher concentration of a reactant relative to a product drives a reaction in the forward direction. If insulin stimulates glucose uptake, as demonstrated in the conscious dog model,4 this would promote glucose phosphorylation and downstream glycolysis and glycogen synthesis. A higher concentration of glucose relative to glucose 6-phosphate would also disfavor glucose dephosphorylation, preventing glycogenolysis or gluconeogenesis. Beyond providing the building blocks of glycogen, higher levels of glycolytic and glycogenic substrates could also serve as allosteric regulators. Specifically, G6P has long been known to allosterically activate glycogen synthase.38 Indeed, mutation of glycogen synthase’s G6P binding site reduced insulin-stimulated glycogen deposition in hepatocytes,39 suggesting that if a “push” mechanism underlies insulin’s acute effect on glucose metabolism, it may also involve allosteric regulation. Another potential mechanism by which insulin regulates hepatic glucose production could be through activation of glucokinase (GCK), which phosphorylates glucose. Metabolic control analysis from hyperinsulinemic clamp data suggests that GCK is the rate-limiting step for insulin-stimulated glycogen synthesis.40 This is not altogether unsurprising, because with bidirectional glucose transport in the liver, glucose is prevented from leaving the cell by its phosphorylation. Furthermore, expression of GCK positively correlates with improved blood glucose control in diabetic and non-diabetic humans.27 One way that insulin could clear blood glucose and promote glycogen storage, therefore, is by increasing glucose 6-phosphate generation. Additionally, primary hepatocyte and rat clamp studies indicate that glycogen synthesis may not be the only beneficiary of insulin-stimulated glucose phosphorylation. Insulin increased total lactate,41,42,43 fructose bisphosphate, glyceraldehyde 3-phosphate, and pyruvate within 30 min,43,44,45 which correlated with a reduction in HGP in one case.45 These studies suggest a relationship between insulin-stimulated glycolysis and suppression of HGP. Here, we used both genetic loss-of-function models and pharmacological inhibition of hepatic AKT to understand the contribution of hepatic insulin signaling via AKT to long and short-term changes to glucose metabolism. Notably, we found that AKT promotes glucose contribution to G6P, UDP-glucose, and glycogen on a rapid timescale of minutes in primary rat hepatocytes. This occurred independently of changes to glycogenolysis, total metabolite levels, and phosphorylation of glycogen synthase, suggesting that insulin regulates glucose phosphorylation and utilization at a step upstream of its canonical regulation of glycogen storage. Secondary to the AKT-driven increase in glucose phosphorylation, gluconeogenic contribution to G6P was suppressed. Together, our data suggest that acute AKT signaling promotes glucose phosphorylation to drive glycogen synthesis, potentially through GCK activation, indirectly limiting gluconeogenic contribution. Results Chronic loss of AKT signaling impacts glucose 6-phosphate levels Previous work by Titchenell et al.24 showed that inducible AKT1 and AKT2 knockout in the liver (L-AKTDKO) reduces G6P levels, suggesting that chronic loss of AKT signaling affects glucose phosphorylation. We conducted analysis on bulk metabolomics data generated from L-AKTDKO livers24 to explore the impact of the loss of AKT on other hepatic metabolites. Notably, we detected a significant decrease in G6P (Figures 1A and 1B), but surprisingly few glycolytic or glycogenic metabolites were significantly changed by 2-fold or greater (Figure S1). Upon conducting pathway analysis, we found that the top pathways affected were related to amino acid metabolism (Figure 1C). Chronic loss of AKT signaling impacts glucose 6-phosphate levels (A) Individual metabolites identified through bulk metabolomics showing metabolites significantly up or downregulated in AKTDKO livers by an average fold-change ±2 with a p value<0.05. Animals were fasted overnight for 16 h and refed for four. Legend is log2 (fold-change). (B) Volcano plot and (C) KEGG pathway analysis performed on bulk metabolomics described in (A). ∗∗p value < 0.005, ∗p value<0.05. Enrichment determined using Fisher’s exact test. N = six control and three AKTDKO mice. AKT directs glucose contribution to glucose 6-phosphate and UDP-glucose While bulk metabolomics allows a global picture of metabolic changes, it does not provide clarity as to how these changes occur. In the liver, altered glucose metabolism can be masked because a decrease in glycolytic contribution to a metabolite’s pool size can be compensated for by increased gluconeogenic or glycogenolytic contribution. For example, glycogen precursors are synthesized directly from glucose and indirectly from gluconeogenesis (Figure 2A). Therefore, large differences in pathway fluxes can underlie seemingly unchanged metabolite pools. Using a stable isotope tracing approach, we evaluated AKT’s acute effect on glucose metabolism with ubiquitously labeled 13C-glucose ([U-13C]-glucose) in primary rat hepatocytes (Figure 2A). An in vitro approach allowed us to collect samples quickly (in minutes) to determine the cell autonomous effects of AKT signaling on glucose metabolism. AKT directs glucose contribution to glucose 6-phosphate and UDP-glucose (A) Schematic illustrating how ubiquitously labeled 13C glucose contributes to glycolytic, glycogenic, and lipogenic intermediates in hepatocytes.13C indicated by blue circles. Reversible reactions indicated by double headed arrows. Dashed arrows indicate multiple reaction steps. (B) Western blot showing that the AKT inhibitor, MK2206, suppresses AKT signaling in primary rat hepatocytes. Cells were fasted overnight in serum- and glucose-free DMEM, pretreated for 30 min with 10 μM MK2206 or DMSO in overnight media, and then treated for 30 min with 10 mM glucose and inhibitor or DMSO. (C) Fractional labeling and (D) fractional contribution of metabolite pools in primary rat hepatocytes treated with 10 mM [U-13C]-glucose and 10 μM MK2206 or DMSO. Cells were treated as described in STAR Methods section [U-13C]-glucose isotope tracing. N = nine replicates from three pooled experiments. p values <0.0005, 0.005, or 0.05 denoted by ∗∗∗, ∗∗, and ∗, determined using two-way ANOVA followed by post-hoc t test in (C) and Student’s t test in (D). UDP-glucose fractional labeling p value <0.0005 for 1.25 and 1.75 min time points in (C). (E) Ion counts from a representative tracing experiment included in (C and D). p values calculated using Student’s t test, values < 0.05 denoted by ∗(M+6), % (M+3), and (M+2), ∼ (M+1), or # (M+0). p values <0.005 or 0.0005 for M = 6 denoted by ∗∗ and ∗∗∗. Error bars represent ± SEM. Time points indicate the time cell extraction buffer was added to wells. We treated primary rat hepatocytes with a potent and selective AKT inhibitor, MK-2206 dihydrochloride (MK2206), which effectively blocks AKT phosphorylation and that of downstream substrates like PRAS40 (Figure 2B). Cells were incubated in glucose-free media overnight prior to a 30-min pre-treatment with 10 μM MK2206 or DMSO vehicle control, after which experimental media was added containing 10 mM [U-13C]-glucose with MK2206 or DMSO. First, we evaluated the metabolites that were completely labeled by 13C, because at early time points this fraction of the pool is likely to be derived directly from the [U-13C]-glucose. For G6P, this means all six carbons are 13C (M+6), while for pyruvate it means all three (M+3). For UDP-glucose, while there are 15 carbons, we considered M+6 UDP-glucose to be “ubiquitously” labeled, as at early time points these carbons likely make up the glucose moiety. AKT inhibition decreased the fraction of the metabolite pool of G6P and UDP-glucose ubiquitously labeled by 13C within 10 min of 13C-glucose addition (Figure 2C). This indicated that AKT signaling either increases glucose contribution to G6P and UDP-glucose or decreases gluconeogenic or glycogenolytic contribution that would otherwise dilute out the 13C label. To discern which pathway was responsible, we queried the fractional contributions of other isotopologues (Figure 2D) and found that AKT inhibition decreased M+6 G6P and UDP-glucose and M+3 glycerol 3-phosphate (G3P). Conversely, it increased the fractional contribution of unlabeled glucose to G6P, UDP-glucose, and G3P pools and partially labeled (M+3) contribution to G6P (Figure 2D). These data indicate that AKT is potentially increasing the glycolytic contribution to these metabolites and suppressing contribution from the indirect pathway. Labeling of glycolytic intermediates and products downstream of G6P, such as pyruvate and lactate were not consistently significantly perturbed by MK2206 (Figure 2D). Looking at total ion counts (Figure 2E), differences in the ubiquitously labeled isotopologues for G6P, UDP-glucose, and G3P largely drove the differences in fractional contributions (Figure 2D). These results are consistent with the model that AKT signaling rapidly promotes glucose phosphorylation and commitment to glycogenic precursors in hepatocytes. Interestingly, AKT inhibition did not affect the total pool size of G6P or UDP-glucose (Figures 2E and S2). This indicates that the hepatocyte maintains its total pool of G6P and UDP-glucose, such that a concomitant increase in endogenous sources of carbon, or decrease in G6P consumption, compensate if glucose contribution decreases. G3P is an interesting exception, as the total pool size of G3P was reduced at multiple time points by a loss in AKT signaling (Figure 2E), suggesting that its metabolism is controlled by AKT through a distinct mechanism. Acute AKT inhibition increases the contribution of endogenous carbon sources to glucose 6-phosphate and intracellular glucose Given that G6P and UDP-glucose pool sizes were maintained with MK2206 treatment, we explored whether endogenous sources of carbon were compensating for the decrease in glucose contribution (Figures 3A–3C). M+1, M+2, and M+3 G6P and glucose can be generated from [U-13C]-glucose being metabolized through some combination of the pentose phosphate pathway, glycolysis, and the citric acid cycle, and then combining with endogenous carbon sources to re-generate G6P and glucose through gluconeogenesis. Notably, the [U-13C]-glucose tracer is not 100 percent M+6 glucose, as we detected 2% M+5 and 0.7% M+4 glucose (Figure S3). The M+4 and M+5 fractions of intracellular and media glucose were not enriched compared to the glucose tracer, so alternate sources of their generation were not considered. Acute AKT inhibition increases the contribution of endogenous carbon sources to glucose 6-phosphate and intracellular glucose (A) Ion counts from a tracing experiment performed as described in Figures 2C–2E and represented in Figure 2E. (B and C) are additional experimental replicates. Time point indications refer to all graphs directly underneath. N = 3 replicates per treatment. p values calculated using Student’s t test, values < 0.05 denoted by ∗ (M+6), ˆ (M+5), @ (M+4), % (M+3), and (M+2), ∼ (M+1), or # (M+0). p values <0.005, 0.0005, or 0.00005 for any MID are denoted by one, two, or three repeats of the symbol. Error bars represent ± SEM. Time points indicate the time cell extraction buffer was added to wells. In intracellular glucose (Figures 3A–3C), we consistently saw increases in M+1, M+2, and M+3 glucose ion counts, and sometimes all three, as early as 5.75 min. In one experiment (Figure 3A), this was accompanied by M+0, M+2, and M+3 glucose accumulating in the media by 30.75–60.75 min with MK2206 treatment. Decreased M+6 G6P (Figures 3A–3C) always preceded the changes to M+1 through M+3 intracellular and media glucose. The appearance of M+1 through M+3 G6P and intracellular glucose between 5.75 and 30.75 min with MK2206 treatment suggests that AKT inhibition causes an increase in the relative contribution of endogenous carbon sources via gluconeogenesis. Because total pool sizes of metabolites affected by MK2206 are unchanged (Figures 2E, 3A–3C, and S2), it follows that endogenous sources, or consumption, changes to compensate for any decrease in glucose contribution to G6P or intracellular glucose, possibly via gluconeogenesis. Acute AKT inhibition does not affect glycolytic or gluconeogenic enzyme protein levels or total pool sizes of glycolytic metabolites Insulin controls glycolytic and gluconeogenic gene expression through AKT,26 specifically Gck, Pck1, and G6pc. Given this, we determined whether acute AKT inhibition altered gene expression and protein levels to drive glucose contribution to G6P and glycogenic intermediates. We treated primary rat hepatocytes with unlabeled 10 mM glucose in the presence or absence of MK2206 for 30 min, using the same experimental paradigm as before (Figure 2). AKT inhibition did not significantly affect gene expression (Figure 4A) or total protein for GCK, PEPCK, and G6PC (Figure 4B), indicating that AKT does not drive glucose phosphorylation through a change in GCK or G6PC protein levels. Acute AKT inhibition does not affect glycolytic or gluconeogenic enzyme protein levels or total pool sizes of glycolytic metabolites (A) Relative RNA expression for glycolytic and gluconeogenic enzymes in primary rat hepatocytes incubated overnight in serum- and glucose-free DMEM, pretreated for 30 min with 10 μM MK2206 or DMSO in overnight media, then treated with 10 mM glucose with or without 10 μM MK2206 or DMSO for 30 min. Housekeeping gene is acidic ribosomal phosphoprotein P0 (36b4). Gck: glucokinase, Pfk1: phosphofructokinase 1, Pfk2: 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase, Pklr: pyruvate kinase, G6pc: glucose 6-phosphatase, Fbp1: fructose-1,6-bisphosphatase 1, Pck1: phosphoenolpyruvate carboxykinase 1, Pc: pyruvate carboxylase. N = 4 experimental replicates averaged from 2 to 4 technical replicates. Error bars represent ± SEM. (B) Protein levels of glycolytic and gluconeogenic enzymes in primary rat hepatocytes treated as described in (A) HSP90 as a loading control corresponds to the blots directly above it. (C) KEGG pathway analysis performed on metabolites identified through bulk metabolomics on primary rat hepatocytes treated with MK2206 for 5 min. Cells were fasted overnight in serum- and glucose-free DMEM and then pretreated for 30 min with 10 μM MK2206 or DMSO. Glucose (10 mM) and 10 μM MK2206 or DMSO was then added to the cells for 5 min before metabolism was quenched and lysates were collected. ∗p value < 0.05, enrichment determined using Fisher’s exact test. (D) Volcano plot and (E) individual metabolites showing metabolites significantly up or downregulated by an average fold-change ±2 with a p value < 0.05 in experiment described in (C). N = three samples per group. Legend in (E) is log2(fold change). Next, given the differential effects of AKT activity on total pool sizes (Figure 2E), we wanted to explore the global effect of acute AKT signaling on hepatocyte metabolism. To this end, we conducted unbiased metabolomics on primary hepatocytes after 5 min’ exposure to 10 mM glucose and MK2206. Cells were treated exactly as described in Figure 2. The top pathway affected was glycerophospholipid metabolism followed by phenylalanine metabolism (Figure 4C). There were trends in enrichment of other lipid and amino acid metabolic pathways, namely ether lipid metabolism, which can be impacted downstream of glycerophospholipid metabolism (Figure 4C). In contrast to the metabolomics conducted on AKT knockout livers (Figure 1C), there was no trend toward enrichment of glycolysis or gluconeogenesis. In evaluating endogenous metabolites impacted by inhibition of AKT, phospholipid precursors stood out. Specifically, CDP-ethanolamine was increased relative to controls and citicoline (CDP-choline) was decreased (Figures 4D and 4E). Each of these compounds contributes to distinct arms of the Kennedy pathway to generate phosphatidylethanolamine (PE) and phosphatidylcholine (PC), respectively. This suggests that AKT may acutely downregulate PE synthesis and favor PC through a post-translational mechanism. Additionally, 1-alkyl-2-acylglycerophosphoethanolamine was also elevated with AKT inhibition (Figures 4D and 4E), which is proposed to be a substrate of ethanolamine phosphotransferase (EPT1), an enzyme that generates PE from 1,2-diacylglycerol (DAG) and CDP-ethanolamine.46 Of note, the total levels of G3P, a precursor of phosphatidylcholine synthesis, trended down in MK2206-treated cells (p value = 0.065, Figure S4C), consistent with the reduction in total ion counts observed with stable isotope tracing (Figure 2E). Notably, glucose 6-phosphate abundance was not impacted by AKT inhibition, consistent with our stable isotope tracing experiments (Figures 2E and S4A). This is further evidence that the reduction in G6P observed in AKT1/2 knockout livers is due to the chronic loss of GCK transcription24 and translation.47 Overall, short-term inhibition of AKT signaling does not affect abundance of glycolytic intermediates. The differences between these results and those of our targeted, isotope tracing approach highlights the ways in which bulk metabolomics conceals dynamic changes in metabolic pathways. AKT directs glucose contribution to glucose 6-phosphate, UDP-glucose and glycogen synthesis independent of glycogen breakdown Our stable isotope tracing data (Figure 2) showed that AKT promotes the fractional contribution of glucose to G6P and UDP-glucose, and this preceded significant changes to endogenous sources of carbon. One way in which this could occur is by decreasing glycogenolysis to prevent the appearance of unlabeled G6P. To test this hypothesis, we evaluated [U-13C]-glucose metabolism in the presence of a glycogen phosphorylase inhibitor. Glycogenolysis accounts for about 45% of glucose production during the first 24 h of fasting in humans19 and an estimated two-thirds of HGP in conscious dogs.7 Insulin suppresses glycogen breakdown48 and promotes its storage.49 AKT controls both glycogen synthesis and breakdown through transcriptional and post-translational mechanisms,34,35,50 and there is no evidence that AKT regulates gluconeogenesis acutely, so we tested the hypothesis that AKT promotes glucose contribution to G6P and UDP-glucose through suppression of glycogen breakdown. We treated cells with glycogen phosphorylase inhibitor (GPI) in the presence or absence of MK2206, using the same experimental approach as in Figure 2. We found that treatment with GPI alone did not significantly increase the fractional contribution of ubiquitously labeled G6P or UDP-glucose to the total pool size, nor did it rescue the effect of AKT inhibition (Figures 5A–5C). Investigating the contribution of other isotopologues (Figures 5B and 5C), we found that AKT inhibition decreased [U-13C]-glucose contribution to G6P and UDP-glucose pools regardless of glycogen breakdown inhibition. Because GPI had minimal effect, we tested its efficacy in primary hepatocytes and found that it increased glycogen levels as expected when compared to glucose alone (Figure 5D). Overall, these data suggest that AKT rapidly promotes glucose contribution to G6P and UDP-glucose through a mechanism independent of glycogen breakdown. AKT directs glucose contribution to glucose 6-phosphate, UDP-glucose, and glycogen synthesis independent of glycogen breakdown (A) Fractional labeling and (B) fractional contribution of metabolite pools in primary rat hepatocytes treated with [U-13C]-glucose with or without 10 μM MK2206 and 5 μM GPI. Cells were treated as described in STAR Methods section [U-13C]-glucose isotope tracing. N = six replicates from two pooled experiments. (C) Ion counts from a representative tracing experiment included in the analysis done in (A and B). p values calculated using a one-way ANOVA followed by a post-hoc t test, values < 0.05 denoted by ∗(M+6), %(M+3), or # (M+0). Values < 0.005 and <0.0005 for M+6 denoted by ∗∗ and ∗∗∗. p values on top of, or adjacent to, specific isotopologues compare MK2206 and Veh, or GPI+MK2206 and GPI. Time points indicate the time cell extraction buffer was added to wells. (D) Total glycogen content in primary rat hepatocytes following treatment with 20 mM glucose with or without 25 μM FSK and 5 μM GPI for 4 h. p values calculated using one-way ANOVA and post-hoc t test, ∗p value <0.05, ∗∗p value<0.005. (E) Glycogen synthesis in primary rat hepatocytes that were fasted overnight in serum and glucose-free media, pretreated for 30 min with 10 μM MK2206 or DMSO in overnight media, and then treated with 1 μCi [U-14C]-glucose and 10 mM cold glucose with inhibitors or DMSO for 30 min. N = four to six replicates pooled from five experiments. ∗p value < 0.05, calculated using a one-way ANOVA followed by a post-hoc t test. (F) Western blot of glycogen synthase (GS) phosphorylation at Ser641 in primary rat hepatocytes incubated overnight in serum- and glucose-free DMEM, pretreated for 30 min in overnight media with 10 μM MK2206 or DMSO, and then treated with 10 mM glucose with inhibitor or DMSO for 30 min. (G) Quantification for blots shown in (F), phospho and total antibodies were normalized to beta-actin and then the phospho-antibody signal was normalized to its respective total protein. Blots and quantification in (F and G) represent 1–2 replicates per treatment group per experiment from 2 independent experiments, with p value calculated using Welch’s t test. All error bars represent ± SEM. Our results from stable isotope tracing (Figures 2 and 5) show a decrease in the synthesis of G6P and UDP-glucose from glucose with AKT inhibition, indicating that there may be a decrease in glycogen synthesis through the direct pathway. To determine the effect of acute AKT inhibition on macromolecule synthesis, we measured glycogen synthesis using the same experimental paradigm as in Figures 5A–5C but with [U-14C]-glucose. We found that AKT inhibition decreased glucose contribution to glycogen following a 30-min incubation with the radiolabel (Figure 5E), too rapid to be accounted for by a transcriptional mechanism. Of note, GPI treatment alone increased glycogen synthesis dramatically, highlighting that significant glycogen cycling occurs in primary hepatocyte culture. Nevertheless, MK2206 treatment suppressed glycogen synthesis in the presence of GPI (Figure 5E), demonstrating that AKT rapidly promotes glucose contribution to glycogen independent of inhibiting its breakdown. There are several known mechanisms that could potentially explain how AKT promotes glycogen synthesis post-translationally. G6P allosterically activates glycogen synthase, increasing glycogen storage.38,39,51 However, our unbiased metabolomics and stable isotope tracing data show that while glucose contribution to G6P is affected by AKT inhibition, the total pool size is unchanged (Figure S2), making this an unlikely explanation. Insulin signaling through AKT can also control glycogen synthesis through reversible phosphorylation of glycogen synthase. AKT inhibits GSK-3β,35 which phosphorylates glycogen synthase to inhibit it.52 Second, AKT can activate the PPP1R3G subunit of the protein phosphatase 1 (PP1) complex,34 stimulating its activating dephosphorylation of glycogen synthase and inhibitory dephosphorylation of glycogen phosphorylase.53 To determine if AKT increases glucose contribution to glycogen on this short timescale through control of glycogen synthase phosphorylation, we blotted our cell lysates for glycogen synthase Ser641 phosphorylation. Glycogen synthase is deactivated through hierarchical phosphorylation,54 in which multiple, sequential phosphorylation events lead to deactivation of the enzyme. The hierarchical phosphorylation of glycogen synthase begins with phosphorylation at Ser698 and Ser8 by protein kinase A (PKA).55 This permits subsequent phosphorylation by GSK-3β at Ser641, Ser645, Ser649, and Ser653.55 Phosphorylation at these sites reduces glycogen synthase enzyme kinetics.56,57,58,59 Insulin treatment reduces phosphorylation at the GSK-3β-regulated sites57 through AKT inhibition of GSK-3β35 or activation of protein phosphatase 1,53 thus an increase in Ser641 phosphorylation with MK2206 treatment would suggest reduced AKT inhibition of GSK-3β and decreased glycogen synthase activity. We found that AKT inhibition failed to increase Ser641 phosphorylation (Figures 5F and 5G), indicating that the decrease in glycogen synthesis could be occurring independent of glycogen synthase phosphorylation. This is consistent with previous reports that mutation of AKT-regulated phosphosites on GSK-3β does not impact glycogen synthesis in vivo.37 Coupled with the absence of consistent changes to the total G6P pool size (Figure S2), the data suggest additional pathways are responsible. The observed decrease in glycogen synthesis from glucose with acute AKT inhibition may be a potential consequence of the upstream defect in glucose contribution to G6P and UDP-glucose. Discussion Insulin’s control of HGP and stimulation of liver glucose uptake is well appreciated.4 Previous studies suggest involvement of hepatic AKT in insulin-mediated control of HGP, whether through suppressing glycogenolysis or gluconeogenesis.26,34,35 Data in this manuscript suggest that AKT signaling may also stimulate glycogen synthesis through the direct pathway, that is, from extracellular glucose and not gluconeogenic intermediates, by increasing glucose contribution to G6P. Previous studies have focused on the importance of HGP suppression for insulin-stimulated liver glucose uptake, specifically, inhibition of glycogenolysis. Data presented here suggest an additional mechanism independent of glycogen breakdown that may precede eventual changes to endogenous glucose production. Broadly, our study reveals an alternative to a substrate push mechanism based on our observation that AKT increases glucose contribution to glycogen synthesis without increasing the G6P or UDP-glucose pool size. The most robust effect of AKT signaling was on glucose contribution to G6P and UDP-glucose, as indicated by changes in the ubiquitously labeled fraction of each metabolite (Figures 2D and 5B). At certain time points, there were increases in the contribution of non-ubiquitously labeled G6P, indicating increased flux through gluconeogenesis (Figures 2D, 3A–3C, and 5B). A substrate push mechanism may be appropriate to explain the effect of MK2206 on endogenous carbon contribution to G6P, whereby transient decreases in the G6P pool caused by decreased glucose phosphorylation could drive an increase in its generation from gluconeogenic precursors. These potential transient changes, however, do not persist long enough to lead to significant differences in pool size (Figure S2A), making G6P allosteric activation of glycogen synthase an unlikely explanation for MK2206’s eventual effect on glycogen synthesis (Figure 5E). An advantage of stable isotope tracing compared to earlier work done with radioisotopes is that it allows a more complete network picture of insulin’s control of glucose metabolism. Earlier studies conducting clamps in the conscious dog model concluded that insulin did not significantly affect gluconeogenic flux,15,16 but authors acknowledged that experimental pitfalls could preclude accurate measurement. First, while insulin might suppress gluconeogenesis, this could be masked by its dominant effect on glycogen breakdown,6 whereby decreased G6P due to decreased glycogenolysis leads to an increase in gluconeogenesis, which would appear as unchanged flux. Second, experiments estimating gluconeogenic flux using 14C-phosphoenolpyruvate or 14C-lactate and measuring incorporation into 14C-glucose in the plasma may underestimate changes in gluconeogenic flux if insulin diverts 14C-G6P derived from HGP to glycogen synthesis instead of glucose export. Third, several studies raised skepticism regarding the accuracy of using 2H2O incorporation into glucose to measure gluconeogenesis vs. glycogenolysis, as glyceraldehyde 3-phosphate (GAP) exchange with fructose 1,6-bisphosphate or retention of a deuterium during conversion of GAP to dihydroxyacetone phosphate (DHAP) can lead to underestimation of gluconeogenesis.60,61 Using our approach, we were able to determine the dilution of our tracer with or without glycogen breakdown to approximate the contribution of gluconeogenesis. We also demonstrate that insulin signaling, at least through AKT, does indeed divert G6P to glycogen rather than downstream to glycolysis. Considerable effort has been made to quantify insulin’s control of gluconeogenesis.6,15,16,17,18 It has been proposed that gluconeogenic flux does not change significantly even with hyperglycemia,62,63 and that perhaps only fatty acid flux controls rates of gluconeogenesis.6,64 Consistent with this notion, Basu et al. showed that type 2 diabetes mellitus patients have elevated glycogenolysis but unchanged gluconeogenesis in response to a hyperinsulinemic clamp compared to mild diabetics or non-diabetic controls.65 Together, the literature suggests that insulin suppresses gluconeogenesis through inhibition of adipose tissue lipolysis and not through direct action at the liver. Our observations (Figures 3A–3C) suggest that insulin signaling via AKT may indirectly decrease gluconeogenesis secondary to promoting glucose phosphorylation. Interestingly, the increased M+1, M+2, and M+3 intracellular glucose in our experiments (Figures 3A–3C) was not at equilibrium with the glucose in the media, as those isotopologues did not become enriched in media glucose until at least 30 min. This indicates that intracellular glucose may become re-phosphorylated more readily than exported in primary hepatocytes. However, it is unclear if this is true in vivo. Our data also shed light on whether glycogen levels in the liver influence glucose phosphorylation and contribution to glycogen precursors, as has been reported in muscle.66 If changing glycogen levels influence glucose contribution to G6P and UDP-glucose, we would expect M+6 labeling of G6P to increase with inhibition of glycogen breakdown. Instead, we observed that GPI had a strong effect on glycogen synthesis (Figure 5E) with little effect on G6P labeling beyond 2 min (Figures 5B and 5C). This suggests that the effect of AKT inhibition on glucose contribution to G6P is independent of glycogen breakdown. Together, these data suggest the possibility of a mechanism where AKT acutely promotes extracellular glucose contribution to glycogen synthesis by rapidly increasing glucose phosphorylation. Previous work proposed that AKT increases glycogen synthesis post-translationally by indirectly controlling phosphorylation of glycogen synthase,35 although we did not detect a significant increase in glycogen synthase phosphorylation in isolated rat hepatocytes (Figures 5F and 5G). Additionally, while MK2206 decreased glycogen synthesis in the presence of GPI, glycogen synthesis in the combination group was elevated compared to MK2206 treatment alone (Figure 5E). This indicates that there is some spare capacity of glycogen synthase during AKT inhibition. If MK2206’s effect on glycogen synthesis from glucose depended on glycogen synthase, GPI treatment with MK2206 would fail to increase glycogen synthesis relative to the control or MK2206 alone. With regard to the mechanism by which AKT acutely promotes glucose contribution to G6P and then to glycogen, the effect of MK2206 on upper glycolysis implicates a change in GCK activity (Figures 2C–2E and 5A–5C). This is consistent with previous work demonstrating that insulin activates GCK.40,42,67,68 Our results are also consistent with a report showing that postprandial GCK activator treatment in rats’ increases glycogen deposition.69 G6Pase overexpression, which reduces G6P levels, reduces glycogen synthesis and glycolysis,70 while GCK overexpression increases G6P, glycogen deposition, and total levels of glycolytic intermediates.71 How AKT might be activating GCK is unclear, but it is well-known that GCK activation is regulated through its subcellular localization. During fasting, when intracellular glucose levels are low, GCK is sequestered in the nucleus by glucokinase regulatory protein (GKRP). Rising blood glucose levels postprandially favor glucose transport into the liver. Glucose diffuses into the nucleus, binds to GKRP, and consequently releases GCK to be transported back into the cytoplasm where it can phosphorylate glucose.72 Interestingly, we did not see a change in GCK localization in primary rat hepatocytes treated with 10 mM glucose with or without 10 μM MK2206 (Figure S5). This suggests that if the mechanism is GCK-dependent, it does not occur through its translocation. Beyond the short-term effects of AKT signaling on upper glycolytic metabolites, we also observed changes to glycerophospholipid metabolism. AKT inhibition increased CDP-ethanolamine and decreased CDP-choline (Figures 4D and 4E), which each contribute to glycerophospholipid synthesis through distinct arms. Thus, AKT may be increasing the phosphatidylethanolamine arm of phospholipid synthesis through post-translational regulation. CCTα is the enzyme catalyzing CDP-choline synthesis, and its activity is regulated by mTORC1 in the liver.73 AKT activates mTORC1 to drive lipogenesis,24 thus it is possible that loss of mTORC1 activity due to AKT inhibition decreased CDP-choline synthesis. It is unclear why a loss in AKT activity leads to an increase in CDP-ethanolamine levels, although it may be due to a compensatory mechanism in which PE synthesis is upregulated to increase PC generation via phosphatidylethanolamine N-methyltransferase (PEMT) activity. We also observed in our stable isotope tracing experiments (Figures 2E and 5C) that acute AKT inhibition decreased G3P levels. This was recapitulated in our bulk metabolomics experiments, where G3P trended to be reduced (p = 0.07, Figures S4C and S4D). G3P is acylated by the GPAT enzymes in the DAG synthesis pathway, which insulin has been shown to activate in adipocytes.74 However, the isoform of GPAT1 expressed in hepatocytes does not contain an AKT consensus phosphorylation motif, so it is unlikely that G3P contribution to DAG synthesis was increased by acute AKT inhibition. G3P is synthesized from DHAP by GPD1, but GPD1 also does not contain an AKT consensus phosphorylation motif. Thus, future studies are needed to investigate how AKT acutely controls G3P levels in hepatocytes. In summary, we have provided evidence that AKT acutely stimulates glucose phosphorylation and contribution to glycogen, independent of glycogen breakdown. This may be occurring independent of glycogen synthase regulation. Future studies are needed to determine the precise mechanism by which AKT is achieving this, with a glucokinase-dependent mechanism possible. Considering that decreased liver GCK expression in humans correlates strongly with worsened blood glucose control,27 and diabetes patients exhibit decreased GCK activity,75 understanding this mechanism likely holds implications for treating hyperglycemia. Limitations of the study With acute inhibition of AKT, we did not observe significant differences in the levels of lactate43,44,45 or pyruvate43 as previous studies had reported, but we did detect the difference in G3P reported by Terrettaz et al.43 Discrepancies may reflect differences between insulin and AKT-specific actions, given these previous reports involved insulin treatment. Additionally, one study described above44 cultured primary hepatocytes for 48 h, at which point primary hepatocytes begin to de-differentiate. There may also be differences in the degree to which insulin signaling is engaged in control cells. In our hands, even serum-starved primary rat hepatocytes exhibit residual AKT phosphorylation (Figure 2B, lane 1), demonstrating that this pathway is difficult to suppress in culture without triggering apoptosis.76 However, without being able to compare phosphorylation of signaling intermediates between studies, it is difficult to discern the root of discrepancies. This is a limitation of our study. A second limitation of our study is that it was conducted entirely in culture to detect immediate changes in metabolism. It is unclear whether our conclusions regarding AKT’s acute control of glucose production or phosphorylation hold true in vivo. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Megan Stefkovich (stefkovichmegan@gmail.com). Materials availability This study did not generate new unique reagents. Data and code availability •Stable isotope tracing and bulk metabolomics data have been deposited into Metabolomics Workbench77 and are publicly accessible as of the date of publication. Study IDs are as follows: ST004471, ST004472, ST004473, ST004474, ST004475, ST004476, and ST004477. Stable isotope tracing and bulk metabolomics data have been deposited into Metabolomics Workbench77 and are publicly accessible as of the date of publication. Study IDs are as follows: ST004471, ST004472, ST004473, ST004474, ST004475, ST004476, and ST004477. •Raw data from Figures 2, 4, 5 and S5 were deposited on Mendeley Data: https://doi.org/10.17632/96z436zcpj.1. Raw data from Figures 2, 4, 5 and S5 were deposited on Mendeley Data: https://doi.org/10.17632/96z436zcpj.1. •This article does not report original code. This article does not report original code. •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.