Tracking the carbons supplying gluconeogenesis

glucose-6-phosphatase.

As the burden of type 2 diabetes mellitus (T2DM) grows in the 21st century, the need to understand glucose metabolism heightens. Increased gluconeogenesis is a major contributor to the hyperglycemia seen in T2DM. Isotope tracer experiments in humans and animals over several decades have offered insights into gluconeogenesis under euglycemic and diabetic conditions. This review focuses on the current understanding of carbon flux in gluconeogenesis, including substrate contribution of various gluconeogenic precursors to glucose production. Alterations of gluconeogenic metabolites and fluxes in T2DM are discussed. We also highlight ongoing knowledge gaps in the literature that require further investigation. A comprehensive analysis of gluconeogenesis may enable a better understanding of T2DM pathophysiology and identification of novel targets for treating hyperglycemia.
Glucose serves as a fuel source for many tissues and is the primary source of energy for neurons, renal medullary cells, and red blood cells (1). Circulating blood glucose levels are maintained in a narrow range (3.9-7.1 mmol/liter), and the liver plays a critical role in maintaining glucose homeostasis (2). The liver stores glucose in the form of glycogen and releases glucose into circulation by either glycogenolysis or gluconeogenesis. In the fed state, hepatic glucose production is suppressed by insulin secretion, and the glucose ingested is stored in part as glycogen.
During a short-term fast, the liver maintains euglycemia through glycogenolysis. During longer periods of fasting, as glycogen stores are depleted, the liver relies on gluconeogenesis to maintain euglycemia (3).
Gluconeogenesis is an intricate process that requires several enzymatic steps (Fig. 1), which are under the regulation of hormones, nutrient intake, stress conditions, and substrate concentrations. Occurring in hepatocytes and renal cortical cells, gluconeogenesis functions as a biosynthetic pathway responsible for countering the glycolytic breakdown of glucose.
T2DM, a chronic medical condition characterized by hyperglycemia, has reached pandemic proportions affecting over 400 million adults globally (4). A major pathophysiological tenet of T2DM is increased hepatic gluconeogenesis with rates elevated up to 40% (5). In T2DM, gluconeogenesis remains a significant contributor to hepatic glucose production both under fasting conditions and after meal intake (6). Hyperinsulinemia during a hyperinsulinemic-euglycemic clamp, where exogenous insulin is infused as supraphysiologic amounts with concurrent infusion of glucose to maintain a certain blood glucose level, completely suppressed glycogenolysis but only reduced gluconeogenesis by about 20% (7). Longstanding hyperglycemia is associated with both macrovascular complications, such as heart attacks and stroke, and microvascular complications affecting retinal, renal, and nerve tissues (8), which help drive the costs of diabetes care to over $322 billion annually in the United States alone (9).
The rise in obesity has led to increased prevalence of T2DM and nonalcoholic fatty liver disease (NAFLD). More than 1 in 3 adult Americans have obesity (10), whereas 1 in 4 have NAFLD (11) and nearly 1 in 10 have T2DM (12). Gluconeogenesis rates are elevated in patients with obesity even without overt diabetes (5) as well in patients with NAFLD (13). Based on these epidemiologic data, most patients with obesity and NAFLD do not develop overt hyperglycemia, highlighting fundamental differences within these patient populations. Understanding gluconeogenesis across distinct but related metabolic conditions might lead to greater insights into underlying pathophysiology and more targeted therapies.
Given the health burden of T2DM and the public health impact, there has been significant research on underlying disease processes leading to varied pharmacologic therapies for the disease. Despite 14 distinct T2DM medication classes currently approved, hyperglycemia remains a persistent challenge for patients, and physicians need to be mindful of avoiding hypoglycemia and minimizing side effects (19,20). Thus, novel therapeutic approaches are warranted. To better target gluconeogenesis, a key question becomes the origin of the carbons that account for the increased glucose production in T2DM. Further, many medications for T2DM affect gluconeogenesis rates directly and indirectly (20), although their mechanisms of action could be better known if we had an accurate assessment of gluconeogenesis flux.
This review discusses the current understanding of gluconeogenic flux based on isotope tracer data primarily from experiments in humans but also from selected animal and in vitro models to fill in where human data are lacking. We focus on how different gluconeogenic precursors contribute to the process, how these contributions may differ in T2DM, and new findings that may question each precursor's relative role in the process. We also discuss how these precursors' concentrations change in T2DM and how precursors themselves may regulate gluconeogenesis.

Metabolomics overview
Given the complexities behind biochemical processes, including gluconeogenesis, researchers have studied metabolites directly to gain insight. The term metabolites refers to all endogenous small molecules (,1,500 Da) involved in metabolic reactions, including substrates, intermediates, and products (21). A metabolite's circulating concentration is based on its synthesis, dietary intake, and degradation as well as uptake and release from other body compartments, such as liver, muscle, and adipose tissue (22). Metabolites most directly reflect physiologic and pathologic conditions in an organism. The entire complement of metabolites in cells, tissues, or whole organisms makes up the metabolome, and metabolomics can measure these molecules with precision and accuracy. There are 6,500 and counting discrete metabolites in the human metabolome (23,24).
Nontargeted, or untargeted, metabolomics compares two different biological conditions, including different disease states, genetic alterations, or drug treatments, and identifies metabolite changes in response to a manipulation. This unbiased approach can generate novel hypotheses regarding metabolites and pathways. However, one needs to study metabolic flux (metabolite flow per time) to fully understand pathway activity (25). To obtain a more thorough understanding of metabolite regulation and quantify fluxes under various conditions, one must introduce a labeled metabolite and "follow the label" (26). Paired with isotope-labeled metabolites, targeted metabolomics can measure metabolic flux as heavy atoms from a labeled substrate are detected in downstream metabolic products across different time points.
Several different methodologies can help with determining metabolic flux. NMR and MS are two commonly used analytical platforms for metabolite detection and quantification. NMR is a highly reproducible technique that can provide fractional abundance of an isotope at a specific atom position (27). For example, a 12 C-1 H interaction gives a different peak than a 13 C-1 H interaction on an NMR spectrum. NMR yields significant structural information about a molecule as adjacent nuclei within that molecule interact via spin-spin coupling to produce distinct peaks. Disadvantages of NMR include low sensitivity, making measurement of metabolites with low concentrations difficult (28). Because there is little sample preparation with NMR, there is no chromatographic separation of structurally similar compounds leading to overlapping resonances, which can make the charting of biochemical pathways difficult. MS is a highly sensitive technique that can detect metabolites even at low concentrations. MS involves fragmenting labeled or unlabeled compounds through ionization by electron impact ionization or chemical impact ionization (29). After going through the ionization source, fragmented ions pass through a mass analyzer with a specific mass/charge (m/z) ratio and retention time (29). MS can detect the subtle mass differences between isotopes. For example, 3-[ 13 C]lactate (m 1 1), which has a label only on lactate's third carbon, has an m/z ratio and retention time in the mass spectrometer different from those of the unlabeled lactate (m 1 0). Chromatographic separation provides high resolution even between structurally similar molecules. Disadvantages of MS include the need for sample derivatization, which can lead to sample loss (30). MS often cannot tell you specifically where in the molecule is the labeled atom (i.e. which carbon is labeled in an M 1 1 lactate molecule).
For a detailed description of the established methods of measuring gluconeogenesis and glycogenolysis using MS and NMR, please refer to a review by Chung et al. (31). Others have written on the practical applications related to in vivo research with metabolomics (32)(33)(34)(35).

Carbon contribution to gluconeogenesis
Given the powerful tools of NMR and MS within metabolomics, one can study how the liver makes glucose under fasting conditions. Glucose is a six-carbon molecule whose concentrations remain relatively constant in the fasted state in metabolically healthy individuals but can rise in subjects with T2DM (1). Gluconeogenic precursors come from noncarbohydrate sources, including lactate, glycerol, and amino acids. The two most relevant amino acids for gluconeogenesis are alanine and glutamine. Glutamine gluconeogenesis is predominantly in the kidney, whereas alanine gluconeogenesis is predominantly in the liver (36).
Infusion of a carbon-labeled precursor of glucose is commonly used to study gluconeogenesis. Using isotope dilution techniques, the ratio of labeled glucose over the labeled precursor equates to the percentage contribution of the precursor to glucose production. Numerous studies assessing substrate contribution to gluconeogenesis in humans were done in the 1960s-1990s using advanced tools for the time. Results vary based on the isotope tracers used, test conditions, and methods of calculation. Although we cannot cover all tracer experiments conducted, we will highlight relevant studies (Table 1) to illustrate key concepts as well as point out inconsistencies in the literature that require reconciliation.

Lactate
As shown in Fig. 2, the Cori cycle depicts shuttling of lactate from anaerobic glycolysis in skeletal muscle cells to the liver to feed gluconeogenesis (37). Many consider lactate the predominant gluconeogenic precursor (38)(39)(40). Studies in healthy humans have shown that lactate contributes as little as 7% (41) to as much as 18% (40) to plasma glucose after an overnight fast. Comparing subjects with T2DM and healthy controls, there was a 2-fold increase in lactate incorporation into glucose in T2DM (42,43).

Alanine
With muscle catabolism, alanine is released into circulation, undergoes deamination in the liver to become pyruvate, and later glucose as depicted in Fig. 2 (44). Studies have shown a 6-11% contribution of the amino acid to glucose production after an overnight fast in healthy humans (40,41,45). The role of alanine's contribution to gluconeogenesis in T2DM remains less clear. Some have documented a 2-fold increase of gluconeogenesis from alanine in subjects with T2DM compared with healthy controls (46,47). However, Consoli et al (42). concluded that those with T2DM did not have an increase in alanine's contribution to glucose production compared with controls. In a separate study, Chochinov et al. (48) concluded that gluconeogenesis from alanine decreased from 11% in controls to just 3% in subjects with T2DM. These conflicting studies make it difficult to assess what role, if any, alanine has in T2DM hyperglycemia.

Glutamine
Glutamine contributes to gluconeogenesis by converting to glutamate, which gets deamidated to a-ketoglutarate (37). Fig.  1 shows how a-ketoglutarate can then enter the Krebs cycle and ultimately feed gluconeogenesis. Glutamine contributed 5-8% to glucose production in healthy humans in prior studies (49,50). With T2DM, the conversion of glutamine to glucose nearly doubled (47).

Glycerol
Lipolysis of triglycerides in adipocytes releases glycerol into circulation, which can become glucose in the liver. The contribution of glycerol to glucose production in metabolically healthy humans ranged from 3 to 7% (51)(52)(53)(54). In T2DM, glycerol's contribution to glucose production increased to 6-10% which was significantly higher compared with healthy controls (52,53).

Direct versus net carbon contribution
Like many biologic processes, the paths of gluconeogenic precursors to glucose production exist in both directions, such that glucose itself can lead to the production of many gluconeogenic precursors. Many tracer studies on gluconeogenesis, including references in this review, primarily report direct carbon contribution of precursors to gluconeogenesis but not net carbon contribution. For example, as in Fig. 3, molecules M and N can contribute to each other's production via reversible reactions. For molecule M, 60% of its flux goes toward the molecule N, whereas 40% goes to the molecule O. For molecule N, only 40% of its flux goes to molecule M, whereas 60% goes to P. If one were to give an isotope-labeled tracer of molecule M, one would see a 60% direct contribution of M to N. However, a dual tracer study with M and N tracers would show smaller net efflux from M to N. In general, introducing a labeled tracer of a molecule can give an idea of where that molecule is going. However, it does not give information about where that specific molecule is coming from. This requires introduction of other labeled substrates to obtain an integrated flux network that can distinguish direct and net contributions.
Glucose itself is major contributor to many gluconeogenic precursors. Perriello et al. (55) showed in metabolically healthy humans that circulating glucose provided 67, 41, and 13% of the carbons for plasma lactate, alanine, and glutamine, respectively. These contributions were via the Cori cycle, glucose-alanine cycle, and glucose-glutamine cycle (Fig. 2). No human data exist to show how much, if at all, glucose contributes to glycerol production. However, studies in metabolically healthy dogs showed that less than 2% of glycerol's carbons come from glucose (56).
Given the reciprocal fluxes between glucose and its precursors, infusing only 13 C tracers of gluconeogenic precursors may not be enough. As the direct and net contribution of a precursor may not be congruent, studies that also give a [ 13 C 6 ]glucose tracer are needed to determine the source of non-glucosederived carbons that fuel gluconeogenesis.
The difference between direct and net contribution may not merely be academic but also practical. A better understanding of gluconeogenesis determinants could lead to more rationally designed and targeted T2DM treatments. For example, one could find a pharmacologic means to stop hepatic conversion of a specific gluconeogenesis substrate to glucose by blocking either key enzymes or substrate-specific transporters in the liver. Alternatively, one could block reabsorption of specific gluconeogenesis substrates in the renal tubules, leading to increased urinary excretion analogous to currently approved sodium-glucose co-transporter-2 inhibitors, which block renal reabsorption of glucose (57). Such theoretical agents would need to be tested in preclinical models and assessed for off-target metabolic effects.

Fluxes between gluconeogenic precursors
Not only can gluconeogenic precursors contribute to glucose production, they can also contribute to the production of other gluconeogenic precursors. Returning to our example, in Fig. 3, for concentrations of molecule M to remain constant, molecule Q must feed molecule M. Otherwise, molecule M concentrations would decrease, given its net effluxes toward molecules N and O.
Lactate may be an important direct carbon contributor to gluconeogenesis, given its high turnover and intimate coupling with glucose via the Cori cycle. However, the Cori cycle is not a net producer of glucose and cannot sufficiently explain the hyperglycemia seen in T2DM (58). Similarly, many argue that the glucose-alanine cycle does not provide net substrate for gluconeogenesis and that its main physiologic function is ammonia transport (44,45). Therefore, for increased gluconeogenesis to contribute to T2DM hyperglycemia, the carbons supplying gluconeogenesis must come from other substrates.
Recently our group has shown in mice, using [ 13 C 3 ]lactate, [ 13 C 3 ]glycerol, and [ 13 C 3 ]glucose tracers administered over different experimental days, that lactate is mainly recycled during the fasting period such that lactate is the largest direct contributor to gluconeogenesis but provides minimal new glucose carbons (59). In the same study, glycerol contributed to the Table 1 Direct contribution of gluconeogenesis precursors and glycogen to hepatic glucose production after an overnight fast in humans as determined by isotope tracer experiments The T2DM column indicates relative changes to glucose contribution from precursor in the setting of T2DM as compared with metabolically healthy controls.   carbons of glucose both by direct conversion in the liver and by first converting to lactate, which then became glucose. Glycerol was the most significant source of new carbons for gluconeogenesis after a 12-h fast, contributing over 50% of the net carbons for gluconeogenesis. Lactate contributed a small proportion to glycerol molecules in mice, and this coincides with rat data using a single intraperitoneal injection of [ 13 C 3 ]lactate that labeled intramuscular glycerol (60). The enzymes glycerol-3phosphate dehydrogenase and glycerol-3-phosphate phosphatase are needed to generate glycerol from the triose phosphate pool (61). To date, no human experiments have assessed how glycerol and lactate may contribute to each other's production. Glycerol may have two different metabolic fates based on the initial site of metabolism, and several tissues have high expression of glycerol kinase, which converts glycerol to glycerol 3phosphate (62). Glycerol 3-phosphate can then enter the triose phosphate pool as an intermediate for glycolysis or gluconeogenesis (Fig. 1). The liver and kidney have high expression of glycerol kinase so that glycerol can contribute directly to glucose production in these two gluconeogenic organs (63). Peripheral tissues such as intestines, lymphatics, and spleen also express glycerol kinase but not gluconeogenic enzymes, allowing glycerol to become a source for lactate (63). The fates of glycerol delivered directly to gluconeogenic and nongluconeogenic organs have not been directly tested in humans. This would require invasive cannulation of certain arteries and veins to isolate certain organs, adding significant risks for subjects.
Infusing [ 13 C 3 ]glycerol into metabolically healthy humans that fasted for 60 h, Landau et al. (64) estimated that the gastrointestinal, renal, and muscle tissues accounted for 63% of the glycerol utilization and that the remaining 37% must be metabolized by other tissues that express the enzyme glycerol kinase. Knowing which tissues in the body process this remaining glycerol can help us further understand glycerol's role in T2DM hyperglycemia. In T2DM, where there is increased lipolysis and circulating glycerol, it is unknown whether that additional circulating glycerol gets evenly distributed between gluconeogenic or nongluconeogenic organs or if one set of organs has an increased metabolism of the substrate.
Alanine can also supplement the lactate pool via a pyruvate intermediate, and one study showed that alanine contributed 16% of the carbons to circulating lactate in healthy humans (45). In contrast, a study in healthy dogs showed that lactate contributed to 70% of the alanine pool (65). There are no known human studies to assess how much glutamine contributes to lactate production or vice versa. However, in vitro work from human fibroblasts has depicted glutamine converting to lactate (66).
Glutamine and alanine can contribute to each other's production as glutamine-derived glutamate can interchange with alanine via the enzyme alanine aminotransferase (67). Using [ 14 C 5 ]glutamate and 3-[ 13 C]alanine tracers at the same time, studies have shown glutamine to be more quantitatively important in delivering protein-derived carbons to glucose in both healthy human subjects (50) and those with T2DM (47).
Finally, it is unknown whether amino acids and glycerol provide any carbon contribution to each other. Prior tracer studies gave labeled gluconeogenic substrates and assessed a limited number of downstream products. However, the metabolome is more interconnected, and gluconeogenic precursors can supply each other with carbons directly or through intermediate metabolites.
Given the various fluxes between gluconeogenic precursors as shown in Fig. 4 in humans with and without T2DM, there is a need for comprehensive experiments spanning multiple tracers in the same human or animal subject to assess the carbon flow. Such experiments may yield information that may lead to certain precursors as the significant culprit carbon contributors. Blocking these substrates pharmacologically from becoming glucose or other gluconeogenic precursors could be a potential treatment strategy. Blocking one pathway may or may not be sufficient to lower hepatic glucose production, as pathways may be redundant or leaky such that overall glucose production may not be affected even if one precursor is prevented from becoming glucose. However, attempts at blocking such pathways can only be made with a sound understanding of carbon flux in the first place.

Changes to gluconeogenic precursor levels in T2DM
Whereas it is critical to study metabolic fluxes of gluconeogenic precursors, it is also pertinent to look at circulating concentrations of precursors under healthy and T2DM conditions. For the liver to make an excess of glucose in T2DM, one might expect circulating precursor levels to change in T2DM compared with metabolically heathy controls. Precursor levels could be elevated to allow for increased hepatic substrate delivery, or precursor levels could decrease due to increased hepatic substrate utilization. Both increased substrate delivery and utilization could also occur without affecting overall circulating precursor levels. Accounting for both metabolite levels and fluxes may yield a more thorough understanding of gluconeogenic changes in T2DM.

Lactate
Plasma lactate levels are elevated in T2DM compared with metabolically healthy humans in some studies (68,69) but not all (42). Lactate turnover, or the amount of lactate appearing in circulation at any given moment, is also increased in T2DM (42). Increased levels of lactate occur in obesity due to decreased blood flow in adipose tissue causing local hypoxia and increased lactate production (70). Insulin resistance in skeletal muscles was associated with decreased oxidative capacity and greater lactate production (71)(72)(73). Despite these mechanisms, the carbon sources for the increased lactate level in T2DM would still be primarily glucose, so lactate alone could not account for hyperglycemia seen in T2DM. Of note, there are no storage reservoirs of lactate in the body compared with amino acids (skeletal muscle tissue) and glycerol (adipose tissue).

Amino acids
Circulating alanine levels were increased in T2DM in some studies (68,74) but not others (42,47,48). Alanine turnover was also increased in T2DM in the studies that have measured it (42,47). Some studies have shown that increased glutamine levels were associated with a decreased risk of developing T2DM (74)(75)(76), whereas others showed no association between glutamine levels and T2DM risk (77)(78)(79). One study comparing glutamine turnover in T2DM and healthy controls showed no difference between the two cohorts (47).
Insulin is an anabolic hormone that promotes protein synthesis and prevents its breakdown (80). Chevalier et al. (81) showed that increased protein catabolism in obese nondiabetic subjects correlated with gluconeogenesis derived from amino acids. However, in T2DM, a condition with insulin resistance and higher compensatory insulin levels, muscle turnover as assessed by leucine turnover was unchanged in two studies (82,83). Under isoaminoacidemic, hyperinsulinemic, euglycemic clamp conditions where amino acid, insulin levels, and glucose levels are held at constant levels by exogenous infusions, protein anabolism was blunted in men with T2DM compared with healthy controls (83). This suggests a defect in protein synthesis under insulin-resistant conditions in men. In women, this effect was not seen, leading to the potential sex differences in protein metabolism under diabetic conditions. T2DM is associated with lower skeletal muscle mass (84,85), and decreased muscle mass is correlated with poorer glycemic control (86). Given these findings of decreased muscle mass with variable amino acid flux in T2DM, it remains difficult to quantify how much amino acids supply gluconeogenic carbons under diabetic conditions.

Glycerol
Circulating glycerol concentrations and glycerol turnover were consistently higher in subjects with T2DM compared with healthy controls in three studies that assessed the parameters (52,53,87). Insulin resistance leads to increased lipolysis, which allows for greater release of glycerol into circulation, and T2DM is routinely linked with increased fat mass (88). This would allow glycerol to be a net carbon contributor to gluconeogenesis in T2DM.

Insights from in vitro experiments
The tracer studies described so far in this review have been in vivo experiments, which are ideally suited to study wholebody metabolism accounting for organ cross-talk via hormones and substrates circulating at physiologic concentrations. However, in vitro experiments have utility, including the ability to more closely control testing conditions. In vitro experiments can be done much more quickly and cheaply to generate hypotheses and assess feasibility prior to scaling up to animal and human studies. Further, in vitro experiments can discern differences in metabolism of metabolites across different tissues without having to invasively cannulate blood vessels.
As an example, investigators can use hepatocytes given labeled precursors and glucose production assays to assess precursor utilization. Kaloyianni et al. (39) used 14 C-labeled precursors at physiologic concentrations in rat primary hepatocytes and showed lactate as the major precursor of glucose, accounting for 60% of the glucose formed. Glutamine and alanine each accounted for ;10% of glucose production, whereas serine, glycine, and threonine accounted for less than 5% each. One notable omission in this model was glycerol.
In contrast, our group showed, using mouse primary hepatocytes given 13 C-labeled substrates at physiologic concentrations, that glycerol accounted for over 75% of the glucose carbons labeled (89). Specifically, labeled glycerol yielded enrichments of m 1 3 and m 1 6 glucose, signifying glycerol as a direct carbon contributor to glucose, whereas labeled pyruvate/lactate yielded a mixed distribution pattern (m 1 1 through m 1 6), suggesting carbon loss via tricarboxylic acid cycle intermediates. This is consistent with findings by Hui et al. (90) that showed that circulating [ 13 C 3 ]lactate primarily labeled Krebs cycle intermediates in fasting mice in all tissues except brain.
Studying fatty livers from rats given a high caloric diet and nonfatty livers from rats given a control diet, Maeda Junior et al. (91) studied glucose production rates by perfusing labeled gluconeogenic precursors. Glycerol infusion led to increased glucose production rates in fatty livers, whereas lactate and lactate plus pyruvate infusions decreased glucose production in fatty livers. The addition of glucagon or the long-chain fatty acid stearate increased glucose production from these substrates, although only in fatty livers. In a separate study using labeled glutamine and labeled alanine, the same research group showed decreased capacity to produce glucose from these two amino acids in fatty rat livers compared with nonfatty rat livers (92). With fatty liver disease, commonly seen in conjunction with T2DM, hepatocytes may have a shift in substrate utilization for gluconeogenesis. However, this requires further exploration in humans.
One must consider how gluconeogenic precursors regulate enzymes relevant to gluconeogenesis. In vitro studies have shown that glycerol induces G6Pase expression in mouse primary hepatocytes (89) and rat hepatoma FAO cells (93). Yoshida et al. (93) showed that glycerol induced G6Pase expression in mouse hepatocytes via binding of the G6Pase promoter region in conjunction with hepatocyte nuclear factor 4a (HNF4a) binding to the promoter region. Glycerol reduced PEPCK expression in some in vitro studies (89) but not all (93). In contrast, lactate and pyruvate did not affect G6Pase and PEPCK expression in mouse primary hepatocytes (89). It is unknown whether amino acids affect expression of these two key gluconeogenic enzymes. Whereas changes in mRNA expression of PEPCK and G6Pase did not correlate with changes in gluconeogenic flux in prior studies (94,95), these changes might provide insight into the substrates used to maintain such fluxes. Glycerol has a much shorter pathway to generate glucose as it enters into the middle portion of the gluconeogenic pathway. Pyruvate, and lactate via pyruvate, enter gluconeogenesis after converting to oxaloacetate in mitochondria and require transport to the cytosol via the malate-aspartate shuttle (96). Glycerol can potentially shift hepatocyte glucose production away from pyruvate and lactate and toward glycerol gluconeogenesis by inducing G6Pase and repressing PEPCK. More investigation is needed to study how T2DM affects gluconeogenic enzymes and whether acute and chronic changes in precursor concentrations, as discussed above, affect gluconeogenic flux.

Gluconeogenesis contribution to glycogen stores
The liver produces glucose for release into systemic circulation as well as storing glucose in the form of glycogen. Hepatic glucose production is a combination of both glycogenolysis and gluconeogenesis. In humans, glycogen is the single greatest source of hepatic glucose production after an overnight fast. However, reports vary on the exact contribution of gluconeogenesis to hepatic glucose output, ranging from 30 to 60% after an overnight fast in healthy humans, depending on the method used (6,(97)(98)(99)(100). In T2DM, glycogen stores and glycogenolysis rates are diminished, and gluconeogenesis accounts for a higher percentage of hepatic glucose output (101). Further, a portion of the glycogen pool undergoes simultaneous synthesis (glycogenesis) and breakdown (glycogenolysis) in a process called glycogen cycling, and a review by Landau discusses different methods to measure glycogen cycling (102). Fig. 1 shows that hepatic glycogen is synthesized from glucose 6-phosphate (103), whose carbons come from an intact glucose molecule (104) or gluconeogenic precursors (105). Thus, it is possible for a gluconeogenic precursor to get stored as glycogen and later be released as a glucose molecule. Hellerstein et al. (106) showed in healthy humans that two-thirds of the glucose produced from gluconeogenesis was released into circulation, whereas one-third remained in the liver for glycogen deposition and cycling after an overnight fast. Thus, current methods that assess precursor contribution to gluconeogenesis underestimate the exact quantity of contribution as a significant portion of the carbons are stored in glycogen for later release. Whether the partitioning of gluconeogenic products between hepatic glucose release and glycogen storage is altered in T2DM also remains unknown.

Untargeted metabolomics studies
Alongside targeted metabolomics studies with isotope tracers, many studies have used untargeted metabolomics to find plasma and urine biomarkers for T2DM, and several reviews expound on this topic (107)(108)(109)(110)(111). Collectively, these studies show that a myriad of metabolites derived from amino acids, lipids, carbohydrates, and nucleotides are altered in T2DM (107), and these metabolites vary across disease progression (109). Despite these broad changes, it is difficult to know what they mean in terms of underlying pathophysiologic mechanisms and whether they contribute to the insulin resistance and hyperglycemia in T2DM or are a byproduct of the underlying disease process. Specifically, we are not aware of any studies that correlate biomarkers directly with gluconeogenic flux as determined by isotope tracer infusion. Such studies could be enlightening as they could inform us of the determinants of hyperglycemia in an individual. Given the phenotypic heterogeneity of T2DM (112), having biomarkers that correlate with increased gluconeogenic flux could lead to more personalized treatments for patients with T2DM that directly target gluconeogenesis.

Perspective and future directions
There is no normal range for carbon contribution to gluconeogenesis. Current analytical techniques offer precise and reproducible measurements but not necessarily absolute measurements. Whereas some studies have been more comprehensive than others, all remain limited in scope, given the complexity of the subject matter. Investigations using multiple tracers in the same subject using the same analytical technique might bring us closer to reconciling direct and net carbon contribution to gluconeogenesis.
Consensus is also needed among thought leaders in the field regarding optimal analytical techniques, sample preparation methods, and data calculations. Subject preparation prior to experimentation, including fasting duration, preceding meal intake, and medication management, need to be addressed to make studies comparable. If investigators conducted isotope tracer experiments in a more uniform fashion, results across different studies could become more comparable. Data from different experimenters could be then integrated into a flux network to better understand carbon flow in metabolism.
In summary, experiments with isotope tracers have led to significant advances in the quantification of gluconeogenic flux. To fully understand hepatic glucose output, one needs an accurate assessment of the input from gluconeogenic precursors and glycogen, which requires ongoing investigation. Such results can shed further insight into human physiology as well as relevant clinical conditions, including T2DM, obesity, and fatty liver disease.
Funding and additional information-F. E. W. was supported by NIDDK, National Institutes of Health, Grant R01DK063349. A. M. S. was supported by NCATS, National Institutes of Health, Grant KL2TR003018. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.