Loss of [13C]Glycerol Carbon via the Pentose Cycle

Whereas many reports substantiated the suitability of using [2-13C]glycerol and Mass Isotoper Distribution Analysis for gluconeogenesis, the use of [13C]glycerol had been shown to give lower estimates of gluconeogenesis (GNG). The reason for the underestimation has been attributed to asymmetric isotope incorporation during gluconeogenesis as well as zonation of gluconeogenic enzymes and a [13C]glycerol gradient across the liver. Since the cycling of glycerol carbons through the pentose cycle pathways can introduce asymmetry in glucose labeling pattern and tracer dilution, we present here a study of the role of the pentose cycle in gluconeogenesis in Fao cells. The metabolic regulation of glucose release and gluconeogenesis by insulin was also studied. Serum-starved cells were incubated for 24 h in Dulbecco's modified Eagle's media containing 1.5 mm[U-13C]glycerol. Mass isotopomers of whole glucose from medium or glycogen and those of the C-1—C-4 fragment were highly asymmetrical, typical of that resulting from the cycling of glucose carbon through the pentose cycle. Substantial exchange of tracer between hexose and pentose intermediates was observed. Our results offer an alternative mechanism for the asymmetrical labeling of glucose carbon from triose phosphate. The scrambling of 13C in hexose phosphate via the pentose phosphate cycle prior to glucose release into the medium is indistinguishable from dilution of labeled glucose by glycogen using MIDA and probably accounts for the underestimation of GNG using 13C tracer methods.

In the liver, the phosphorylation of glycerol and its subsequent conversion to glyceraldehyde phosphate (GAP) 1 and dihydroxyacetone phosphate (DHAP) provides the basis of an important isotopomer method for the determination of gluconeogenesis using 13 C-labeled glycerol (1). Whereas many reports substantiated the suitability of using [2-13 C]glycerol and MIDA for gluconeogenesis (2,3), the use of [U-13 C]glycerol had been shown to give lower estimates of gluconeogenesis (4). The rea-son for the underestimation has been attributed to the cycling of triose phosphate and glycerol and the lack of complete equilibrium between GAP and DHAP leading to asymmetric isotope incorporation as well as zonation of gluconeogenic enzymes and a [ 13 C]glycerol gradient across the liver during gluconeogenesis. How these different factors may affect the precision and accuracy of the MIDA approach in the estimation of gluconeogenesis has not been demonstrated (1,4,5).
When rat liver was perfused after a 2-day fast with [U-13 C]glycerol, M1 and M2 glucose mass isotopomers were found in addition to the M3 and M6 of glucose predicted on the basis of the combination of two m3 triose-P molecules. Since M1 and M2 isotopomers were also seen in triose-P and in phosphoenolpyruvate (PEP) (4), the source of M1 and M2 isotopomers of glucose was assumed to come from the loss of carbon via the tricarboxylic acid cycle. Since glyceraldehyde phosphate is also an intermediate of the pentose cycle, the cycling of glycerol carbons through the pentose cycle pathways produces asymmetry in glucose carbon labeling pattern and can also lead to the formation of M1 and M2 glucose. We present here a study of the role of the pentose cycle in gluconeogenesis in Fao cells, in which the cycling of [U- 13 C]glycerol carbon between glycerol and PEP was minimized with the dilution of PEP from unlabeled gluconeogenic precursors, pyruvate and glutamine. Fao hepatoma cells are derived from the Reuber H35 rat hepatoma cell line (6,7) and show stable expression of a number of liver-specific functions including the expression of the gluconeogenic enzymes PEP carboxykinase and fructose 1,6bisphosphatase. Since Fao cells can grow in glucose-free media and have a uniform enzymatic profile, unlike the mixture of periportal and perivenous cells seen in primary hepatocyte culture, Fao cells were used to study gluconeogenesis. In addition, Fao cells have been found to have specific and high affinity binding of insulin at physiologic concentrations. The metabolic regulation of glucose release and gluconeogenesis by insulin was also studied.
Tissue Culture Conditions-The Fao cells were serum-starved for 18 h in low glucose (1 g/liter) DMEM, containing an additional 1 mM glutamine, 1 mM sodium pyruvate, and 1% penicillin/streptomycin before the isotope study. The medium was then changed to DMEM with no glucose, which contains sodium pyruvate, 1 mM glutamine (2 mM final), and penicillin/streptomycin as in the low glucose condition. Dexamethasone was then added to a final concentration of 10 Ϫ6 M, and the final concentration of the ETOH carrier for dexamethasone was 0.02%. Two substrate conditions (2 mM xylitol or 12.5 mM glucose) and two hormone conditions (in the presence or absence of insulin) were used giving rise to four incubation conditions. Undiluted [U-13 C]glycerol was added to all culture plates to a final [U-13 C]glycerol of 1.5 mM to start the experiment. For incubation with insulin, the cells were stimulated with dexamethasone for 2 h before the addition of insulin. The final concentration of insulin was 10 Ϫ8 M. The plates were incubated for a total of 24 h after the time the media were collected.
Sample Processing-At the end of experiment, cells were harvested in ice-cold phosphate-buffered saline, spun down, resuspended in a total volume of 50 l with phosphate-buffered saline, and stored at Ϫ80°C until processed. This pellet was sonicated prior to measurement of glycogen or ribose. Glycogen glucose was isolated from cell pellets after treatment with amyloglucosidase (8). RNA was extracted from cell pellet after treatment with ice-cold Trizol 1:2%, v/v (9, 10). Incubation medium from each culture plate was analyzed for lactate, glucose, and glutamate using previously established methods. Lactate was first extracted from the acidified medium using ethyl acetate. Glucose, glutamine, and glutamate were separated using ion-exchange chromatography. (8) Glucose and xylitol were co-eluted in the neutral fraction from the medium.
Gas Chromatography-Mass Spectrometry Analysis-Medium or glycogen glucose was derivatized as its penta-acetate or aldonitrile pentaacetate derivative for gas chromatography-mass spectrometry analysis (11). Under either derivatization condition, xylitol was converted to its acetate derivative by the acetic anhydride reaction. GC conditions were as follows: capillary column (HP5); carrier gas, helium at flow rate of 1 ml/min, initial oven temperature 220°C for 2 min and a temperature gradient of 10°C per min until final temperature of 250°C, xylitol penta-acetate and glucose aldonitrile penta-acetate were separated. Chemical ionization (CI) was used to give the molecular ion (C-1-C-6) of the glucose molecule at m/z 331 for its penta-acetate and m/z 328 for its aldonitrile penta-acetate. Electron impact ionization (EI) of the aldonitrile derivative was used to characterize glucose positional isotopomers at m/z 187 for C-3-C-6 and m/z 242 for C-1-C-4 fragments. Ribose was converted to its aldonitrile acetate derivative and lactate to its heptafluorobutyrate n-propylamide derivative as described previously (12).
Results of the mass isotopomers in glucose or lactate are reported as molar fractions of m0, m1, m2, etc., according to the number of labeled carbons in the molecule (13). The sum of all isotopomers of the molecules, ⌺m i for I ϭ 1 to n (n ϭ 3, 5, or 6 for lactate, ribose, and glucose, respectively), equal to 1 or 100%. The labeled isotopomer fractions are also reported as m i /⌺m i , in percent, which is a distribution not affected by the dilution with unlabeled compounds. The enrichment, ⌺mn, is the weighted sum of the labeled species (⌺mn ϭ m1 ϫ 1 ϩ m2 ϫ 2 ϩ m3 ϫ 3 . . . ). The relative molar enrichment equals enrichment in the product, ⌺mn, divided by the enrichment in the infused precursor, (ϭ m1⅐1 ϩ m2⅐2 ϩ m3⅐3 ϭ average 13 C/molecule) ϭ 3 for [1,2,3-13 C 3 ]glycerol. The relative enrichment is equivalent to relative molar specific activity, the ratio of 13 C activity in a product to the 13 C specific activity of the infused glycerol.
Data Interpretation-Glucose has been considered to be a dimer resulting from the condensation of GAP and DHAP molecules. When the rate of formation of glucose is much slower than the rate of isomerization of these triose-P molecules, the labeling pattern of the top half (C-1-C-3) of the glucose molecule is expected to be the same as that of the bottom half (C-4 -C-6). Asymmetry in the glucose labeling pattern can result when the two triose-P molecules are produced or removed at unequal rates as demonstrated by Previs et al. (4). Asymmetry of labeling of the top and bottom half of the glucose molecule can occur through the loss of [ 13 C]carbon via the pentose cycle. An account of the fate of [1,2,3-13 C 3 ]glucose-6-P carbon going through the pentose cycle is provided in Fig. 1. [1,2,3-13 C 3 ]glucose (M3 top) can be oxidized by the reaction of glucose-6-phosphate dehydrogenase (G6PDH) resulting in [1,2-13 C 2 ]pentose-5-P (either ribose-5-P or xylulose-5-P). The action of transketolase on xylulose-5-P and ribose-5-P leads to the formation of [1,2,3,4-13 C 4 ]-, [1,2-13 C 2 ]-, and [3,4-13 C 2 ]sedoheptulose-7-P. Subsequently, due to the action of transaldolase, [1,2,3-13 C 3 ]-, [1,2-13 C 2 ]-, and [3-13 C]fructose-6-P, and [1-13 C]erythrose-4-P and unlabeled erythrose-4-P are formed. M3 glucose can rapidly become M1 and M2 glucose. The action of the pentose cycle leads to the formation of M2 and M1 glucose at the expense of M3 label in the C-1-C-3 position. As shown in Fig. 1, the bottom three carbons, on the other hand, cycle via the transketolase and transaldolase reactions without being changed. Thus [4,5,[6][7][8][9][10][11][12][13] C 3 ]glucose (M3 bottom) is expected to remain as M3. As a result, M3 bottom (C-4 -C-6) is always greater than M3 top (C-1-C-3), and the opposite is true for M2 and M1. Asymmetry in glucose labeling pattern can be demonstrated by comparing the distribution of isotopomers of the C-1-C-4 fragment (by EI) to that of the C-1-C-6 fragment (by CI) as illustrated in Fig. 2.
As shown in Fig. 2, if only the sole action of the pentose cycle is considered, a loss in the M3 isotopomers by EI, due to cleaving off carbons 5 and 6 of glucose, results in a gain of M1. However, some randomization of top and bottom of the glucose molecule does occur, secondary to the action of aldolase on fructose-1,6-P, in which case proportionate amount of M3 and M2 should be lost, if the label in M3, M2, and M1 glucose is symmetrically distributed (see "Results").

RESULTS
Distribution of Lactate Isotopomers-In order to minimize the effect of the tricarboxylic acid cycle on the final isotopomer distribution in glucose and ribose, unlabeled gluconeogenic pre- The orientation of all molecules is such that carbon 1 (C-1) position is on the top. Asymmetry of labeling of the top and bottom half of the glucose molecule can result when the two triose-P molecules are produced or removed at unequal rates. Successive loss of labeled glucose carbon at C-1-C-3 can occur through the loss catalyzed by glucose-6-phosphate dehydrogenase via the oxidative limb of the pentose cycle producing M2 and M1 glucose. [ 13 C]Carbon in the lower half of the glucose molecule that cycles through the non-oxidative limb of the pentose cycle (glucose 6-phosphate 3 ribose/xylulose phosphates 3 glyceraldehyde 3-phosphate ϩ fructose 6-phosphate) remains intact. Asymmetry in glucose labeling pattern reflecting the sole action of pentose cycle can be demonstrated by comparing the distribution of isotopomers of the C-1-C-4 fragment (by EI) to that of the C-1-C-6 fragment (by CI).
cursors, pyruvate and glutamine, were added to the incubation media to a final concentration of 1 mM. This had the effect of diluting any 13 C label coming from the randomization of the 13 C label through tricarboxylic acid cycle. Under such conditions, mass isotopomer distribution in lactate is the result of conversion of glycerol to lactate diluted by unlabeled lactate from other sources. Table I shows that lactate produced from glycerol by cultured Fao cells is greatly diluted by lactate produced from pyruvate or glutamate. This observation suggests that very little 13 C-labeled glycerol went into the tricarboxylic acid cycle, and this was also supported by the fact that 13 C isotopomer of glutamate was not detectable in medium glutamate (data not shown). Therefore our experimental conditions allowed us to focus on the interactions of glycolytic/ gluconeogenic, glycogen, and pentose phosphate pathways in this Fao model.
Effect of Insulin on Glucose Release-The release of glucose into the medium was quantitated using the medium xylitol as a recovery standard. Xylitol is co-isolated with glucose and is converted to penta-acetate during the derivatization of glucose. The derivative of xylitol shares a common C-3-C-6 fragment (at m/z 187) with the glucose aldonitrile penta-acetate derivative but has a different retention time than that of the glucose derivative (Fig. 3). The inhibition of glucose release into the medium by insulin is clearly demonstrated by the reduction of the glucose peak. The inhibition can be quantitated by examining the GC ratio of glucose to that of xylitol using a standard curve. Results are shown in Fig. 4. The release of glucose into the medium was reduced from 15.1 mol/24 h with no insulin to 4.2 mol of glucose with insulin ( Fig. 4).
Distribution of Glucose Isotopomers-The mass isotopomer distribution in medium glucose (new glucose) is shown in Table  II. The enrichment in glucose was 0.42 and 0.44 in the insulintreated and untreated cells, being about four times that of lactate. Thus, [U-13 C]glycerol contributed to 12-16% of carbon atoms of the new glucose molecules. Regardless of insulin treatment, 8 -10% of glucose in the medium is labeled with 3 [ 13 C]carbon atoms, 2.7-3% with 2 [ 13 C]carbon atoms, and 6.6 -7.1% with 1 [ 13 C]carbon atom per molecule. This pattern is significantly different from that in lactate suggesting sources of M1 and M2 isotopomers of glucose other than the tricarboxylic acid cycle. The mass isotopomer distribution in the C-1-C-4 fragment of glucose aldonitrile penta-acetate is shown in the lower half of Table II. The number of molecules containing 3 [ 13 C]carbon atoms was substantially reduced when carbon 5 and carbon 6 of glucose were cleaved by EI. Only 1.7-1.8% of the molecules contains 3 [ 13 C]carbon atoms in the C-1-C-4 fragment. By comparing the M3 molar fraction of the whole glucose molecule as determined by CI and that in the C-1-C-4 fragment determined by EI, it can be estimated that the glucose molecule was asymmetrically labeled with 20% of M3 labeling C-1-C-3 and 80% labeling C-4 -C-6. The molar fractions of M2 glucose (C-1-C-6) in Table II were comparable to the M2 of the C-1-C-4 glucose fragment (Table II, bottom). The distribution of glucose mass isotopomers deduced from the difference between isotopomers in C-1-C-6 and C-1-C-4 fragment is shown in Fig. 5. Some randomization of top and bottom of the glucose molecule does occur, secondary to the action of aldolase on fructose-1,6-P (18). If the label in M3, M2, and M1 glucose is symmetrically distributed, M1 can be in any of the C-1-C-6 positions, but M2 can only be in C-1-C-2 or C-5-C-6 and M3 in C-1-C-3 or C-4-C-6 ( Fig. 5). Therefore, a proportionate amount of M3 and M2 should be lost, greater than the loss in M1, by cleaving off carbon 5 and 6 of glucose by EI. However, if the loss of M3 is more than the loss of M1 and M2 of the   Fig. 3 for details). The height of the peaks in Fig. 3 is proportional to the concentration of glucose or xylitol in the media. The xylitol peak of the integrated EI spectra at m/z 187, shown in Fig. 3, was taken to be an internal standard for this calculation, as we have found there is no change in the xylitol concentration during the experiment for the two conditions (with and without insulin). C-1-C-4 fragment, it is an indication of M3 in C-4 -C-6 being greater than M3 in C-1-C-3 of glucose, and M2 and M1 in C-4-C-6 being less than that in C-1-C-3 glucose. The degree of loss of [ 13 C]glycerol carbon depends on the magnitude of the non-oxidative and oxidative fluxes, relative to the gluconeogenic flux and input of 5 carbon precursors, such as xylitol carbon. These considerations are illustrated by our findings, showing that the reduction in M3 is associated with a slight decrease in M2, as predicted in Fig. 5. The asymmetry in the labeling pattern can only be the result of loss of [ 13 C]carbon through the pentose cycle during gluconeogenesis. Table III compares the relative enrichments in glycogen glucose derived from Fao cells incubated in the presence of 12.5 mM glucose Ϯ 10 Ϫ8 M insulin to that derived from Fao cells incubated in 2 mM xylitol Ϯ 10 Ϫ8 M insulin with no cold glucose added. The enrichment in glycogen glucose from Fao cells treated with 12.5 mM glucose was 0.349 which was significantly higher than the enrichment of 0.261 from cells treated with 2 mM xylitol (p Ͻ 0.005). Enrichment of M3 glucose was also higher for incubation in 12.5 mM glucose than in 2 mM xylitol (6.7 Ϯ 0.7% compared with 5.3 Ϯ 0.5%). The lower enrichment in glucose of the xylitol-treated case may indicate that some of the 2 mM xylitol was converted into glucose carbons via equilibration with the transaldolase and transketolase reactions of the pentose cycle diluting the [ 13 C]carbon label coming from glycerol. Furthermore, this observation is consistent with the absence of glucokinase in Fao cells as reported previously (14). If glucokinase were appreciably expressed and active, the enrichment in glycogen glucose from 12.5 mM glucose-treated cells would be lower than that in glycogen from cells incubated with xylitol and no added glucose. However, the enrichment in the glucose-treated cells was higher than that of the xylitoltreated cells supporting the lack of glucokinase activity.
As in the case of enrichment in new glucose from gluconeogenesis (Table II), no difference was observed between enrichment in glycogen from insulin and no insulin-treated cells of either medium condition. Thus, insulin has no effect on the influx of unlabeled precursor in the dexamethasone-treated Fao cells.
Distribution of Pentose/Ribose Isotopomers- Table IV compares the relative enrichments in ribose derived from Fao cells incubated in the presence of 12.5 mM glucose Ϯ 10 Ϫ8 M insulin to Fao cells incubated in 2 mM xylitol Ϯ 10 Ϫ8 M insulin with no cold glucose added. Again, as in the enrichment of glycogen, enrichment in ribose of the xylitol-treated cells was significantly lower than that of the glucose-treated cells. This indicates that xylitol from the medium is phosphorylated to xylulose-5-P (Xu5P) which equilibrates with the intracellular ribose pool, diluting the labeled ribose originated from the [U-13 C]glycerol. Since labeled glycerol can be incorporated into M3 ribose by the transketolase/transaldolase reaction or recycled from glucose labeled in the C-4 -C-6 positions, M3 ribose was labeled only in positions C-3-C-5, which was confirmed by the loss of M3 in the C-1-C-4 fragment of ribose (data not shown). Since 13 C atoms are conserved by the TK/TA reactions, the existence of M1 and M2 ribose in relatively high amounts can only be the result of G6PDH oxidation of the [1,2,3-13 C 3 ]glucose and subsequent cycling via the non-oxidative pathways. DISCUSSION The loss of [ 13 C]glycerol carbon during gluconeogenesis has been reported previously (4). When rat liver was perfused with 0.1-1.5 mM [U-13 C]glycerol, glucose isotopomers with 1-6 [ 13 C]carbon substitutions were obtained. The symmetry of labeling in the glucose molecule was not determined. However, the enrichment and pattern of these mass isotopomers were different from those of PEP or of triose-P and were not compatible with the synthesis of glucose from a simple combination of these 3-carbon precursors. Such a discrepancy was attributed to compartmentalization or heterogeneity or isotope gradient of these precursor pools. In a later study, the lost of [ 13 C]glycerol carbon during gluconeogenesis was again observed in isolated hepatocytes (5). When hepatocytes were incubated with [3-13 C 1 ]lactate/pyruvate mixture in the presence of unlabeled glycerol, substantially less 13 C was found in the C-1-C-3 fragment of glucose as compared with the C-4 -C-6 fragment. The loss of [ 13 C]carbon from the C-1-C-3 position of glucose was also observed, but to a lesser degree with [2-13 C 1 ]glycerol incubation (Fig. 5    plained by the loss of 13 C via the tricarboxylic acid cycle. Neither can it be explained by the lack of equilibrium in triose-P isomerase. That label in the C-1 and C-2 position of glucose is lost to a different degree is compatible with the action of G6PDH and cycling between hexose-P and pentose-P. In the present study, we observed loss of [ 13 C]glycerol carbon mainly from C-1-C-3 of the glucose molecules, a pattern consistent with the loss of labeled carbon via the pentose cycle. The dilution of label in glucose by xylitol is further evidence for the participation of the pentose pathway in gluconeogenesis. In light of such evidence, gluconeogenesis is not only the result of the condensation of two triose-P molecules but also the result of the pentose cycle depending on substrate availability. Under conditions where the pentose cycle is active, glucose molecule is not just a dimer of triose-P molecules, and the application of MIDA in the study of gluconeogenesis assuming dimerization of triose-P is clearly problematic. The underestimation of gluconeogenesis by MIDA has been an enigma. It has been mathematically proven that the approach tolerates a wide range of disequilibrium between GAP and DHAP (1). A 50% asymmetry between label in C-1-C-3 and C-4 -C-6 of glucose results in less than 5% underestimation of precursor enrichment. The underestimation of precursor enrichment can only falsely increase the estimation of GNG, a condition that may be difficult to detect. Therefore, asymmetry of labeling is not the reason for underestimation of GNG. The MIDA method is, however, modestly sensitive to isotope gradient. By using the model of reference (Fig. 5 in Ref. 4), one can show that a gradient of 25% in glycerol enrichment (from 0.2 to 0.15) results in 2% underestimation of GNG. Substantial underestimation of GNG occurs when the gradient is 75%. The effect of cycling of glucose carbon through the PPP has not been studied. An example of such an effect is provided under the "Appendix." There are three major variables that can influence the distribution of mass isotopomers in glucose. These are the oxidation of glucose-6-P by G6PDH, the dilution of glucose carbon by unlabeled pentose carbon, and finally, the substrate flux via the non-oxidative branch of the PPP relative to that of gluconeogenic flux. These three processes all have the effect of altering the mass isotopomer distribution after glucose is synthesized (by dimerization of triose-P) and can contribute to the underestimation of gluconeogenesis.
The Fao hepatoma cell is an 8-azaguanine-resistant, ouabain-resistant clone derived from the Reuber H35 rat hepatoma cell line (6,7). These Fao cells show stable expression of a number of liver-specific functions. These include the following: 1) the expression of the gluconeogenic enzymes PEP carboxykinase and fructose-1,6-bisphosphatase (7,15), and 2) the ability to grow in the absence of glucose and to release glucose into the medium. In addition, Fao cells have been found to have specific and high affinity binding of insulin at physiologic concentrations and to respond to insulin and dexamethasone similar to other primary hepatocytes (16,17). In the present study, the effect of insulin on the inhibition of gluconeogenesis was demonstrated. Since insulin treatment did not alter the enrichment or labeling pattern in glucose, the effect of insulin is probably to decrease the activity of glucose-6-phosphatase. The Fao cells TABLE V Mass isotopomer distribution in pentose phosphate from the action of G6PDH on glucose 6-phosphate without dilution by unlabeled pentose phosphate In this and subsequent tables, x and o represent 13 C and 12 C, respectively. The two m3 mass isotopomers are designated as m3 top (m3t) and m3 bottom (m3b) depending on whether the 13 C is in C-1-C-3 or C-4 -C-6. The t and b are used in combination with m1, m2, and m3 throughout to designate the corresponding carbon position of the label in glucose. Please note, the mass isotopomer distribution is that from 100% GNG.
x-x-x-x-x (m5) [1,2,3-13 C 3 ]Glucose 25 [1,[2][3][4][5][6][7][8][9][10][11][12][13] x-x-o-o-o (m2t) [4,5,6-13 C 3 ]Glucose 25 [4,5,[6][7][8][9][10][11][12][13]   are known to lack glucokinase (14). We did not find any glucose uptake even in the presence of 12.5 mM glucose in the medium. We demonstrate for the first time a very active pentose phosphate cycle in these cells which is an integral part of substrate flux in gluconeogenesis. Traditionally, the pentose phosphate cycle has been considered to be relatively unimportant in hepatic glucose metabolism. This conclusion is based on the following: 1) that the flux through the oxidative branch of the pentose cycle accounts for a small part (less than 10%) of the glucose uptake, and 2) that the quantity of pentose produced by the action of G6PDH exceeds the net synthesis of ribose in RNA. Combining data from [1-14 C]-, [2-14 C]-, and [6-14 C]glucose studies on the pentose cycle, Katz and Rognstad (18) showed that the flux of hexose-P through the non-oxidative branch could be substantial equaling 2-3 times that of the glucose uptake. This was confirmed in a recent study using [1,2-13 C 2 ]glucose (19). We show in the present study that during gluconeogenesis, the flux through the pentose cycle could be equally substantial. The enrichment of M1 and M2 isotopomers suggests the flux to be 2-3 times that of the gluconeogenic flux. In addition to the production of xylulose-5-P and its action on the bifunctional enzyme, we believe that the pentose cycle may play other roles in the regulation of glycolysis/gluconeogenesis. The role of the pentose cycle in the regulation of hepatic glucose metabolism therefore deserves further study.

APPENDIX
Although hexose-P flux through the oxidative branch of the pentose pathway represents only a small portion of the glycolytic/gluconeogenic flux, hexose-P flux through the non-oxidative branch has been shown to be substantial, equaling to 2-3 times that of the glucose uptake (18).
In addition, Crawford and Blum (20) showed that the magnitude of the bidirectional flux through transketolase or transaldolase exceeds the flux through the limbs of the key substrate cycles of glycolysis/gluconeogenesis (glucokinase/ glucose-6-phosphatase, 6-phosphofructokinase/fructose-1,6bisphosphatase, and pyruvate kinase/phosphoenolpyruvate carboxykinase) and glycogen storage by 2-10-fold in fed hepatocytes.
The operation of the pentose pathway therefore can substantially influence the mass isotopomer distribution in glucose and on the calculation of gluconeogenesis based on the enrichment of 13 C-labeled precursors. There are three major variables that can influence the distribution of mass isotopomer in glucose. These are the oxidation of glucose-6-P by G6PDH, the dilution of glucose carbon by unlabeled pentose carbon, and the substrate flux via the non-oxidative branch of the PPP relative to that of gluconeogenic flux. In this example, we follow glucose isotopomers formed from the combination of 50% enriched [U-13 C]triose-P through the textbook account of the pentose cycle assuming no tracer dilution by unlabeled pentose (Table  V). From the combination of two triose-P, we have four distinct mass and positional isotopomers, which are converted to their respective pentose phosphate counterparts. By inspection, carbons of C-1-C-2 position in the pentose phosphate are 50% 13 C and 50% 12 C, or 50% m2 and 50% m0. At equilibrium, there are eight isotopomers of sedoheptulose-7-P and five isotopomers of erythrose-P (Table VI). The distribution of these isotopomers can be calculated by their respective combinatorial probabilities. For example [U-13 C 7 ]sedoheptulose-7-P is the result of combining m2 in C-1-C-2 position with m5 in the pentose-5-P, and the probability is the product of 50 and 25% giving 12.5%.
The labeled sedoheptulose-7-P and erythrose-4-P eventually return to fructose-6-P by either combining the top three carbons of sedoheptulose-7-P with glyceraldehyde phosphate or condensing the top 2 carbons of pentose-5-P with erythrose-4-P. The resulting distribution is again dictated by combinatorial probability (Table VII). The example is simplified by the fact that both the enrichment in glyceraldehyde phosphate (m3b) and C-1-C-2 of pentose phosphate (m2t) happen to be 50%. To facilitate the calculation of distribution from combinatorial probability using information of Table VI, we designate the origin of the precursor either as labeled top or labeled bottom (m3t, m3b, etc.). Thus, m6 glucose can be derived from combining m3t of sedoheptulose-7-P and m3b of glyceraldehyde-3-P or m2t from xylulose-5-P and m4b of erythrose-4-P. m3t in sedoheptulose-7-P is 25% and m3b in glyceraldehyde-3-P is 50%. Therefore, the probability of m6 (m3tm3b) is 12.5% or (0.125). The isotopomer distribution in glucose after passing through the pentose cycle is the average of those from both of these two transketolase reactions and is presented in column 2 in Table  VIII.
The final mass isotopomer distribution in glucose can be considered to be the result of mixing glucose from the combination of two triose-P with the glucose recycled from the pentose cycle. The final mass isotopomer distribution depends on the relative contribution of these two pools. We consider two possibilities, one having 30% gluconeogenic flux to pass through the pentose cycle (30%) and the other 60%, shown in the 3rd and 4th columns of Table VIII. The assumption in these calculations of 30 -60% of gluconeogenic flux transversing the pentose cycle may be conservative, considering the estimations of bidirectional transketolase and transaldolase flux by Crawford and Blum (20). The mass isotopomers in column 3 are the result of adding 30% of column 2 to 70% of column 1; and the isotopomers in column 4 are the result of adding 60% of column 2 to 40% of column 1. When the precursor enrichment (p) and the fractional gluconeogenesis (GNG) are calculated using MIDA, we confirm the minimal influence of the asymmetry on the calculation of p. However, substantial underestimation of GNG occurs because of the effect of the pentose cycle. Note that the asymmetry of the labeling pattern in glucose due to the pentose cycle is readily detected as indicated by m3t/m3b ratios. The asymmetry in our experiment is 1.7:9.2 (Fig. 5), suggesting that the gluconeogenic flux passing through the pen-tose cycle is very large.
In this example, we assume possible tracer dilution by unlabeled pentose to be relatively small. We have not discussed the effect of isotopic equilibration between the pentose phosphate and the hexose phosphate pools by just the transketolase and transaldolase reactions without the action of G6PDH. It should be pointed out that such equilibrium reactions have the same effect in introducing asymmetry in symmetrically labeled glucose precursor, i.e. M1, M2, and M3 glucose are generated from [1,2,3-13 C 3 ]glucose but not from [4,5,6-13 C 3 ]glucose. Their effects on the glucose mass isotopomer distribution and calculations of MIDA are expected to be similar. As shown in the above examples (Tables V-VIII), the calculation of GNG is subject to compensating errors of overestimation of GNG due to underestimation of p and underestimation of GNG due to scrambling of isotope. From the results of Ref. 5, we can conclude that the effect of isotope scrambling dominates even in the studies using [2-13 C]glycerol causing significant underestimation of GNG.