Endocrinology and Metabolism

Use of 2H2O for estimating rates of gluconeogenesis: determination and correction of error due to transaldolase exchange

Jeffrey D. Browning, Shawn C. Burgess

Abstract

The use of deuterated water as a method to measure gluconeogenesis has previously been well validated and is reflective of normal human physiology. However, there has been concern since the method was first introduced that transaldolase exchange may lead to the overestimation of gluconeogenesis. We examined the impact of transaldolase exchange on the estimation of gluconenogenesis using the deuterated water method under a variety of physiological conditions in humans by using the gluconeogenic tracer [U-13C]propionate, 2H2O, and 2H/13C nuclear magnetic resonance (NMR) spectroscopy. When [U-13C]propionate was used, 13C labeling inequality occurred between the top and bottom halves of glucose in individuals fasted for 12–24 h who were weight stable (n = 18) or had lost weight via calorie restriction (n = 7), consistent with transaldolase exchange. Similar analysis of glucose standards revealed no significant difference in the total 13C enrichment between the top and bottom halves of glucose, indicating that the differences detected were biological, not analytical, in origin. This labeling inequality was attenuated by extending the fasting period to 48 h (n = 12) as well as by dietary carbohydrate restriction (n = 7), both conditions associated with decreased glycogenolysis. These findings were consistent with a transaldolase effect; however, the resultant overestimation of gluconeogenesis in the overnight-fasted state was modest (7–12%), leading to an error of 14–24% that was easily correctable by using either a simultaneous 13C gluconeogenic tracer or a correction nomogram generated from data in the present study.

  • deuterated water
  • transaldolase
  • stable isotope
  • gluconeogenesis
  • glycogenolysis

a simple, noninvasive, and widely used method for measuring the contribution of gluconeogenesis to glucose production in humans using deuterated water (2H2O) was first introduced by Landau et al. (21) in 1995. The approach was adapted from a tritiated water (3H2O) method (18) and, given certain reasonable assumptions, overcame a number of limitations that accompany the use of 13C tracers (16). The fact that a proton (1H) from water is added to the carbons of glucose at specific steps of its synthesis forms the basis of the method: 1) during the conversion of pyruvate to glucose, protons are added to carbon 6 during either the transamination of pyruvate with alanine or the formation of oxaloacetate and subsequent rapid equilibration with malate and fumarate within the mitochondrial tricarboxylic (TCA) cycle; 2) the conversion of dihydroxyacetone phosphate (DHAP) to glyceraldehyde 3-phosphate (GAP) by the enzyme triose-phosphate isomerase (TPI) leads to the protonation of carbon 5, regardless of whether glucose is synthesized from phosphoenolpyruvate carboxykinase (PEPCK)-dependent (i.e., pyruvate, lactate, amino acids) or -independent substrates (i.e., glycerol); and 3) conversion of fructose 6-phosphate (F6P) to glucose 6-phosphate (G6P) contributes a proton to carbon 2, an exchange that occurs during both gluconeogenesis and glycogenolysis via the common intermediate of G6P and F6P. When body water is enriched with tracer amounts of 2H2O, a deuteron (2H) is incorporated at these steps, thereby allowing estimation of the fractional contribution of gluconeogenesis and glycogenolysis to glucose production via examination of the 2H enrichment of carbons 2, 5, and 6 of glucose.

The validity of these labeling events is supported by classic radioisotope experiments, as reviewed by Landau (19). Contemporary experiments also support the accuracy of the deuterated water method; fractional glycogenolysis determined by deuterium enrichment of carbon 5 decreases at the precise rate as the depletion of hepatic glycogen content (24), and the enrichments at carbon 2 and 5 become equally enriched after hepatic glycogen is depleted. In addition, mice that do not express the gluconeogenic enzyme PEPCK in liver and therefore cannot synthesize glucose from pyruvate precursors also do not incorporate deuterium on carbon 6 of glucose (6). Finally, the method provides identical results regardless of whether deuterium enrichment is detected by gas chromatography-mass spectrometry (GC-MS) or nuclear magnetic resonance (NMR) (7, 17). However, a number of critical assumptions were originally pointed out by Landau et al. (21), including the requirement for negligible transaldolase exchange, a process that could lead to an overestimation of gluconeogenesis by the 2H2O method.

Transaldolase is an integral enzyme of the nonoxidative pentose phosphate pathway that catalyzes the exchange of the bottom four carbons of sedoheptulose 7-phosphate (C4-C7) with GAP to form F6P and erythrose 4-phosphate. More importantly, transaldolase also has a collateral activity that exchanges the bottom three carbons of F6P (C4–C6) with GAP from the triose pool (23). This function appears to be metabolically inert (no new metabolites are formed and no cofactors are consumed) except that the carbons of GAP carry any isotopic label introduced by a tracer experiment. Given that glucose derived from glycogen is in equilibrium with F6P, such an exchange would artificially enrich carbon 5 (and possibly 6) of glycogen, thereby making glycogen-derived glucose molecules appear to have come from gluconeogenesis (Fig. 1A). Thus, transaldolase exchange would lead to the overestimation of gluconeogenesis.

Fig. 1.

Mechanisms by which unequal enrichment in the top and bottom halves of glucose can occur when a gluconeogenic tracer is given. Under such conditions, the tracer enters the gluconeogenic pathway as phosphoenolpyruvate (PEP). A: transaldolase (TA) is an integral enzyme of the pentose phosphate pathway that can exchange the bottom 4 carbons of sedoheptulose 7-phosphate (C4–C7) and, more importantly, the bottom 3 carbons of fructose 6-phosphate (F6P; C4–C6) with glyceraldehydes 3-phosphate (GAP) from the triose pool (23). In order for TA exchange to dilute enrichment in the top half of glucose, an unlabeled pool of glucose must be available. Under the conditions of study, only glycogen could provide such a pool. Given that glucose derived from glycogen is in equilibrium with F6P, TA exchange would serve to enrich glucose carbons 4, 5, and 6, yielding higher enrichments in the bottom half of glucose relative to the top half. This effect should be minimal or absent under conditions of glycogen depletion. B: triose phosphate isomerase (TPI) is responsible for the isomerization of GAP to dihydroxyacetone phosphate (DHAP) and vice versa. When a gluconeogenic tracer is used, it is assumed that TPI provides rapid equilibration of the label in GAP to DHAP and therefore equivalent enrichment in the top and bottom halves of glucose. Incomplete equilibration of the label in GAP would lead to greater enrichment in glucose carbons 4, 5, and 6 compared with carbons 1, 2, and 3. The unlabeled DHAP is derived, in part, from glycerol.

Quantifying this overestimation has been problematic. Landau's group (2, 3) originally infused [3,5-2H]glucose or [3,5-2H]galactose to estimate the error, anticipating that transaldolase exchange would replace carbons 4–6 of the hexose tracer with an unlabeled GAP, thereby leading to labeling inequality between the top and bottom halves of glucose (C3 > C5). Indeed, the observed C5:C3 ratio was 67–81%; however, the cause of the labeling inequality when these tracers were used could have resulted from either transaldolase exchange or a kinetic isotope effect known to be substantial for the cleavage of 2H by TPI (22). Other approaches to defining the impact of transaldolase exchange have used gluconeogenic tracers as opposed to hexose tracers. Basu et al. (1) used [1-13C]acetate and found the 13C labeling inequality between the top and bottom halves of glucose to be an astounding 56–66%, suggesting that nearly one-half of the glycogen released underwent transaldolase exchange. However, none of the above studies accounted for the impact of nonisotopic equilibrium at the level of TPI on the observed labeling inequality (Fig. 1B). Only the study by Jones et al. (14), using the gluconeogenic tracer [1,2,3-13C]glycerol, ensured isotopic equilibrium at TPI. Although that study demonstrated that the labeling inequality between the top and bottom halves of glucose was 75–77%, metabolic heterogeneity in the utilization of glycerol by liver may have lead to an overestimation of transaldolase exchange.

Inasmuch as the magnitude of the error introduced by transaldolase exchange is dependent on the rate of glycogenolysis, the error should be maximal in the postabsorptive state and progressively diminish with fasting until completely eliminated upon glycogen depletion. To date, no study has taken advantage of this fact to define the error due to both isotope effects and transaldolase activity. In the present study, we examined the impact of transaldolase exchange on the estimation of gluconenogenesis using the deuterated water method under a variety of physiological conditions in humans. First, we examined how labeling inequality between the top and bottom half of glucose changed over the course of a 48 h fast using the gluconeogenic tracer [U-13C]propionate, 2H2O, and 2H/13C nuclear magnetic resonance (NMR) spectroscopy. This approach allowed us to 1) quantify the error introduced by transaldolase exchange and determine how this error was attenuated by reduction in the relative rate of glycogenolysis, and 2) estimate the contribution of nonisotopic equilibrium and a deuterium isotope effect at TPI to unequal glucose labeling by performing studies when glycogenolysis, and hence the effect of transaldolase exchange, must be excluded. Second, we performed similar studies in subjects after weight loss via dietary calorie and carbohydrate restriction for 2 wk, two conditions where the relative rate of glycogenolysis differs dramatically after a simple overnight fast (5). These data were combined to provide an estimate of the error introduced by transaldolase exchange to estimates of gluconeogenesis over a wide physiological range.

MATERIALS AND METHODS

Participants.

Individuals were recruited for study at University of Texas Southwestern Medical Center. Six healthy subjects were studied after an overnight fast (3 women and 3 men, age 29 ± 3 yr, BMI 22 ± 1 kg/m2; means ± SE). These subjects consumed a weight-maintaining diet that was 40% carbohydrate, 30% fat, and 30% protein for 3 days prior to study. Twelve healthy subjects were studied during a 48-h fast (7 women and 5 men, age 22 ± 1 yr, BMI 23 ± 1 kg/m2) in parallel with a study of the effects of fasting on glucose, lipid, and energy metabolism in humans (4). These subjects consumed a weight-maintaining diet composed of 48% carbohydrate, 38% fat, and 14% protein for 3 days prior to the fast. Data from 14 overweight/obese subjects who had previously been studied (5) after weight loss via 2 wk of calorie restriction (5 women and 2 men, age 43 ± 4 yr, BMI 32 ± 2 kg/m2) or carbohydrate restriction (5 women and 2 men, age 44 ± 5 yr, BMI 33 ± 1 kg/m2) were also used. Diet composition for the calorie-restricted group was 47% carbohydrate, 36% fat, and 17% protein with basal energy intake reduced by 800 kcal/day. Diet composition for the carbohydrate-restricted group was 5% carbohydrate, 61% fat, and 34% protein. The protocol and consent form were approved by the UTSW Institutional Review Board, and all participants provided written informed consent prior to enrollment.

Design.

Participants were admitted to the Clinical and Translational Research Center (CTRC), where they were fasted either overnight or for 48 h. Between 2200 and 0900, subjects received two tracers orally: divided doses of 70% [2H]water (5 g/kg body water, calculated as 60% of body weight in men and 50% of body weight in women) at 2200, 0200, and 0600 and [U-13C]propionate (∼400 mg) at 0800, 0830, and 0900. Subjects were allowed to drink 0.5% [2H]water ad libitum for the remainder of the fast. Although subjects were also given a 2.25 mg/kg bolus of [3,4-13C]glucose intravenously at 0900 followed by a 2-h infusion (0.0225 mg·kg−1·min−1), absolute flux data are not presented. At the end of the infusion period (1100), a 50-ml blood draw was performed. These procedures, except for overnight loading with 70% [2H]water, were repeated on the second admission day in subjects fasted for 48 h.

Isotopes and other materials.

Seventy percent [2H]water and 99% [U-13C]propionate (sodium salt) were obtained from Cambridge Isotopes (Andover, MA). Sterility- and pyrogen-tested [3,4-13C]glucose was obtained from Omicron Biochemicals, (South Bend, IN). Other common reagents were purchased from Sigma (St. Louis, MO).

Measurement of glucose isotopomers.

Deuterium and 13C isotopomers of plasma glucose were evaluated by 2H- and 13C-NMR as previously described (7–8, 21). Briefly, plasma was extracted with perchloric acid, and the glucose was purified (7, 21). Purified plasma glucose was converted to 1,2-isopropylidene glucofuranose [monoacetone glucose (MAG)] before 2H- and 13C-NMR analysis, as detailed previously (7–8, 21). Samples were analyzed on a 14.1 Tesla Varian Inova spectrometer (Varian Instruments, Palo Alto, CA) equipped with a 3-mm broadband probe tuned to 92 MHz for 2H spectra or 150 MHz for 13C spectra. Resonance areas were determined using ACD/Labs 12.0 (Advanced Chemistry Development, Toronto, ON, Canada). [3,4-13C]glucose with a range of enrichments (0.0, 0.1, 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, and 5.0%) were made using simple dilution. These standards were converted to MAG and used to determine if differences in C2 and C5 13C enrichment occurred as a result of the analytical approach used (12).

The ratio of 13C enrichment in carbons 2 and 5 of glucose (C2:C5; Fig. 2) were used to correct the measurement of gluconeogenesis determined from the ratio of 2H enrichment at carbons 2 and 5 of glucose (H5:H2) for transaldolase exchange as follows (1):

Fig. 2.

Glucose isotopomers derived from [U-13C]propionate and their analysis by 13C-NMR. After ingestion, propionate is avidly taken up by liver and enters the TCA cycle as [1,2,3-13]succinyl-CoA. This labeled intermediate can take several routes on its way to becoming a triose (GAP and/or DHAP), resulting in formation of a variety of triose isotopomers that are then combined to form the top and bottom halves of glucose. Assuming that there is no TA exchange and that the 13C label is equilibrated between the trioses by TPI, 13C enrichment in the top and bottom halves of glucose should be similar. Due to spin-spin coupling, the 13C resonance of glucose C2 and C5 after [U-13C]propionate ingestion is a multiplet composed of a singlet (S), two doublets (D), and a quartet (Q). Together, the multiplet provides information regarding the 13C enrichment of glucose C2 and C5 and allows isotopomers derived from the gluconeogenic tracer to be easily distinguished from primarily natural abundance 13C. Use of the [3,4-13C]glucose tracer did not interfere with this analysis.

H5:H2corrected=H5:H2measured×C2:C5

Statistical analysis.

Statistical analyses were performed using SigmaPlot 11.0 (Systat Software, San Jose, CA). Differences between two groups were evaluated using unpaired t-tests. Comparisons of multiple measurements were performed using one- and two-factor repeated-measures analysis of variance (ANOVA). Regression lines were compared using analysis of covariance (ANCOVA). Unless otherwise indicated, values are presented as means ± SE. Statistical significance was taken at P < 0.05.

RESULTS

13C enrichment in the top and bottom halves of glucose is unequal in overnight-fasted subjects.

The 13C-NMR resonances of plasma glucose C2 and C5 after [U-13C]propionate ingestion are both multiplets composed of a singlet (S), two doublets (D), and a quartet (Q) (Fig. 2) (15). Together, the relative areas of the multiplets report the 13C enrichment of glucose C2 and C5 and allow isotopomers derived from the gluconeogenic tracer (mainly doublets and quartet) to be easily distinguished from naturally abundant 13C (singlet). Analysis of glucose standards revealed no significant difference in the total C2 and C5 13C enrichment (P = 0.924; C2:C5 ratio: 103 ± 1%, mean ± SE), indicating that differences detected in this ratio must be biological, not analytical, in origin.

To examine glucose C2 13C enrichment relative to C5 (C2:C5) under conditions typically used in human studies, six healthy, weight-stable subjects were studied after an overnight fast. The C2:C5 enrichment data are presented in Table 1 and compared with the C2:C5 enrichment ratios of glucose standards. The data demonstrated a small labeling inequality (∼6%, P = 0.001) between the top and bottom halves of glucose (top < bottom) for total resonance and the majority of isotopomers derived from [U-13C]propionate. The ratio of the singlet at glucose C2 and C5 was the exception and was 1:1. Inasmuch as this resonance was dominated by a naturally abundant isotopomer, we also compared the total resonance of these carbons, excluding the singlet. This indicated a slightly more pronounced difference in enrichment between the top and bottom halves of glucose (∼12%, P = 0.001). These findings were consistent with transaldolase activity (Fig. 1), but were a much more modest effect than reported earlier (1–3, 14).

Table 1.

Glucose C2:C5 13C enrichment ratios after ingesting [U-13C]propionate in 6 overnight-fasted healthy subjects with stable weight

Fasting equalizes 13C enrichment in the top and bottom halves of glucose.

If transaldolase exchange was responsible for the difference in enrichment between the top and bottom halves of glucose, then this difference should be diminished by fasting. Therefore, data from 12 healthy subjects studied after a 24- and 48-h fast were analyzed. After the 48-h fast, plasma glucose concentration fell (89 ± 2 to 72 ± 3 mg/dl, P < 0.001) and plasma ketone concentration rose more than 25-fold (95 ± 22 to 2,601 ± 725 μM, P < 0.001). The C2:C5 enrichment ratios for each subject at 24- and 48-h fasted are presented in Table 2. Total 13C enrichment in the top half of glucose (C2) was less than that in the bottom half (C5) at 24 h but became similar at 48-h fasted. An analogous enrichment pattern was observed when the minor resonances in the C2 and C5 multiplets were compared except for the ratio of D23 to D45 and the singlet. The D23:D45 resonance ratio tended to increase with fasting (P = 0.093); however, the enrichment levels did not equalize by 48-h fasted. Conversely, the singlets of C2 and C5 were similar at 24- and 48-h fasted.

Table 2.

Glucose C2:C5 13C enrichment ratios after ingesting [U-13C]propionate in 12 healthy subjects fasted for 24 and 48 h

When the singlet was excluded from the total resonance, the C2:C5 enrichment ratio was significantly lower than that of the total resonance at 24-h fasted (P = 0.002) but not at 48-h fasted (P = 0.956). Taken together, these data indicated unequal enrichment between the top and bottom halves of glucose (top < bottom) in subjects with active hepatic glycogenolysis (i.e., overnight or 24-h fasted) but not in subjects who were glycogen depleted (i.e., 48-h fasted). This finding is consistent with the process of transaldolase exchange, which would artificially enrich the bottom half of unlabeled glucose (i.e., glucose derived from glycogenolysis), but not with other processes such as a kinetic isotope effect or isotopic disequilibrium, which would be constitutively present (Fig. 1). As a result, we further investigated the role of glycogenolysis in the observed 13C labeling inequality.

Negative energy balance and dietary carbohydrate restriction equalize 13C enrichment in the top and bottom halves of glucose.

The canonical role of transaldolase in the pentose phosphate pathway and, thus, anabolic pathways such as lipogenesis led us to investigate the possibility that the exchange effect would be lessened in subjects undergoing weight loss. Glucose C2:C5 13C enrichment ratios were analyzed in a group of subjects who had undergone weight loss via either calorie or carbohydrate restriction for 2 wk (5). We previously reported that in these subjects carbohydrate restriction induced more weight loss (−4.4 ± 0.6 vs. −2.3 ± 0.6 kg, P = 0.024) and greater ketonemia (1,280 ± 199 vs. 436 ± 72, P = 0.006) than calorie restriction (5).

Glucose C2:C5 13C enrichment ratios for these subjects are presented in Table 3. After an overnight fast, the total resonances of C2 and C5 were comparable in individuals who had undergone weight loss via carbohydrate restriction, and the C2:C5 total resonance ratio was significantly higher than that in individuals who had lost weight via calorie restriction (P = 0.012). The C2:C5 13C enrichment ratio for D12 and D56 demonstrated a similar pattern, with nearly identical glucose top- and bottom-half enrichments in subjects after carbohydrate restriction but not calorie restriction (P = 0.035). As with subjects who underwent a short-term fast, the C2:C5 enrichment ratio for D23 and D45 did not differ between dietary interventions. The ratio of the C2 and C5 quartets also did not differ between the groups. Excluding the singlet from the total resonance of each carbon resulted in a tendency for the C2:C5 13C enrichment ratio of the low-carbohydrate group to be greater than that of the calorie-restricted group (P = 0.082). These data suggested that dietary carbohydrate restriction, but not weight loss per se, reduced the inequality of 13C enrichment in the top and bottom halves of glucose.

Table 3.

Glucose C2:C5 13C enrichment ratios after ingesting [U-13C]propionate in 14 overnight-fasted obese subjects after 2 wk of weight loss by calorie or carbohydrate restriction

Unequal 13C enrichment in the top and bottom halves of glucose is related to glycogenolysis.

In order for transaldolase exchange to alter tracer labeling in the bottom half of glucose, an unlabeled pool of glucose must be available (Fig. 1A); otherwise, the exchange would occur with an equally enriched triose pool and no anomaly would be detected. Under the conditions of most studies, only glycogen could provide such a pool. Due to the fact that short-term fasting and weight loss secondary to carbohydrate restriction equalized the C2:C5 13C enrichment ratio of glucose, we determined whether this was related to changes in glycogen breakdown as assessed by the deuterated water method. As described in the introduction, the ratio of deuterium enrichment at glucose carbon 5 (H5) to that at glucose carbon 2 (H2) provides a measure of gluconeogenesis and glycogenolysis (H5:H2 = gluconeogenesis, 1 − H5:H2 = glycogenolysis) (21). The H5:H2 enrichment ratios for all subjects are presented in Table 4. Percent glycogenolysis decreased and gluconeogenesis increased significantly with fasting or carbohydrate-restricted weight loss but not with calorie-restricted weight loss (5). For all subjects studied (n = 44), there was a significant relationship between the C2:C5 13C enrichment ratio and the H5:H2 2H enrichment ratio (r = 0.551, P < 0.001): greater inequality in 13C enrichment between the top and bottom halves of glucose occurred in those with greater fractional glycogenolysis. These data demonstrated that the disparity in labeling between the top and bottom halves of glucose was related to the activity of glycogenolysis, an effect entirely consistent with transaldolase exchange. Additionally, this effect appeared otherwise systematic or at least independent of nutritional state.

View this table:
Table 4.

Glucose H5:H2 2H enrichment ratios after ingesting 2H2O under various conditions

Correcting the deuterated water method for transaldolase exchange.

Inasmuch as the effect of transaldolase exchange was systematic, it could also be corrected. Glucose from glycogen that undergoes transaldolase exchange, in addition to diluting 13C enrichment in glucose carbons 1, 2, and 3, will become enriched with deuterium at carbon 5 and appear isotopically to have come from gluconeogenesis (i.e., gluconeogenesis is overestimated). Assuming the glucose C2:C5 13C enrichment ratio provides an estimate of transaldolase exchange, the H5:H2 ratio can be corrected to provide a better estimate of gluconeogenesis and glycogenolysis. Measured H5:H2 ratios for all subjects (n = 44) were plotted against corrected H5:H2 ratios (Fig. 3A). For this analysis, the ratio of the total resonance at glucose carbon 2 and 5 with and without the singlet were used to provide a better representation of the potential error due to transaldolase exchange. Regression lines were fitted to these data and then plotted separately as a correction nomogram with a reference line (y = x) (Fig. 3B). The slopes of the two regression lines were similar (P = 0.331) but differed significantly from 1 (P < 0.001). The adjusted means of the two regression lines also differed significantly (P = 0.035).

Fig. 3.

Correction of the relative rate of gluconeogenesis determined by the deuterated water method for TA exchange. A: In the presence of deuterated water, the ratio of 2H enrichment at glucose carbon 5 to that of carbon 2 (H5:H2) provides an estimate of the rate of gluconeogenesis relative to glucose production. The measured H5:H2 ratios for all data points (n = 44) were corrected for TA exchange using 2 different metrics and the measured values plotted against the corrected values. B: the relationship between measured and corrected H5:H2 ratios was derived using linear regression and used to create a correction nomogram. The unity line (x = y) is provided for reference. The slopes of the 2 regression lines were similar (P = 0.331) but differed significantly from unity (P < 0.001). The adjusted means of the 2 regression lines also differed significantly (P = 0.035).

DISCUSSION

The exchange of triose intermediates with the carbons of glucose precursors introduces a potential error into all tracer-based estimates of gluconeogenesis. Here, we report an integrated assessment of the impact of transaldolase exchange on the estimation of gluconeogenesis when using the deuterated water method (21). The error introduced by transaldolase is dependent on the presence of an unlabeled pool of glycogen. We therefore exploited normal human physiology under conditions that alter glycogenolysis, such as fasting and negative energy balance, to determine by the deuterated water method the degree to which this exchange reaction led to the overestimation of gluconeogenesis. When the gluconeogenic tracer [U-13C]propionate was used, 13C labeling inequality occurred between the top and bottom halves of glucose in individuals fasted overnight or for 24 h. This inequality diminished with the attenuation of glycogenolysis that occurs with 48-h fasting and dietary carbohydrate restriction, consistent with a transaldolase effect. However, the effect was relatively modest (between −7 and −12%) and systematic across nutritional states. This led to two important conclusions: 1) the effect does not invalidate previous tracer-based measurements of gluconeogenesis carried out under well-controlled conditions, since all measurements would be increased systematically; and 2) the effect can be corrected by combining a 13C measurement of the labeling inequality in the top and bottom halves of glucose simultaneously with the measurement of gluconeogenesis using 2H.

We considered two plausible explanations for the unequal labeling between the top and bottom halves of glucose: 1) unlabeled glycogen was undergoing transaldolase exchange with the pool of labeled GAP, leading to preferential enrichment of the bottom half of glucose (Fig. 1A); or 2) the enzyme TPI failed to bring DHAP and GAP into 13C isotopic equilibrium, thereby yielding greater enrichment in GAP and, hence, the bottom half of glucose (Fig. 1B). 13C isotope effects of this degree are uncommon, and the relationship between the magnitude of 13C labeling inequality in glucose and relative rate of glycogenolysis strongly suggested that transaldolase exchange was causative. Additionally, the fact that 13C enrichment in the top and bottom halves of glucose as well as 2H enrichment at carbons 5 and 2 of glucose were similar after a 48-h fast or 2 wk of carbohydrate restriction also suggested that DHAP and GAP were in isotopic equilibrium during these studies.

On the basis of data from 44 subjects under a wide range of nutritional states, we determined a correction nomogram for the fractional rate of gluconeogenesis when using the deuterated water method (Fig. 3B). On the assumption that the fractional rate of gluconeogenesis when consuming a “Western” diet and after an overnight fast is ∼50% (8, 13, 15, 21), this value corrects to 38–43% due to an overestimation by transaldolase exchange (Fig. 3B). Inasmuch as the majority of studies using the deuterated water method are carried out under overnight-fasted conditions, we believe 14–24% to be a reasonable maximal value for the error introduced by transaldolase activity when the deuterated water method is used to estimate gluconeogenesis. Indeed, more prolonged fasting resulted in a diminution in this error such that as fractional gluconeogenesis approached 100% the error approached zero.

Several previous investigations support our findings (Table 5). Although these studies were not designed to examine transaldolase exchange per se, the 14C enrichments in carbons 1–6 of glucose were documented after administration of a variety of gluconeogenic tracers. After an overnight fast, the 14C enrichment ratio (C123/C456) ranged from 74 to 86% (11), closely matching our measurements (Table 1). Similar studies in 60-h-fasted subjects demonstrated that the 14C enrichment ratio ranged from 92 to 98% (20, 26), findings that are also in agreement with the present study (Table 2). Taken together, these data suggested that transaldolase exchange led to a modest overestimation of gluconeogenesis when the deuterated water method was used, an overestimation that was easily corrected. Likewise, the correction applied in the present study should be translatable to other tracer techniques.

View this table:
Table 5.

Historical glucose 14C enrichment ratios using a variety of gluconeogenic tracers

In contrast, the magnitude of 13C labeling inequality in glucose we observed and those reported by prior studies (11, 20, 26) were significantly less than that reported when [1-13C]acetate was used (1). The use of acetate as a tracer of hepatic metabolism has been criticized because it is predominantly metabolized outside of liver (26). Nonetheless, the labeling inequality in the top and bottom halves of glucose observed with [2-14C]acetate after 60 h of fasting (26) was similar to our observations after a 48-h fast using [U-13C]propionate. Indeed, prior work using [1-14C]acetate (27, 28) failed to demonstrate such a large differences in 14C labeling between glucose C3 and C4 in overnight-fasted subjects as those observed by Basu et al. (1). Regardless, the main difference between acetate labeled at carbon 2 and at carbon 1 is that the former generates glucose isotopomers labeled primarily in carbons 1, 2, 5, and 6, whereas the latter generates glucose isotopomers that are labeled at carbons 3 and 4. From our data, an unexplained observation was the much more marked inequality between D23 and D45 enrichments compared with the other glucose isotopomers (Tables 13). However, since the D23:D45 enrichment ratio did not approach unity with fasting or carbohydrate restriction (Tables 2 and 3), we concluded that a phenomenon other than transaldolase exchange or analytical error contributed to the inequality in these isotopomers. One possibility is a 13C isotope effect on the enzyme responsible for combining DHAP and GAP to form a hexose (fructose-1,6-bisphosphate aldolase) that discriminates against 13C enrichment in glucose carbon 3 (10); however, other isotopomers, such as those indicated by the Q2 and Q5 (Tables 2 and 3) that also originate from glucose carbons 3 and 4, are inconsistent with this explanation. Nonetheless, the study of transaldolase exchange using acetate labeled at carbon 1 (1) did not investigate glycogen-depleted subjects, so it is impossible to confirm whether the labeling inequality from these studies was due solely to transaldolase exchange. Indeed, there is evidence that, unlike other tracers (Tables 2 and 5), the labeling inequality between glucose C3 and C4 persists with fasting when 1-labeled acetate is the gluconeogenic tracer (Fasted 12–18 h: 83 ± 7% vs. Fasted 24–40 h: 80 ± 9%, P = 0.829) (27, 28).

Prior studies using the hexose tracers [3,5-2H]glucose, [U-2H]glucose, or [3,5-2H]galactose to investigate transaldolase exchange found labeling inequality between the top and bottom halves of glucose to be 67–81% (2, 3, 14), indicating an error in the estimation of gluconeogenesis of 19–33% under overnight-fasted conditions. However, these tracers are subject to a kinetic isotope effect at TPI (22) that accentuates differences between C5 and C3 (Fig. 4) and thus overestimates transaldolase exchange, as previously discussed (2, 3). Determining the magnitude of the deuterium isotope effect on TPI in vivo has been impossible due to the competing effect of transaldolase exchange in subjects with active glycogenolysis (i.e., overnight-fasted subjects). Examination of the glucose H3:H5 2H enrichment ratio under conditions absent of transaldolase effect (i.e., glycogen-depleted subjects; Table 2) provided an estimate of the TPI deuterium isotope effect. Data from Table 6 indicate that this kinetic isotope effect is responsible for as much as 13% of the inequality of 2H labeling in carbons 3 and 5 when the deuterated water method is used. Assuming that this deuterium isotope effect on TPI is constant, prior estimates of transaldolase exchange using 2H hexose tracers (2, 3, 14) should be adjusted lower by this amount. Such a correction makes those estimates of error due to transaldolase exchange (6–20%) remarkably similar to the estimates reported here (14–24%).

Fig. 4.

Addition of protons to the carbons of glucose by TPI. The trioses DHAP and GAP are numbered according to their final position in glucose. In isomerization of GAP to DHAP, a proton is added to what will become glucose carbon 3 (arrow). In isomerization of DHAP to GAP, a proton is added to what will become glucose carbon 5 (arrow). In the presence of 2H2O, 2H can be added to these carbons during isomerization via TPI, thereby enriching carbons 5 and 3 of glucose, respectively. Likewise, when hexose tracers with 2H at carbons 3 and 5 are broken down to the level of a triose, the 2H label should be lost equally due to the isomerization reaction. However, enzyme kinetic studies have shown that the free energy of adding 2H to DHAP is high, leading to preferential protonation of this carbon by TPI (12). The free energy of adding 2H and 1H to GAP is similar, and TPI demonstrates no preference for protonation or deuteration of this carbon. As a result, the ratio of 2H enrichment at glucose carbon 3 (H3) to that of carbon 5 (H5) with the deuterated water method is nearly always less than 100% (see Table 6). Conversely, this effect will lead to preferential retention of 2H on carbon 3 when a [3, 5-2H]hexose tracer is used and contribute to the overestimation of TA exchange.

View this table:
Table 6.

Glucose H3:H5 2H enrichment ratios after ingesting 2H2O under various conditions

In summary, transaldolase exchange led to a modest overestimation of gluconeogenesis when the deuterated water method was used as originally conjectured by Landau et al. (21). The transaldolase exchange that occurred in overnight-fasted individuals resulted in a 14–24% overestimation of gluconeogenesis. However, this error was systematic, diminished under conditions of reduced glycogen breakdown (short-term fasting, dietary carbohydrate restriction), and was easily correctable (Fig. 3B). In confirmation of prior work (9, 25), we found no evidence that 13C and 2H isotopic disequilibrium at the level of TPI contributed to labeling inequality between the top and bottom halves of glucose. However, we did find evidence for a deuterium isotope effect on TPI that limits deuterium exchange on carbon 3 of glucose. The error introduced by transaldolase exchange does not invalidate the accuracy of the deuterated water method, and it remains the tracer approach of choice for the study of human physiology.

GRANTS

This research was supported by a Clinical and Translational Science Award at UT Southwestern (UL1 RR-024982), the Task Force for Obesity Research (TORS) at UT Southwestern (UL1 DE-019584), the TORS Human Biology Core (PL1 DK-081183), and the TORS Molecular and Metabolic Mouse Phenotyping Core (PL1 DK-081182). Individual investigators were supported as follows: NIH RL1 DK-081187 (J. D. Browning, S. C. Burgess), NIH K23 DK-074396 (J. D Browning), NIH R01 DK-087977 (J. D. Browning), NIH RR-02584 (S. C. Burgess), NIH R01 DK-078184 (S. C. Burgess), and American Diabetes Association 7-09-BS-24 (S. C. Burgess).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: J.D.B. and S.C.B. conception and design of research; J.D.B. and S.C.B. performed experiments; J.D.B. and S.C.B. analyzed data; J.D.B. and S.C.B. interpreted results of experiments; J.D.B. and S.C.B. prepared figures; J.D.B. and S.C.B. drafted manuscript; J.D.B. and S.C.B. edited and revised manuscript; J.D.B. and S.C.B. approved final version of manuscript.

ACKNOWLEDGMENTS

We thank Jeannie Davis, Sonya Rios, and Janet Jerrow from UT Southwestern Medical Center for aid with study conduct. We also thank Visvanathan Chandramouli from Case Western Reserve Medical School for critical reading of the manuscript.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
View Abstract