The habitual diet of many inhabitants of the industrialized world has been frequently associated with deleterious health consequences (Anderson 1995, Byers 1993, Stephen et al. 1995). Most concerns have focused on the role of the high intake of saturated triglyceride associated with these diets. The potential metabolic and functional consequences of the concomitantly lower intake of carbohydrate, other than dietary fiber, have received less attention. Separate from its role as a primary energy source, glucose plays an important role as a biosynthetic precursor, including a potential role in the synthesis of nonessential and conditionally essential amino acids (Consoli et al. 1990, Jahoor et al. 1994).

Quantitative aspects of the metabolic pathways that underlie glucose homeostasis remain surprisingly controversial. Published estimates of the metabolic recycling of glucose and the contribution of gluconeogenesis to blood glucose turnover (Katz et al.1989 and 1993, Landau et al. 1995a and 1995b, Neese et al. 1995, Tayek and Katz 1996, Tserng and Kalhan 1983) and hepatic glycogen synthesis (Huang and Veech 1988, Katz et al. 1991, Mitrakou et al. 1991, Shulman et al. 1987) differ widely and continue to generate substantial argument (Landau et al. 1995a, Neese et al. 1995, Previs et al. 1995).

An important reason for the lack of systematic information is technical. Because of the central metabolic role of glucose, its catabolic pathways intersect with those of virtually all other organic nutrients, thus posing substantial problems for the interpretation of the data from carbon-isotope studies (Di Donato et al. 1993, Landau et al. 1995a, Previs et al. 1995). Some 10-15 years ago, the potential advantages of uniformly (U) ¹³C-labeled glucose as a tracer were emphasized (Kalderon et al. 1986, Tserng and Kalhan 1983). In principle, the combination of this tracer with suitable analysis of the distribution of ¹³C-isotopomers in glucose and its metabolic products allows the simultaneous calculation of glucose turnover, the contribution of glucose to the 3-carbon and Krebs cycle intermediate pools as well as the rate of gluconeogenesis. This approach has been successfully used in studies of glucose metabolism in infants (Bougneres et al. 1995, Keshen et al. 1997, Tserng and Kalhan 1983) and children (Kalderon et al. 1986).

The literature on the subject of gluconeogenesis is dominated by studies in the fasted state, which used intravenous tracers of glucose metabolism. With some notable exceptions (Moore et al. 1994, Taylor et al. 1996), there is less information on the disposition of oral tracer glucose, particularly when given as part of a mixed meal. Furthermore, most attention has focused on determination of the net extraction of portal glucose by the liver and on glycogen synthesis rather than on glucose metabolism and gluconeogenesis.

In the present study, we added [U-¹³C]-glucose to the diet and measured the labeling of blood glucose, hepatic glycogen and the 3-carbon metabolites of glucose in both blood and liver. We wished to achieve the following two main objectives: first, to quantify the extent of first-pass splanchnic metabolism and incorporation into hepatic glycogen of enteral glucose; and second, to examine the impact of a low carbohydrate/high fat diet on these pathways of glucose disposition. A subsidiary objective, not dealt with in this paper, was to quantify the relative rates of glucose metabolism via acetyl CoA and oxaloacetate in different tissues. In designing the studies, we hypothesized that glucose metabolic cycling and gluconeogenesis would be maintained in the fed state and that these pathways would not be altered by carbohydrate restriction.

MATERIALS AND METHODS

Table 1. Composition of the high (HCD) and low (LCD) carbohydrate diets offered to the mice [View Table]

Before amino acid isolation and derivatization, a portion of the liver (~500 mg) was homogenized with 5 mL 0.5 mol/L perchloric acid. The acid-soluble fraction was separated by centrifugation, at 3000 × g, neutralized and applied to a Dowex AG-50 WX8 (H+ form) column. As above, the amino acids were eluted with NH₄OH (2 mol/L). Liver glycogen was precipitated from a separate portion with 95% ethanol and hydrolyzed to glucose with amyloglucosidase (Sigma Chemical, St. Louis, MO). The liberated glucose was then converted to the penta-acetate derivative.

Mass spectrometry was performed on a Hewlett-Packard 9890 gas chromatograph quadropole mass spectrometer (Hewlett-Packard, Palo Alto, CA). We used methane negative chemical ionization for the determination of alanine (mass/charge 307-310) and lactate (mass/charge 131-134) labeling and methane positive chemical ionization for glucose (mass/charge 331-337). The crude ion yields were converted into tracer:tracee ratios using the matrix approach of Brauman (1966) as used in previous papers (Berthold et al. 1995).

A portion of the [M + 3]-oxaloacetate synthesized from [M + 6]-glucose is then decarboxylated to phosphoenolpyruvate and this is the precursor for the resynthesis of ¹³C-glucose. However, this resynthesized glucose contains fewer than six ¹³C-atoms. Thus, during the administration of [M + 6]-glucose, blood glucose becomes a mixture of two labeled forms, i.e., [M + 6]-glucose that has derived from the administered tracer, and [M + 1]-, [M + 2]- and [M + 3]-glucose derived from labeled oxaloacetate. Selected ion monitoring mass spectrometry can quantify the different isotopomers of glucose molecules and the ratio of the isotopic enrichments of the newly labeled ([M + 1]-, [M + 2]- and [M + 3])-glucose to that of the tracer [M + 6]-glucose can be used to calculate estimates of gluconeogenesis expressed as a proportion of the glucose entry rate (fractional gluconeogenesis).

The calculation of gluconeogenesis can then be approached in two ways; the first uses the ¹³C-isotopic enrichment, and the second, the tracer:tracee ratios of the [M + 3]- and [M + 6]-isotopomers of glucose.

1) Calculations based on 13C- recycling. The ¹³C-isotopic enrichment of any labeled molecule was calculated from the sum of the products of the tracer:tracee ratio of each isotopomer and the number of ¹³C atoms in the isotopomer in question. Thus: in which [M + 1], [M + 2] ... are the tracer:tracee ratios of [M + 1]-glucose, [M + 2]-glucose ... .

Equation (3) estimates the total ¹³C-isotopic enrichment of glucose, and Equation (4) measures ¹³C-isotopic enrichment of the glucose molecules that have become labeled via recycling of the glucose tracer.

A similar calculation for the ¹³C-enrichment in the 3-carbon pool derived from the glycolytic metabolism of the tracer glucose can be made: The minimum estimate of fractional gluconeogenesis (¹³C- recycling; GC) is as follows: i.e., Equation (4)/Equation (3).

This expression underestimates the true rate of gluconeogenesis because the ¹³C-pyruvate and oxaloacetate, the precursors of [M + 1]-, [M + 2]- and [M + 3]-glucose, are diluted with unlabeled molecules derived from amino and fatty acid carbon entry into the 3-carbon and Krebs cycle intermediate pools. Dilution of ¹³C can be estimated from the precursor product relationship between the precursor [M + 6]-glucose and its products [M + 1]- to [M + 3]-pyruvate/lactate/alanine and between 3-carbon precursors (i.e., [M + 1]- to [M + 3]-pyruvate/lactate/alanine) and their product [M + 1]- to [M + 3]-glucose.

Thus, the ¹³C-dilution of glucose carbon in the 3-carbon pool is given by: The factor of two in the denominator accounts for the production of two labeled pyruvate molecules per glucose molecule metabolized.

There is also dilution of the ¹³C in the Krebs cycle. In the present work, we used the equation proposed by Tayek and Katz (1996) to estimate this isotopic dilution as follows: in which the numerator is the sum of glucose molecules labeled with one, two and three ¹³C atoms (i.e., the number of oxaloacetate molecules metabolized to glucose), and the denominator represents the total ¹³C contained in these isotopomers. The equation thus calculates the average ¹³C-isotopic enrichment of any given molecule of oxaloacetate from which labeled glucose was derived. Thus fractional gluconeogenesis is Equation (6) × Equation (7) × Equation (8), i.e., ¹³C-recycling × 3-carbon dilution factor × Krebs cycle dilution.

2) Calculations from isotopomer recycling. This approach relies on measurements of two unique isotopomers of glucose, i.e., [M + 6]-glucose, which can derive only from the tracer, and [M + 3]-glucose, which can derive only from the direct metabolism of [M + 3]-pyruvate synthesized from the tracer glucose. In this approach, the minimum estimate of fractional gluconeogenesis is given by the [M + 3]- isotopomer recycling (GIR) as follows: The factor of 2 in the denominator accounts for the fact that each molecule of [M + 6]-glucose yields 2 molecules of [M + 3]-glucose.

This expression is also an underestimate of fractional gluconeogenesis because of the ¹²C-dilution discussed above. In this approach the dilution in the 3-carbon pool is given by the following: However, in using only the [M + 3]-isotopomer in the calculation, it must be noted that, because of metabolic cycling between oxaloacetate and fumarate, [M + 3]-oxaloacetate becomes a mixture of two forms, one labeled in carbons 1-3 (which yields [M + 3]-glucose) and one labeled in carbons 2-4 (which yields [M + 2]-glucose). At equilibrium, there is an equimolar mixture of the two forms of [M + 3]-oxaloacetate, so that Equation (9) underestimates true recycling by a factor of 2. Therefore, applying the Krebs cycle dilution factor [Equation (8)] as before, fractional gluconeogenesis becomes the following: [2 × Equation (9)] × Equation (8) × Equation (10), i.e., (2 × isotopomer recycling) × 3-carbon dilution × Krebs cycle dilution.

RESULTS

The combined values for the tracer:tracee ratios of the ¹³C-isotopomers in blood glucose in the mice that received the tracer for 1, 2 and 5 d, are shown in Table 2. The tracer:tracee ratio of [M + 6]-glucose [1.34 mol/100 mol (HCD) and 0.84 mol/100 mol (LCD)] was substantially lower than that of the diet (5 mol/100 mol). There was a significant diet effect (P < 0.025). Even after a 4-h feeding period, the tracer:tracee ratio of blood [M + 6]-glucose (3.4 ± 0.5 mol/100 mol) was only 68% of that of the dietary tracer. Circulating glucose was highly enriched with the [M + 3]-isotopomer; after 4 h, the tracer:tracee ratio of [M + 3]-glucose was 40 ± 7% of [M + 6]-glucose, and by 24 h of feeding, the tracer:tracee ratio of [M + 3]-glucose was 61 (HCD) and 86% (LCD) of that of the [M + 6]-isotopomer of blood glucose (diet effect P < 0.01).

Table 2. Molar tracer:tracee ratios and 13C-enrichments of cardiac blood glucose obtained from two groups of fed mice offered, for 1-5 d, diets containing high (HCD) or low (LCD) carbohydrate contents and [U-13C]-glucose1 [View Table]

Values for glucose turnover and recycling calculated from the degree of isotopic dilution of the [M + 6]-glucose tracer are summarized in Table 3. The apparent glucose entry rates [7.9 (HCD) and 7.1 g/d (LCD)] were 3.4- (HCD) and 5.5-fold (LCD) higher than the total carbohydrate intake of the two groups of mice (P < 0.001). Thus, apparent endogenous glucose "production" contributed 67 (HCD) and 81% (LCD) of the total blood glucose turnover. The diet effect was significant (P < 0.01). A minimum [Equation (9)] of 25 (HCD) and 30% (LCD) (diet effect P < 0.05) of the glucose [M + 6]-tracer had recycled directly via oxaloacetate. The proportion of the total ¹³C-glucose ingested by the mice that had recycled [Equation (6)] was even higher [44 (HCD) and 50% (LCD); diet effect P < 0.05].

Table 3. Estimated apparent glucose entry rate, endogenous glucose production and label recycling as measured by blood glucose labeling in two groups of fed mice offered, for 1-5 d, diets containing high (HCD) or low (LCD) carbohydrate and [U-3C]-glucose1 [View Table]

The labeling of circulating lactate and blood and hepatic alanine is shown in Table 4. By 24 h, blood lactate and alanine were in isotopic equilibrium. After 4 h of feeding [U-¹³C]-glucose, the tracer:tracee ratio of blood [M + 3]-alanine (2.68 ± 0.16) was 66% of [M + 6]-glucose + (0.5 × [M + 3]-glucose) [see Equation (10)], and by 24 h, glucose and alanine had reached full isotopic equilibrium. The tracer:tracee ratio of hepatic [M + 3]-alanine was significantly (P < 0.05) higher than that of blood [M + 3]-alanine.

Table 4. Molar tracer:tracee ratios and 13C-enrichments of cardiac blood lactate and blood and hepatic alanine obtained from two groups of fed mice offered, for 1-5 d, diets containing high (HCD) or low (LCD) carbohydrate contents and [U-13C]-glucose [View Table]

In Table 5, we summarize the estimates of the isotopic dilution of the hepatic 3-carbon, using hepatic alanine as the basis of the calculation, and of Krebs cycle dilution using the Tayek and Katz (1996) formula. The Krebs cycle dilution factor was 1.8, suggesting that 55% (i.e., 1/1.8) of the hepatic pool of oxaloacetate derived from pyruvate carboxylation. Table 5 also shows the two estimates of fractional contribution of gluconeogenesis to blood glucose turnover, after adjustment for the label dilution in the 3-carbon and oxaloacetate pools. Gluconeogenesis, as measured by glucose ¹³C-recycling, accounted for 63 (HCD) and 74% (LCD) of total glucose turnover (diet effect P < 0.05). The corresponding values derived from isotopomer recycling gave significantly higher values [72 (HCD) and 85% (LCD); diet effect P < 0.01]. Thus the estimates of gluconeogenesis from tracer dilution [67 (HCD) and 81%(LCD)] and from label reincorporation [HCD (63-72%), LCD (74-85%)] were similar.

Table 5. 3-Carbon and Krebs cycle dilution and gluconeogenesis calculated from 13C-recycling and isotopomer recycling in blood glucose in two groups of fed mice offered, for 1-5 d, diets containing high (HCD) or low (LCD) carbohydrate contents and [U-13C]-glucose1 [View Table]

The labeling of hepatic glycogen is shown in Table 6. Unlike blood glucose and its metabolites, the tracer:tracee ratio in all glucose ¹³C-isotopomers in hepatic glycogen increased with time. These data were compatible with a fractional rate of hepatic glycogen turnover of 114%/d in the HCD and 108%/d in the LCD group (diet effect P > 0.05, not significant). In both groups, [M + 6]-glycogen glucose had a significantly (P < 0.01) higher tracer:tracee ratio [+54% (HCD) and +115% (LCD); diet effect P < 0.001] than [M + 6]-glucose in blood. [M + 3]-glycogen glucose had a generally lower tracer:tracee ratio [5% (HCD) and 26% (LCD); diet effect P < 0.05] than [M + 3]-blood glucose. The difference between [M + 3]-glycogen glucose and [M + 3]-blood glucose was significant (P < 0.05) in the LCD group, but was not significant in the HCD group. Isotopomer recycling (17 ± 5%) and ¹³C-recycling (31 ± 2%) determined from the labeling of hepatic glycogen were also significantly different (P < 0.001) than those in blood glucose and, unlike blood glucose labeling, the contribution of recycled glucose to glycogen was unaffected by the diet. Applying the dilution factors for the 3-carbon and Krebs cycle suggested that 44% of hepatic glycogen derived from recycled glucose.

Table 6. Time course of isotopic enrichment of the 13C-isotopomers of hepatic glycogen-bound glucose from two groups of fed mice offered, for 1-5 d, diets containing high (HCD) or low (LCD) carbohydrate contents and [U-13C]-glucose1 [View Table]

DISCUSSION

The results that we obtained were notable in the following respects: Although it is relatively simple to measure systemic glucose entry rates with ²H- or ³H-glucose, the interactions between glucose intermediary metabolism and the metabolism of other organic macronutrients pose substantial challenges to the design of experiments aimed at the measurement of pathways of glucose metabolism. The fact that despite many years of investigation, there is still considerable argument regarding the exact rate of gluconeogenesis in vivo, even after a prolonged fast (Hellerstein et al. 1995, Landau et al. 1995a; Neese et al. 1995), provides a good indication of these difficulties.

There are two measurements that are important for gluconeogenesis, namely, the measurement of the overall rate of glucose synthesis, irrespective of its source of carbon and the measurement of the relative contribution of glucose recycling (i.e., the Cori cycle; see Consoli et al. 1990) and of "new" glucose, synthesized from amino acids and glycerol, to total glucose synthesis. Total gluconeogenesis can now be quantified with the deuterium incorporation method (Landau et al. 1995b); quantifying metabolic recycling, however, is generally expedited by the use of carbon tracers. We are of the opinion that, even with the difficulties of determining some key intracellular isotope dilution factors, the use of [U-¹³C]-glucose has some specific advantages.

The first advantage is that the tracer:tracee ratio of circulating [M + 6]-glucose provides a direct estimate of the systemic entry rate of glucose not bearing six ¹³C-atoms. Indeed, it was the observation in the present work that the labeling of [U-¹³C]-glucose in arterial blood was much less than that of the oral [U-¹³C]-glucose tracer in the present study that gave the first indication that the first-pass metabolism of the oral tracer (estimated at between 65 and 80% of dose) was much more extensive than would have been anticipated from a consideration of the previous literature (Moore et al. 1991 and 1994, Shulman et al. 1987).

In interpreting this observation, it is critical to emphasize two points. First, peripheral glucose utilization will not alter the isotopic enrichment of the circulating tracer glucose. Thus, the fall in the tracer:tracee ratio of [M + 6]-glucose between the diet and the cardiac blood is the result of entry between the two sites of glucose that is not labeled with six ¹³C-atoms, i.e., glucose entry in the tissues of the splanchnic bed. Second, when [U-¹³C]-glucose is the tracer, the entry of any glucose molecule that is not labeled specifically with six ¹³C-atoms will dilute the tracer:tracee ratio of the [M + 6]-isotopomer. Thus, the "dilution" of [M + 6]-glucose could result from the entry either of unlabeled glucose (for example from glycogen) or of glucose labeled with less than six ¹³C-atoms, i.e., isotopically labeled glucose that has been resynthesized from the tracer.

This second point highlights other advantages of using [U-¹³C]-glucose as a tracer. Because [U-¹³C]-glucose introduces ¹³C into intermediary metabolic pathways in a predictable pattern, it leads to the synthesis of molecules whose distribution of ¹³C-isotopomers (i.e., ¹³C₁ , ¹³C₂ ...) is also predictable from a knowledge of the metabolic pathways that are potentially involved. For example, the appearance of the [M + 3]-isotopomer of glucose can occur only via the direct operation of the pathway [M + 6]-glucose[M + 3]-pyuvate[M + 3]-oxaloacetate[M + 3]-glucose. As such, the level of labeling of [M + 3]-glucose gives a minimum value for the proportion of the glucose tracer that has recycled, at some site, via pyruvate and oxaloacetate. In addition, measurements of the labeling of the [M + 3]-isotopomers of pyruvate, lactate and alanine give an unequivocal estimate of the contribution of glucose to the 3-carbon metabolite flux and hence the entry of new carbon into the gluconeogenic precursor pool.

Both of these advantages become apparent when we consider the other aspects of the isotopic data. In the present work, the tracer:tracee ratio of the [M + 3]-isotopomer of glucose was remarkably high in relation to previous observations from experiments with intravenous [U-¹³C]-glucose (e.g., Katz et al. 1989, Keshen et al. 1997). This shows that much of the "dilution" of the [M + 6]-glucose in blood represented the reincorporation of labeled carbon from the tracer glucose. The high recycling of the oral tracer glucose suggested by this result is in keeping with data obtained in dogs over a 6-h period after the ingestion of [U-¹⁴C]-glucose as part of a mixed meal (Moore et al. 1994). These authors found that the measured systemic appearance of unlabeled glucose from a mixed meal (60% of intake) was less than that measured from the appearance of an oral [U-¹⁴C]-glucose load (82% of dose). Because the measurement of the specific radioactivity of ¹⁴C-glucose measured all ¹⁴C-glucose, rather than just [U-¹⁴C]-glucose, it included all of the recycled label. The result suggests that ~35% of the oral glucose had been recycled in first pass. This value is similar to our estimate (36%) based on isotopomer recycling after 4 h of feeding [U-¹³C]-glucose.

The advantage of [U-¹³C]-glucose is also illustrated by the data on the labeling of the 3-carbon pools. In our previous intravenous tracer studies in piglets (Jahoor et al. 1994) and low birth weight infants (Keshen et al. 1997), we found that blood lactate, pyruvate and alanine were in isotopic equilibrium with one another, but that their tracer:tracee ratios were only about 25% of that of plasma [M + 6]-glucose. However, in the present experiment with oral [U-¹³]-glucose, blood lactate and alanine were in isotopic equilibrium with blood glucose. In addition, the tracer:tracee ratio of hepatic [M + 3]-alanine was higher than that of plasma [M + 3]-alanine. It would appear therefore that the oral glucose tracer revealed substantial first-pass splanchnic conversion of portal glucose to the hepatic 3-carbon pool that was not seen with an intravenous tracer. The site of the production of 3-carbon metabolites of the enteral tracer glucose is not certain at this stage and could occur in the intestine or the liver. We believe, however, that the major site is the liver (see also Mitrakou et al. 1991, Moore et al. 1991, Shulman et al. 1987) because although isolated enterocytes will produce lactate from glucose (Wu et al. 1995), and in vivo evidence shows lactate production from enteral (Vaugelade et al. 1995) or systemic (Windmueller and Spaeth 1978) glucose, the proportion of the glucose dose converted to lactate in vivo is small. Nevertheless, this remains an important question with both nutritional and regulatory implications. However, irrespective of the site of first-pass metabolism of enteral glucose, label recycling suggested that "gluconeogenesis" contributed between 63 and 72% (HCD) and 74 and 85% (LCD) of total glucose turnover, whereas isotopic dilution of the oral tracer suggested very similar values of 67 (HCD) and 81%(LCD).

The evidence that has accumulated over a number of years (Katz and Lee 1991, Katz and McGarry 1984, Katz et al. 1989, Mitrakou et al. 1991, Moore et al. 1994, Shulman et al. 1987 and 1990) suggests that, even for animals in the fed state, hepatic glycogen is continually broken down and resynthesized and that a nutritionally relevant portion of the synthesis also derives from glucose that has been synthesized from 3-carbon precursors. In the present experiment, the results suggested that the fractional rate of hepatic glycogen turnover was approximately 100%/d (114%/d, HCD and 108%/d, LCD). This corresponds to the incorporation of 83 (HCD) and 57 mg (LCD) of glucose/(mouse·d), accounting for 3.6 and 4.6%, respectively, of total carbohydrate intake of the mice.

In the present work, we also applied the tracer recycling approach to the calculation of the contribution of the direct (i.e., oral glucosehepatic glucose 6 phosphateglycogen) and the indirect (oral glucosepyruvateglucose 6 phosphateglycogen) pathways. Applying the dilution factors for the hepatic 3-carbon and oxaloacetate pools led to the conclusion that the indirect pathway contributed 44% of the glucose incorporated into hepatic glycogen. This was significantly (P < 0.025) lower than the contribution of gluconeogenesis to plasma glucose (63-85%) and was, moreover, unaffected by habitual dietary carbohydrate. This leads to the conclusion that there is channeling of dietary glucose directly into hepatic glycogen even in the fully fed state and that this pathway is preserved, and probably enhanced, when dietary carbohydrate is in restricted supply.

At present, the functional importance of what appears to be a substantial splanchnic glucose cycle can be a subject only of speculation, but we believe that it may reflect the intersection of glucose and amino acid metabolism. In this context, it is important to stress that in the present study the mice consumed an adequate and mixed (i.e., protein-containing) diet. Moore et al. (1994) have emphasized that the metabolic disposition of glucose derived from a mixed meal may differ from that derived from a test feeding of glucose alone, in part because of the likelihood of substantial synthesis of lactate and alanine from dietary protein in the intestinal mucosal cells. This increases the flux of gluconeogenic precursors, especially alanine, to the liver and may necessitate an increase in intrahepatic gluconeogenesis (Jungas et al. 1992). Moreover, it appears that the cycle is maintained in the face of a substantial restriction in dietary carbohydrate intake. However, both groups of mice received diets that provided protein in considerable excess over their metabolic requirements and, as a consequence, the mice were catabolizing substantial quantities of amino acids. Thus, if Jungas et al. (1992) are correct in their important proposal that the hepatic catabolism of amino acids involves an obligatory synthesis of glucose, then it is possible that the metabolic cycle revealed by the present and earlier results relates more to the regulation of the oxidation of excess dietary protein than to glucose metabolism itself.