## Abstract

There are conflicting reports concerning the reliability of mass isotopomer distribution analysis (MIDA) for estimating the contribution of gluconeogenesis to total glucose production (f) during [^{13}C]glycerol infusion.^{1}We have evaluated substrate-induced effects on rate of appearance (R_{a}) of glycerol and glucose and f during [2-^{13}C]glycerol infusion in vivo. Five groups of mice were fasted for 30 h and then infused with [2-^{13}C]glycerol at variable rates and variable ^{13}C enrichments (*group I*: 20 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C;*group II*: 60 μmol ⋅ kg^{−1} ⋅ min^{−1}, 60% ^{13}C;*group III*: 60 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C;*group IV*: 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, 40% ^{13}C; or*group V*: 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C). The total glycerol R_{a} increased from ∼104 to ∼157 and to ∼210 μmol ⋅ kg^{−1} ⋅ min^{−1}as the infusion of [2-^{13}C]glycerol increased from 20 to 60 and to 120 μmol ⋅ kg^{− 1} ⋅ min^{−1}, respectively. As the amount of 99% enriched [2-^{13}C]glycerol increased from 20 to 60 and to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}(*groups I*,*III*, and*V*, respectively), plasma glycerol enrichment increased from ∼21 to ∼42 and to ∼57% and the calculated f increased from ∼27 to ∼56 and to ∼87%, respectively. Similar plasma glycerol enrichments were observed in*groups I*,*II*, and*IV* (i.e., ∼21–24%), yet f increased from ∼27 to ∼57 and to ∼86% in*groups II* and*IV*, respectively. Estimates of absolute gluconeogenesis increased from ∼14 to ∼33 and ∼86 μmol ⋅ kg^{−1} ⋅ min^{−1}as the infusion of [2-^{13}C]glycerol increased from 20 to 60 and 120 μmol ⋅ kg^{−1} ⋅ min^{−1}. Plausible estimates of f were obtained only under conditions that increased total glycerol R_{a}∼2-fold (*P* < 0.001) and increased glucose R_{a} ∼1.5-fold (*P* < 0.01) above basal. We conclude that in 30-h fasted mice, *1*) estimates of f by MIDA with low infusion rates of [2-^{13}C]glycerol yield erroneous results and *2*) reasonable estimates of f are obtained at glycerol infusion rates that perturb glycerol and glucose metabolism.

- stable isotopes
- mass spectrometry
- triose phosphate turnover
- diabetes
- mice

strong et al. (29) originally reported that it is possible to measure the biosynthesis of polymeric molecules if a labeled precursor is administered and the mass isotopomer distribution of the polymer is determined. More recently, mass isotopomer distribution analysis (MIDA) was proposed as a method for estimating the fractional synthetic rate of various biopolymers including cholesterol, fatty acids, glucose, and DNA (11). (For a comprehensive review of the application of isotopomer analysis methods for studying physiological problems see Ref. 3.)

Glucose can be considered as a dimer formed from the condensation of two triose phosphate molecules. Thus MIDA of glucose made from a^{13}C-labeled gluconeogenic precursor(s) has been proposed as a method for estimating the contribution of gluconeogenesis to total glucose production (f; Refs.20, 21). MIDA permits the determination of the triose phosphate enrichment. In contrast to other tracer methods, MIDA calculations of f are not subject to artifacts of isotope exchange or dilution, which have limited investigations of gluconeogenesis (5, 14, 15). The main underlying assumption of MIDA, however, is that the triose phosphate pool(s) in all gluconeogenic cells must be at similar^{13}C enrichments (16, 24, 25). If this assumption is not valid, f is underestimated (16, 24, 25).

There are conflicting opinions regarding the general applicability of MIDA for estimating f during the infusion of [^{13}C]glycerol in vivo. Investigators have infused [2-^{13}C]glycerol and concluded that it is possible to correctly estimate f (12, 20, 21, 23). However, other investigators have infused [U-^{13}C_{3}]glycerol and concluded that f was underestimated and that MIDA is not a reliable method for estimating f (16, 24). A recent report (25) has shown that [2-^{13}C]glycerol and [U-^{13}C_{3}]glycerol provide the same estimation of f if used under identical conditions in isolated hepatocytes. In addition, this in vitro study demonstrated that the relative contribution of [^{13}C]glycerol vs. other gluconeogenic precursors plays a role in determining f, such that f increased as the contribution of [^{13}C]glycerol increased (25). Lastly, the hepatocyte experiments demonstrated that glucose production increased as the supply of glycerol increased. Thus it appears that the amount of [^{13}C]glycerol infused, and not the type of [^{13}C]glycerol, may be an important factor for applying MIDA to studies of gluconeogenesis.

Because of the increasing application of the MIDA technique (7, 10, 13,28, 30) and the uncertainty of its reliability in vivo (16, 22, 24), we have asked two questions. First, are there substrate-induced effects on glycerol and glucose metabolism during [2-^{13}C]glycerol infusion in vivo? Second, do substrate-induced effects have an impact on MIDA and estimations of f ?

## MATERIALS AND METHODS

*Materials*. Unless noted, chemicals were purchased from Sigma-Aldrich. [2-^{13}C]glycerol (99 atom percent excess) was purchased from Cambridge Isotopes (Andover, MA). [3-^{3}H]glucose was purchased from New England Nuclear (Boston, MA). Gas chromatography-mass spectrometry (GC-MS) supplies were purchased from Hewlett-Packard (Wilmington, DE).*N*-methyl-*N*-(tert-butyldimethylsilyl)trifluoroacetamide + 1% tert-butyldimethylchlorosilane (TBDMS) and bis(trimethylsilyl)trifluoroacetamide + 10% trimethylchlorosilane (TMS) were purchased from Pierce (Rockford, IL).

*In vivo studies*. Male BALB/C mice (Charles River, 20–25 g) were fitted with jugular vein catheters and allowed a 4-day recovery. Mice were fasted 30 h and infused with [2-^{13}C]glycerol at variable rates and variable ^{13}C enrichments (*group I*: 20 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C;*group II*: 60 μmol ⋅ kg^{−1} ⋅ min^{−1}, 60% ^{13}C;*group III*: 60 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C;*group IV*: 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, 40% ^{13}C; or*group V*: 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99%^{13}C). [3-^{3}H]glucose was infused at 1.0 μCi ⋅ kg^{−1} ⋅ min^{−1} to estimate total glucose production.

During preliminary experiments (Table 1; Fig. 1), blood (∼65 μl) was sampled at 0, 60, 100, 140, and 180 min. These studies revealed that the plasma glycerol enrichment and calculated values of f reached a steady state before 140 min. During all additional studies, blood (∼65 μl) was sampled via tail-tip bleeds at 0, 140, and 180 min for measurement of the concentration and labeling of glycerol, glucose, and lactate.

*Analyses*. Plasma samples obtained at 140 and 180 min were processed for^{3}H counting as follows. Briefly, a 5-μl aliquot of plasma was deproteinized by adding 10 μl of 10% trichloroacetic acid. A 10-μl portion of the supernatant was dried to remove^{3}H_{2}O. The residue was dissolved in 100 μl of water. After addition of 2 ml of scintillation fluid (Ultima Gold, Packard Instrument, Meriden, CT), samples were counted for 1 min.

GC-MS analyses were conducted with a Hewlett Packard 5973 MSD equipped with a Hewlett Packard 6980 GC. All samples were analyzed with split injection (10:1) on a Hewlett Packard 5MS capillary column (30 m ⋅ 0.25 mm ⋅ 0.25 μm) maintained at a constant helium flow (1.2 ml/min). Plasma samples obtained at 0, 140, and 180 min were processed for GC-MS analysis of the mass isotopomer distribution of glucose as follows. A 10-μl plasma sample was added to 100 μl of 100% methanol. The supernatant was evaporated to dryness. The mass isotopomer distribution of glucose was determined after conversion to its aldonitrile pentaacetate derivative and analysis with chemical ionization (methane gas). This sample was also analyzed with electron impact ionization to determine^{13}CO_{2}reincorporation in one-half of the glucose (2, 19).^{13}CO_{2}reincorporation could lead to the formation of a double-labeled triose phosphate, and if uncorrected for, gluconeogenesis would be underestimated (19). We did not find any*M* + 2_{Glc} in the fragment corresponding to C-4_{Glc} to C-6_{Glc}.

We determined the concentration of glucose and the enrichment and concentration of glycerol and lactate as follows. Briefly, a known amount of [U-^{13}C_{6}]glucose, [^{2}H_{5}]glycerol, and [U-^{13}C_{3}]lactate was added to 25 μl of plasma. The sample was added to 200 μl of 100% methanol. The supernatant was split into three parts (150, 30, and 30 μl), and each was evaporated to dryness. Plasma glycerol concentration and ^{13}C enrichment were determined on the 150-μl fraction, as its TMS derivative with electron ionization. The glycerol concentration was calculated from the mass-to-charge ratio (*m*/*z*) (205 + 206)/208 signal, and the glycerol enrichment was calculated from the*m*/*z*206/(205 + 206) signal. A 30-μl fraction was used to determine the concentration of glucose, as its aldonitrile pentaacetate derivative with chemical ionization. The glucose concentration was calculated from the*m*/*z*(328 + 329 + 330)/334 signal. The second 30-μl fraction was used to determine the concentration and^{13}C enrichment of plasma lactate, as its TBDMS derivative with electron ionization conditions. The lactate concentration was calculated from the*m*/*z*(261 + 262)/264 signal, and the lactate enrichment was calculated from the*m*/*z*262/(261 + 262) signal.

All samples were analyzed in duplicate. Background mass isotopomer distributions were corrected (9, 27). Data are presented as the means ± SE. Statistics were calculated with one-way ANOVA, and Tukey’s post hoc testing was used to determine significance.

*Calculations*. The total rate of appearance (R_{a}) of glycerol, in μmol ⋅ kg^{−1} ⋅ min^{−1}, equals
Equation 1 The endogenous R_{a} of glycerol, in μmol ⋅ kg^{−1} ⋅ min^{−1}, equals
Equation 2 The R_{a} of glucose, in μmol ⋅ kg^{−1} ⋅ min^{−1}, equals
Equation 3 The fractional triose phosphate enrichment equals
Equation 4 where*M* + 1_{glucose} and*M* + 2_{glucose} are the molar percent excess values of glucose with one and two^{13}C atoms, respectively. The contribution of gluconeogenesis to f equals
Equation 5 where p is the fractional triose phosphate enrichment. The absolute rate of gluconeogenesis, in μmol ⋅ kg^{−1} ⋅ min^{−1}, equals
Equation 6

## RESULTS

Data from preliminary studies revealed that steady-state concentrations and enrichments of plasma glycerol were observed under the conditions tested. For example, Fig. 1 shows the profile of plasma glycerol concentration and ^{13}C enrichment in mice infused with [2-^{13}C]glycerol (20 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C). Similar steady-state data were obtained from mice infused at the other rates of glycerol (not shown).

Table 1 shows the mass isotopomer distribution of plasma glucose and the contribution of gluconeogenesis to total glucose production from 100 to 180 min in mice infused with [2-^{13}C]glycerol (20 μmol ⋅ kg^{−1} ⋅ min^{−1}, 99% ^{13}C). Calculated values of f were in steady state by 100 min. Similar steady-state data were obtained from mice infused at the other rates of glycerol (not shown).

Table 2 shows the effect of increasing glycerol infusion to 30-h fasted mice. Data were averaged from samples obtained at 140 and 180 min. Despite a threefold increase in plasma glycerol concentration and a twofold increase in total glycerol R_{a} when the glycerol infusion rate was increased from 20 to 120 μmol ⋅ kg^{− 1} ⋅ min^{−1}, the endogenous glycerol R_{a} was similar in each group. Thus the increase in total glycerol R_{a} was induced by the exogenous glycerol infusion. As the glycerol infusion increased from 20 to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, the plasma glucose concentration increased from ∼133 to ∼214 mg/dl (*P* < 0.001) and the glucose R_{a} increased from ∼63 to ∼105 μmol ⋅ kg^{−1} ⋅ min^{−1}(*P* < 0.01).

The triose phosphate enrichment and f were calculated from the corrected mass isotopomer distributions of plasma glucose observed in each group (Table 3). The increase in the calculated triose phosphate enrichment was proportional to the infusion rate of 99% enriched [2-^{13}C]glycerol (*groups I*,*III*, and*V*). Similar triose phosphate enrichments were maintained in *groups I*,*II*, and*IV* (i.e., ∼14–15%) by adjusting the ^{13}C enrichment of the glycerol infusate. However, despite the low triose phosphate enrichments, f increased as the infusion rate of glycerol increased (compare *groups I*,*II*, and*IV*). Estimations of absolute gluconeogenesis, calculated as the product of the glucose R_{a} (Table 2) and f (Table 3), increased as the glycerol infusion rate increased. There was a linear increase in both f (*y* = 0.583*x* + 17.5;*r*
^{2} = 0.98) and absolute gluconeogenesis (*y* = 0.732*x* − 4.53;*r*
^{2} = 0.97) as the infusion rate of glycerol increased (Fig.2). These data demonstrate that estimates of gluconeogenesis are dependent on the load of glycerol and independent of the calculated triose phosphate enrichment.

Table 4 shows the effects of variable [2-^{13}C]glycerol infusion on plasma lactate ^{13}C enrichment and concentration. As the infusion rate of 99% enriched [2-^{13}C]glycerol increased from 20 to 60 and to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, plasma lactate enrichment increased from ∼9 to ∼19 and to ∼30%, respectively. Plasma lactate enrichment did not change in the*groups I*,*II*, and*IV*, which had similar plasma glycerol enrichments. Lactate concentration remained relatively constant in all groups.

## DISCUSSION

In theory, after the infusion of a^{13}C-labeled gluconeogenic precursor (e.g., [^{13}C]glycerol), the MIDA of plasma glucose can be used to estimate contribution of gluconeogenesis to f (11). An example may clarify how MIDA works for calculating f. Consider the synthesis of glucose during the infusion of [2-^{13}C]glycerol in vivo. The triose phosphate pool contains species at*m* + 0 (unlabeled triose phosphate from endogenous gluconeogenic substrates) and*m* + 1 (triose phosphate with 1^{13}C from the infused [2-^{13}C]glycerol). Depending on the relative amount of *m*+ 0 vs. *m* + 1 triose phosphate, newly synthesized glucose is either molecular weight 180 (*M* + 0), 181 (*M* + 1), or 182 (*M* + 2). From the ratio of*M* + 2_{glucose} to*M* + 1_{glucose}, one can infer the triose phosphate enrichment (i.e., *Eq*.*
4
*) and then calculate f (i.e.,*Eq*.*
5
*). If [U-^{13}C_{3}]glycerol is infused, the triose phosphate pool contains species at*m* + 0 to*m* + 3 and newly synthesized glucose contains species at *M* + 0 to*M* + 6 (16, 24, 25). The principle of the calculation of the triose phosphate enrichment and f is the same as described above, although the equations are more complex and generally require a computer algorithm to solve for the triose phosphate mass isotopomer distribution (16, 24, 25).

Currently, there are contrasting opinions regarding the validity of MIDA for estimating f during [^{13}C]glycerol infusion in vivo. Investigators have infused [2-^{13}C]glycerol and concluded that it is possible to correctly estimate f (20, 21, 23). However, other investigators have infused [U-^{13}C_{3}]glycerol and concluded that f was underestimated (16, 24). It was proposed that the apparent underestimation of f could be related to a transplanchnic (or transhepatic) decrease in the concentration and enrichment of [^{13}C]glycerol (16). Such gradients were directly measured with arteriovenous isotope balance measurements (17, 24, 26).

Data from the study of Ekberg et al. (8) indirectly support the stable isotope data demonstrating a transsplanchnic gradient of glycerol concentration and enrichment in vivo (17, 24, 26). Briefly, 36-h-fasted subjects ingested 0.5 g of acetaminophen and were infused with [2-^{14}C]glycerol and [1-^{14}C]lactate. The^{14}C distribution in blood glucose and urinary acetaminophen glucuronide was measured. [2-^{14}C]glycerol labels C-2 and C-5 of blood glucose and urinary glucuronide. [1-^{14}C]lactate labels C-3 and C-4 of blood glucose and urinary glucuronide. The labeling ratio (C-2 + C-5) to (C-3 + C-4) was higher in the blood glucose than in the urinary glucuronide. Presumably, the acetaminophen sampled a hepatic pool of glucose 6-phosphate in which there was more label from lactate than from glycerol. It was concluded that the concentration of glycerol decreased much faster than the concentration of lactate as blood passed from periportal to perivenous hepatocytes.

More recently, in vitro studies demonstrated that MIDA estimates of f are dependent on the relative flux of [^{13}C]glycerol vs. other gluconeogenic substrates and independent of the type of [^{13}C]glycerol infused (25). This raised the question as to whether estimates of f are affected by increasing the infusion rate of [2-^{13}C]glycerol in vivo. We found that as the infusion rate of [2-^{13}C]glycerol increased from 20 to 60 and to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}that estimates of f increased from ∼27 to 56 and to 86%, respectively (Table 3). These findings suggest that not all gluconeogenic cells are equally labeled when low doses of [2-^{13}C]glycerol are infused. Presumably, as the infusion rate of [2-^{13}C]glycerol increased, a more homogeneous gluconeogenic precursor enrichment was achieved. In addition, as the infusion rate of [2-^{13}C]glycerol increased from 20 to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, estimates of total glucose production increased from ∼63 to ∼105 μmol ⋅ kg^{−1} ⋅ min^{−1}, respectively. In this study, estimates of absolute gluconeogenesis increased from ∼14 to ∼33 and to ∼86 μmol ⋅ kg^{−1} ⋅ min^{−1}, as the infusion rate of [2-^{13}C]glycerol increased from 20 to 60 and to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}, respectively (Table 3).

MIDA of gluconeogenesis is limited to the extent that the^{13}C-gluconeogenic precursor does not enter all gluconeogenic cells equally. Metabolic zonation of gluconeogenesis can lead to heterogeneity in labeling the triose phosphate pool. Several mechanisms have been presented by which metabolic zonation can occur (8, 16, 17, 24-26). These include*1*) a decrease in substrate concentration across a gluconeogenic organ,*2*) dilution of tracer across a gluconeogenic organ, *3*) inflow of substrate close to or below the Michaelis-Menten constant (*K*
_{m}) of the first rate-limiting enzyme that acts on the substrate,*4*) multiple gluconeogenic organs, and/or *5*) relative intracellular flux into the triose phosphate pool is not constant in all gluconeogenic cells. In particular, metabolic zonation of the triose phosphate labeling can be induced by transhepatic changes in the concentrations of various gluconeogenic precursors (e.g., glycerol and lactate). Most likely, in previous studies, periportal hepatocytes were exposed to more [U-^{13}C_{3}]glycerol than perivenous hepatocytes (16, 24). As a result, gluconeogenesis occurring in the perivenous zone of the liver was erroneously ascribed to glycogenolysis. Similar substrate gradients may develop across the kidney (17, 26) and could lead to heterogeneity of renal glucose production. Renal glucose production may be important during some conditions (1, 4, 6, 22). Presumably, in our current study the increase in total glycerol concentration must have resulted in blunting of transhepatic (and transrenal) gradients in the concentration and^{13}C enrichment of glycerol. Consequently, a more homogeneous triose phosphate enrichment was maintained.

The groups of Hellerstein and co-workers [Dekker et al. (7) and Siler et al. (28)] and Peroni et al. (23) have suggested that perhaps methodological error(s) was responsible for the underestimation of f when [U-^{13}C_{3}]glycerol was infused in vivo. However, in previous studies the mass isotopomer distributions of blood glucose and urinary acetaminophen glucuronide (analyzed as glucose) from humans infused with [U-^{13}C_{3}]glycerol were determined by independent laboratories with different derivatization techniques and analytical conditions (16). Despite the use of different methodologies, the calculated f values were similar and unexpectedly low (16). This indirectly argues against the occurrence of systematic error related to the use of [U-^{13}C_{3}]glycerol.

Recently, Dekker et al. (7) stated that “the use of [U-^{13}C_{3}]glycerol and [U-^{13}C_{3}]lactate leads to predictable problems due to enormous dynamic range for abundances of *M*
_{+6}compared to *M*
_{0},*M*
_{+1}...the ratio of*M*
_{+6}:*M*
_{0}with [U-^{13}C_{3}]glycerol or [U-^{13}C_{3}]lactate is greater than 1,000:1… ”. Although calculations of f are sensitive to analytical error and the use of U-^{13}C_{3}-labeled tracers requires a complex data fitting algorithm, the statement made by Dekker et al. (7) is unfounded. When one more closely compares the*M*
_{+6}-to-*M*
_{0}ratio with estimates of f, it is clear that there is no correlation between the*M*
_{+6}-to-*M*
_{0}ratio and estimates of f (24). For example, the largest*M*
_{+6}-to-*M*
_{0}ratios can be found in the in vivo studies (i.e., rats or monkeys; Ref.24). Glucose from rats infused with [U-^{13}C_{3}]glycerol had a*M*
_{+6}-to-*M*
_{0}ratio of ∼686:1 and f was ∼75%. Glucose from rats infused with [U-^{13}C_{3}]lactate had a*M*
_{+6}-to-*M*
_{0}ratio of ∼1,696:1, and f was ∼97%. Thus, under identical experimental conditions in rats, the greatest*M*
_{+6}-to-*M*
_{0}ratio was observed during [U-^{13}C_{3}]lactate infusion; yet, f was in the expected range considering the nutritional status (i.e., 48 h fasted). In monkeys, the*M*
_{+6}-to-*M*
_{0}ratio was similar during [U-^{13}C_{3}]glycerol or [U-^{13}C_{3}]lactate infusion, yet the f calculated during [U-^{13}C_{3}]lactate infusion was approximately twofold higher than f calculated during [U-^{13}C_{3}]glycerol infusion (80 vs. 48%). Furthermore, the value calculated for f was independent of the type of [^{13}C]lactate infused because similar f values were calculated in monkeys infused with [U-^{13}C_{3}]lactate (f ∼80%) or [3-^{13}C]lactate (f ∼81%; see Ref. 25). The f calculated during the infusion of [^{13}C]lactate to monkeys is reasonable considering their nutritional status (18 h fasted). Also, Lee et al. (18) obtained reasonable estimates of f during the infusion of [^{13}C]lactate to humans.

From the accumulated data (8, 16, 17, 24-26), one can conclude that a possible cause of the discrepancy in the estimates of f observed at various glycerol infusion rates may lie not in the isotopomer ratios and analytical techniques but in the substrate concentrations. In fact, with anesthetized rats, Peroni et al. (21) concluded that accurate in vivo measurements of f could be made at the expense of some perturbation of the metabolic pathway studied. In our study, we attempted to estimate gluconeogenesis at variable [2-^{13}C]glycerol infusion rates and independent of triose phosphate labeling. We found [2-^{13}C]glycerol infusion rates that maintained plasma glycerol enrichments at approximately the same level (i.e., 21–24%;*groups I*,*II*, and*IV*) and allowed us to give increasing quantities of glycerol. When either 60 or 99% enriched [2-^{13}C]glycerol was infused at 60 μmol ⋅ kg^{−1} ⋅ min^{−1}, we observed that the plasma glycerol concentration increased approximately twofold over basal (Table 2). However, despite the different plasma glycerol enrichments (24 vs. 42%) and calculated triose phosphate enrichments (14 vs. 23%), similar values of f were calculated (57 vs. 56%). Likewise, similar metabolic effects were observed when either 40 or 99% enriched [2-^{13}C]glycerol was infused at 120 μmol ⋅ kg^{−1} ⋅ min^{−1}(Table 2). Different plasma glycerol enrichments were imposed (24 vs. 57%), and different triose phosphate enrichments were calculated (15 vs. 45%), yet similar values of f were calculated (86 vs. 87%). Our findings indicate that by increasing the infusion rate of [2-^{13}C]glycerol from 20 to 120 μmol ⋅ kg^{−1} ⋅ min^{−1}that estimates of f increased from ∼27 to ∼87%, respectively. The increase in f was independent of the calculated triose phosphate enrichment (i.e., *groups I*,*II*, and*IV*). These data show that calculations of f were apparently limited by the supply of glycerol.

Lastly, our findings point to another potential limitation of MIDA regarding the calculation of the triose phosphate R_{a}. Neese et al. (20) proposed that it is possible to estimate the triose phosphate R_{a} during the infusion of [2-^{13}C]glycerol. However, to calculate the triose phosphate R_{a} one must assume that [2-^{13}C]glycerol is only taken up by the liver and that no other^{13}C-labeled compounds contribute to the influx to triose phosphates (20). Others have demonstrated that extrahepatic glycerol utilization is significant (17, 26) and that plasma lactate becomes labeled during the infusion of [U-^{13}C_{3}]glycerol (24). Under the conditions tested here, we found that plasma lactate becomes significantly enriched during the infusion of [2-^{13}C]glycerol (Table4). In particular, the lactate enrichment increased with increasing amounts of 99% enriched [2-^{13}C]glycerol. Thus our findings demonstrate that it is not possible to directly calculate the triose phosphate R_{a}, because one must account for [^{13}C]lactate flux to the triose phosphate pool.

On the basis of results of this study and earlier studies, we conclude that MIDA with [2-^{13}C]glycerol is not a reliable method for estimating f. Estimates of f by MIDA with low infusion rates of [2-^{13}C]glycerol yield erroneous results. Errors in MIDA are most likely due to metabolic zonation of gluconeogenesis, which leads to heterogeneity in labeling the triose phosphate pool. Reasonable estimates of f are obtained at high glycerol infusion rates that, under normal conditions, perturb glycerol and glucose metabolism. Presumably, transhepatic (and transrenal) glycerol gradients are decreased during high glycerol flux, and therefore a more homogeneous triose phosphate enrichment is maintained. Could there be conditions of increased glycerol flux during which the infusion of [^{13}C]glycerol and MIDA would provide reliable estimates of gluconeogenesis? One might expect an increased glycerol flux during total parenteral nutrition with lipid emulsions. However, detailed studies of MIDA estimates of gluconeogenesis have not yet been performed under such conditions.

## Acknowledgments

We thank Laura A. Burden, Veronika Walton, and Ying Zhu for expert technical assistance.

## Footnotes

Address for reprint requests and other correspondence: G. I. Shulman, Howard Hughes Medical Institute, Yale Univ. School of Medicine, 333 Cedar St., New Haven, CT 06520.

↵1 In the context of this report, the word “isotopomer” refers to mass isotopmers.

*M*+ X_{Glc}refers to a mass isotopomer of glucose, where X is the number of excess^{13}C atoms. Positional isotopomers are not considered here.This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases Grants R01-DK-40936 and P30-DK-45735. G. I. Shulman is an investigator of Howard Hughes Medical Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “

*advertisement*” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

- Copyright © 1999 the American Physiological Society