Under the growth conditions of these experiments, the Rh30 cells double in about 24 h in the presence of serum. The growth rate in the absence of serum (0.5% BSA) was about 30% slower, indicating only a weak dependence on growth factors. In contrast, the primary myocytes grew much more slowly, with a doubling time of approximately 50 h in the presence of serum growth factors. This suggests that the cancer cells have an overall net higher demand for metabolic energy to drive macromolecular biosynthesis as the cells proliferate. This is in addition to the normal basal cell metabolic requirements.

The amount of glucose remaining in the medium and the amounts of lactate and Ala that were secreted by the cells are readily measured by ¹H NMR and/or GC-MS. Furthermore, as NMR can distinguish between ¹³C-labeled and ¹²C lactate and glucose (see below), it was straightforward to determine the fraction of lactate that derived from the supplied ¹³C-glucose, or from other sources, according to Eq. (2). Table 1 shows the data for glucose consumption and secretion of lactate and Ala into the medium from the two cell types.

The consumption of [U-¹³C]-glucose by the Rh30 cells at 24 h was comparable to the consumption by the primary myocytes in 48 h, reflecting the slower growth rate of the latter. The amount and degree of ¹³C enrichment in lactate secreted was very much higher for the Rh30 cells than the primary myocytes (Table 1 and Figure 4). Since the dry weight of the myocytes at 48 h was comparable to that of the Rh30 cells at 24 h, this implies an 8-fold higher lactate production rate in the Rh30 cells than the myocytes. The amount and degree of ¹³C enrichment of Ala in the Rh30 medium was also much higher than measured in the myocyte medium, although Ala secretion constituted a smaller fraction of the labeled glucose consumption (Table 1 and Figure 4). Thus, the Rh30 cells converted a much higher fraction, F (cf. Eq. 2 and Table 1) of the ¹³C glucose consumed into lactate (50%) than the myocytes (10%). We have observed that typically 30–50% of the glucose consumed was converted to lactate by various epithelial derived cancer cell lines, (A.N. Lane, T. WM Fan, unpublished data). This is in contrast to a recent report on glioblastoma cells, in which more than 95% of the glucose consumed was converted to lactate and Ala 38. The high rate of production of labeled lactate in Rh30 and other cancer cells indicates that the pyruvate kinase (PK) step in glycolysis (Fig. 1) must be active, in contrast with the reports of defective PK in some tumor cells 28.

The high degree of lactate labeling in Rh30 cells indicates that the major source of carbon in lactate and Ala was indeed glucose. However, the fraction of ¹³C Ala labeling in the myocyte was essentially at natural abundance, reflecting its primary origin from the 10% FCS present in the medium. Thus, not only did the Ala labeling pattern in myocytes differ from that in Rh30 cells but also it was distinct from the lactate labeling pattern in both cell lines. As the immediate precursor to Ala and lactate is pyruvate, these differences in labeling patterns could indicate the presence of distinct intracellular pools of pyruvate as observed in other cells 3940 and see below. Moreover, a significant fraction of glucose taken up by Rh30 (48%) and myocytes (90%) was converted to metabolites other than lactate (cf. Table 1). The fate of some of the remaining glucose carbon was traced in the cellular components is described below

Thirty four metabolites in the TCA extracts of the cell biomass were analyzed using GC-MS (Figures 2, 3), twenty of which were also analyzed by ¹H NMR (Figure 5, 6). An additional eight compounds were determined by ¹H NMR alone for a total of forty two identified metabolites. The profiles of selected metabolites in Rh30 and myocytes determined by GC-MS are compared in Figures 2 and 3. Lactate, Gly, Thr, Glu, glutathione (GSH), Arg, and phosphocholine (P-choline), taurine and myo-inositol were among the more abundant metabolites (cf. Figures 2, 3, 4, 5). There was good agreement between GC-MS and NMR determinations, indicating high analytical reliability. Figs 2, 3, 4, 5 also illustrate that there are significant differences in levels of numerous metabolites between the myocytes and Rh30 cells. Notably there was a much higher content of amino acids, GSH, taurine, P-choline, myo-inositol, adenine nucleotides (5'-AXP), and uracil nucleotides in the Rh30 cells. The high abundance of P-choline in Rh30 compared with the myocytes (Figure 5) appears to be a distinct feature of proliferating cells 3537414243444546.

Figure 6 shows an expansion of the lactate region for the Rh30 cells shown in Fig 5. The 3-methyl resonances of both Ala and lactate (doublets at 1.53 and 1.4 ppm, respectively) display pairs of ¹³C satellite peaks displaced 126 Hz apart. These satellite peaks have a complex structure, which can be accounted for by considering additional couplings of these protons to ¹³C2 and ¹³C1 of the compounds (i.e. uniformly labeled) 3547. The presence of uniformly ¹³C-labeled lactate and Ala was also evident in the high-resolution 2-D ¹H-¹³C HSQC or the 1-D ¹³C projection spectra of Rh30 and myocyte cell extracts, as illustrated in Figure 7. The respective doublet and triplet patterns of the β-(C3) and α-(C2) carbons of lactate and Ala (due to ¹³C-¹³C coupling) 36 is further consistent with the ¹H TOCSY satellite patterns for these two metabolites (cf. Fig. 8). The NMR data were corroborated by the GC-MS analysis, where the m+3 (lactate+3) mass isotopomers predominated. These data together indicate that the three ¹³C in both lactate and Ala derived directly from the [U-¹³C]-glucose precursor via glycolysis.

Detailed quantification of the ¹³C-isotopomers of intracellular lactate and Ala by NMR and GC-MS is given in Table 2 and Fig. 3. As observed for the medium (see above), the myocytes showed a lower degree of enrichment than the Rh30 cells, especially for Ala, consistent with a substantially lower glycolytic rate in the primary cells. The ¹³C-Ala enrichment in the Rh30 cells grown in the absence of growth factors from FCS was even lower than that in the myocytes. There was no unlabeled Ala present in this medium indicating that the cells under growth factor depleted conditions were not actively transaminating labeled pyruvate (cf. Figure 1). Thus, the unlabeled Ala fraction may reflect contributions from other metabolic processes such as protein turnover or glutaminolysis 30 (cf. Figure 1) and possibly involving a different pool of pyruvate as has been observed in other cells 3940. However, the amount of protein turnover or glutaminolysis required to account for the production of unlabeled Ala was small owing to its low concentration (cf. Figures 4, 5).

Several other critical metabolites in the cells were also labeled with ¹³C as shown by the 2-D ¹H TOCSY spectrum of an example Rh30 cell extract in Figure 8. The cross-peak patterns are characteristic of particular metabolites, such as the ribose moieties of nucleotides, lactate, Ala and Glu, as annotated. Many of these cross-peaks are surrounded by satellite peaks that denote the presence of ¹³C. These are made clearer in the expansion of the spectral region of Glu, Ala and Lac (Fig. 9). It is clear that Glu was substantially labeled at the C4 (2.65 ppm) and C2 (4.08 ppm) positions and to a lesser extent at the C3 (2.22 ppm) position. Figure 9 also revealed the presence of four different Glu isotopomers, i.e. ¹³C-2,3-Glu, ¹³C-2-Glu, ¹³C-3-Glu, ¹³C-2,4-Glu, ¹³C-4-Glu, and ¹³C-3,4-Glu. Additional isotopomer information was obtained from analysis of the HSQC spectrum (Fig. 7). The doublet pattern of the γ- (C-4) and α- (C-2) carbons of Glu suggests the presence of ¹³C-4,5-Glu and/or ¹³C-3,4-Glu as well as ¹³C-1,2-Glu and/or ¹³C-2,3-Glu isotopomers, respectively. However, the ¹³C satellite pattern for the TOCSY cross-peak of H3 to H2 of Glu (2.22 to 4.08 ppm in Fig. 9) shows a significant amount of ¹³C-2,3-Glu. Similarly, the dominance of ¹³C-3,4-Glu is clear from the ¹³C satellite pattern for the cross peak of H4 to H3 of Glu (2.65 to 4.08 ppm in Fig. 9) but the presence of ¹³C-4,5-Glu or ¹³C-1,2-Glu could not be ruled out with the present data.

The above results are clear evidence of de novo synthesis of Glu from labeled glucose (10 mM), even though unlabeled Gln (2 mM) and Glu (0.13 mM) were both present in the medium. Further examination of the labeling pattern revealed evidence for the route of synthesis for the different Glu isotopomers. Glutamate is derives from 2-oxoglutarate by transamination or reductive amination (Glu dehydrogenase) in mammalian cells. There are two routes of label incorporation from [¹³C-U]-glucose into 2-oxoglutarate via pyruvate (Figures 1, 10). As shown in detail in Figure 10, one route (Fig. 10A) involves the pathway pyruvate → acetylCoA → citrate → 2-oxoglutarate, which leads to the production of ¹³C-4,5-Glu. The other route (Fig. 10B) utilizes the anaplerotic reaction of pyruvate carboxylase to produce OAA directly from pyruvate with the subsequent reaction sequence of citrate → 2-oxoglutarate, which generates ¹³C-2,3-Glu. The presence of ¹³C-2,3-Glu shown in Figures 7 and 9 is consistent with the activity of pyruvate carboxylase. The redistribution of labels into different carbons of Glu is presumably the result of carbon scrambling at the succinate dehydrogenase (SDH) step due to the symmetry of the succinate molecule. Moreover, the presence of multiple ¹³C-Glu isotopomers described above supports multiple turns of the cycle with ¹³C label input (cf. Figure 10).

It is clear from Figure 7 that the Rh30 cells had a much higher buildup of the different ¹³C isotopomers of Glu (than the myocytes. This implicates a higher turnover of the Krebs cycle and specifically a higher capacity for anaplerotic pyruvate carboxylation, as evident by the enhanced buildup of ¹³C-4,5-Glu and ¹³C-2,3-Glu, respectively.

Malate and aspartate also showed substantial ¹³C enrichment (Figure 3B) in both doubly and triply labeled mass isotopomers based on the GC-MS analysis. The positional isotopomer of the triply labeled malate was deduced from the 2-D HSQC analysis, as shown in Figure 7. The doublet of C3 or β-malate and the triplet of C2 or α-malate indicate the prevalence of [¹³C-1,2,3]-malate. The doubly labeled malate isotopomers can be [¹³C-2,3]-malate, [¹³C-1,2]-malate, and/or [¹³C-3,4]-malate, the abundance of which could not be ascertained with the present data. By comparing Rh30 cells with myocytes in terms of the ¹³C-malate isotopomer patterns in Fig. 3B and 7, it is clear that Rh30 cells synthesized a much higher amount of doubly and triply labeled malate, the latter of which was identified as [¹³C-1,2,3]-malate. Similar considerations apply to Asp i.e. the doubly and triply labeled ([¹³C-1,2,3]-Asp) isotopomers were more abundant in Rh30 cells than in myocytes (Figure 3B).

According to the schemes in Figs 1 and 10B, [¹³C-1,2,3]-malate and [¹³C-1,2,3]-Asp can only be synthesized from [U-¹³C]-pyruvate via anaplerotic pyruvate carboxylation (PC). PC can also lead to the synthesis of the doubly labeled [¹³C-2,3]-malate or Asp after one turn of the Krebs cycle, as outlined in Fig. 10B. In contrast, the doubly labeled malate species (as [¹³C-1,2]- and [¹³C-3,4]-malate) or Asp should derive from the normal Krebs' cycle sequence with input of [U-¹³C]-pyruvate, as illustrated in Fig. 10A, or the doubly labeled [¹³C-2,3]-malate from pyruvate carboxylation, as outlined in Fig. 10B. Thus, the combined GC-MS and NMR determination of malate and Asp isotopomers not only revealed an enhanced activity of the Krebs' cycle but also activation of pyruvate carboxylation in the Rh30 cells over the primary myocytes. Such anaplerotic reactions are essential for actively dividing cells, as the production of important metabolites such as pyrimidine nucleobases and reduced glutathione deplete Krebs' cycle intermediates (Fig. 1 and see below), which must be replenished.

Krebs cycle intermediates may also be replenished via glutaminolysis 303848, which has been shown to be activated in glioblastoma cells, based on the production of labeled lactate derived from labeled Gln 38. The role of glutaminolysis in anaplerosis in Rh30 cells could not be ascertained here for lack of carbon tracing from labeled Gln. However, the comparable production of unlabeled lactate by Rh30 cells and myocytes (Figures 1, 4 and see above) suggests a minor role for glutaminolysis in activating anaplerosis in Rh30 cells.

The sustained activity of the Krebs' cycle in the Rh30 cells requires a functional respiratory chain to regenerate NAD⁺ and FAD in the mitochondria. The same chain also reoxidizes cytosolic NADH produced by glycolysis, and transported into the mitochondria via the malate-aspartate shuttle (cf. Fig. 1), that is not involved in reducing pyruvate to lactate (and see above).

Furthermore, transformed Rh30 cells were much more active in de novo GSH synthesis than the primary myocytes. GSH was labeled in the Glu residue in Rh30 cells and the extent of this labeling was much stronger in Rh30 cells than in the myocytes (cf. Figs. 7 and 8), just as the case for its precursor Glu (Table 4). Although the cysteine supply can be rate limiting in GSH synthesis, in this case the medium contained an adequate supply of cystine (0.2 mM). The incorporation of labeled glutamate indicates a substantial rate of de novo GSH synthesis in the Rh30 cells.

Asp is a direct crucial precursor to pyrimidine biosynthesis (Fig. 1). We have previously shown that the pyrimidine rings of nucleotides become ¹³C labeled in the Rh30 cells 35. All possible isotopomers of 5'-UXP C5C6 were observed as shown in the TOCSY spectra (Figure 9B). This ¹³C satellite pattern was similar to that of Asp (Figure 8), which is consistent with Asp being a direct precursor to the de novo synthesis of 5'UXP ring (Figures 1, and 10B). The ¹³C abundance and extent of the labeling in uracil ring of 5'-UXP (Table 3) and its precursor Asp (Figures 3B and 7). was much higher in Rh30 cells than in myocytes. Thus, it is clear that Rh30 cells exhibited a higher capacity for the de novo synthesis of pyrimidine nucleotides than myocytes.

As Asp is derived from the transamination of the Krebs' cycle intermediate, OAA, increased diversion of Asp to pyrimidine ring biosynthesis would ultimately deplete OAA, which must be replenished to sustain the Krebs' cycle activity. As described above, this can occur by enhanced anaplerosis via pyruvate carboxylation in Rh30 cells. Furthermore, labeled Asp synthesized in the mitochondria is presumably transported to the cytoplasm for pyrimidine biosynthesis via the Asp/malate shuttle (Fig. 1). The enhanced rate of pyrimidine synthesis in Rh30 cells may well reflect an increased Asp/malate shuttle activity. This shuttle also facilitates the transfer of NADH produced in the cytoplasm to the mitochondrion for ATP production via oxidative phosphorylation. Although pyrimidine biosynthesis is mainly cytoplasmic, one step occurs within the mitochondrion and involves coupling to the electron transfer chain. Collectively these all point to metabolically functional mitochondria in the cancer cells.

The higher capacity for pyrimidine synthesis in Rh30 cells is corroborated by the¹³C labeling data in the ribose moiety of 5'-UXP, where all five carbons of the ribose residue were uniformly labeled in ¹³C. This was evident from the TOCSY ¹³C satellite patterns of the 5'-UXP cross-peaks (e.g. Fig. 9B for H1' → H2' of 5'-UXP) and ¹³C spin coupling patterns of C1' to C4' of ribose in 5'-UXP (Fig. 8). The abundance of [U-¹³C-ribose]-UXP (Fig. 9B) and % ¹³C enrichment (Table 3) were much higher in Rh30 cells than in myocytes, consistent with an enhanced rate of 5'-UXP synthesis. Also evident in Fig. 8 is the higher rate of synthesis for purine nucleotides (i.e. 5'-AXP and 5'-GXP), as shown by the higher abundance of [U-¹³C-ribose]-5'-AXP and 5'-GXP. This difference in ¹³C labeling of nucleotides is consistent with a higher metabolic demands for macromolecular (e.g. RNA) biosynthesis for the transformed cells compared with slowly dividing primary cells (also see below).

It should also be noted that the ¹H NMR resonances of 5'-AXP and 5'-UXP in Fig. 8 were in fact mixtures of 3 resonances, which correspond to the tri, di and mono phosphates species, with intensities in the order 5'-ATP > 5'-ADP > 5'-AMP and similarly for 5'-UTP, 5'-UDP and 5'-UMP (data not shown).

Additional 2-D ¹H NMR experiments helped identify ¹³C-labeled glucose-6-phosphate (G6P) and/or glycogen in the Rh30 and myocyte extracts. These included ¹H HCCH-TOCSY and ¹H-¹³C HSQC-TOCSY. The HCCH-TOCSY spectrum of the myocyte extract (Figure 11) displayed covalent connectivities of protons attached to consecutively ¹³C-labeled carbons of various metabolites such as lactate, Ala, Glu, β-glucose-6-phosphate (βG6P), glycogen, and AXP. The Rh30 spectrum (not shown) also exhibited similar covalent networks except for the absence of connectivities for glycogen, lower abundance connectivities for βG6P, and more prominent connectivities for 5'-UXP and 5'-GXP. The connectivity patterns for these metabolites is consistent with the presence of [¹³C-1-5]-βG6P and [U-¹³C-glucose]-glycogen, in addition to [U-¹³C]-lactate, [U-¹³C]-Ala, and [¹³C-2,3,4]-Glu.

To confirm further the assignment for G6P (as opposed to glucose), 2-D high-resolution ¹H-¹³C HSQC-TOCSY spectra were analyzed and that for the myocytes is illustrated in Figure 12. In this spectrum, covalent connectivities from ¹³C1 to H1 then relayed to H2 and H3 as well as ¹³C3 to H3 then to H4 of βG6P were clearly observed. Also evident was the set of ¹³C6 to H6 connectivities of βG6P, thus providing unequivocal identification of βG6P in the crude extract of the myocytes. Moreover, the high-resolution spectrum revealed ¹³C coupling patterns of βG6P carbons, i.e. doublet for C1, triplet for C3-5, as well as doublet for C6, which indicate the presence of uniformly labeled βG6P and thus its origin from [U-¹³C]-glucose.

Free glucose is not expected to be present in cells where the activity of hexokinase exceeds that of glucose transport 49. The 1D proton spectra (Fig. 5A) show that the concentrations of glucose and G6P were negligible in the Rh30 cells which indicates that the steps downstream of G6P in glycolysis are not flux limiting. The low concentration of free glucose is a feature of cancer cells which together with the low level of G6P reflects the high glycolytic flux, making glucose transport rate limiting 50, despite derepression of GLUT4 expression by mutated p5312. In contrast, the concentration of G6P in the myocytes was significant (Fig 7A), which suggests that for these cells, the uptake of glucose or hexokinase was not limiting glycolysis in these cells. Rather, a step downstream of G6P, such as PFK1, limits flux leading to an accumulation of G6P. This is consistent with typical estimates of intracellular G6P in skeletal muscle (ca. 200–400 μM)51. The much higher demand from glycolysis and the PPP (see above) in Rh30 cells, compared with the myocytes, may well be limiting glycogen synthesis, which competes for the same glucose precursor. This may underlie the lack of newly synthesized glycogen in Rh30 cells versus its significant presence in the myocytes (cf. Fig. 11).

Finally, the extent of ¹³C labeling in lactate, Ala, and pyrimidine nucleotides was higher in Rh30 cells grown in fetal calf serum (FCS) than cells grown in BSA (Table 3). The low enrichment of the intracellular Ala may be related to differences in pyruvate production from sources other than glucose between the FCS-stimulated and unstimulated Rh30 (BSA) cells (see above). For Glu, the two growth conditions gave a comparable extent of ¹³C labeling, which was significantly higher than that for the myocytes. Thus, even without the influence of the growth factor in FCS, Rh30 cells grown in BSA still exhibited a greater turnover of [U-¹³C]-glucose through glycolysis, Krebs' cycle, and pyrimidine nucleotide biosynthesis than the primary myocytes.

Initial experiments were carried out with unlabeled glucose, followed by full metabolite profile experiments using [U-¹³C]-glucose. In one experiment, both the Rh30 and myocytes were cultured for 24 h in labeled glucose. In a second experiment to account for the differences in cell doubling rates, the Rh30 cells were cultured for 24 h, and the myocytes for 48 h in the labeled glucose. In this work, ¹³C atoms in various metabolites deriving exclusively from [U-¹³C]-glucose were monitored by NMR or mass spectrometry, which represented de novo metabolic transformations originating from glucose. Growth rates and viability were measured by direct cell counting on a hemocytometer using 0.4% Trypan Blue.

After growth for 24 or 48 h the cells were harvested by trypsinization, scraping followed by low spin centrifugation at 4°C. The pellet was washed twice with ice-cold PBS to remove contaminating medium, pelleted again, flash frozen in lN₂ and lyophilized. The dry cell mass was recorded for normalizing metabolite concentrations before extractions twice with ice-cold 10% trichloroacetic acid (TCA), followed by lyophilization as previously described3753. The dry mass of the cells was recorded for normalizing metabolite concentrations. The medium was similarly treated with TCA in the same way for measuring secreted metabolites (predominantly lactate and Ala), and to assay the glucose utilization during the experiment. ¹³C-glucose was quantified at t = 0 and at harvest (t) by¹H NMR using the well-resolved ¹³C1 satellite signals of -α glucose centered at 5.22 ppm. This accounts for 36% of the total glucose. The ¹³C and ¹²C lactate and Ala concentrations were determined by integration of the methyl peak and its satellites and calibrated against the concentration of lactate determined independently by GC-MS. From this, the amount of glucose consumed, ΔGlc, and the fraction converted to lactate plus Ala, F, could be estimated, according to Eq. (1) and Eq. (2) 35:

ΔGlc = n(¹³C-Glc)⁰-n(¹³C-Glc)^t]

F = [n(¹³C-lac) + n(¹³CAla)]/2ΔGlc

n is the number of moles of the compound in parentheses.

The factor of 2 accounted for the fact that one molecule of glucose gives rise to two molecules of lactate or Ala.

1-F then represents the glucose carbon that enters other metabolites and macromolecules in the cell mass or otherwise not accounted for. The fractions were not been corrected for the small (5%) contribution from unlabeled glucose present in the medium.

NMR spectra were recorded at either 14.1 T or 18.8 T on Varian Inova NMR spectrometers at 20°C using a 90° excitation pulse. One-dimensional NMR spectra were recorded with an acquisition time of 2 sec and a relaxation delay of 3 sec. Under these conditions, the protons were essentially full relaxed, as determined by independent measurements of the spin-lattice relaxation time, T₁. For identifying metabolites in the extracts and determining the positional enrichment with ¹³C we used a suite of 2D experiments including TOCSY (or DQ COSY), HSQC, HCCH-TOCSY and HSQC-TOCSY as previously described 35363753. The latter two experiments make use of the isotope editing function. HCCH-TOCSY specifically selects for those molecules in which at least two adjacent carbon atoms are ¹³C whereas HSQC or HSQC-TOCSY detects protons directly attached to ¹³C. The TOCSY experiments were recorded with a spectral width of 6000 Hz in F2, 0.341 s acquisition time in t₂ and 0.05 s in t₁, 1.9 s interpulse delay, 50 ms mixing time, and an 8 kHz B₁ field strength. The HSQC-TOCSY experiments were recorded with an acquisition time of 0.12 s in t₂ and 12 ms in t₁ using a 50 ms proton spin lock at a strength of 8 kHz. For the HCCH-TOCSY experiments, the acquisition times were 0.12 s in t₂, 0.05 s in t₁ using an 8 kHz ¹³C spin lock field centered in the aliphatic region of the spectrum for a duration of 12 ms.

Metabolites were assigned based on their ¹H chemical shift, TOCSY connectivity patterns and correlation with ¹³C, using our in-house database as described previously 3654. All metabolites except choline, sugars and the nucleotides were quantified from GC-MS data 55 whereas choline was quantified from the peak at 3.22 ppm in the 1-D NMR spectra, as described previously 56.

¹³C enrichment in lactate and Ala was determined from 1D ¹H NMR experiments, as the methyl resonances of these metabolites were well resolved. The peaks were integrated, and the areas of the protons attached to ¹²C (central peak) and to ¹³C (satellite peaks) were recorded. The ¹³C content was then calculated as 35:

%¹³C = 100*area(¹³C satellites)/[area(¹²C)+area(¹³Csatellites)]

For other metabolites, which were not resolved in 1D NMR experiments, we volume-integrated cross-peaks in TOCSY spectra after correcting the base-planes. The various isotopomer enrichments were calculated as for the 1D spectra, i.e. the peak volume of the particular ¹³C isotopomer divided by the total volume of the cross peak including the satellites. In the TOCSY experiments, the protons were partially saturated owing to the shorter recycle time (2 s) of these experiments. Thus, the actual peak volumes were corrected according to the differential T₁ values of protons attached to ¹³C or ¹³C; as:

M(true) = M(obs)/[1-exp-t/T₁]

Where M(true) is the corrected area, M(obs) is the observed area, and t is the recycle time. As the relaxation during the spin lock has the opposite effect on peak intensities, this correction was small 35. Effective T₁ values were determined on standards recorded under the same solvent conditions.

Following NMR analysis, an aliquot (50–100 μl) of the NMR sample was re-equilibrated with H₂O and lyophilized to remove any deuteration, then silylated with 25–50 μl 1:1 (v/v) acetonitrile:MTBSTFA (N-methyl-N-[tert-butyldimethylsilyl]trifluoroacetamide, (Regis Chemical, Morton Grove, IL) by 3 h of sonication followed by standing overnight 5758. The solution was directly analyzed on a PolarisQ GC-ion trap MSn (ThermoFinnigan, Austin, TX) using a 0.5 μl injection volume. The column was 0.15 mm i.d. × 50 m fused silica open-tubular, coated with 0.2 μm BPX-5 (5% phenyl-methylsiloxane). The following conditions were employed for GC-MS. Injector at 280°C, column at 60°C for 2 min, followed by a 20°C/min ramp to 150°C, then 6°C/min to 300°C, splitless vent held for 1.5 min, He carrier gas velocity 30 cm/s at 60°C, transfer line = 280°C, electron energy = 70 eV, source heated to 190°C, automatic gain control target value = 50, maximum inject time = 25 ms, He damping gas = 0.3 ml/min, full scan acquisition from 140 to 650 m/z at a rate of five spectra/s which were averaged into one, and mass calibration preformed byperfluorotributylamine.

Metabolites were identified and quantified automatically using Xcalibur software (ThermoFinnigan), based on their GC retention times and mass fragmentation patterns matched against an in-house database and external standards. Identities were also verified by manual inspection. GC-MS quantification of total abundance of metabolites was accomplished by comparing the m/z 147 ion response for each metabolite in the samples with that for the corresponding standard of known concentration. The m/z 147 ion was chosen for quantification of labeled amino acids because this ion fragment did not contain ¹³C as it originated entirely from the silyl moiety. Conversely, for ¹³C isotopomers, the pseudo-molecular ion clusters of authentic standards were used to obtain the empirical ion ratios as the basis of the calculation for label abundance, as described in detail previously 37.