Pyruvate carboxylase activity and pyruvate carboxylation
Pyruvate carboxylase is an astrocyte-specific enzyme (Yu et al. 1983). When primary astrocytes are incubated with [1–13C]glucose, its activity is evidenced by different enrichments at glutamine C2 and C3 (Martin et al. 1993). This asymmetrical labelling demonstrates that the pyruvate carboxylation product ([3–13C]oxaloacetate) does not equilibrate with fumarate after its entry into the TCA cycle. A 39% partial cycling between oxaloacetate and fumarate was deduced from glutamine isotopomer analysis (Merle et al. 1996a,b).
Pyruvate carboxylation has also been found to occur in primary neurons by evaluating carbon enrichments in amino acids or by analyzing glutamate isotopomer after incubating cerebellar granule cells with either [2 or 3–13C]pyruvate (Brand et al. 1992) or [1–13C]glucose (Martin et al. 1993), respectively. Glutamate labelling was characterized by equivalent enrichments at C2 and C3, which thus implied complete equilibration of the pyruvate carboxylation product with fumarate (Martin et al. 1993).
Pyruvate carboxylation was more recently reported to occur in neurons both in vitro and in vivo and was proposed to be mediated by malic enzyme (ME) (Hassel and Brathe 2000a,b). The occurrence of ME in neurons is well documented (Cruz et al. 1998), particularly that of the mitochondrial enzyme (Vogel et al. 1998) which presents a high activity in synaptic terminals (McKenna et al. 2000), and it may have a cataplerotic function. ME could make TCA cycle intermediates available (Bukato et al. 1995) and thereby could be involved in the neuronal pyruvate recycling pathway described by Cerdan et al. (1990) and could contribute to the regeneration of NADPH required for mitochondrial glutathione disulphide reduction (Vogel et al. 1999). However, the pyruvate carboxylation carbon flux in primary neurons was found to correspond to 23–33% of the flux through PDH (Brand et al. 1992; Merle et al. 1996b; Hassel and Brathe 2000a), thus demonstrating the intrinsic anaplerotic capacity of ME in immature neurons. In the adult brain, although the reversibility of the ME reaction makes neuronal pyruvate carboxylation possible according to the metabolic needs (as discussed by Hassel 2001), the anaplerotic role of the enzyme remains controversial.
In the light of the above considerations, the methods used to determine PC/PDH in brain astrocytes need re-evaluation. Since pyruvate carboxylation in primary neurons from a labelled substrate generates glutamate equally labelled at C2 and C3 (Martin et al. 1993; Zwingmann et al. 2000a), a difference in brain glutamine (or glutamate) C2 and C3 (or GABA C4 and C3) enrichments evidences glial PC activity. Therefore, the method for evaluating PC/PDH based on the difference in carbon enrichment seems relevant. However, owing to a possible cycling of oxaloacetate to fumarate, the method may under evaluate the contribution of PC. In comparison, the method used by Lapidot and Gopher (1994) based on amino acid isotopomer analysis after metabolism of [U-13C]glucose appears unable to discriminate between neuronal and glial pyruvate carboxylation. For PC/PDH evaluation, they considered as equivalent the glutamate (or glutamine) isotopomers labelled at C2 and C3 and those labelled at C1, C2 and C3, whereas the occurrence of the latter implied fumarase activity. In their metabolic model, they assumed total cycling of oxaloacetate to fumarate. As a consequence, the contribution of PC deduced from their study might be overestimated. The same remark holds for the work of Künnecke et al. (1993) who used [1,2–13C2]glucose, a precursor which cannot generate per se asymmetrical labelling of oxaloacetate C2 and C3 after pyruvate carboxylation (in fact, the authors assumed randomization of the oxaloacetate label in the malate-fumarate equilibrium).
The experimental discrimination between neuronal (via ME) and astrocytic (via PC) pyruvate carboxylation in brain appears to be an interesting challenge. If oxaloacetate and malate, the two products of pyruvate carboxylation, are really characterized by different yields of equilibration with fumarate, isotopomer analysis after [U–13C]glucose metabolism (as in Lapidot and Gopher 1994) might provide useful information. Indeed, if we assume that oxaloacetate does not equilibrate with fumarate whereas malate does so fully, then glutamate and glutamine isotopomers labelled at C1, C2 and C3 would be specific of ME activity. According to the data of Lapidot and Gopher (1994), it appears that in the rabbit brain and near steady state, 90% of the glutamate and glutamine label which entered through pyruvate carboxylation was randomized, thereby suggesting a neuronal metabolism. The result would probably be different in rats or mice because at steady state, glutamine C2 and C3 labelling from [1–13C]glucose metabolism are far from the same.
The issues of the occurrence of neuronal anaplerosis and the discrimination between neuronal and glial pyruvate carboxylation seem of importance for the general understanding of brain metabolism. For example, in two recent studies that aimed at providing an overview of brain metabolism using mathematical modelling, neuronal pyruvate carboxylation was not considered. Sibson et al. (2001) used [2–13C]glucose as substrate to determine the rate of different brain metabolic pathways. Although they considered the possibility of oxaloacetate C2 and C3 label equilibration in their model, they assumed that glutamate and glutamine C2 and C3 labelling from [2–13C]glucose specifically resulted from glial PC activity, i.e. the contribution of any neuronal pyruvate carboxylation to amino acid labelling was thus ignored. Under this assumption, the contribution of astrocytic anaplerosis to glutamine synthesis might be overestimated. On the contrary, Gruetter et al. (2001) used a model to analyse [1–13C]glucose metabolism in human brain wherein reverse flux from oxaloacetate to fumarate was neglected. Evaluation of the pyruvate carboxylation flux was thus based on the difference in glutamine C2 and C3 enrichments. Therefore, this flux might be underestimated. Notwithstanding the important intrinsic differences in modelling, flux values estimated from these two studies were unexpectedly broadly similar. These two studies were based mainly on NMR data acquired in vivo. Considering the importance of the early labelling phase as shown in the present study, the signal to noise ratio and time resolution in data from in vivo experiments appear to be crucial parameters for obtaining relevant data.
Time dependence of PC/PDH
Under the present experimental conditions, PC/PDH was around 80% after 10 min [1–13C]glucose infusion and around 25–30% during the 20–60-min period, as deduced from glutamine labelling. In contrast, no significant contribution of PC to glutamate and GABA labelling was evidenced. From a general point of view, these results are in agreement with previous findings of a much higher contribution of PC to glutamine metabolism than to that of other amino acids (Shank et al. 1993; Hassel et al. 1995; Aureli et al. 1997). However, a time-dependent decrease of PC contribution to glutamine labelling was already noted by Hassel et al. (1995) and the same phenomenon could be deduced from the data of Aureli et al. (1997). This decrease was interpreted as reflecting the scrambling of the label between the oxaloacetate C2 and C3 positions due to equilibration of oxaloacetate with the symmetrical fumarate (Hassel et al. 1995). More likely, this dependence can be due to the different labelling kinetics of the two glutamine precursor glutamate types, i.e. the glial and the neuronal glutamate. The glial glutamate is in direct connection with the astrocytic TCA cycle and then rapidly incorporates the label entered in that cycle via PC and PDH. The neuronal glutamate corresponds to the transmitter taken up by the astrocytes after its release from the neuronal vesicles, so it is not directly connected to the neuronal TCA cycle. With time, labelling of the small astrocytic glutamate pool is masked by labelling of the large neuronal glutamate pool, and glutamine labelling from its astrocytic precursor is obscured by the occurrence of the glutamate-glutamine cycle. An estimate of the relative flux on the two pathways of glutamine labelling can be obtained from the time evolution of the PC/PDH value. Indeed, assuming that the value at 10 min (84%) results only from the astrocytic TCA cycle activity and that the value at 60 min (28%) results from the steady state where both astrocytic and neuronal glutamate contribute to glutamine labelling, the relative contributions of the astrocytic TCA cycle and the glutamate-glutamine cycle to glutamine labelling might be about 1/3 and 2/3, respectively.
As discussed above, the 80% contribution of PC to glutamine enrichment in the early labelling phase represents a minimal evaluation if the possibility of partial cycling of oxaloacetate to fumarate is considered. Moreover, the masking effect due to the contribution of neuronal glutamate to glutamine labelling might not be negligible even after 10 min Since the contribution of substrates other than pyruvate to acetyl-CoA synthesis was rather low (from the glutamine enrichment pattern, acetyl-CoA C2 enrichment was in the range 22–24% while glucose C1 enrichment was around 46%), this means that most, if not all, of the glial TCA cycle activity was devoted to anaplerosis.
Most of the net glutamine synthesis is assumed to be related to brain ammonia detoxification (Sibson et al. 2001), which results in a net flux of brain glutamine exportation to blood. However, the release of glutamine is associated with an entry of blood glutamine into the brain, which leads to different labelling of Glu and Gln C4 at steady state. From the data in Table 2, it may be estimated that after 60 min infusion, blood glutamine represents at least 30% of the brain glutamine (a value of 26% was reported in human by Shen et al. 1999). In view of the various glutamine sources from the whole body, labelling of blood glutamine is likely to be slow. For example, blood glutamine C4 enrichment was only 7% after 1 h [1–13C]glucose infusion in glioma-bearing rats (Bouzier et al. 1999). Therefore, in the presteady-state phase of brain glutamine labelling, the consequence of the blood glutamine influx is a simple isotopic dilution not affecting the evaluation of PC/PDH. However at steady state, the blood glutamine enrichment pattern (which may differ from that of brain-synthesized glutamine) may interfere with that of brain glutamine, depending on the origin (kidney, liver…) of the labelled glutamine in the blood.
If the anaplerotic flux through the astrocytic TCA cycle is considered to represent the net glutamine synthesis, then this could lead to an overestimation of the latter, at least under normoammonemic conditions. Indeed, if the enrichment pattern of 2-oxoglutarate is transferred to astrocytic glutamate (by isotopic exchange) and then to glutamine, the net glutamine synthesis may be lower than pyruvate carboxylation, depending on the anaplerotic needs of the cells. For example, in primary astrocytes, anaplerosis was found to be much more due to the synthesis of citrate than to that of glutamine (Martin et al. 1997). Moreover, the anaplerotic flux may be partly required for the replenishment of neurotransmitter amino acid pools if some of them re-enter the TCA cycle (e.g. to provide the molecules for pyruvate recycling).
Glutamine release from brain is particularly necessary under hyperammonemic conditions to avoid the occurrence of a brain oedema generated by an excess of glutamine acting as an osmoregulator (Olafsson et al. 1995; Zwingmann et al. 2000b). Under such conditions, astrocytic anaplerosis could be more directly associated with ammonia detoxification via glutamine synthesis and efflux from the brain. A 1.5-fold increase in PC/PDH activity under hyperammonemic conditions was reported by Gopher and Lapidot (1991). In their experiments, PC/PDH was evaluated from glutamine labelling 40 min after [U–13C]glucose infusion. Considering the already high PC/PDH value determined in the present study under normoammonemic conditions, it may be proposed that the increase in PC/PDH reflects both an intrinsic increase in the ratio (i.e. PC versus PDH activity) and an increase in the astrocytic TCA cycle flux, the consequence of the latter being an increase in the contribution of the astrocytic glutamate to glutamine synthesis, compared with the neuronal glutamate. This assumption is strengthened by the results of Zwingmann et al. (1998) who found both an increase in PC/PDH and an increase in glutamate and glutamine labelling in primary astrocytes incubated with [1–13C]glucose under hyperammonemia.
Although the PC catalyzed reaction is acknowledged as the main anaplerotic pathway in the brain, the occurrence of cytosolic ME in astrocytes (Kurz et al. 1993) raises the question of its contribution to anaplerosis, as discussed by Vogel et al. (1999) and Zwingmann et al. (2000a). However, in the present study, the contribution of this pathway cannot be evaluated if the malate generated by the cytosolic ME equilibrates with fumarate after its entry into mitochondria and TCA cycle.
In conclusion, this study emphasizes the importance of the presteady-state phase of amino acid labelling in the investigation of brain metabolism using labelled substrates. Under our experimental conditions, astrocytic TCA cycle activity and anaplerosis were found to be in the same range, thereby indicating that TCA cycle function is more closely related to anaplerotic than to energy requirements.