Fast exchange fluxes around the pyruvate node: a leaky cell model to explain the gain and loss of unlabelled and labelled metabolites in a tracer experiment

This work demonstrated significant exchange of pyruvate and lactate between extracellular and intracellular pools, which affected the enrichment of pyruvate and downstream metabolites. Intracellular pyruvate enrichment is therefore a function of extracellular lactate and pyruvate enrichment, not just incoming glucose and glutamine. The large extracellular lactate pool appeared to partially explain the dampened dynamics seen in pyruvate enrichment compared to other glycolytic intermediates in a labelled glucose experiment. Although medium refresh is recommended to minimise interference caused by metabolic exchange of metabolites, including lactate [34], the use of serum makes it impossible to completely eliminate these issues. Unless defined medium or dialysed serum is used when performing ¹³C-MFA, flux analysis should account for a reversible lactate exchange. The absence of large initial extracellular lactate will significantly shorten the time required to achieve isotopic steady state for pyruvate and downstream metabolites. The results from the current study indicates that because of an apparent free exchange between pools, when performing ¹³C flux analysis in cultured cells, extracellular enrichment data should be considered, in addition to intracellular data.

There were two main tracer findings in the present work that confirmed a fast lactate exchange flux. Firstly, both extracellular pyruvate and lactate enrichment fractions showed similar trajectories. Although the direct inter-conversion of extracellular pyruvate and lactate could explain this observation, this phenomenon is very unlikely to occur. Intracellular pyruvate and lactate enrichments, which also showed similar trajectories, were key to confirm that the intracellular route was taken to convert lactate to pyruvate and vice-versa. The second evidence was that more labelled glucose did not lead to more labelled pyruvate. Phosphoenolpyruvate, which mirrors glucose enrichment, was essentially feeding into a large pyruvate node that includes both extracellular and intracellular pyruvate and lactate. The rate of intracellular pyruvate enrichment is therefore dependent on the abundance of extracellular lactate and pyruvate. The flux modelling performed is a form of hypothesis testing, whereby a reversible lactate exchange, i.e., a missing reaction [38], must be provided to adequately fit the FM and EM enrichment data.

There is a growing recognition that lactate is both a substrate as well a by-product. Literature has suggested that tumours can utilise lactate [10], and this phenomenon is often referred to by different terms, such as metabolic symbiosis [39], two-compartment nutrient-sharing model [2], reverse Warburg effect [18], and tumour-stroma co-evolution [40, 41]. Cells switched to lactate consumption upon glucose depletion without any appreciable metabolic adjustment [42], and lactate can contribute to the by-products alanine and glutamate in mammary carcinomas even in the presence of glucose [43]. Tumours are known for their ability to scavenge nutrients, such as fatty acids, branched-chain amino acids and acetate [1, 20]; pyruvate serves as an even more versatile energy and biomass precursor. Here, we showed that cells essentially have a large pyruvate pool that extends beyond the cell membrane. Rather than being deliberately up-regulated, the tumour cells’ ability to dynamically access extracellular lactate to refill intracellular pyruvate may a persistent background process.

Glutamine, catabolised via the TCA cycle, has a very diverse contribution to energy, biosynthesis and redox balance [44]. By tracer experiments, the nutrient has been linked to a compensatory role when mitochondrial pyruvate carrier or pyruvate dehydrogenase was suppressed [14, 23, 24], to the increased invasiveness of ovarian cancer [12], and to lipogenesis in hypoxic condition [45]. As it appears that careful maintenance of TCA metabolite pools would be crucial to cell proliferation, the detection of 10–20 % enrichment in TCA cycle intermediates in a labelled glucose experiment—extracellularly—was therefore unexpected. This means that, for the aforementioned labelled glutamine experiments, if labelled metabolites are lost to the medium, then the flux of labelled metabolites will taper off with increasing distance from 2-oxoglutarate due to leakages. The anapleurosis role of glutamine is therefore even more pervasive than the current estimates if compensation for these losses is necessary. However, there may be other substrates that can refill TCA cycle metabolites. For example, in the control C2C12 myoblasts, TCA cycle metabolites succinate, fumarate, malate and 2-oxoglutarate were in total 55 to 70 % labelled by both glucose and glutamine [24], with the balance unaccounted for. We raised the exchange issue regarding TCA cycle metabolites with the HPDE vs. PANC-1 experiment, where enriched TCA cycle metabolites were found not only in the medium but were also differentially enriched between the normal and cancerous cell lines. More data, however, will be required to quantify individual metabolites’ exchange dynamics.

The “leaky” cell model can be used to our advantage to continuously monitor metabolism without harvesting cells, simply by collecting small amounts of culture medium over time. We showed that extracellular pyruvate and lactate closely mirrored intracellular enrichments, albeit the latter displayed a slight lag. Under this framework, TNF-? was used as a model to investigate effects of chronic inflammation on cancer progression and the metabolic reprogramming involved in metastasis. With ¹³C enrichment and abundance of glucose, lactate and pyruvate, PANC-1 cells treated with TNF-? were shown to have increased glycolysis and lactate flux, as well as the conversion of non-glucose substrates to pyruvate represented by ME1 flux. Note that the increase in pyruvate oxidation was inferred by flux balancing (i.e., lactate and pyruvate net production rates were constrained) and therefore will require further validation. In prostate and breast epithelial cells, TNF-? was shown to increase aerobic glycolysis but not the TCA cycle activity [37, 46].

Tracer experiments are very effective grounds for testing metabolic hypothesis, particularly in light of the near-boundless design space and the ever-expanding analytical coverage [47–49]. Parameter identifiability is affected by the input substrate specification, and therefore identifying the range of impinging substrates can broaden our design scope. For example, this study suggests the possibility of using labelled lactate to quantify TCA cycle fluxes even if the cell is actively producing lactate.