Metabolic reprogramming through lysine harvesting strengthens microbial stress defences

Single and multicellular organisms depend on anti-stress mechanisms to deal with sudden changes in the environment. This includes exposure to oxidants, which are required in redox reactions and signalling but cause oxidative stress, manifested by the accumulation of reactive oxygen species (ROS), when their concentration exceeds normal physiological levels.

ROS disrupt many essential cellular processes, and cells react to their build-up by arresting in the cell cycle, inducing antioxidant proteins, and adjusting their metabolism. Metabolism produces both oxidising and reducing metabolites, therefore directly influencing the redox balance, and flux through the metabolic network is intimately linked with the antioxidant machinery: notably, a highly conserved first-line response to oxidative insults is a flux redirection from glycolysis to the NADPH-generating pentose phosphate pathway, to meet the increased demand for NADPH by the antioxidant machinery.

Rather than thinking about cell metabolism as a collection of individual reactions, Markus Ralser and his group seek to understand metabolism as a dynamic, interconnected network of processes that adapts in response to changes and stresses in the environment, and that evolves and functions as an entity. They use yeast as a model system, allowing them to study how complex metabolic processes are controlled, and how they are reconfigured in response to environmental changes.

In 2019, the Ralser lab found a novel metabolic adaptation that protects microbes in stress situations — a process they christened lysine harvesting (Olin-Sandoval et al, 2019). The work has engendered a new concept: that cells not only sense and take up metabolites to enable cell growth, but also — when the metabolic environment is favourable — can pre-emptively reprogramme their metabolism to become more robust in the face of oxidative stress.

The group came to lysine harvesting indirectly, from an earlier discovery that the yeast polyamine transporter TpoIp has a role in the oxidative stress response (Krüger et al, 2013). TpoIp is an exporter protein that removes polyamines from cells, and Ralser’s group showed that cells lacking it are more sensitive to oxidative stress. Why this should be was not immediately clear, so to shed more light on the problem, the group re-analysed an earlier proteomic time-course experiment and looked for differentially expressed proteins. Surprisingly, they found one specific metabolic adaptation: in the absence of TPOI, some of the enzymes of the lysine biosynthetic pathway were upregulated more quickly and in a more pronounced fashion in response to a classic oxidative stress inducer, hydrogen peroxide.

In thinking why Tpo1p might induce such a phenotype, the group realised that there could be a chemical link between the polyamine and lysine biosynthetic pathways: decarboxylation of lysine produces cadaverine (NH₂(CH₂)₅NH₂), a foul-smelling compound which is a close structural analogue of the equally pungent putrescine (NH₂(CH₂)₄NH₂). Putrescine is the canonical precursor of spermine and spermidine, and all three are substrates of Tpo1p, so it was possible that cadaverine might also be a Tpo1p client protein. However, although lysine decarboxylation had been observed in bacteria and plants, S. cerevisiae had no dedicated lysine decarboxylase enzyme, adding to the mystery.

Was something generating cadaverine? The group quantified intracellular levels of cadaverine, and saw only trace amounts if cells were cultured in minimal medium; however, upon supplementation of the medium with lysine, there was a 22-fold increase. Isotope tracing confirmed the result. Clearly, lysine could be decarboxylated, but by what means?

There is a structural similarity between lysine and the putrescine precursor ornithine, so the group tested whether the enzyme ornithine decarboxylase (Spe1p in yeast) was involved. In cells where Spe1p was either deleted or reduced by feedback inhibition, the increase in cadaverine seen upon lysine supplementation was lost, putting Spe1p firmly in the picture.

Determination of the 3-D structure of Spe1p by homology modelling (using Leishmania donovani ornithine decarboxylase as a template) demonstrated that it was theoretically possible to molecularly dock lysine into the catalytic site, although it was a poor fit; it would need to overcome steric hindrance in order to share the catalytic site with the cofactor pyridoxal phosphate, so would likely bind with far lower affinity than ornithine.

The prediction was correct: Spe1p could decarboxylate both lysine and ornithine, but the Michaelis-Menten constant for lysine decarboxylation was in the millimolar range, 100 times higher than for ornithine. This neatly explained why detectable levels of cadaverine were only found when yeast cells were grown in lysine-supplemented medium: with an external source of lysine, there was a nearly 72-fold increase in intracellular lysine, upping its concentration to a level that would allow the low affinity decarboxylation to take place.

This extravagant importation of lysine fitted with a previous observation: yeast amino acid transporters of the APC superfamily are responsible for moving amino acids into and out of the cell, but there is an exception to this two-way traffic: the S cerevisiae L-lysine proton symporter Lyp1p moves lysine into cells, but not out again, allowing lysine to accumulate in submolar concentrations. Such unusually high levels are not reached in yeast by other amino acids. This was a puzzle, as yeast are prototrophic for lysine, and their lysine biosynthetic capacity is already sufficient for exponential cell growth. Ralser and colleagues therefore suggested a new concept: that yeast could take up lysine for purposes other than growth, something they christened “lysine harvesting”.

What effect on metabolic networks would lysine harvesting have? The lysine biosynthetic pathway is prone to feedback inhibition, meaning that harvester cells switch from self-synthesis to consumption. To determine the metabolic changes induced, the group ran a flux balance analysis—a mathematical method for simulating metabolism in genome-scale reconstructions of metabolic networks. This predicted that in the presence of harvested lysine, there would be major changes within NADPH-forming reactions, reflecting the fact that the switch away from lysine biosynthesis removed the need for the NADPH molecules that drive the pathway.

If less NADPH is required for growth, there are two possibilities: firstly, reduced need could mean reduced production, manifested as a reduced flux to NADPH- producing pathways, including the pentose phosphate pathway; secondly, production could remain the same, but excess NADPH could be repurposed.

The first model could be readily tested: knocking out glucose-6-phosphate dehydrogenase (a component of the pentose phosphate pathway) results in cells become auxotrophic for methionine, because the high demand for NADPH in methionine biosynthesis cannot be met. This auxotrophy could be complemented by lysine harvesting, indicating there was more, not less NADPH available in the lysine harvesters, a result confirmed by an increase in lysine harvesters in the redox potential of NADPH upon exposure to hydrogen peroxide.

Increased NADPH levels can stimulate the biosynthesis of the antioxidant glutathione, and further mathematical modelling suggested that a switch to extracellular lysine would result in a seven-fold increase in flux through glutathione oxidoreductase. This was confirmed experimentally: lysine harvesting promoted a 7.85-fold increase in the concentration of free glutathione (GSH) being reconstituted from ‘spent’ glutathione disulphide (GSSG).

Due to this increase in glutathione levels, lysine harvesting rendered yeast cells more resistant to the oxidants diamide and hydrogen peroxide, and was far more effective than methionine supplementation, previously shown to improve the anti-oxidant response.

This novel form of stress protection is conserved across billions of years of evolution and several clades of life: two highly diverged yeast strains, Pichia pastoralis and Candida tropicalis, as well as the bacterium Bacillus subtilis, are also able to harvest lysine and use it to protect themselves from oxidative stress. Mammalian cells, however, are unaffected, as they are already auxotrophic for lysine, meaning they cannot redirect NADPH in the same way.

Lysine harvesting is a metabolic reconfiguration that channels more NADPH into glutathione metabolism, reducing levels of ROS and increasing tolerance to stress. It is one of a number of ways to stimulate NADPH production and prevent imbalances in the redox state under oxidative conditions, but is more efficient than all the previously reported mechanisms: its ability to induce an approximately 8-fold increase in the pool of reduced glutathione — already one of the most concentrated cellular metabolites — is unrivalled. Quantitatively, therefore, lysine harvesting may be one of the most powerful preventative metabolic antioxidant strategies available to microbial cells.

What of the initial observation that drew the group into this work: that loss of the polyamine exporter TpoIp makes cells more sensitive to oxidative stress? It’s not yet obvious how TpoIp, and by extension polyamines, might regulate expression of the lysine biosynthetic enzymes, but ironically, this is almost a side issue. Rather, the oxidative stress response is inhibited in the absence of Tpo1p because cadaverine, the waste product of lysine decarboxylation, cannot be removed from cells; build-up of intracellular cadaverine causes feedback inhibition and consequently far less lysine can be harvested.

The discovery of lysine harvesting is of fundamental interest, demonstrating how seemingly unrelated metabolic pathways can interact, and that a simple process can drive major metabolic changes. It is also important for commercial reasons, as it suggests a straightforward way to increase the efficiency of cell factories—genetically engineered cells cultured on an industrial scale which can produce pharmaceuticals, food, fuels and chemicals.