We thank the authors for their detailed response.

We now report that trehalose and α-glucan mediate distinct abiotic stress responses in Pseudomonas aeruginosa PAO1 [Woodcock 2021]. Biochemical and genetical evidence for all of the enzyme activities of the GlgE pathway in P. aeruginosa PAO1 are consistent with those predicted from the genome. This has revealed the configuration of the trehalose/alpha-glucan metabolic network in this bacterium. Trehalose protects against osmotic stress in 0.85 M NaCl and GlgE-derived alpha-glucan, whether linear or branched, protects against desiccation stress at 75% relative humidity. Thus, osmotic and desiccation stresses are distinct, as are the measures that the bacterium takes to mitigate them.

The configuration of the GlgE pathways of P. aeruginosa [Woodcock 2021] and Mycobacterium tuberculosis [Koliwer-Brandl 2016] are indeed quite different. There is evidence that M. tuberculosis Rv3032 does not contribute to alpha-glucan biosynthesis so this enzyme is presumably only involved in the biosynthesis of MGLPs. In addition, this bacterium is viable when devoid of alpha-glucan. The synthetic lethality of a treS and Rv3032 double mutant is most likely a consequence of the accumulation of toxic metabolic intermediates, analogous to the self-poisoning phenotypes of glgE and otsB2 single mutants [Kalscheuer 2010, Korte 2016].

Now that the metabolic network involving trehalose and alpha-glucan has been elucidated in P. aeruginosa PAO1, it is timely to re-evaluate the work reported by Djonovic ́et al on the very similar P. aeruginosa PA14 strain. Without going through each and every experimental result, the present discussion will focus on the key data that either support or refute key conclusions of the paper, together with some open questions that remain for future study.

1. Is trehalose a virulence factor in P. aeruginosa?

A treS mutant in the context of the newly elucidated metabolic network would be expected to be able to generate trehalose via TreY/TreZ but not produce GlgE-derived alpha-glucan because of the loss of Pep2, which is fused to the TreS protein. The lack of a substantial growth phenotype of a treS P. aeruginosa PA14 mutant (2.3-fold vs a 19.7-fold loss with a glgA mutant) in a plant leaf implies that trehalose is indeed a virulence factor. This evidence therefore does support one of the main conclusions of the paper even though the configuration of the metabolic network is now known to be different to that assumed at the time.

2. Could intermediates of this metabolic network other than trehalose contribute to virulence?

Maltose and maltotriose are almost as effective as trehalose at suppressing the delta42 mutant phenotype. There is a higher amount of maltose as well as trehalose in the treS mutant in the PAO1 strain compared with the wildtype strain [Woodcock 2021]. This implies that maltose and maltotriose (and possibly longer maltooligosaccharides) could also be virulence factors even if trehalose is the primary virulence factor owing to its much higher abundance and slightly greater biological effect.

3. Does TreS support the biosynthesis of GlgE-derived alpha-glucan or trehalose in P. aeruginosa PA14?

The equilibrium of the TreS-catalysed reaction is slightly in favour of trehalose over maltose (between 2:1 and 5:1 depending on the conditions [Miah 2013]. When studied in isolation, TreS therefore generates mostly trehalose at equilibrium – hence its name. The thermodynamics of the GlgE pathway are such that each of the three steps beyond the TreS reaction are thermodynamically very favourable, leading to an alpha-glucan metabolic sink [Syson 2011]. The laws of thermodynamics dictate that flux trough TreS must be from trehalose to maltose in the context of the GlgE pathway, assuming all enzymes of the pathway are active and ATP is available for the second Pep2 maltose kinase step. Indeed, flux through TreS from trehalose to maltose has been shown to occur in mycobacteria [Miah 2013].

Does the GlgE pathway operate in P. aeruginosa as predicted from the operon that codes for all four of the relevant enzymes? There are now several independent biochemical and genetic lines of evidence that the pathway is operational in the PAO1 strain [Woodcock 2021]. Therefore, flux through TreS in the context of the GlgE pathway is from trehalose to maltose in P. aeruginosa PAO1.

Is it possible that the Pep2 activity could be allosterically regulated to allow flux through TreS to be reversed to generate trehalose? Regulation of the individual TreS/Pep2 fusion enzyme activities is possible. However, regulation of an enzyme cannot alone reverse the equilibrium and flux of the reaction it catalyses given the constraints of the laws of thermodynamics. Furthermore, product inhibition is a manifestation of either an inherent chemical equilibrium, which in the case of Pep2 favours the product maltose-1-phosphate, or sometimes an enzyme mechanism. Either way, the impact is relatively small, particularly when the consumption of maltose-1-phosphate by GlgE is also favourable given its chemical equilibrium.

Changing the rate-limiting steps of a pathway can modulate the steady-state concentrations of intermediates. It is therefore feasible that a substantial blockage of Pep2 could allow the steady state concentration of trehalose to increase if MalQ generates maltose faster than Pep2 can consume it. However, the kcat/Km of TreS is several orders of magnitude lower than that of Pep2 [Woodcock 2021]. The implication of this difference in catalytic efficiency is that flux through TreS is rate-limiting, leading to the steady-state accumulation of trehalose generated by TreY/TreZ despite alpha-glucan being a thermodynamic sink. This is indeed observed in both the PAO1 and PA14 strains. Similarly, the maltose supplied by MalQ would be expected to be converted by Pep2 to maltose-1-phosphate rather than to trehalose.

Is such a lack of conversion of maltose to trehalose supported by another experiment? A treZ mutant in the PAO1 strain cannot make trehalose through the TreY/TreZ route. However, it remains capable of generating maltose via MalQ. The prediction is that this maltose would not be converted to trehalose, but yield maltose-1-phosphate and alpha-glucan. Indeed, no trehalose was detected in the treZ mutant of PAO1, maltose-1-phosphate was still produced and more alpha-glucan accumulated than in the wildtype stain [Woodcock 2021]. Crucially, a treZ mutant in the PA14 strain also didn’t produce trehalose. The evidence therefore strongly refutes the hypothesis that flux through TreS is towards trehalose in either P. aeruginosa strain.

The circumstances in which TreS could reverse its flux to generate trehalose from maltose are when there is no ATP available or when the relative catalytic efficiencies of TreS and Pep2 are reversed by shifting one or other by many orders of magnitude. Neither seem likely and, indeed, neither appear to occur.

One can therefore conclude that TreS does not support the biosynthesis of trehalose in P. aeruginosa PA14. Rather it is more likely that it supports the biosynthesis of GlgE-derived alpha-glucan, as is the case in the PAO1 strain.

Quite why trehalose levels decreased by half in the PA14 treS mutant strain but increase 7-fold in the corresponding PAO1 strain remains unclear. Perhaps the more of the carbon flows through to maltose via MalQ in the PA14 treS mutant strain than in the corresponding PAO1 strain.

4. Could GlgE-derived alpha-glucan contribute to virulence?

The loss of the GlgE pathway operon leads to a 5.9-fold reduction in growth of PA14. While this is less severe than the 28.7-fold loss of the glgA operon that would block the formation of both trehalose and alpha-glucan, it shows the loss of GlgE-derived alpha-glucan does seem to have some impact. The contribution of alpha-glucan to virulence could be associated with its role in mitigating desiccation stress [Woodcock 2021].

5. Other questions for the future.

It is curious that trehalose may need to be exported by P. aeruginosa to support virulence. It would be interesting to test whether the wildtype strain exports trehalose in vitro. On a separate but related note, there is evidence that pseudomonads export their alpha-glucan [Quiles 2012, 2014].

If nitrogen were limiting for the detla42 mutant in a leaf, it would be interesting to test whether nitrogen is also limiting for the wildtype strain.

Glucose and sucrose do not exhibit the same suppressive effect of the mutant phenotype as trehalose, maltose or maltotriose. It would therefore appear that the mechanistic basis for these virulence factors does not involve them being either sugars or osmolites. It remains to be seen what the mechanism is.

References

Woodcock 2021 https://journals.plos.org...
Syson 2011 https://pubmed.ncbi.nlm.n...
Kalscheuer 2010 https://pubmed.ncbi.nlm.n...
Koliwer-Brandl 2016 https://pubmed.ncbi.nlm.n...
Korte 2016 https://pubmed.ncbi.nlm.n...
Miah 2013 https://pubmed.ncbi.nlm.n...
Quiles 2012 1 https://pubmed.ncbi.nlm.n...
Quiles 2014 https://pubmed.ncbi.nlm.n...

Stephen Bornemann