An evolutionary conserved detoxification system for membrane lipid–derived peroxyl radicals in Gram-negative bacteria

How double-membraned Gram-negative bacteria overcome lipid peroxidation is virtually unknown. Bactericidal antibiotics and superoxide ion stress stimulate the transcription of the Burkholderia cenocepacia bcnA gene that encodes a secreted lipocalin. bcnA gene orthologs are conserved in bacteria and generally linked to a conserved upstream gene encoding a cytochrome b561 membrane protein (herein named lcoA, lipocalin-associated cytochrome oxidase gene). Mutants in bcnA, lcoA, and in a gene encoding a conserved cytoplasmic aldehyde reductase (peroxidative stress-associated aldehyde reductase gene, psrA) display enhanced membrane lipid peroxidation. Compared to wild type, the levels of the peroxidation biomarker malondialdehyde (MDA) increase in the mutants upon exposure to sublethal concentrations of the bactericidal antibiotics polymyxin B and norfloxacin. Microscopy with lipid peroxidation–sensitive fluorescent probes shows that lipid peroxyl radicals accumulate at the bacterial cell poles and septum and peroxidation is associated with a redistribution of anionic phospholipids and reduced antimicrobial resistance in the mutants. We conclude that BcnA, LcoA, and PsrA are components of an evolutionary conserved, hitherto unrecognized peroxidation detoxification system that protects the bacterial cell envelope from lipid peroxyl radicals.


Lipidomics:
In their lipidomics experiments, the authors find no significant changes in the amount of PE, but increased levels of CL and PG, as a result of polymyxin B treatment (Fig. 9). This appears partially similar to a previous observation in ref. 24, where the authors claimed larger CL/PE ratios after treatment with ciprofloxacin, kanamycin, and ampicillin in E. coli. However, in that work, PG does not seem to be significantly increased as a result of treatment. Could the authors better compare and contrast their lipidomics results to these and other results present in the literature?
In the study highlighted by the reviewer and referenced in the manuscript, the authors quantified the ratio of CL/PE in bacterial membranes after antibiotics insults using the semi-quantitative thin layer chromatography method. In our study, we used HPLC-MS for lipidomics, which provides better sensitivity, but more importantly allowed us to quantify the changes of each individual lipid species using an internal standard (SPE). Therefore, we could test whether polymyxin treatment had an impact on CL or PE since the increase of CL/PE ratio as shown in ref 24 could be due to either increase in CL or a decrease in PE. Our results (RV, S11 Fig) indicate that polymyxin treatment affected CL but not PE. We also enclose a figure below for the reviewer in which we expressed our results in CL/PE ratios; the data indicate that polymyxin only affected the CL/PE ratios in the WT and possibly in the BcnA but not in the other two mutants. This is consistent with our idea that anionic phospholipids including cardiolipin may be primarily targeted by peroxidation damage over PE.

5.
Line 194: Please name the gene before describing the knockout, or revise the presentation of this paragraph. The indicated knockout strain name was nowhere to be found in Fig. 2! The reviewer was correct; it appeared confusing as originally written. The text has been edited to reflect the new gene name, which has been introduced much earlier. See RV, lines 186-188.

6.
Line 232, the reference should be to Fig. 2B. This is now Fig. 1B; see RV, line 227.

7.
Line 290, since the authors only measure MDA, "byproducts" should not be plural. Corrected; see RV, line 286. Fig. 3, are the MDA levels normalized to bacterial protein concentrations? The methods mention a Bradford assay, but no normalization is mentioned in the figure legend. 9. Fig. S5, presenting the figures normalized to mean untreated intensities as they vary along the cellular axis would make it easier for readers to compare the treatment effects. This is a very good point. We have normalized the fluorescence distribution along the axis of bacterial cells by the average fluorescence of the untreated control. The method has been explained in Material and Methods (RV, lines 737-742) and information was also added to the figure legend (RV, S8 Fig).

8.
10. Fig. 6, what is the rationale for changing the concentration to 75% MIC here? As mentioned above, the authors should perform dose-response experiments to confirm these results across a range of concentrations. 75% MIC was out standard treatment for WT and mutants to allow for growth under antibiotic stress using sublethal amounts of antibiotic according to the MIC of each mutant that was lower than that of WT. 14. The CFU/mL ranges in Figs. 2 and 4 seem somewhat small, as they only span at most one log. These differences may be smaller than what some readers might be used to, for instance when examining treatment of other pathogens like E. coli and S. aureus with antibiotics (where differences in survival are typically at least one, if not multiple, logs). Perhaps the authors could comment on this difference if they feel it could improve the accessibility of their work. B. cenocepacia is known to be susceptible to cold. In general, colonies on LB plates and stored in the fridge become not viable after a week. It is not surprising to us that the recovery is low, but still clear differences are present. We see similar results with bcnA mutants in Klebsiella and Enterobacter cloacae which we have recently constructed, but we these results are part of a different study beyond the scope of this one.

Comments from Academic Editor Bert van den Berg:
We thank Professor van den Berg's comments indicating our work is novel, interesting and opens new avenues of research. Concerning the length of the manuscript for a Discovery Report, our manuscript was intended as a full research article. Upon submission it was decided by the PlosBiology editorial team to proceed to review with the submitted version with the understanding that the manuscript would be consider as a Discovery Report.
Concerning the other questions and comments: 1. Since I'm not very familiar with the topic, I was wondering how important the prevention of lipid peroxidation is under "regular" conditions. It seemed to me that the chemicals and conditions used to induce the effects of BncA etc are somewhat exotic. Can the authors make a reasonable argument that the main role of these proteins is indeed to limit the harmful effects of lipid peroxidation? This is an excellent point. BcnA was discovered by us as a protein of unknow function during a research project investigating bacteria heteroresistance to antibiotics. This is a situation poorly understood mechanistically but commonly associated to high levels of intrinsic resistance whereby a seemingly isogenic bacterial population possesses mix populations of bacteria with different levels of resistance including higher than the MIC of the population (which is only an average value after all). We also found that polyamines are relevant in this context. The absence of reported function for bacterial lipocalins prompted us to investigate more and we realised that these proteins could be involved in providing a means for bacteria to withstand peroxidation. Due to the location of the protein in the periplasm (often secreted too in the case of Burkholderia), the presence of a ubiquinone in other bacterial BcnA homologues that were crystallized, and the association with a membrane cytochrome suggested to us that this set of proteins could form a system to provide a level of protection by detoxification of radicals formed in the outer membrane under extreme stress conditions such as bacterial exposure to sublethal and lethal antibiotic concentrations. We are currently investigating the function of the quinone by a combination of mutagenesis, chemical probes and molecular modelling and starting to characterize the cytochrome in detail. We propose that the quinone in the lipocalin is oxidised by the hydroperoxyl radicals and the cytochrome can reduce it, so the cycle starts again, and the two proteins work together as a peroxyl radicals sink. Any other products, like MDA are membrane permeable and they can be dealt with by aldehyde reductases such as PsrA. We allude to this model the end of our discussion in this manuscript, but we would like to have more solid evidence before we can present it in a more forceful way to the scientific community.

line 83: what is being scavenged by the lipocalins? Do the authors think it is feasible in future work to determine directly (eg by mass spec) what kind of compounds are bound under various conditions?
We think that lipocalins bind ubiquinone and are currently performing experiments, including MS to validate this. Please refer to the Discussion in the RV, lines 471-473 and 487-495.

line 109: "co-regulated". Is this up-or down-regulation or either, depending on conditions?
The genes are upregulated under antibiotic pressure and paraquat (methylviologen), a pesticide that induces endogenous hydroxyl radicals' formation. We have changed the text to "upregulated"; RV, lines 109-110.

line 123: do LcoA/BncA/B orthologs occur in different phyla than the Proteobacteria?
BcnA/B occur in most bacteria including Gram-positives. The main difference is the Gram-positive orthologues lack a signal peptide, suggesting that the proteins function in the bacterial cytoplasm. We have some suggestions based on bioinformatics that in S. aureus, the lipocalin may work in concert with lipid synthesis pathways but have not done any studies towards this end. We mainly analysed Proteobacteria to be keep the focus on the story of the manuscript since a detailed phylogenetic analysis encompassing all phyla would be outside of this work 5. Given that orthologs exist in "model" bacteria, why do the authors perform these experiments in an exotic bug like Burkholderia cenocepacia? As indicated in our response to comment 1, we discovered the system in B. cenocepacia. This microorganism is a model opportunistic pathogen. In enteric bacteria such as E. coli, there are 2 lipocalins: a BcnA homologue and a second one, called Bcl, which have been only partially studied and proposed to be a phospholipid transported but with very little direct evidence of this function. We are extending our research onto enteric bacteria (Klebsiella and Enterobacter) to directly establish the function of the two proteins.
6. line 157: does the DPPP=O fluorescence originate from the IM, OM or both and does this matter? DPPP=O fluorescence originate from any membrane since the probe requires a hydrophobic environment. It would matter for us to eventually define that the peroxidation of the outer membrane lipids drives the system, since we expect that the outer membrane has less compensatory mechanisms to replace damaged lipids than the inner membrane, which is in contact to cytoplasmic enzymes, etc. At present, this is the best we can do with this probe. We would require a probe that differentially integrates in one membrane or the other, which to our knowledge is currently lacking.

line 174: are the cells fine at higher temperatures? What is the relevance of doing the experiments at 8 degrees C
given that orthologs are also present in Enterobacteria that are unlikely to be exposed to such low temperatures?
We explained in the manuscript that peroxidation is known to increase in plants under low temperatures and this process requires lipocalins that are from the same family as BcnA. We have indicated this in the RV (see line 224-225). We expect multiple rounds of peroxidation will produce a range of reactive aldehydes (including malondialdehyde) leading to lipid degradation. If a lipid experiences multiple rounds of oxidation, it is degraded and will not be captured by lipidomics since this solvent extraction methods only captures native intact lipids, such as PG, PE, and CL as we have shown. However, we are exploring way to capture the peroxidised lipids by optimising the extraction protocols 16. line 471: replace :"under no" by "without". done; See RV, line 436.
17. The discussion seems overly long and could probably be condensed.
We have edited the discussion eliminating repetition and reducing the text to about 2/3 of the original version.
We have also introduced minor edits to the original version to improve clarity, reduced repetition and rearranged some sections in view of the new distribution of figures. As a Discovery Report, the main text has only 4 figures and there 11 supplementary figures plus 3 supplementary tables. We do not have Western blots or gels in this work.
Overall, we thank both reviewers and the technical editorial team for their help improving our manuscript and hope it now meets the standard for acceptance to PlosBiology.
Miguel A. Valvano