4-Hydroxy-2-nonenal antimicrobial toxicity is neutralized by an intracellular pathogen

Acceptance summary:

As more is discovered about host-pathogen wars, this study reveals another host defense mechanism that can be neutralized by an intracellular pathogen. In this case, the under-appreciated toxic alpha-beta carbonyl 4-hydroxy-2-nonenal that is produced by ROS-mediated oxidation of host membrane poly-unsaturated fatty acids was found to be neutralized by two genes from Listeria monocytogenes when expressed in an avirulent bacteria.

Decision letter after peer review:

Thank you for submitting your article "4-hydroxy-2-nonenal antimicrobial toxicity is neutralized by an intracellular pathogen" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Wendy Garrett as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Camille Danne (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

As the editors have judged that your manuscript is of interest but, as described below, essential additional experiments are required before it is published, we would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). First, because many researchers have temporarily lost access to the labs, we will give authors as much time as they need to submit revised manuscripts. We are also offering, if you choose, to post the manuscript to bioRxiv (if it is not already there) along with this decision letter and a formal designation that the manuscript is "in revision at eLife". Please let us know if you would like to pursue this option or use these reviews to revise and submit elsewhere.

Summary:

In this work, Tabakh et al. studied the ability of L. monocytogenes to detoxify 4- hydroxy-2-nonenal (4-HNE), a well-studied compound in inflammatory disease but a rather unexplored metabolite during infection. In this study, 4-HNE is shown to accumulate in mouse hepatocytes and spleens during infection by L. monocytogenes and is inhibitory to the growth of Firmicutes bacteria. However, L. monocytogenes is somewhat resistant to this compound as it encodes a detoxification system comprised of rha1 and rha2, which neutralize the ROS activity of 4- HNE via NADPH dependent reduction.

This study is intriguing, mostly well controlled and potentially opens up a new field of 4-HNE metabolism in infection. However, the mechanism by which rha1 and 2 mediate the turnover of 4-HNE needs to be further strengthened wit rigorous biochemistry. In addition, while the Rha proteins appear to affect 4-HNE, some of the claims need to be tempered based on the data presented. The most important being that this study does not convincingly show that 4-HNE production by the host contributes to L. monocytogenes control given that there was no phenotype in immunocompetent mice for the rha double mutant v. WT L. monocytogenes strains.

Essential revisions:

The following experiments are required for acceptance:

1. Rha1/Rha2 role as NADPH dependent oxidoreductases needs to be further established with rigorous biochemical studies. The in vitro 4-HNE L. monocytogenes resistance phenotype of the rha1 and rha2 mutants is somewhat convincing, however the authors should have followed up these results either by measuring 4-HNE levels or adducts in L. monocytogenes or by better delineating an enzyme mechanism in vitro. These data would be more supportive of the model that Rha1/2 breaks down 4-HNE or its derivatives in a physiological setting. It appears Rha1 and especially Rha2 are not very active under the conditions tested, barely showing a curve representative of catalytic activity. Characterization of this enzyme requires changes in assay conditions. and more rigorous analysis.

2. The authors proposed that Rha1/2 could specifically reduce 4-HNE based on similarity to various enone reductases, but 4-HNE does not contain an enone group. Also, although it was shown that other aldehydes did not induce rha1/2 expression, another aldehyde as potential Rha1/2 substrates in this assay could have been tested to support the specificity claim. Also, is it feasible to measure depletion of 4-HNE alongside NADPH oxidation over the course of the assay?

3. Are the rha1/rha2 genes conserved? What is the evolutionary picture for these genes. Some bioinformatics is warranted to understand how prevalent is this protection mechanism. Are there homologs found in B. subtilis? Did you test 4-HNE toxicity on bacterial species from a different phylum, such as Bacteroidetes or Proteobacteria? Another intracellular bacteria?

4. Many genes encoding general protein quality control factors were also strongly induced upon 4-HNE exposure (clpC, groS, dnaK among others). Based on the data presented, it is possible that Rha1 and Rha2 contribute to L. monocytogenes stress responses by another mechanism and not necessarily through the specific detoxification of 4-HNE. Does 4-HNE treatment make bacteria more susceptible to well-characterized proteotoxic stresses like heat shock given that several well known HSP genes were induced? Does 4-HNE lead to accumulation of insoluble proteins? How could you explain the interplay between rha1 and rha2? Do you think they have the same redundant function or rather a complementary function? If redundant functions, how do you explain that the expression of rha2 in B. subtilis does not provide at least partial resistance to 4-HNE?

5. In Figure 1A, a positive control with direct exposure to 4-HNE should be included along with infection to provide comparable levels of adduct formation during other inflammatory diseases.

6. In Figure 4A, a positive control should be included.

7. Figure 2 Show a gene that is not impacted by 4-HNE to highlight the specificity of the response.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "4-hydroxy-2-nonenal antimicrobial toxicity is neutralized by an intracellular pathogen" for further consideration by eLife. Your revised article has been evaluated by Wendy Garrett as the Senior Editor and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Overall, the reviewers think the authors addressed the majority of concerns.

However, the text appears to be overstating results and unless complementation experiments are provided the text must correctly reflect the data. Please revise text and/or data and highlight modified text/data changes.

For Figure 1, the authors are incorrectly interpreting the histopathology; they claim the staining of 4HNE is around bacteria but what I am seeing is tons of 4HNE staining everywhere; The reddish color in the infected tissues clearly shows diffuse staining everywhere. It doesn't take away from their conclusion at all, they just need to reword their text to make them not try to claim the staining is only around the bacteria.

Concerning characterization of bacterial sensitivity to 4-HNE toxicity, authors use log to describe the reduction in bacterial CFU, except for Lmo where they use %. Please use consistent labeling of axis so comparisons can be made.

The abstract should be rewritten as the authors say that Rha1 and Rha2 mutations do not impact Lmo infectious potential, but in the Results section the message is not that clear. The finding are fine but the conclusions are over stated as it might be due to redundancy in stress response proteins. The authors need to rephrase to avoid excessive overstating conclusions.

Finally, the data is not convincing to support the statement that rha genes provide any fitness advantage in vivo to Lmo as all the other animal data are negative. The authors would need a complementation assay for panel F to rule out something else minor like polarity. In the absence of this data, the authors must temper their conclusions based on that figure. The Bacillus result is more robust if they want to claim Rha proteins CAN be protective but there is no convincing data to show they are necessary.

Essential revisions:

The following experiments are required for acceptance:

1. Rha1/Rha2 role as NADPH dependent oxidoreductases needs to be further established with rigorous biochemical studies. The in vitro 4-HNE L. monocytogenes resistance phenotype of the rha1 and rha2 mutants is somewhat convincing, however the authors should have followed up these results either by measuring 4-HNE levels or adducts in L. monocytogenes or by better delineating an enzyme mechanism in vitro. These data would be more supportive of the model that Rha1/2 breaks down 4-HNE or its derivatives in a physiological setting. It appears Rha1 and especially Rha2 are not very active under the conditions tested, barely showing a curve representative of catalytic activity. Characterization of this enzyme requires changes in assay conditions. and more rigorous analysis.

“…however the authors should have followed up these results either by measuring 4-HNE levels or adducts in L. monocytogenes…”

We thank the reviewers for their suggestion. We have now added new data in Figure 4D to the manuscript showing a ~50% increase in 4-HNE adducts in L. monocytogenes∆∆rha1/2 mutant compared to WT upon incubation with 640µM 4-HNE by dot blot and that complementation with either gene had modest effects. Given the nominal magnitude of adduct accumulation it is difficult to draw much from this data. Because expression of these genes in B. subtilis had much larger impacts on adduct accumulation, these findings suggest that other mechanisms for clearing 4-HNE adducted proteins may be present in Listeria, which may mask the effects of Rha1/2. This is also consistent with the significant level of resistance that Listeria has over B. subtilis even when the latter organism expresses Rha1/2. Clearly Rha1/2 are only the beginning of the story for 4-HNE resistance in Listeria and future studies are warranted to explore this resistance in further detail.

“…by better delineating an enzyme mechanism in vitro.”

We agreed with the reviewers about our initial enzymatic data. As such, we assessed purified enzyme turnover in numerous conditions, including in bacterial lysates, in sterile bacterial media, with the addition of various salts, metals, crowding agents, and different buffers. From this comprehensive screening of in vitro biochemical conditions, we found that the addition of 20% PEG 8000 to the reactions significantly decreased the Km of both Rha1 and Rha2 to within the broad physiological range of 4-HNE (Rha1 134 µM, Rha2 367 µM). We have added these data as new Figure 5B. We suspect PEG-induced crowding is decreasing the Km of both the enzymes, either by shifting the protein to its predicted active dimer form (Rha1) or perhaps by generally reducing protein instability and therefore increasing the potential for successful 4-HNE-protein association (Ma and Nussinov, 2013). To provide a comparison to a known 4-HNE metabolizing enzyme, we tested turnover of the known human 4-HNE NADPH-dependent detoxification enzyme AKR1C1 under the same conditions. While AKR1C1 exhibits a lower Km, the turnover rates of Rha1/2 (kcat) are higher than the mammalian counterpart under these conditions. These data are now presented as new Figure 5C.

2. The authors proposed that Rha1/2 could specifically reduce 4-HNE based on similarity to various enone reductases, but 4-HNE does not contain an enone group.

Although the reviewers are correct in noting that 4-HNE does not contain an enone group, it does contain an enal group, which is often reduced by a similar enzymatic mechanism to enone reduction. Ene-reductases specifically have reduction activity against carbon-carbon double bonds both in enal and enone contexts and we have tried to clarify this in the text (Toogood and Scrutton, 2018).

Also, although it was shown that other aldehydes did not induce rha1/2 expression, another aldehyde as potential Rha1/2 substrates in this assay could have been tested to support the specificity claim.

As suggested by the reviewers, we expanded the panel of aldehydes we tested as potential substrates for both Rha1 and Rha2. Our findings largely mirrored observations of rha1 and rha2 gene induction by this same panel (Figure 3B), in which activity was not observed for any combination with one exception, Rha2 exhibited some turnover capacity with the lipid oxidation product of n-3 fatty acids 4-hydroxy 2hexenal. These observations are in line with our assignment of Rha1/2 as reductases of the αβ-double bond of 4-HNE, which is absent in other aldehydes tested in this panel. Collectively these observations are consistent with these genes functioning in response to enal induced stress and toxicity and that perhaps Rha2 is more promiscuous than Rha1. Of course, the conclusions of specificity for 4-HNE are limited by the substrates tested and we have toned down the language relating to specificity. In fact, we hypothesize that these enzymes may indeed have other αβ-carbonyl containing substrates, either endogenous (i.e. quinones) or other αβ-unsaturated compounds that induce electrophilic stress. These data are now presented as new Figure 5D.

Also, is it feasible to measure depletion of 4-HNE alongside NADPH oxidation over the course of the assay?

We successfully visualized Rha1 and Rha2 enzymatic conversion of 4-HNE using thin-layer chromatography with phosphomolybdic acid visualization. These data are now presented in new Figure 5E. As a control for direct comparison on the TLC plate, we included AKR1C1, known to generate the alcohol 1,4-dihydroxynonene (1,4-DHN) and P1-ZCr, an Arabidopsis thaliana enzyme known to perform a hydrogenation reaction to saturate 4-HNE to 4-hydroxynananal (4-HNA). We found 4-HNE was converted to 1,4-DHN by AKR1C1 as reported, to 4-HNA by P1-ZCr as reported and both Rha1 and Rha2 enzymes converted 4-HNE to 4-HNA. These studies were not done in parallel with NADPH consumption due to technical limitations. Specifically, the amount of enzyme and NADPH in the experiments for monitoring NADPH consumption is not sufficient to detect the 4-HNE and reaction product 4-HNA using TLC. As such, the two experiments were performed with different amounts of enzyme and NADPH. However, we found no 4-HNE conversion with any enzymes in the absence of NADPH. Overall, we were able to successfully demonstrate NADPH-dependent 4-HNE reduction to 4HNA by both Rha1 and Rha2.

3. Are the rha1/rha2 genes conserved? What is the evolutionary picture for these genes. Some bioinformatics is warranted to understand how prevalent is this protection mechanism.

We performed sequence homology analysis for both rha1 and rha2 across the set of bacteria we tested for 4-HNE sensitivity to obtain a sense of the distribution of both genes across a variety of prokaryotes. Author response table 1 reports the sequence identity of the closest matching Rha1 and Rha2-like proteins from tested organisms. E. faecalis, which we found to be both the second most 4-HNE tolerant bacteria we tested, had the closest homologs to both Rha1 and Rha2. Among other organisms, we found that the association between 4-HNE survival and closeness or presence of protein homologs was weak. This is not surprising, as functional homology and sequence homology are often not very strongly correlated among oxidoreductase enzymes (Todd, Orengo and Thornton, 2001). This is due to the fact that very small changes in the binding pocket completely alters substrate specificity and it’s very difficult, if not impossible, to predict with any confidence the substrate of many oxidoreductase enzymes from simple sequence homology. We speculate that some close relatives of L. monocytogenes, including perhaps E. faecalis, may utilize similar mechanisms of 4-HNE detoxification and that other bacteria that are distantly related, like P. aeruginosa likely have other enzymes with limited sequence homology to Rha1/2 that are capable of performing similar functions. Rigorous transcriptional, biochemical and other experimental analysis will be required to say with any confidence if other bacteria code for enzymes with similar 4-HNE detoxification roles.

Are there homologs found in B. subtilis? Did you test 4-HNE toxicity on bacterial species from a different phylum, such as Bacteroidetes or Proteobacteria? Another intracellular bacteria?

We thank the reviewer for their suggestions. To address this point, we expanded our killing assay to include a set of Gram-negative bacteria: E. coli DH10b (K12), Pseudomonas aeruginosa PA01 and Francisella novicida U112, which is also an intracellular pathogen. These data are presented as new Figure 2A. We found that even among this expanded bacterial cohort, L. monocytogenes still showed the highest survival capabilities. Interestingly, F. novicida was uniquely sensitive to 4-HNE toxicity. Reports that F. novicida actively blocks ROS generating pathways during infection (Mohapatra et al., 2010) suggests that this organism may avoid 4-HNE mediated toxicity by halting its production rather than specific detoxification or resistance mechanisms. We have added discussion of these points to the Results and Discussion sections.

4. Many genes encoding general protein quality control factors were also strongly induced upon 4-HNE exposure (clpC, groS, dnaK among others). Based on the data presented, it is possible that Rha1 and Rha2 contribute to L. monocytogenes stress responses by another mechanism and not necessarily through the specific detoxification of 4-HNE.

Induction of heat shock proteins in response to electrophilic stress has been previously reported in B. subtilis (Huyen et al., 2009). We agree with the reviewers that while it is still formally possible that Rha1/2 contribute to the Listeria stress response through an indirect pathway to 4-HNE resistance rather than direct detoxification, our improved in vitro enzymology (new Figure 5) suggests the most parsimonious explanation is that they play a role in directly de-toxifying 4-HNE. However, to provide a more comprehensive picture of Rha1 and Rha2 expression in bacterial stress responses, we measured rha1/2 expression in response to a panel of non-aldehyde stressors, including heat, diamide induced disulfide stress, and the RNS agent nitric oxide. We found that NO was unable to induce either gene, but that heat did induce both genes by approximately 30-fold, and the disulfide stress agent diamide induced rha1 by 11-fold and rha2 by 100-fold. However, the most robust induction of both rha1 and rha2 was with 4-HNE, as we observed previously with our aldehyde panel in Figure 3B. Induction by heat and diamide suggests that rha1/2 are induced during other stress responses in L. monocytogenes. Despite their induction under both heat and diamide induced thiol stress, loss of each gene had no discernible impact on bacterial survival in these conditions. These data are now presented in Figure 3C and Figure 4B,C.

Does 4-HNE treatment make bacteria more susceptible to well-characterized proteotoxic stresses like heat shock given that several well known HSP genes were induced?

We found that indeed, 4-HNE treatment prior to heat shock (50°C for 10 minutes) does increase bacterial death. This is consistent with 4-HNE impacting proteotoxic stress responses. This data is now presented in Figure 2E. Given that many of the proteases involved in the heat shock require active site nucleophiles that are likely highly susceptible to 4-HNE adduction, it is feasible that 4-HNE poisons these proteins from functioning to clear damaged proteins during elevated temperature. While these observations are intriguing, detailing the mechanism of this synergy is beyond the scope of this manuscript.

Does 4-HNE lead to accumulation of insoluble proteins?

We found that in WT L. monocytogenes 4-HNE does not lead to significant total protein aggregation compared to untreated control, especially compared to the positive control of elevated temperature exposure. These data are now presented in Figure 4E. In addition, the ∆∆rha1/2 mutant does appear to have only a modest (10-15%) and not statistically significant increase in protein aggregation accumulation with 4-HNE treatment compared to WT. This suggests that the difference in WT versus ∆∆rha1/2 survival in 4-HNE is not driven by global adduct accumulation or protein aggregation but rather other molecular targets susceptible to 4-HNE reactivity. Such targets may be other macromolecules (i.e. lipids, nucleic acids, etc) or adduction to specific cellular proteins, namely inactivation of specific essential proteins that are not reflected in the global analyses conducted here.

How could you explain the interplay between rha1 and rha2? Do you think they have the same redundant function or rather a complementary function? If redundant functions, how do you explain that the expression of rha2 in B. subtilis does not provide at least partial resistance to 4-HNE?

Although we do not know exactly what the relationship between the two proteins is, we suspect that the two enzymes serve complimentary functions. Both enzymes reduce 4-HNE to 4-HNA (new Figure 5E) however, they appear to function non-redundantly, with both enzymes being expressed heterologously in B. subtilis allowing for greater than simply additive survival for the expressing B. subtilis (Supplemental Figure 3B, Figure 6A). The enzymes have different Km affinities for 4-HNE (Figure 5)

as well as different 4-HNE turnover specificity. Rha1 is seemingly specific for 4-HNE as a substrate for NADPH oxidation while Rha2 is able to perform the reduction of both 4-HNE and 4-HHE (Figure 5D). The differential induction of rha1 versus rha2 with our aldehyde and stress panel (Figure 3B-C; see our response to point 4 above) suggests that the two proteins may be regulated by different inputs or play specific roles during 4-HNE exposure possibly through distinct protein localization within the cell.

5. In Figure 1A, a positive control with direct exposure to 4-HNE should be included along with infection to provide comparable levels of adduct formation during other inflammatory diseases.

We repeated the TIB73 L. monocytogenes infection experiments with the included positive control of TIB73s that were treated for 10 minutes with 10µM 4-HNE in PBS. These data are presented in new Figure 1A. We found that the treated cells accumulate adducts at a similar level to 6 hours of L. monocytogenes infection. Although that suggests overall low levels of 4-HNE, we believe that this is due to the combination of (1) a small amount of cells infected with L. monocytogenes at 6 hours post infection and (2) the highly segregated nature of 4-HNE accumulation. We believe that 10µM overall 4HNE accumulation masks the much higher localized levels of 4-HNE in cells. Additionally, measurement of 4-HNE levels in cells actively producing the metabolite, as during infection, reflects the amount of protein conjugates that are accumulating and being degraded.

6. In Figure 4A, a positive control should be included.

We included the positive control of the human aldo-keto reductase AKR1C1 that is well-known to perform NADPH-dependent oxidation of 4-HNE. This data is now in the new Figure 5C.

7. Figure 2 Show a gene that is not impacted by 4-HNE to highlight the specificity of the response.

We included the gene rplD whose expression we found to not be impacted by 4-HNE exposure. This data is now in new Figure 2D.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Overall, the reviewers think the authors addressed the majority of concerns.

However, the text appears to be overstating results and unless complementation experiments are provided the text must correctly reflect the data. Please revise text and/or data and highlight modified text/data changes.

For Figure 1, the authors are incorrectly interpreting the histopathology; they claim the staining of 4HNE is around bacteria but what I am seeing is tons of 4HNE staining everywhere; The reddish color in the infected tissues clearly shows diffuse staining everywhere. It doesn't take away from their conclusion at all, they just need to reword their text to make them not try to claim the staining is only around the bacteria.

We completely agree that there is an increase in 4-HNE staining throughout the spleen following L. monocytogenes infection as indicated by the overall reddish color between the infected and uninfected tissue sections shown in Figure 1B and C, respectively. With increased magnification as shown in Figure 1E, there are clearly regions of the tissue section that stand out as having a much darker staining than the diffuse background seen in much of the tissue section. These cells have such a strong signal for the 4HNE that they are very dark brown. We have added arrows to Figure E to highlight these regions. In comparison to the L. monocytogenes imaging in Figure 1D at the same magnification, the staining pattern looks similar. At 100X magnification, the 4-HNE labeled cells exhibiting the darkest staining have a punctate pattern (Figure 1G) that looks strikingly similar to the L. monocytogenes staining (Figure 1F). We completely agree that this comparison of staining is clearly not sufficient evidence to conclude that the bacteria are being directly labeled by the metabolite. It is equally possible that there is a subset of spleen cells that simply produce significantly elevated levels of 4-HNE. As such, we have removed the statement that this increased staining is around the bacteria (Lines 106-109) and added arrows to Figure 1E to further direct the reader’s attention to the enhanced staining in specific regions of the tissue.

Concerning characterization of bacterial sensitivity to 4-HNE toxicity, authors use log to describe the reduction in bacterial CFU, except for Lmo where they use %. Please use consistent labeling of axis so comparisons can be made.

We have changed the wording on Lines 130-131 to use log reduction to describe L. monocytogenes viability following 4-HNE exposure.

The abstract should be rewritten as the authors say that Rha1 and Rha2 mutations do not impact Lmo infectious potential, but in the Results section the message is not that clear. The finding are fine but the conclusions are over stated as it might be due to redundancy in stress response proteins. The authors need to rephrase to avoid excessive overstating conclusions.

We have included a statement highlighting that Rha1 and Rha2 are not necessary for L. monocytogenes infectious potential (Lines 22-23).

Finally, the data is not convincing to support the statement that rha genes provide any fitness advantage in vivo to Lmo as all the other animal data are negative. The authors would need a complementation assay for panel F to rule out something else minor like polarity. In the absence of this data, the authors must temper their conclusions based on that figure. The Bacillus result is more robust if they want to claim Rha proteins CAN be protective but there is no convincing data to show they are necessary.

We agree that the reduction in CFU following macrophage infection is not sufficient to make any claim about in vivo fitness associated with these genes in L. monocytogenes. We have made changes to the text to reflect this (Lines 240-244).