TCA cycle tailoring facilitates optimal growth of proton-pumping NADH dehydrogenase-dependent Escherichia coli

ABSTRACT The bacterial lifestyle is plastic, requiring transcriptional, translational, and metabolic tailoring for survival. These dynamic cellular processes are energy intensive; therefore, flexible energetics is requisite for adaptive plasticity. An intricate network of complementary and supplementary pathways exists in bacterial energy metabolism. There are two main entry points for electrons in the aerobic electron transport system, NADH dehydrogenase (NDH) and succinate dehydrogenase (SDH), receiving electrons from NADH and succinate, respectively. Aerobic bacterial phyla have a non-proton-pumping NADH dehydrogenase, which is often the primary dehydrogenase under aerobiosis. Here, we report adaptive changes supporting growth restoration in an Escherichia coli strain lacking the primary dehydrogenase. Growth optimization is achieved by reducing the activity of succinate dehydrogenase, and thus we demonstrate a physiological discord between proton-pumping NADH dehydrogenase and succinate dehydrogenase in supporting growth. Beyond the fundamental understanding of the bioenergetic network, identifying this compensatory feature provides impetus to rational antimicrobial combinations for targeting the non-proton-pumping dehydrogenase. IMPORTANCE Energy generation pathways are a potential avenue for the development of novel antibiotics. However, bacteria possess remarkable resilience due to the compensatory pathways, which presents a challenge in this direction. NADH, the primary reducing equivalent, can transfer electrons to two distinct types of NADH dehydrogenases. Type I NADH dehydrogenase is an enzyme complex comprising multiple subunits and can generate proton motive force (PMF). Type II NADH dehydrogenase does not pump protons but plays a crucial role in maintaining the turnover of NAD+. To study the adaptive rewiring of energy metabolism, we evolved an Escherichia coli mutant lacking type II NADH dehydrogenase. We discovered that by modifying the flux through the tricarboxylic acid (TCA) cycle, E. coli could mitigate the growth impairment observed in the absence of type II NADH dehydrogenase. This research provides valuable insights into the intricate mechanisms employed by bacteria to compensate for disruptions in energy metabolism.

The study is well designed and logically organised.The manuscript is well written and cites the appropriate literature.
However, in order to definitively validate their observations, the authors should consider: 1) Measure the AXP pool and determine the adenylate energy charge to understand the growth advantage observed that eΔndh-A in M9 as Figure 1 panel B as well as in the GMOS strain.
2) Perform targeted 13C stable isotope metabolomics to validate the figure 1 panel G and confirm that indeed carbon flux is redirected in eΔndh-A compared to Δndh Reviewer #2 (Comments for the Author): Goel et al present an interesting observation on how Escherichia coli can rewire oxidative phosphorylation to adjust to the loss of type-2 dehydrogenases.The manuscript is clear, and the conclusions are generally supported by the data presented.Nevertheless, a few issues should be addressed.1.Given that there is no complemented strain for ndh, and that the genome was sequenced, it would be important to show that the unevolved ndh does not harbor secondary mutations that may account for the observed growth defect.2. The deletion of sdhA in the ndh background was not compared with the e ndh lineages with point mutations in sdhA (Figure 1E), but it seems like ndh sdhA growth does not reach WT/ e ndh.A comparison between ndh sdhA and e ndh, and some discussion on why the strains don't behave similarly would be important.3. Is the difference in growth between ndh and ndh sdh significant?4. Does ndh sdhA have the same adaptive regarding the nuo operon?This would strengthen the proposed causality.5.One hypothesis that is not clearly stated is that the rescue effect observed in e ndh could be, at least in part, due to nuo overexpression.6.The authors propose a "a general increase in the central carbon metabolism in the evolved strains" based on a metabolic model.In the next paragraph, the authors state "The selective silencing of the NADH-producing reactions could be an adaptive response to maintain a growth conducive redox environment.".Given that the central carbon metabolism is the main generator of NADH, these statements are somewhat contradictory.It would be important to clarify which NADH producing reactions are predicted to be inhibited and which specific pathways from CCM have increased flux.

Reviewer #3 (Comments for the Author):
The manuscript "TCA cycle tailoring facilitates optimal growth of proton pumping NADH dehydrogenase-dependent Escherichia coli" by Goel et.al., has used adaptive laboratory evolution to study the consequence of deleting type II NADH dehydrogenase (ndh) on the growth and metabolism of E. coli.They show that mutants lacking ndh are defective for growth on glucose.Interestingly, the ndh deletion strain overcomes growth defect in 300 generations by: i) accumulating mutations that decrease the activity of succinate dehydrogenase (SDH), and ii) upregulating type I NADH dehydrogenase (Nuo).The authors suggest that reduction in the activity of succinate dehydrogenase optimises the growth of a ∆ndh strain.Major comments: 1.Some discussion regarding how lowering of SDH activity might optimise the growth of ∆ndh strain is required.2.More replicates are required for SDH activity assay.In Figure 1D, for Λndh and eΛndh-D, 2 of the 3 replicates have SDH activity lower than the WT. 3. In Figure 1F, authors have checked the expression of components of NADH dehydrogenase I (nuo).The nuo genes are upregulated after adaptive evolution of a Λndh strain.To support the claim that in the evolved Λndh strain, SDH and Nuo activities are antagonistic, it will be important to compare Nuo activity in the evolved and unevolved Λndh strains.4. In Supplementary figure 1A and B, data is presented with standard error of the mean.However, for all other growth curve data, data is presented with standard deviation.Both for consistency and better representation of the data, the data should be presented with standard deviation.5.In Supplementary figure 1B, Λndh shows growth defect in succinate.Some explanation of this phenotype is required in the text: Is this because of less energy generation, redox imbalance, or due to a decrease in SDH activity (Figure 1D: 2 of the 3 replicates have SDH activity lower than the WT). 6. Page 4, Lines 129-131, the authors state that 'The selective silencing of the NADH-producing reactions could be an adaptive response to maintain a growth conducive redox environment.'Here they should comment on the evolved Λndh-D strain where sucA is mutated.SucA is a component of 2-oxoglutarate dehydrogenase, and is involved in the NADH generating step in TCA cycle.They should also comment on why other steps of NADH generation in TCA cycle were not affected in the evolved lineages, e.g., malate dehydrogenase.
Minor comments 1.For both main and supplementary figures depicting growth curve data, please provide details of the media used for monitoring growth of strains in the figure legends, especially since in different experiments either glucose or succinate is used as the carbon source.2. Supplementary Figure 1A and B, Since the ΛsdhA strain fails to grow, there is no data corresponding to this strain in these figures.Hence, the label for ΛsdhA shown with the figures is misleading, and can be omitted.A mention of the phenotype in the text would be sufficient.

Preparing Revision Guidelines
To submit your modified manuscript, log onto the eJP submission site at https://spectrum.msubmit.net/cgi-bin/main.plex.Go to Author Tasks and click the appropriate manuscript title to begin the revision process.The information that you entered when you first submitted the paper will be displayed.Please update the information as necessary.Here are a few examples of required updates that authors must address: • Point-by-point responses to the issues raised by the reviewers in a file named "Response to Reviewers," NOT IN YOUR COVER LETTER.
• Upload a compare copy of the manuscript (without figures) as a "Marked-Up Manuscript" file.
• Each figure must be uploaded as a separate file, and any multipanel figures must be assembled into one file.For complete guidelines on revision requirements, please see the journal Submission and Review Process requirements at https://journals.asm.org/journal/Spectrum/submission-review-process.Submissions of a paper that does not conform to Microbiology Spectrum guidelines will delay acceptance of your manuscript." Please return the manuscript within 60 days; if you cannot complete the modification within this time period, please contact me.If you do not wish to modify the manuscript and prefer to submit it to another journal, please notify me of your decision immediately so that the manuscript may be formally withdrawn from consideration by Microbiology Spectrum.
If your manuscript is accepted for publication, you will be contacted separately about payment when the proofs are issued; please follow the instructions in that e-mail.Arrangements for payment must be made before your article is published.For a complete list of Publication Fees, including supplemental material costs, please visit our website.
Corresponding authors may join or renew ASM membership to obtain discounts on publication fees.Need to upgrade your membership level?Please contact Customer Service at Service@asmusa.org.
Thank you for submitting your paper to Microbiology Spectrum.
Goel et al present an interesting observation on how Escherichia coli can rewire oxidative phosphorylation to adjust to the loss of type-2 dehydrogenases.The manuscript is clear, and the conclusions are generally supported by the data presented.Nevertheless, a few issues should be addressed.
1. Given that there is no complemented strain for Δndh, and that the genome was sequenced, it would be important to show that the unevolved Δndh does not harbor secondary mutations that may account for the observed growth defect.2. The deletion of sdhA in the Δndh background was not compared with the eΔndh lineages with point mutations in sdhA (Figure 1E), but it seems like ΔndhΔsdhA growth does not reach WT/ eΔndh.A comparison between ΔndhΔsdhA and eΔndh, and some discussion on why the strains don't behave similarly would be important.3. Is the difference in growth between Δndh and ΔndhΔsdh significant?4. Does ΔndhΔsdhA have the same adaptive regarding the nuo operon?This would strengthen the proposed causality.5.One hypothesis that is not clearly stated is that the rescue effect observed in eΔndh could be, at least in part, due to nuo overexpression.6.The authors propose a "a general increase in the central carbon metabolism in the evolved strains" based on a metabolic model.In the next paragraph, the authors state "The selective silencing of the NADH-producing reactions could be an adaptive response to maintain a growth conducive redox environment.".Given that the central carbon metabolism is the main generator of NADH, these statements are somewhat contradictory.It would be important to clarify which NADH producing reactions are predicted to be inhibited and which specific pathways from CCM have increased flux.

Comments for the authors
The manuscript "TCA cycle tailoring facilitates optimal growth of proton pumping NADH dehydrogenase-dependent Escherichia coli" by Goel et.al., has used adaptive laboratory evolution to study the consequence of deleting type II NADH dehydrogenase (ndh) on the growth and metabolism of E. coli.They show that mutants lacking ndh are defective for growth on glucose.Interestingly, the ndh deletion strain overcomes growth defect in 300 generations by: i) accumulating mutations that decrease the activity of succinate dehydrogenase (SDH), and ii) upregulating type I NADH dehydrogenase (Nuo).The authors suggest that reduction in the activity of succinate dehydrogenase optimises the growth of a ∆ndh strain.
Major comments: 1. Some discussion regarding how lowering of SDH activity might optimise the growth of ∆ndh strain is required.2.More replicates are required for SDH activity assay.In Figure 1D, for Dndh and eDndh-D, 2 of the 3 replicates have SDH activity lower than the WT. 3. In Figure 1F, authors have checked the expression of components of NADH dehydrogenase I (nuo).The nuo genes are upregulated after adaptive evolution of a ∆ndh strain.To support the claim that in the evolved ∆ndh strain, SDH and Nuo activities are antagonistic, it will be important to compare Nuo activity in the evolved and unevolved ∆ndh strains.4. In Supplementary figure 1A and B, data is presented with standard error of the mean.
However, for all other growth curve data, data is presented with standard deviation.Both for consistency and better representation of the data, the data should be presented with standard deviation.5.In Supplementary figure 1B, Dndh shows growth defect in succinate.Some explanation of this phenotype is required in the text: Is this because of less energy generation, redox imbalance, or due to a decrease in SDH activity (Figure 1D: 2 of the 3 replicates have SDH activity lower than the WT). 6. Page 4, Lines 129-131, the authors state that 'The selective silencing of the NADHproducing reactions could be an adaptive response to maintain a growth conducive redox environment.'Here they should comment on the evolved ∆ndh-D strain where sucA is mutated.SucA is a component of 2-oxoglutarate dehydrogenase, and is involved in the NADH generating step in TCA cycle.They should also comment on why other steps of NADH generation in TCA cycle were not affected in the evolved lineages, e.g., malate dehydrogenase.
Minor comments 1.For both main and supplementary figures depicting growth curve data, please provide details of the media used for monitoring growth of strains in the figure legends, especially since in different experiments either glucose or succinate is used as the carbon source.
2. Supplementary Figure 1A and B, Since the ∆sdhA strain fails to grow, there is no data corresponding to this strain in these figures.Hence, the label for DsdhA shown with the figures is misleading, and can be omitted.A mention of the phenotype in the text would be sufficient.We are very excited to submit the revised version of our manuscript, "TCA cycle tailoring facilitates optimal growth of proton pumping NADH dehydrogenase-dependent Escherichia coli," which, we believe, has benefited significantly from the review process.We would like to thank the reviewers for their time and effort in providing a constructive evaluation and insightful review of the manuscript.Our response to each remark is presented below.The study by Goel and colleagues shed light on the bioenergetics flexibility of the model organism E. coli through an adaptive evolution experiment.
The study is well-designed and logically organized.The manuscript is well-written and cites the appropriate literature.
Thank you for your appreciation and the encouraging review!However, in order to definitively validate their observations, the authors should consider: 1) Measure the AXP pool and determine the adenylate energy charge to understand the growth advantage observed that eΔndh-A in M9 as Figure 1 panel B as well as in the GMOS strain.
We performed the adenylate charge calculations, but the values were not significantly different between evolved vs unevolved strains.We further went on to calculate ATP to ADP ratios.Although ATP:ADP ratios were also not significantly different, we observed a trend towards an increase in the ratio upon evolution.
The adenylate charge and ATP/ADP ratios were calculated for WT, GMOS, Δndh, eΔndh-A, and ΔndhΔsdhA using UPLC.A total of five biological replicates were used for each strain.The solid bar for each strain in the violin plot represents the median while the dotted lines represent the lower and upper quartiles.
We noticed that the adenylate charge is reported to be unaffected by growth rate and several other perturbations (1-3).Here, the growth advantage upon evolution appears to coincide with a higher ATP:ADP ratio.
2) Perform targeted 13C stable isotope metabolomics to validate the figure 1 panel G and confirm that indeed carbon flux is redirected in eΔndh-A compared to Δndh Our flux map for the central carbon metabolism uses a ME-model or a model of metabolism and expression.These models are capable of predicting the optimal macromolecular (protein, nucleotide, cofactor) expression required for growth.Our models are constrained based on the exchange rates of several key metabolites like glucose, acetate, lactate, succinate, formate, and ethanol.Notably, succinate, formate, and ethanol were below the detection limit.We are also constraining the model with measured growth rates and expression data.Thus, we can be more confident in the simulated metabolic fluxes as opposed to just using the HPLC data or more basic metabolic models.
Having said that, we recognize the value of the suggested approach, but 13C-based metabolomics is outside our experimental expertise, and we would like to be excused from this line of experimentation for this case.We always attempt to produce lucid manuscripts and are glad that our efforts are recognizable.Nevertheless, a few issues should be addressed.1.Given that there is no complemented strain for ndh, and that the genome was sequenced, it would be important to show that the unevolved ndh does not harbor secondary mutations that may account for the observed growth defect.
Before the start of our laboratory evolution experiments, we carefully examine the starting strain.As noted by you, we sequenced the whole genome of the strain.We have created the largest database of mutations from adaptive laboratory evolution experiments, and we refer to this database (ALEdb) to ensure the validity of strains (4).We did not find any mutations in the unevolved Δndh that are consequential in nature.
2. The deletion of sdhA in the ndh background was not compared with the endh lineages with point mutations in sdhA (Figure 1E), but it seems like ndhsdhA growth does not reach WT/ endh.A comparison between ndhsdhA and endh, and some discussion on why the strains don't behave similarly would be important.Yes, ΔndhΔsdhA does not reach eΔndh growth but their growth overlaps with WT.
Growth curve of WT, Δndh, eΔndh-A, and ΔndhΔsdhA on M9 minimal media with glucose.The data represent a mean of five biological replicates (with three technical replicates each) and the error bars show the standard error of mean.
The evolved Δndh lineages have additional mutations responsible for the growth optimization on minimal media in an aerobic environment.Two replicates have mutations that Growth rates of Δndh and ΔndhΔsdhA.For each strain, 5 biological replicates are used for growth rate calculations.The solid line inside the box plots represents the median while the whiskers represent the maximum and minimum data points.* P value < 0.05, two-tailed Mann-Whitney test.
4. Does ndhsdhA have the same adaptive regarding the nuo operon?This would strengthen the proposed causality.
Our current understanding is based on growth profiling, but we do not have RNAseq data for ΔndhΔsdhA strain.Your suggestion motivated us to look for some alternate explanations as well.Interestingly, the inhibition of flux through SDH (complex II) is reported to prevent ROS production at NDH-1 (complex I) by minimizing the production of metabolites contributing to reverse electron flux (8).
We intend to follow up our Observation article with a detailed work looking into this aspect and a characterization of every TCA reaction in the background of ndh deletion.
5. One hypothesis that is not clearly stated is that the rescue effect observed in endh could be, at least in part, due to nuo overexpression.Thank you for this observation.We mentioned this observation in a subtle manner.We have now elaborated on this possibility further.
6.The authors propose a "a general increase in the central carbon metabolism in the evolved strains" based on a metabolic model.In the next paragraph, the authors state "The selective silencing of the NADH-producing reactions could be an adaptive response to maintain a growth conducive redox environment.".Given that the central carbon metabolism is the main generator of NADH, these statements are somewhat contradictory.It would be important to clarify which NADH producing reactions are predicted to be inhibited and which specific pathways from CCM have increased flux.
The hypothesis about selective silencing of NADH-producing reactions is based on an Escher-based examination of the impact of SDH (SUCD) and AKGDH deletion on the TCA and the ME-model based flux simulation (9).Both deletions appear to reduce flux through the AKGDH-catalyzed reaction.We are motivated to probe this hypothesis deeper.The manuscript "TCA cycle tailoring facilitates optimal growth of proton pumping NADH dehydrogenase-dependent Escherichia coli" by Goel et.al., has used adaptive laboratory evolution to study the consequence of deleting type II NADH dehydrogenase (ndh) on the growth and metabolism of E. coli.They show that mutants lacking ndh are defective for growth on glucose.Interestingly, the ndh deletion strain overcomes growth defect in 300 generations by: i) accumulating mutations that decrease the activity of succinate dehydrogenase (SDH), and ii) upregulating type I NADH dehydrogenase (Nuo).The authors suggest that reduction in the activity of succinate dehydrogenase optimises the growth of a ∆ndh strain.
Thank you for the overview of our work.
Major comments: 1.Some discussion regarding how lowering of SDH activity might optimize the growth of ∆ndh strain is required.Among the two NADH-dehydrogenases of E. coli, NDH-2 (ndh) has a higher catalytic turnover for NADH oxidation.In Δndh, the oxidation of NADH might get compromised.Since the TCA cycle contributes significantly to NADH production, lowering flux through NADH producing reactions can help mitigate this redox imbalance.This can explain the presence of sdhA mutations in the evolved strains.Additionally, the inhibition of flux through SDH is reported to prevent ROS production at NDH-1 by minimizing the production of metabolites contributing to reverse electron flux (8).We have elaborated these possibilities in the revised manuscript.
2. More replicates are required for SDH activity assay.In Figure 1D, for Λndh and eΛndh-D, 2 of the 3 replicates have SDH activity lower than the WT.
We have generated more replicates and hope you will find the revised data satisfactory.
In vitro SDH activity of ∆sdhA, WT, ∆ndh (unevolved and evolved) strains.All seven replicate values are displayed with standard error.A significance test using one-way ANOVA was performed compared to ΔsdhA.
3. In Figure 1F, authors have checked the expression of components of NADH dehydrogenase I (nuo).The nuo genes are upregulated after adaptive evolution of a Λndh strain.To support the claim that in the evolved Λndh strain, SDH and Nuo activities are antagonistic, it will be important to compare Nuo activity in the evolved and unevolved Λndh strains.
Our proposition is based on the increase in the expression of nuo genes, as there is no change in the protein sequence.We attempted to perform the Nuo activity assay but could not get presentable data (10) (11) 4. In Supplementary figure 1A and B, data is presented with standard error of the mean.However, for all other growth curve data, data is presented with standard deviation.Both for consistency and better representation of the data, the data should be presented with standard deviation.
This difference was due to the use of two different platforms for growth profiling (micro-well plate based assay and culture-tube based assay).In micro-well plate based assay we had three technical replicates for each biological replicate, thus we presented standard error of mean.Whereas culture-tube based assay had biological replicates only and, therefore, we presented standard deviation.We have re-generated all the data using micro-well plate based assay read using plate reader to ensure uniformity across the manuscript.5.In Supplementary figure 1B, Λndh shows growth defect in succinate.Some explanation of this phenotype is required in the text: Is this because of less energy generation, redox imbalance, or due to a decrease in SDH activity (Figure 1D: 2 of the 3 replicates have SDH activity lower than the WT).
Δndh shows growth defects on both glucose and succinate.As compared to glucose, succinate is more of a respiratory metabolite than fermentative.This respiro-fermentative bias will exert extra pressure on ETS functioning in Δndh growing on succinate.The growth defect on succinate could be for the same reason as that for glucose.Also, Δndh has reduced SDH activity compared to WT, which is potentially aiding the growth defect of Δndh on succinate.We have elaborated this part in the revised manuscript.
6. Page 4, Lines 129-131, the authors state that 'The selective silencing of the NADH-producing reactions could be an adaptive response to maintain a growth conducive redox environment.'Here they should comment on the evolved Λndh-D strain where sucA is mutated.SucA is a component of 2-oxoglutarate dehydrogenase, and is involved in the NADH generating step in TCA cycle.They should also comment on why other steps of NADH generation in TCA cycle were not affected in the evolved lineages, e.g., malate dehydrogenase.
A similar clarification has been sought by the second reviewer.We are copying the response here with additional thoughts for easy reference- We do not have a clear explanation for why other NADH producing reactions were not targeted for adaptation.The answer might lie in the metabolic inter-connections or simply adding more lineages in the future evolution experiment.We are motivated to probe this hypothesis deeper.
Minor comments 1.For both main and supplementary figures depicting growth curve data, please provide details of the media used for monitoring growth of strains in the figure legends, especially since in different experiments either glucose or succinate is used as the carbon source.
Appropriate details are added in the revised manuscript.1A and B, Since the ΛsdhA strain fails to grow, there is no data corresponding to this strain in these figures.Hence, the label for ΛsdhA shown with the figures is misleading, and can be omitted.A mention of the phenotype in the text would be sufficient.

Supplementary Figure
There is data for ΔsdhA in the plot.The strain did not show an increase in OD and essentially is getting masked with eΔndh-A.We have now plotted the time-separated alternate OD values for ΔsdhA and eΔndh-A so that both data can be noticed in the plot.

Page 4, Line 99, Provide full form for eΛndh
We have added a scheme in the revised manuscript to elaborate on the labeling scheme.

3 .
Page 4, Line 99, Provide full form for eΛndh Staff Comments: • Manuscript: A .DOC version of the revised manuscript • Figures: Editable, high-resolution, individual figure files are required at revision, TIFF or EPS files are preferred