Dioxygen reductase heterogeneity is crucial for robust aerobic growth physiology of Escherichia coli

Summary

The development of a system to leverage molecular oxygen for energy-efficient pathways required several molecular adaptations. The enzymatic reduction of dioxygen to water is one such prominent evolutionary molecular trait. Microbes evolved several enzymes capable of reducing dioxygen and, interestingly, retained multiples of them in their genomes. While their structure and biochemical functions are well-studied, understanding their degeneracy and co-operativity in the system remains elusive.

We used genetic engineering and evolutionary repair approaches to examine the impact of the high oxygen affinity cytochrome bd oxidase deficiency in Escherichia coli aerobic growth. We found a crucial role of cytochrome bd oxidases in the robustness of aerobic physiology. Evolutionary repair experiments alleviated growth defects in bd oxidase-deficient strains by ArcAB system dysregulation at the cost of impaired stress response pathways. Energy generation pathways are potential antimicrobial targets, and understanding collateral phenotypes is crucial in designing therapeutic approaches that reduce antimicrobial resistance development.

Graphical abstract

Highlights

• •Deficiency of cytochrome bd oxidases impairs the aerobic growth of E. coli
• •Loss of ArcAB function alleviates growth defect in cytochrome bd oxidase-deficient E. coli
• •Adaptive rewiring results in collateral compromise in the bacterial stress response pathway

Biological sciences; Biochemistry; Microbiology

Introduction

Oxygen, which colloquially synonymizes life, was a nonexistent molecule in primitive Earth’s environment. The rise of oxygen in the environment necessitated several molecular changes in microbial physiology to withstand oxygen toxicity and exploit this high reduction potential molecule for evolving a superior energetics system.¹ The evolution of dioxygen reductases (O₂REDs) is suggested to be an adaptive mechanism to reduce oxygen to water, which later became a hallmark of aerobic respiration.²

There are several types of O₂REDs that differ in the redox centers and, consequently, in their reduction potentials.³ The primary role of the O₂REDs is to facilitate the aerobic flow of electrons in the electron transport system (ETS), and the range of reduction potential enables their functioning in a wide range of oxygen partial pressures. Two major classes of O₂REDs are heme center-based and heme-copper hybrid center-based oxidoreductases.⁴ Heme-center-based oxidoreductases are more widely distributed than heme-copper center-based oxidoreductases; often, both O₂REDs are present within the same bacterium.^5,6 Dioxygen reductases appear to be functionally substitutable for bacteria in an aerobic environment.^7,8,9,10 The significance of accumulating multiple O₂REDs in an otherwise conservative bacterial genome is incomprehensible.¹¹

Escherichia coli, a facultative anaerobe, has both the O₂REDs: (i) cytochrome bo₃ oxidase consisting of heme b, heme o₃, and copper redox centers and (ii) two types of cytochrome bd oxidases (bd-I and bd-II) consisting of heme b and heme d redox centers (Figure 1A).¹² The two O₂REDs differ in their oxygen affinities. Cytochrome bo₃ oxidase has lower oxygen affinity and is reported to be the primary O₂RED in aerobic environments.¹³ Cytochrome bd oxidases have higher oxygen affinity and are observed to be majorly expressed in microaerobic environments.¹⁴ These O₂REDs have different abilities to support proton motive force (PMF) generation. Cytochrome bo₃ oxidase yields a PMF of 2H⁺/e⁻ by pumping a proton and translocating another vectorially. In contrast, cytochrome bd oxidases do not actively pump protons but still contribute to a PMF of 1H⁺/e⁻ through the vectorial translocation of protons.^15,16 The deficiency of neither cytochrome bo₃ oxidase nor cytochrome bd oxidase is reported to impact the aerobic growth of E. coli.^7,17,18

Figure 1.

Contribution of dioxygen reductases in aerobic growth of E. coli

(A) Schematic representation of dioxygen reductase diversity in E. coli. E. coli has a low oxygen affinity cytochrome bo₃ oxidase and two high oxygen affinity bd oxidases: bd-I and bd-II (only bd-II shown above for representation purposes). Cytochrome bo₃ oxidase pumps one proton and translocates another vectorially to create a proton motive force of 2H⁺/e⁻ effectively. Cytochrome bd oxidases contribute to a proton motive force of 1H⁺/e⁻ by vectorial translocation. (Color scheme - cytochrome bo₃ oxidase: CyoB-Peach, CyoA-Green, CyoC-Blue, CyoD-Pink; cytochrome bd-II oxidase: AppB-Green, AppC-Peach, AppX-Orange).

(B) Influence of dioxygen reductase deficiency on the aerobic growth of E. coli. The growth curve shows a mean of three biological replicates (with three technical replicates each), and the error bars show the standard error of the mean.

(C) Growth rates of dioxygen reductase deficient E. coli strains grown in aerobic conditions. The plot shows the mean of three biological replicates (each with three technical replicates (in the same color)), and the error bars show the standard deviation. The Kruskal-Wallis test was performed to determine the significance of growth rate differences compared to wild-type (WT).¹⁹ Cytochrome bo₃ oxidase deficient (ΔcyoB), cytochrome bd-I oxidase deficient (ΔcydB), cytochrome bd-II oxidase deficient (ΔappC), unevolved cytochrome bd oxidase deficient (uΔCBD). See also Figures S1 and S2.

Interestingly, cytochrome bd oxidases are also implicated in redox homeostasis, pathogenesis, and antimicrobial tolerance, suggesting a broader role of these oxidases in bacterial pathophysiology.^{20,21,22,23,24} Therefore, cytochrome bd oxidases are being explored as potential antimicrobial targets. To understand cytochrome bd oxidases' contributions to aerobic growth physiology and bacterial adaptive response to their loss, we genetically engineered an E. coli strain lacking both cytochrome bd oxidases. We performed an evolutionary repair experiment to resolve the proximal and distal response to the loss of high oxygen affinity oxidases in an aerobic environment.²⁵ We found¹: while there is a minor growth defect for the loss of individual O₂REDs, the deficiency of both cytochrome bd oxidases decreases the aerobic growth significantly,² evolutionary repair experiments could restore the growth of the strain,³ the evolved strains alleviate ArcAB repression for aerobic growth optimization,⁴ the evolved strains have better carbon utilization efficiency and oxidative phosphorylation dependency, and⁵ the evolutionary growth optimization sensitizes the strain to agents generating reactive oxygen species. We report the bioenergetic superiority and associated tradeoffs of cytochrome bd oxidase-deficient strain. We demonstrate the criticality of O₂RED degeneracy in robust aerobic physiology.

Results and discussion

Influence of dioxygen reductase deficiencies on aerobic growth and evolutionary repair

O₂REDs complete the aerobic flow of electrons through ETS by transferring them to oxygen. While cytochrome bo₃ oxidase is reported to be the preferred O₂RED,²⁶ we examined the influence of all three O₂REDs of E. coli on aerobic growth. We engineered four strains differing in their O₂RED compositions: (a) cytochrome bo₃ oxidase deficient (ΔcyoB), (b) cytochrome bd-I oxidase deficient (ΔcydB), (c) cytochrome bd-II oxidase deficient (ΔappC), and (d) cytochrome bd oxidase deficient (ΔcydBΔappC, hereafter labeled as ΔCBD; we will be using uΔCBD specifically for unevolved/starting strain) (Figure S1). The deficiency of any individual O₂RED resulted in a similar drop in exponential growth compared to the wild-type (WT) (Figures 1B and 1C). Notably, consistent with the report of reduced biomass, ΔcyoB enters the stationary phase earlier than other strains.²⁷ ΔcydB and ΔappC showed similar growth profiles, although bd-I oxidase seems more critical than bd-II oxidase. However, it is the deletion of both cytochrome bd oxidases that caused the most significant growth retardation (Figures 1B and 1C). Thus, while bd-I and bd-II oxidases are substitutable, at least one of the cytochrome bd oxidases is required in an aerobic environment. As expected, complementing the appC gene in ΔCBD improves its aerobic growth (Figure S2).

Since uΔCBD has functional cytochrome bo₃ oxidase, which is reported to be the primary aerobic oxidase, we wanted to examine whether the strain can recover its full growth potential.²⁶ Therefore, we performed an evolutionary repair experiment by continuously passaging ΔCBD in an aerobic growth condition until the increase in growth rate plateaued. We observed similar evolution trajectories for this strain’s four independently evolved lineages (eΔCBD-A-D) (Figure 2A; Table S1). In about 350 generations, all four lineages achieved growth rates (∼0.9 h⁻¹) similar to a wild-type strain that evolved under the same conditions.^28,29 This growth rate improvement indicated the existence of compensatory mechanisms in cytochrome bd oxidase-deficient strain.

Figure 2.

Genetic basis for the growth optimization of cytochrome bd oxidase-deficient E. coli

(A) Aerobic growth optimization of cytochrome bd oxidase-deficient strain by evolutionary repair experiment.

(B) List of genes mutated in the evolved ΔCBD strains compared to the unevolved ΔCBD strain. See also Table S1 and Figure S3.

Genetic basis for evolutionary repair of cytochrome bd oxidase deficiency in E. coli

The restoration of stress-induced growth defects in bacteria is frequently attributable to some adaptive genetic changes. We performed the genome sequencing of the ΔCBD strains to determine the genetic basis of their growth improvement. Each evolved lineage acquired mutations that have been characterized to improve the growth of E. coli on the M9 minimal medium (Figure 2B). Three of the eΔCBD strains mutated RNA polymerase subunits. These mutations enhance the transcriptional rate and, in turn, improve growth.^28,30 Intergenic mutations of pyrE-rph are also associated with enhancing growth by alleviating orotate phosphoribosyltransferase deficiency and restoring pyrimidine biosynthesis.²⁸ These mutations are acquired by a glucose minimal media-optimized strain (GMOS), a WT strain evolved under a similar condition.²⁸

Along with these well-characterized growth-promoting mutations, all evolved lineages acquired mutations in the ArcAB two-component system (Figure 2B). This anoxic redox control (Arc) system consists of a sensor kinase (ArcB) and a response regulator (ArcA).³¹ The ArcAB system regulates respiro-fermentative metabolism.³² Among several genes repressed by this system, the genes coding for cytochrome bo₃ oxidase are notable in the context of this study. The functioning of this oxidase should be relevant in the absence of cytochrome bd oxidases. We, therefore, explored the nature of the observed mutations. ArcAB is reported to be constitutively expressed, and we also observed no significant difference in the expression level of these genes³³ (Figure S3). In eΔCBD-A, an approximately 16 kb part of the genome was deleted, encompassing the arcB gene; therefore, we do not observe the expression of this gene (Figures 2B and S3).

The complete deletion of the arcB gene in the eΔCBD-A strain indicated that the mutations observed in the evolved strains potentially result in a loss of function of the ArcAB system. The evolved lineages C and D had point mutations in the arcB gene, while lineage B had a point mutation in the arcA gene. We investigated the effect of these mutations on the ArcA and ArcB proteins by analyzing amino acid properties, sequence homology, and overall structural stability.^34,35,36,37 In the eΔCBD-B strain, we identified the ArcA-R67C mutation in the N-terminal response regulator domain (Figure 3A). This mutation replaces a charged, hydrophilic residue with a neutral, hydrophobic residue, potentially destabilizing the protein and impairing its function. The mutations in eΔCBD-C and eΔCBD-D strains led to the formation of two ArcB protein variants: A8V-ArcB and I290S-ArcB (Figure 3B). The A8V mutation, located in the transmembrane domain, introduces a valine residue. Valine has a lower propensity to contribute to a helical structure and can disrupt protein folding.^38,39 The I290S variant introduces a serine residue, which is less hydrophobic than the wild-type residue, potentially affecting the protein’s intrinsic interactions. Notably, the I290S mutation is near a key residue involved in the ArcB phosphorelay mechanism (H292),³³ likely impacting the protein’s functionality. To experimentally support the interpretation from protein structure simulation, we introduced the WT copy of the arcB gene in evolved lineage D harboring a point mutation in this gene. We observed a growth defect upon the restoration of arcB (Figure S4).

Figure 3.

Validation of causality and nature of mutation in evolved cytochrome bd oxidase-deficient E. coli

(A) The point mutation observed in the eΔCBD-B strain mapped on the ArcA protein from E. coli (AlphaFold3 structure rendered using sequence from Uniprot ID P0AEC3) (Color scheme: Response regulatory domain-Blue, Magnesium ions-Orange).

(B) The point mutations observed in the eΔCBD-C and eΔCBD-D strains mapped on the ArcB protein from E. coli (AlphaFold3 structure rendered using sequence from Uniprot ID P0A9Q1) (Color scheme: Histidine kinase domain-Yellow, Response regulatory domain-Green, Hpt domain-Blue). In the zoomed-in inset of Figures 3A and 3B, green represents wild-type amino acid residue, while red represents mutant amino acid residue.

(C) Effect of arcB gene deletion on the growth profile of ΔCBD strains. The growth curve shows a mean of three biological replicates (with three technical replicates each), and the error bars show the standard error of the mean. See also Table S2 and Figures S4 and S5.

To strengthen the causality of the arcB mutations in improving the growth of uΔCBD, we engineered ArcB deficiency in uΔCBD by knocking out the arcB gene. The uΔCBDΔarcB strain showed growth improvement compared to the uΔCBD strain (Figure 3C), suggesting that arcB loss of function is responsible for improving the growth. Interestingly, the deletion of the arcB gene in the WT strain showed growth retardation, reestablishing the benefit of arcB loss of function being specific to the ΔCBD strain (Figure S5). However, the eΔCBD-A strain showed better growth than uΔCBDΔarcB, potentially due to additional contributions from mutations responsible for facilitating faster growth on the M9 minimal medium. We probed the additive interactions between these two classes of mutations by picking an intermediate strain (iΔCBD) from the evolutionary lineage that had media-responsive mutations only (Table S2). We then knocked out arcB in iΔCBD-A to create iΔCBD-A_ΔarcB. The iΔCBD-A_ΔarcB strain grew similarly to eΔCBD-A (Figure 3C). The growth rate trend of uΔCBD < uΔCBDΔarcB < iΔCBD-A < eΔCBD-A ∼ iΔCBD-A_ΔarcB established the additive effect of ArcAB and media-responsive mutations in improving the growth of uΔCBD.

Metabolic adjustments underlying the growth optimization of cytochrome bd oxidase-deficient strain

Genetic adaptation acts through metabolic rewiring to improve phenotypes. The ArcAB system influences a wide range of bacterial physiology, with respiration being a primary target. ArcAB inhibits the expression of cytochrome bo₃ oxidase and promotes the expression of cytochrome bd oxidases.⁴⁰ We examined the expression levels of genes of subunit I (cyoB) and II (cyoA) of cytochrome bo₃ oxidase in the ΔCBD strains (Figure 4A). uΔCBD strain showed a significant increase in the expression as compared to WT. We used an evolved strain of WT as a reference to examine expression response to an increase in growth rate. GMOS decreased the expression of cyoAB.²⁹ Interestingly, in the eΔCBD strain, the expression of bo₃ oxidase was maintained to a level similar to uΔCBD, suggesting that the loss of function of ArcAB might alleviate repression on the expression of cytochrome bo₃ oxidase. While we observed a trend toward higher expression of cyoAB in eΔCBD, the difference was not statistically significant.

Figure 4.

Metabolic basis for the growth optimization of cytochrome bd oxidase-deficient E. coli

(A) Relative expression of the cytochrome bo₃ oxidase genes (cyoA and cyoB) in GMOS and ΔCBD strains with respect to WT. All six replicate values are displayed. Whiskers represent the maximum and minimum data points, while the solid line inside the boxplots represents the median. Significance was determined using the Mann-Whitney test.⁴¹

(B) The percentage of carbon secreted as acetate is determined using the strains’ corresponding glucose uptake rate and acetate secretion rate.

(C) Estimation of oxygen uptake normalized to biomass in the strains determined using fluorescence sensor-based oxygen measurements. Two independent replicates are displayed independently. Growth rates of cytochrome bd oxidase-deficient E. coli in the following media conditions: (D) M9 minimal medium supplemented with succinate as sole carbon source and (E) M9 minimal medium supplemented with 4 g/L glucose and 5μM Paraquat (PQ). The plot shows the mean of three biological replicates (each with three technical replicates (in the same color)), and the error bars show the standard deviation. The Kruskal-Wallis test was performed to determine the significance of growth rate differences.¹⁹ Panel (D) and (E) share the y axis description.

(F) Growth profiles of cytochrome bd oxidase-deficient E. coli in M9 minimal medium supplemented with 4 g/L glucose and ciprofloxacin. The growth curve shows a mean of three biological replicates (with three technical replicates each), and the error bars show the standard error of the mean. See also Table S3 and Figure S6.

We estimated a panel of respiratory exometabolites (glucose, acetate, lactate, succinate, and formate) in the growth medium to probe the metabolic rewiring in the evolved strains. We observed an increased glucose uptake rate in strains with higher growth rates (GMOS and eΔCBD-A) (Table S3). However, unlike GMOS, the eΔCBD strain did not show a proportional increase in acetate secretion rate. The eΔCBD strain showed a lower carbon loss in the form of acetate with no detectable rise in other fermentation metabolites (Figure 4B; Table S3). The most efficient carbon utilization route is to flux the acetate to the TCA cycle, which feeds into the ETS. We compared the oxygen consumption profile of the evolved and unevolved ΔCBD strains to assess if the rise in ETS operation is responsible for efficient carbon utilization. The eΔCBD strain showed a higher oxygen consumption per biomass than uΔCBD (Figure 4C). These results indicate that the growth improvement of the uΔCBD strain is driven by efficient oxidative phosphorylation and not due to reliance on acetate overflow.

Collateral phenotypes of cytochrome bd oxidase-deficiency

Glucose is a respiro-fermentative metabolite; therefore, glucose-based energetics can have contributions from both respiratory and fermentative metabolism. We then used succinate as the carbon source to compare the contributions of respiratory energetics toward the growth of ΔCBD strains. We observed the retention of growth superiority of eΔCBD compared to uΔCBD when fermentative energetics are restricted (Figures 4D and S6A).

Beyond regulating respiration, the ArcAB system contributes to redox homeostasis, and its dysregulation can affect stress response capacities.³¹ We examined the sensitivity of ΔCBD strains to superoxide by treating them with paraquat. uΔCBD showed a lower reactive oxygen species (ROS) sensitivity compared to the WT strain. The evolved strain showed the most significant drop in growth rate on paraquat exposure among these strains, suggesting eΔCBD to be more sensitive to superoxide (Figures 4E and S6B). Notably, such differences in ROS sensitivity could also arise due to the differences in their growth rates. Such fitness tradeoffs can further compromise tolerance to antibiotics working by generating ROS. We probed this hypothesis by challenging the strains with ciprofloxacin, a fluoroquinolone antibiotic known to elevate cellular ROS in eliciting its bactericidal activity.⁴² We observed a relatively higher sensitivity to ciprofloxacin in the evolved ΔCBD strain (Figures 4F and S6C). Interestingly, the SOS response induced by ciprofloxacin may cause some mutagenic changes, which might have a confounding impact on the data interpretation.⁴³

Our observations suggest a critical role of cytochrome bd oxidase even in aerobic physiology and strengthen its candidacy as an antibiotic development target. However, the underlying adaptive rewiring is crucial in designing therapeutic approaches. We are restricting this study to the proposition that antibiotics targeting cytochrome bd oxidase in combination with ROS-generating antibiotics may limit resistant strain development. Parallely, the carbon-efficient nature of the evolved strains encourages the proposal of the strains' bioproduction applications since the reduction in carbon loss as acetate has been reported to improve recombinant protein production.^44,45,46

Limitations of the study

The study aimed to investigate the importance of cytochrome bd oxidase deficiency on the aerobic growth of E. coli and the compensatory response to the deficiency of these oxidases. Following up on the relative contributions and cooperation between various dioxygen reductases in aerobic conditions can provide crucial insights into bioenergetics. Also, the upregulation of cytochrome bo₃ oxidase expression in the unevolved strain presents an intriguing avenue for future research, especially to uncover the molecular mechanisms behind this adaptive response. Further, the bioproduction and biotherapeutic claims of the study remain to be validated.

Resource availability

Lead contact

Further information and requests for reagents may be directed to, and will be fulfilled by, the corresponding author, Dr. Amitesh Anand (amitesh.anand@tifr.res.in).

Material availability

All bacterial strains generated in this study are available from the lead contact with a completed Materials Transfer Agreement.

Data and code availability

• •DNAseq data supporting this study are deposited in the NCBI Sequence Read Archive (PRJNA1127851).
• •This article does not report any original code.
• •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

This work was supported by the DAE-Tata Institute of Fundamental Research Grant (19P0120) and DBT-Ramalingaswami Fellowship (21X432) to Amitesh Anand. We thank Prof. Bernhard Palsson (Systems Biology Research Group, University of California, San Diego) for sharing the evolutionary repair experiment resources. We would like to express our heartfelt gratitude to Mr. Sebastian S. Cocioba (Binomica Labs, New York) for providing the pECXA-FuGFP+KanR plasmid used in this study.

Author contributions

A.A. designed the study. A.A. and A.P. performed the experiments. A.A., A.P., and A.S. analyzed the data and wrote the article.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
Escherichia coli K12 MG1655	ATCC	700926
Escherichia coli K12 MG1655_ΔcyoB	This study	N/A
Escherichia coli K12 MG1655_ΔcydB	This study	N/A
Escherichia coli K12 MG1655_ΔappC	This study	N/A
Escherichia coli K12 MG1655_ΔcydBΔappC	This study	N/A
Escherichia coli K12 MG1655_ΔcydBΔappC_A13F16I1R1	This study	A13F16I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A14F16I1R1	This study	A14F16I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A15F16I1R1	This study	A15F16I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A16F16I1R1	This study	A16F16I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A13F58I1R1	This study	A13F58I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A14F60I1R1	This study	A14F60I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A15F57I1R1	This study	A15F57I1R1
Escherichia coli K12 MG1655_ΔcydBΔappC_A16F56I1R1	This study	A16F56I1R1
Escherichia coli K12 MG1655_ΔcydBΔappCΔarcB	This study	N/A
Escherichia coli K12 MG1655_ΔarcB	This study	N/A
Escherichia coli K12 MG1655_ΔcydBΔappC_A13F16I1R1_ΔarcB	This study	N/A
Chemicals, peptides, and recombinant proteins
CaCl2.2H2O	Sigma Aldrich	C3306
MgSO4.7H2O	SRL	1344116
Na2HPO4	Sigma Aldrich	S9763
KH2PO4	Sigma Aldrich	P5655
NaCl	MP Bio	194848
NH4Cl	Sigma Aldrich	A4515
FeCl3.6H2O	Sigma Aldrich	F2877
ZnCl2 anhydrous	Sigma Aldrich	Z1052
CoCl2.6H2O	Sigma Aldrich	C8661
Na2MoO4.2H2O	Sigma Aldrich	M1651
CuCl2.2H2O	Sigma Aldrich	C3279
H3BO3	USB	76324
Concentrated HCl	Sigma Aldrich	H1758
Glucose	Sigma Aldrich	49139
Sodium succinate dibasic	Sigma Aldrich	14160
Luria-Bertani broth	Himedia	M575
Luria-Bertani agar	Himedia	M1151
Kanamycin	Sigma Aldrich	K1377
Ampicillin	Sigma Aldrich	A9518
Ciprofloxacin	Sigma Aldrich	17850
Paraquat	Sigma Aldrich	36541
BsaI-HFv2	NEB	R3733S
rCutsmart 10X	NEB	B6004S
TRIzol reagent	Thermo Fisher Scientific	15596018
Chloroform	SDFCL	3771OUR
Absolute ethanol	Honeywell	32221
H2SO4	Sigma Aldrich	40325
Sodium acetate	Sigma Aldrich	791741
Sodium lactate	Sigma Aldrich	71718
Sodium formate	Sigma Aldrich	798630
Critical commercial assays
Nucleospin Tissue kit	Macherey Nagel	740952.50
Nextera XT kit	Illumina	FC-131–1024
First Strand cDNA synthesis kit	Thermo Scientific	18091050
SsoAdvanced Universal SYBR Green Supermix	BioRad	1725270
Deposited data
DNAseq data	NCBI Sequence Read Archive	PRJNA1127851
Oligonucleotides
k1: CAGTCATAGCCGAATAGCCT	Sigma Aldrich	Customised
cyoB FP: CTAGCGAATACAACCAGGTG	Sigma Aldrich	Customised
cydB FP: CGAACTCGTCACTGACCGCA	Sigma Aldrich	Customised
appC FP: TTTGGTCGACTCTGCGCAGC	Sigma Aldrich	Customised
appC RP: CCAGACCTGGTTGCCTTCCC	Sigma Aldrich	Customised
appC FP with BsaI extension: TATAGGTCTCGAA TGTGGGATGTCATTGATTT	Sigma Aldrich	Customised
appC RP with BsaI extension: TATAGGTCTCAAA GCTTACCCCTGTTGCTGCGTCG	Sigma Aldrich	Customised
arcB FP with BsaI extension: TATAGGTCTCGAAT GAAGCAAATTCGTCTGC	Sigma Aldrich	Customised
arcB RP with BsaI extension: TATAGGTCTCAAAG CTCATTTTTTAGTGGCTTTTG	Sigma Aldrich	Customised
Software and algorithms
Graph Pad Prism	Version 10.0.2	Graph pad.com
gcplyr	https://github.com/mikeblazanin/gcplyr	N/A
PyMol	Version 3.0	N/A
AlphaFold3	https://doi.org/10.1038/s41586-024-07487-w	N/A

Experimental model and subject details

Bacterial strains and growth conditions

Bacterial strains and primers used in this study are listed in the key resources table. Escherichia coli K12 MG1655 (ATCC 700926) was used as the wild-type strain. Knockout strains were generated using the P1 phage transduction method, with Keio collection strains as donors for gene knockout cassettes.^47,48 All the growth assays were performed using the Tecan Spark multimode microplate reader at 37^oC and atmospheric oxygen partial pressure using 96 well plate with 200μL culture per well. Three biological replicates (each with three technical replicates) were performed for each experiment. Unless stated otherwise, M9 minimal medium supplemented with 4 g/L glucose was used for bacterial growth. For the growth profiling with media variation, the concentration of succinate used is calculated according to the carbon equivalence of 4 g/L glucose while paraquat and ciprofloxacin were added to M9 minimal medium supplemented with 4 g/L glucose at a final concentration of 5 μM and 0.03 μg/mL, respectively. Growth rate calculations are done using gcplyr: an R package for microbial growth curve data analysis.⁴⁹

Method details

Genetic complementation of mutants

The plasmids, pECXA-appC and pECXA-arcB, were constructed using golden gate assembly.⁵⁰ The plasmids were made using pECXA-FuGFP+KanR, kindly shared by Sebastian S. Cocioba (Binomica Labs, New York), containing minimal constitutive promoter J23100. FuGFP+KanR was replaced with the gene of interest (oligonucleotides are listed in the key resources table). Growth characterization of the transformed strains was performed as mentioned in the preceding section, with ampicillin in the medium as a selection marker for the plasmid.

Evolutionary repair experiment

The evolutionary repair experiment was performed using four independent replicates of the ΔcydBΔappC strain. Cultures were serially propagated in M9 minimal medium supplemented with 4 g/L glucose at 37°C and atmospheric oxygen partial pressure, using an automated system that passed the cultures to fresh tubes once they had reached an OD₆₀₀ of 0.3 (Tecan 26 Sunrise plate reader, equivalent to an OD₆₀₀ of ∼1 on a traditional spectrophotometer with a 1 cm path length). The cultures were well-mixed for proper aeration. Cultures were always maintained in excess nutrient conditions assessed by non-tapering exponential growth. The laboratory evolution was performed for a sufficient interval to allow the cells to reach their growth rate plateau.

DNA resequencing

DNA resequencing was performed on a clone from the endpoints of evolved strains. Total DNA was sampled from an overnight culture and immediately centrifuged for 5 min at 8,000 r.p.m. The supernatant was decanted, and the cell pellet was frozen at −80°C. Genomic DNA was isolated using a Nucleospin Tissue kit (Macherey Nagel 740952.50) following the manufacturer’s protocol, including treatment with RNase A. Resequencing libraries were prepared using a Nextera XT kit (Illumina FC-131–1024) following the manufacturer’s protocol. Libraries were run on a HiSeq and/or NextSeq (Illumina).

Sequencing reads were filtered and trimmed using the software AfterQC version 0.9.623. The breseq bioinformatics pipeline version 0.31.1 was used to map sequencing reads and identify mutations relative to an E. coli K-12 MG1655 reference genome (NC_000913.3) amended to reflect the starting strain best.

Structure preparation and visualization in PyMol

Protein structures of cytochrome bo₃ oxidase (PDB ID: 8F68) and cytochrome bd-II oxidase (PDB ID: 7OY2) of E. coli K-12 were acquired from the Protein DataBank (PDB). These structures were rendered using PyMol to facilitate detailed visualization.

Protein structure simulation to perform mutation analysis

The sequence for wild-type ArcA and ArcB proteins of E. coli K-12 was retrieved from the UniProt database under the identifier numbers P0A9Q1 and P0AEC3, respectively. The protein structure for ArcA and ArcB was created using AlphaFold3³⁵ and used to analyze the impact of mutations. PyMol was used to introduce mutated residues and create the mutant protein structures.

RNA isolation and qPCR

Bacteria were grown in an M9 minimal medium supplemented with 4 g/L glucose till their mid-exponential phase. The culture was harvested and immediately processed for RNA isolation. The cells were dissolved in a TRIzol and incubated at 65^oC for 15 min. RNA was obtained in an aqueous layer upon the addition of chloroform, which was then precipitated using chilled isopropanol. This RNA pellet was washed with alcohol and finally resuspended in DEPC-treated water. RNA was quantified using a nano spectrophotometer before being frozen at −80°C for later use. cDNA synthesis was performed using Thermo Scientific RevertAid First Strand cDNA synthesis kit using random hexamer primer, according to the manufacturer’s protocol. Quantitative real-time PCR was carried out with BioRad SsoAdvanced Universal SYBR Green Supermix in CFX Opus 96 real-time PCR system (BioRad).

Phenotype characterization

Culture density was measured at 600 nm absorbance with a spectrophotometer and correlated to cell biomass. Samples for exometabolite estimation were filtered through a 0.22 μm filter (PVDF, Millipore) and measured using refractive index detection by HPLC (Agilent 12600 Infinity) with a Bio-Rad Aminex HPX87-H ion exclusion column. The HPLC method was the following: injection volume of 10 μL and 5 mM H₂SO₄ mobile phase set to a flow rate and temperature of 0.5 mL/min and 45°C, respectively, for a run time of 20 min per sample.

The oxygen uptake of each culture was determined by measuring the depletion of dissolved oxygen using the Presens oxygen sensor system (SFR vario). The cultures grown in M9 minimal medium supplemented with 4 g/L glucose at 37^oC/300 r.p.m in the SFR vario sensor shake flasks were used to measure oxygen and biomass changes.

Quantification and statistical analysis

Prism (GraphPad) version 10 was used for quantification and generation of plots. All experiments in this study were performed at least in triplicate. The growth curves show a mean of three biological replicates (with three technical replicates each), and the error bars show the standard error of the mean. The growth rate plots show the mean of three biological replicates (each with three technical replicates), and the error bars show the standard deviation. The Kruskal-Wallis test was performed to determine the significance of growth rate differences. The significance for the relative expression of the cytochrome bo₃ oxidase genes was determined using the Mann-Whitney test. The figure legends of each plot mention the statistical details and tests used.

Published: November 28, 2024