Respiration is essential for aerobic growth of Zymomonas mobilis ZM4

ABSTRACT Zymomonas mobilis is an alpha-proteobacterium that is a promising platform for industrial scale production of biofuels due to its efficient ethanol fermentation and low biomass generation. Z. mobilis is aerotolerant and encodes a complete respiratory electron transport chain, but the benefit of respiration for growth in oxic conditions has never been confirmed, despite decades of research. Growth and ethanol production of wild-type Z. mobilis is poor in oxic conditions indicating that it does not benefit from oxidative phosphorylation. Additionally, in previous studies, aerobic growth improved significantly when respiratory genes were disrupted (ndh) or acquired point mutations (cydA and cydB), even if respiration was significantly reduced by these changes. Here, we obtained clean deletions of respiratory genes ndh and cydAB, individually and in combination, and showed, for the first time, that deletion of cydAB completely inhibited O2 respiration and dramatically reduced growth in oxic conditions. Both respiration and aerobic growth were restored by expressing a heterologous, water-forming NADH oxidase, noxE. Oxygen can have many negative effects, including formation of reactive oxygen species (ROS) or directly inactivating oxygen sensitive enzymes. Our results suggest that the effect of molecular oxygen on enzymes had a greater negative impact on Z. mobilis than formation of ROS. This result shows that the main role of the electron transport chain in Z. mobilis is reducing the intracellular concentration of molecular oxygen, helping to explain why it is beneficial for Z. mobilis to use electron transport chain complexes that have little capacity to contribute to oxidative phosphorylation. IMPORTANCE A key to producing next-generation biofuels is to engineer microbes that efficiently convert non-food materials into drop-in fuels, and to engineer microbes effectively, we must understand their metabolism thoroughly. Zymomonas mobilis is a bacterium that is a promising candidate biofuel producer, but its metabolism remains poorly understood, especially its metabolism when exposed to oxygen. Although Z. mobilis respires with oxygen, its aerobic growth is poor, and disruption of genes related to respiration counterintuitively improves aerobic growth. This unusual result has sparked decades of research and debate regarding the function of respiration in Z. mobilis. Here, we used a new set of mutants to determine that respiration is essential for aerobic growth and likely protects the cells from damage caused by oxygen. We conclude that the respiratory pathway of Z. mobilis should not be deleted from chassis strains for industrial production because this would yield a strain that is intolerant of oxygen, which is more difficult to manage in industrial settings.

an interesting model for testing assumptions about bacterial physiology and metabo lism.For example, unlike other preferentially anaerobic organisms, Z. mobilis uses the Entner-Doudoroff pathway for glycolysis (one ATP per glucose) rather than the Embden-Meyerhof-Parnas pathway (two ATPs per glucose) (1).Additionally, attempts to genetically modify Z. mobilis have yielded evidence that this organism carries multiple copies of its genome, with some estimates as high as 50 copies per cell (2,3).The aerobic metabolism of Z. mobilis is also unusual and has been the subject of debate.Although Z. mobilis has a documented respiratory electron transport chain (ETC), it grows more quickly and to a higher density under anoxic conditions than under oxic conditions (4,5).The poor aerobic growth of Z. mobilis is surprising because in most other organ isms, respiration generates more ATPs per glucose than fermentation, and therefore, facultative aerobes typically have a higher growth yield (mg biomass dry weight/mg glucose) in the presence of oxygen.The negative effect of oxygen on Z. mobilis has been attributed to accumulation of the toxic metabolic intermediate acetaldehyde (6,7).While some of the acetaldehyde is converted to acetate by an acetaldehyde dehydrogen ase, residual acetaldehyde is often observed in aerobic cultures of Z. mobilis (8).Better understanding of the respiratory chain in Z. mobilis will allow us to determine whether ETC genes should be deleted from chassis strains (as suggested by Kalnenieks et al.), (9) and provide insight into why seemingly harmful genes are maintained in the genome.
The respiratory chain of Z. mobilis has a low predicted coupling efficiency; i.e., few protons are pumped across the inner membrane per O 2 consumed, possibly leading to a smaller than usual benefit for using the ETC.Based on the genome sequence, Z. mobilis is predicted to have four dehydrogenases that donate electrons to the quinone pool: NAD(P)H dehydrogenase (Ndh, ZMO1113), quinone-linked lactate dehydrogenase (Ldh, ZMO0256), glucose dehydrogenase (Gdh, ZMO0072), and succinate dehydrogenase (Sdh, ZMO0568 and ZMO0569).A bd-type terminal oxidase (CydAB, ZMO1571-1572) and a bc1 complex (quinol:cytochrome c oxidoreductase, ZMO0956-0958) are also encoded in the genome, but there is no known cytochrome c oxidase to complete the bc 1 pathway.Because none of the predicted ETC complexes are proton pumping, proton translocation only occurs through the oxidation and reduction of respiratory quinones at active sites on opposite sides of the membrane (Fig. 1).Compared with ETC configurations in other organisms that may translocate as many as 10 H + /2e − , the coupling efficiency of 2 H + /2e − in Z. mobilis is very low and may not appreciably contribute to ATP synthesis by the F o F 1 ATP synthase (10,11).Indeed, previous work indicates that the F o F 1 ATP synthase operates in the direction of ATP hydrolysis during aerobic growth by Z. mobilis (12).
Multiple research groups have previously shown that disruption of the gene encoding the NADH dehydrogenase in Z. mobilis (ndh) improves aerobic growth (13)(14)(15)(16).The improved growth of respiratory mutants has raised the question of why ETC genes are conserved in the genome, since they appear to have no effect under anoxic conditions and to be deleterious under oxic conditions.Jones-Burrage et al. (13) found that while respiratory chain activity does not enhance growth, it contributes to survival in station ary phase in minimal media, as previously observed by Rutkis et al. (16).Because ndh mutants exhibited clear growth phenotypes and dramatically decreased oxygen consumption, they were assumed to represent the phenotype of a total lack of respira tion.However, there are other entry points into the electron transport chain, including Gdh, Sdh, and a respiratory lactate dehydrogenase (ZMO0256), indicating that disrupting Ndh does not completely block the ETC (17).Further, in the absence of Ndh, the respira tory lactate dehydrogenase and the cytosolic (NAD + -linked) lactate dehydrogenase (ZMO1257) may form a bypass whereby NADH can be oxidized and electrons enter the respiratory chain in the absence of functional Ndh (18).
There are also unusual aspects of the latter half of the ETC.The Z. mobilis genome encodes a bd oxidase and a bc 1 complex but no cytochrome c oxidase that would be predicted to terminate the bc 1 branch.Sootsuwan et al. proposed that a peroxidase could terminate this branch using either hydrogen peroxide or O 2 as the electron acceptor (19).A later study determined that while the peroxidase PerC contributed to hydrogen peroxide tolerance, it did not accept electrons from the bc 1 complex or contribute significantly to the oxygen consumption rate of Z. mobilis (20).Strazdina et al. previously disrupted genes in the bc 1 complex (cytB) and bd complex (cydB) by a chloramphenicol resistance marker insertion (21).These mutants had very low oxygen consumption rates initially but recovered wild type (WT)-level oxygen consumption rates after 11-12 hours of aerobic growth.This could have been due to changes in regulation or instability of the mutation.
In this study, we deleted ndh and cydAB individually and in combination and measured growth, glucose metabolism, oxygen consumption rates, and reactive oxygen species (ROS) formation in the mutant strains.In contrast to previous studies, we found that ETC flux is essential to Z. mobilis in oxic conditions.We rescued growth of an ETC-deficient mutant with a water-forming NADH oxidase (NoxE), which oxidizes NADH and consumes O 2 but does not contribute to proton-motive force to drive oxidative phosphorylation.Results with NoxE indicated that oxygen removal from the environ ment is a key role of the ETC but not the sole benefit it provides to Z. mobilis.

Construction of mutant strains
To investigate the role of the electron transport chain in Z. mobilis, we constructed deletion mutants of the genes encoding the NADH dehydrogenase (ndh, ZMO1113) and the bd quinol oxidase (cydAB, ZMO1571-72) using a homologous recombination method (22).The procedure was performed in anoxic conditions as much as possible to avoid selection against the mutant strains, although exposure to oxygen was necessary at some points (see Materials and Methods).Deletion of ndh by this method was straight forward, but deletion of cydAB was more challenging.Putative mutant colonies were screened for deletion by PCR amplification of the region of interest.For the cydAB deletion, we observed that most colonies that produced the PCR amplicon consistent with deletion also produced an amplicon consistent with the WT sequence.Of 39 colonies screened by PCR, 6 showed the deletion band and only 1 of these also lacked the WT band (Fig. S1, isolate 5).For comparison, 8 colonies of 11 screened for ndh knockout showed the deletion band, and none of these contained the wild-type band with the deletion band.The ΔcydAB strain was also checked by Illumina re-sequencing and we found no reads aligned in the deleted region (Fig. S2).Re-sequencing indicated two point mutations across the entire genome, one in an oligosaccharide flippase family protein and one in an intergenic region of a plasmid.We also generated a ΔcydABΔndh deletion strain by deleting ndh from the ΔcydAB strain.

Growth and metabolism of mutant strains
As expected, the growth of both Δndh and ΔcydAB strains was indistinguishable from WT ZM4 in anoxic conditions (Fig. S3).We observed that deletion of ndh improved growth in oxic-rich medium (Fig. 2A, ZM4 final OD 600 = 4.54 ± 0.06, Δndh final OD 600 = 5.18 ± 0.15).In contrast, deletion of cydAB dramatically reduced growth in the same conditions (final OD 600 = 0.29 ± 0.01).Complementation of the deletions with the corresponding genes expressed from a plasmid returned each strain to the WT phenotype, decreasing growth of Δndh and increasing growth of ΔcydAB (final OD 600 = 4.63 ± 0.13 and final OD 600 = 4.48 ± 0.14, respectively).Note that to facilitate comparisons, the non-com plemented strains in Fig. 2 contain the vector, pRL814, an inducible plasmid carrying gfp that was used to construct both complementing plasmids.Here, we will refer to pRL814 as "vector." This plasmid does not change the growth of WT or knockout strains when uninduced (data not shown).We observed the most complete growth comple mentation without isopropyl-beta-D-thiogalactoside (IPTG) induction and increasing IPTG concentration made complementation less effective (Fig. S4).We have previously observed that expression from pRL814 is leaky, so some expression of the gene of interest occurs even when inducer is not added (23).
Consistent with the improved growth, deletion of ndh increased glucose consump tion and ethanol production and reduced acetate production compared with WT (Fig. 2B).The ΔcydAB strain showed low glucose consumption and ethanol and acetate production, consistent with the poor growth.In oxic conditions, both mutant strains produced ethanol more efficiently than WT (Table 1).Thus, although the ΔcydAB strain produced less ethanol than WT, it converted glucose to ethanol more efficiently.Again, both strains were returned to the WT phenotype (i.e., low efficiency ethanol produc tion) by expressing the deleted gene from a plasmid.Neither mutant strain produced detectable acetaldehyde, while ZM4 accumulated 25-30 mM, possibly explaining the increase in ethanol production efficiency for the mutants (Fig. S5).
We measured the oxygen consumption rate of each strain using microplates that contained optical sensors integrated into the bottom of the wells to facilitate highthroughput monitoring (OxoPlates).Oxygen partial pressure (pO 2 ) is measured as changes in fluorescence of the optical sensor.Cultures grown anaerobically overnight were diluted in fresh aerobic medium to an OD 600 appropriate for measuring pO 2 over time (see Materials and Methods).Cultures were aerated for 30 min by shaking in a plate reader; this time should be sufficient to induce any ETC genes that are upregulated by oxygen exposure (24).We found that ZM4 consumed oxygen at a rate of 0.941 ± 0.017 mg/L/min, while Δndh consumed oxygen at 13% of the WT rate and ΔcydAB consumed oxygen at 0.6% of the WT rate (Fig. 3).Complementing Δndh with ndh returned the oxygen consumption rate to 60% of the WT rate, and complementation of ΔcydAB with cydAB returned oxygen consumption to 106% of the WT rate (representa tive dissolved oxygen profiles in Fig. S6).
We also assessed the ability of ZM4 and the mutants to generate proton-motive force (PMF) using a cationic voltage indicator dye, thioflavin T (ThT).This dye freely diffuses into bacterial cells, and its fluorescence intensity increases as PMF increases (25).We observed that when cells were provided with glucose (Fig. 4B), ThT fluorescence rapidly increased in ZM4 (6012 ± 2319 AU/hour).Both mutant strains showed a very different response to glucose, with an initial decrease in fluorescence, followed by a slower increase (2,646 ± 513 and 1,449 ± 159 AU/hour for Δndh and ΔcydAB, respectively).All strains showed low, stable fluorescence in the absence of glucose (Fig. 4A).

Analysis of a ΔcydABΔndh double mutant
We also generated a double mutant lacking both ndh and cydAB.As expected, the oxygen uptake rate of the double mutant was negligible, as in the ΔcydAB strain.Complementing with pRLndh did not improve the oxygen consumption, and comple menting with pRLcydAB increased it to a level similar to the Δndh strain (Fig. 5, represen tative profiles in Fig. S7).As in previous figures, strains containing the vector were used as the negative control for complementation.
We observed that the double mutant had the same phenotype as ΔcydAB, growing very poorly in oxic conditions (Fig. 6A).Similar to the observation for oxygen con sumption, complementation with cydAB rescued growth, while complementation with ndh had no effect on growth.Glucose consumption, ethanol production, and acetate production for the double mutant and complemented strains were all consistent with the growth, essentially as in the ΔcydAB strain (Fig. 6B).This result confirms that Ndh and CydAB are operating in the same metabolic pathway (ETC).

Growth rescue by a water-forming NADH oxidase
CydAB may be important to Z. mobilis for different reasons, including reoxidation of quinols in the inner membrane, removal of O 2 from the environment, or contribution to PMF generation for ATP synthesis or other processes.To determine whether oxy gen consumption was the key role of CydAB, we recombinantly expressed NoxE from  Lactobacillus brevis, a water-forming NADH oxidase that oxidizes NADH and reduces O 2 to H 2 O in the cytoplasm (26,27).NoxE consumes oxygen without contributing to PMF formation.When noxE was expressed in Z. mobilis ΔcydAB, we observed IPTG-dependent oxygen consumption, indicating that functional NoxE was produced (Fig. 7; Fig. S8).Maximal oxygen consumption was observed with induction with 100-µM IPTG.At this induction level, oxygen consumption of the ΔcydAB strain producing NoxE was 56% of  the WT rate.We also measured growth at a range of IPTG concentrations and observed that NoxE expression moderately improved growth of ΔcydAB (Fig. 8).

Intracellular reactive oxygen species
To better understand the growth defect of ΔcydAB, we assessed formation of ROS in the WT and mutants using a ROS sensing dye, CellROX Green.For this experiment, cells were grown in rich medium in anoxic conditions overnight.Cultures were washed in PBS and diluted to OD 600 = 0.1 in PBS with glucose.We did not observe increased CellROX Green fluorescence in the ΔcydAB mutant, suggesting that the growth defect is not due to increased ROS formation.To confirm that the dye could detect ROS in all strains, we used PBS with menadione (a redox-cycling ROS generator) as a positive control.All strains showed high CellROX Green fluorescence in the presence of menadione (Fig. 9B).We also compared ROS formation for all strains in PBS without glucose to observe the effect of respiratory activity.In ZM4 and Δndh, we observed that more cells had high CellROX fluorescence without glucose than with glucose, suggesting that oxygen consumption was protective against ROS.Glucose also decreased the number of cells with high CellROX fluorescence in ΔcydAB, suggesting that other ROS protection mechanisms such as the respiratory peroxidase were active when glucose was present (20).

DISCUSSION
As previously observed by other groups, we found that deletion of ndh from the Z. mobilis genome resulted in enhanced growth in oxic conditions.However, when we deleted cydAB from the genome, we observed a severe growth defect in oxic condi tions.These results suggest that contrary to the previous hypotheses that respiration is harmful to Z. mobilis growth, the electron transport chain is essential to aerobic growth.
We speculate that all previous respiratory deficient conditions, whether induced by a mutation or inhibitor, likely still had significant respiratory activity, resulting in the lack of a growth defect in previous studies.In the case of Ndh mutation, Strazdina et al. found that removing Ndh alone is insufficient to block respiratory activity because the NAD + -linked and quinone-linked lactate dehydrogenases create a bypass that allows entry of electrons into the ETC (18).Our oxygen consumption results corroborate their findings, indicating that electrons can still pass through the ETC in the absence of Ndh.
The respiration deficient mutants isolated by Hayashi et al. included some large deletions or stop codons in ndh but only point mutations in cydA and cydB, indicating that all these mutants likely had residual respiratory activity.The point mutations in cydA and cydB were missense mutations leading to single amino acid substitutions in each case.Similar to Ndh disruption, cyanide addition to Z. mobilis cultures reduces the respiratory rate but does not block the ETC completely and leads to enhanced growth in oxic conditions (6).
Our results show that the residual respiratory activity in all of these cases was essential to aerobic growth.Strazdina et al. previously generated a cydAB mutant via insertion of a chlorampheni col resistance cassette but found that the mutant retained respiratory capacity (21).They suggested that an alternative pathway involving the bc1 complex could compensate for the lack of the bd oxidase and thus respiration would still occur if only one of the pathways is removed.Our results show that only the CydAB pathway is active in ZM4 under these conditions, leaving the possible role of the bc1 complex unknown.In Strazdina et al., growth data for the cydAB mutant strain were not presented, so it is difficult to compare our strain with theirs, although we suspect that, due to chromo somal polyploidy of Zymomonas (2, 3), they may have had a mixture of the disrupted and WT sequences.Other groups have also observed that intended gene deletions in Z. mobilis actually function as knockdowns rather than knockouts because each cell carries multiple genome copies, and in some cases not all copies are disrupted (2).This was observed during whole genome transposon mutagenesis of Z. mobilis ZM4, where many essential genes received transposon insertions, but the insertion strains were heterozygous (28).Insertions were detected in cydA and cydB in the transposon library, but they had strong fitness defects under all oxic conditions, suggesting that they were heterozygous (29).A strong fitness defect and heterozygous insertion was observed for known essential genes, including rpoB and ftsZ.The polyploidy of Z. mobilis likely led to the hybrid PCR results we observed during construction of ΔcydAB, but PCR indicated that we obtained a colony with all copies deleted (Fig. S1).Further, genome sequencing did not detect any reads in the ΔcydAB open reading frame, indicating that we generated a true deletion mutant (Fig. S2).
Comparing the growth and oxygen consumption rates of WT, Δndh, and ΔcydAB, it appears that even the low level of oxygen consumption of Δndh is very important to growth in aerobic conditions, although additional flux is not helpful.Although the ETC appears to be suboptimally regulated in laboratory conditions, we found that it is necessary for oxic growth.It remains puzzling why ndh has been conserved in the Z. mobilis genome when it seems deletion has no effect or has improved growth in the conditions tested in our study and others.Jones-Burrage et al. found that Ndh contributes to survival during stationary phase in oxic minimal medium, suggesting that there may be an important role under environmental conditions (13).Future work to simultaneously delete multiple respiratory dehydrogenases may shed light on the specific role of Ndh.
Previous work suggested that aerobic respiration is harmful to Z. mobilis because of acetaldehyde formation (4) and that ndh mutants grow better than WT because of reduced acetaldehyde concentrations.However, we observed that the ΔcydAB strain grew very poorly despite producing no acetaldehyde (Fig. S5).This indicates that acetaldehyde toxicity alone is insufficient to explain the poor aerobic growth of Z. mobilis or the growth enhancement of Δndh strains.To test whether oxygen itself is the cause of poor growth in ΔcydAB, we expressed a heterologous water-forming NADH oxidase, NoxE.We found that NoxE improved the final culture density to ~75% of the WT level, indicating that reducing O 2 concentration in the culture is one of the key functions of the ETC.
Oxygen can have a range of negative effects on cells, including formation of ROS and inactivation of oxygen-sensitive enzymes.We performed intracellular ROS measure ments using flow cytometry and a ROS-sensing dye to explore which aspect of oxygen is harmful to Z. mobilis.We found that intracellular ROS was lower in ZM4 with glucose than in starved ZM4, suggesting that respiration has a protective effect on WT, likely by reducing the oxygen concentration in the cell.Similarly, ΔcydAB has lower ROS levels with glucose than without.Although this strain could not have protected itself from ROS by oxygen consumption, we speculate that electrons were still fed to the cytochrome c peroxidase, leading to the protective effect of glucose (30).Overall, the ΔcydAB strain exhibited lower intracellular ROS than WT with or without glucose, suggesting that while respiration protects against ROS, CydAB is also a significant source of ROS in ZM4.Based on these results, it appears more likely that NoxE benefitted ΔcydAB not by protection from ROS but by protection of oxygen-sensitive enzymes from O 2 .Previous multi-omic analysis of oxygen exposure in Z. mobilis indicated that iron sulfur cluster enzymes were quickly inactivated by O 2 , leading to major disruptions in central metabolism (24).By expressing the electron transport chain, Z. mobilis can quickly remove oxygen from the environment and begin repairing iron sulfur cluster enzymes to return to metabolic homeostasis (within minutes in typical culturing conditions [in a culture tube at OD 600 = 1.0,Fig. S9]).
Although our results show that oxygen consumption is likely the main role of the ETC, there was no condition under which NoxE fully rescued growth, suggesting that the ETC may also contribute other benefits, possibly including formation of PMF.Although the oxygen consumption rate of the NoxE producing strains was lower than WT, the oxygen consumption for Δndh was also lower, and this strain grew better than WT.Therefore, slower oxygen consumption cannot fully explain why NoxE did not fully rescue the cydAB deletion.Measurements using a membrane voltage indicator dye suggest that the ETC does contribute to PMF generation.The difference in dye fluorescence with and without electron donor (glucose) showed the same trend as fluorescence with and without an electron acceptor in the well-studied respirer Shewanella oneidensis, indicating that the dye is functioning as a reporter of PMF generation by the ETC (25).Our results confirm previous work showing that Z. mobilis ETC components are capable of generating PMF (31).While previous work indicated that PMF generated by the ETC could power ATP synthesis in vesicles or starved cells, it remains unclear whether PMF drives ATP synthesis during aerobic growth in this organism (31).
Although respiration is generally considered a source of ATP via oxidative phosphory lation, our results support previous findings that energy conservation is not the only or even the major role of the electron transport chain in Z. mobilis.Evidence that oxygen consumption is the main role of the ETC in Z. mobilis helps to explain why complexes with low coupling efficiency are used.If proton translocation by the ETC is not needed, complexes with low coupling efficiency may be used without any growth penalty.At the same time, use of less efficient ETC components can be advantageous because these tend to be smaller and simpler than their efficient counterparts and therefore incur less metabolic cost to produce.For example, the proton-translocating NADH dehydrogenase in E. coli consists of 14 subunits, while the uncoupling version has only one.Therefore, by using a low efficiency electron transport chain, Z. mobilis can gain the benefits of oxygen consumption with the minimum metabolic cost.This has also been observed in Azotobacter vinelandii, which uses a low efficiency electron transport chain to reduce intracellular oxygen levels to protect its nitrogenase from oxygen (32).Low efficiency electron transport complexes are found in a wide variety of bacteria, and future work will further expand our understanding of a likely wide range of functions they can contribute to beyond oxidative phosphorylation.
ZM4 deletion mutants were constructed as described before (18) using oligonucleo tides listed in Table 3.Briefly, chromosomal 500-bp regions directly upstream and downstream of a target gene (flanks) were amplified from ZM4 using Q5 polymerase; primers are listed in Table 3.Both fragments were inserted into a SpeI site of non-rep licating plasmid pPK15534 (GFP, CmR) by Gibson assembly (22).The reaction mixture was transformed into chemically competent E. coli Mach 1. Sequences of the plasmids, isolated from chloramphenicol-resistant E. coli Mach 1 transformants, were confirmed by Sanger sequencing.pPK15534 bearing sequences flanking a target gene was introduced to ZM4 by conjugation using WM6026 (DAP−) as a donor strain as described in Felczak et al (35).Selection for colonies with the plasmid integrated into chromosome was for Cm resistance and DAP independence.One colony of a ZM4-CmR conjugant was used to inoculate ZRMG without the antibiotic, and culture was grown for at least 10 generations.10 4 CFUs were spread on 100 ZRMG plates without chloramphenicol and left for 48 hours to grow.Colonies were screened for loss of GFP fluorescence using a blue light illuminator.Conjugation to ZM4 and resolution of primary integrants to obtain ZM4Δndh by recombination were performed in aerobic conditions.To obtain ZM4ΔcydAB and ΔcydABΔndh, conjugation and the following steps were performed in an anaerobic chamber.In this case, plates containing single colonies were left in 4°C outside of the chamber for 24 hours for maturation of fluorescence before screening for non-fluorescent colonies.Colony PCR from primers flanking a target gene was used to confirm the gene deletion in non-fluorescent colonies.The double mutant, ΔcydABΔndh, was constructed by conjugating pPK15534 bearing regions flanking ndh to ZM4ΔcydAB strain followed by selection as for single knockout strain.
Plasmids pRLndh, pRLcydAB, and pRLnoxE were constructed by introducing the indicated gene into pRL814 to replace gfp.All primers used in cloning are listed in Table 2. Specifically, ndh (ZMO1113) was amplified from ZM4 genomic DNA using primers which added NdeI and BamHI restriction sites at 5′ and 3′ ends, respectively.The fragment was then digested with the above enzymes and introduced to NdeI/BamHI digested pRL814 by ligation.cydAB (ZMO1571-1572) was amplified from ZM4 gDNA using primers MF_13 and MF_14 and introduced to pRL814 amplified with primers MF_15 and MF_16 by Gibson assembly.A gBlock (IDTDNA) was used for cloning of noxE from Lactobacillus brevis (AF536177.1)to pRL814.ggagatatacat and gcttgatatcga overhangs were added at the 5′ and 3′ termini, respectively, of noxE gene sequence.The fragment was cloned by Gibson assembly to pRL814 amplified using primers MF_48 and MF_49.All reaction mixtures were routinely transformed to E. coli Mach 1, and colony PCR was used to confirm insertion of the gene into pRL814.Finally, recombined pRL814 plasmids were sequenced by the Sanger method to confirm the correctness of the construct.

Bacterial growth
Strains were grown in indicated media in an anaerobic chamber overnight.Spectinomy cin was added when indicated.In the morning, cultures were diluted to OD 600 = 0.1 in 5 mL of the same media.For aerobic growth, bacteria were grown in glass tubes covered loosely with plastic caps.For anaerobic growth, cultures were grown in Hungate tubes.In this case, after dilution of overnight cultures with fresh, anaerobic media, the tubes were closed with rubber stoppers and secured with aluminum crimps in anaerobic chamber (gas composition ~1% H 2 , balance N 2 ).Aerobic and anaerobic cultures were then incubated with shaking at 275 rpm at 30°C outside of the chamber.Samples were taken periodically for cell density and HPLC analysis.

Rescue of ΔcydAB growth from pRLnoxE
Growth rescue with pRLnoxE was performed in ZRDM because we observed growth inhibition of ZM4 after expression of pRLnoxE in the standard ZRMG medium.ΔcydAB or ZM4 bearing the vector or the noxE plasmid were grown in ZRDM supplemented with spectinomycin, overnight, in an anaerobic chamber.In the morning, cultures were diluted to OD 600 = 0.1 in fresh aerobic ZRDM with spectinomycin.IPTG was added to final concentrations as indicated.Wells of 96-well, clear, flat-bottomed microplates (Greiner) were filled with 200 µL of culture in triplicate.Bacteria were grown aerobically in a 30°C incubator with shaking at 500 rpm using a portable shaker (IKA MS three digital) for 48 hours.Absorbance at 600 nm was measured periodically by plate reader (Synergy H1 BioTek).At the end of the experiment, cultures from technical replicates were combined, and OD 600 was measured by a spectrophotometer with a 1-cm path length.

HPLC analysis
Samples were taken at the times indicated and stored in −20°C until used.Just before HPLC analysis, samples were thawed and centrifuged at maximum speed in a benchtop microcentrifuge for 10 min, and clear supernatants were used for analysis.HPLC chromatography was performed on a Shimadzu 20A HPLC equipped with an Aminex HPX-87H, 300.0 × 7.8-mm column, at 50°C.Metabolites were eluted with 5-mM sulfuric acid at 0.6 mL/min, and detection was by refraction index based on the standards run alongside.

Optical sensor microplate
Bacteria were grown overnight in ZRMG or ZRDM at 30°C in an anaerobic chamber.Cultures were diluted in fresh, aerobic medium to the initial OD 600 indicated in the experiments.The dilution factor was determined experimentally to adjust for O 2 uptake rate characteristic for each strain that allowed for near 100% air saturation after 30 min of shaking in a plate reader.The vial was closed and both standards were used immediately.The plate was covered with the lid and was shaken in a plate reader in aerobic conditions for 30 min at 30°C to stabilize the temperature and to allow for expression of oxygen-inducible genes.

Manual optical sensor
Cultures were grown in ZRMG, at 30°C in anoxic conditions, overnight.After diluting in fresh ZRMG to OD 600 = 1.0, 40 mL was aerated in 250-mL flasks by vigorous shaking at 30°C for 30 min.After this time, 30 mL of culture was immediately transferred into a 50-mL VWR tube, and dissolved oxygen was measured every 30 seconds for 10 min, with a pre-equilibrated optical probe (InLab OptiOx Optical Dissolved Oxygen Sensor, Mettler Toledo).

Thioflavin T fluorescence
ZM4, Δndh, and ΔcydAB were grown in 3 mL of ZRMG in anoxic conditions overnight.Cells were pelleted by centrifugation, and resuspended in an equal volume of PBS.The centrifugation step was repeated, and pellets were resuspended in 1 mL of PBS.Cells were diluted to OD 600 = 0.5 in PBS, and thioflavin T was added to the final concentration of 10 µM.Six wells of a 96-well microplate (clear/black Greiner 655090) were filled with 0.2 mL of each bacterial suspension, and ThT fluorescence was measured by microplate reader for 2 hours at excitation/emission 460/528 nm, respectively (Synergy H1, BioTek).Optical density (A 600 ) was measured at the same time.After this time, glucose was added to three out of six wells to 2%, and fluorescence was measured for an additional 2 hours.

Flow cytometric analysis of ROS
Strains were grown in ZRMG in an anaerobic chamber overnight.Cultures were centrifuged at 14,000 rpm for 1 min at room temperature, resuspended in equal volume of PBS and diluted to OD 600 = 0.1 in PBS with 0.5-µM CellROX Green (Invitrogen) ROS detection reagent.Glucose was added to the final concentration of 2%, and menadione was added to a concentration of 100 µM from a 10-mM stock solution in dimethyl sulfoxide (DMSO), as indicated in the experiments.All samples were incubated at 30°C for 30 min and subsequently kept on ice before being subjected to flow cytometric analysis.Sytox Red (Invitrogen) viability dye at 10-nM concentration was added just prior to acquisition.CellROX Green (BL1-A) and Sytox Red (RL1-A) fluorescence was assessed on an Attune CytPix (ThermoFisher Scientific) flow cytometer in the MSU Flow Cytometry Core Facility.Post-acquisition analysis was performed using FCS Express (v.7,De Novo Software).ROS was assessed based on CellROX Green fluorescence in bacteria gated on a FSC-A/SSC-A cellular population, singlets, and live cells (Sytox Red ꟷ ).Gating placement for the percentage of CellROX Green + bacteria was determined based on a fluorescence minus one control (Sytox Red stained only).The fold change in CellROX Green + percentages were calculated with respect to the average percentage CellROX Green + cells ZM4 for each sample (% CellROX Green + / average % of CellROX Green + in PBS-treated ZM4 cells for each experiment = relative fold change).

FIG 2 (
FIG 2 (A) Growth (log scale) and (B) substrate and product concentrations for ZM4 and Δndh and ΔcydAB in oxic-rich medium, with and without complementation of deleted genes.ZM4 and Δndh and ΔcydAB bearing the vector or respective complementing plasmids were grown in Zymomonas rich medium with spectinomycin in glass tubes covered with loose caps with shaking at 30°C at ambient oxygen pressure.OD 600 was measured and samples were analyzed by high performance liquid chromatography (HPLC) as described in Materials and Methods.Points are the average of three biological replicates, and error bars are standard errors.Complement: pRLndh or pRLcydAB.

FIG 3
FIG 3 Oxygen consumption by Z. mobilis ZM4 and Δndh and ΔcydAB mutants with and without complementation.Overnight cultures were diluted in fresh Zymomonas rich medium with spectinomycin to an OD 600 appropriate for oxygen measurement; ZM4/vector and Δndh and ΔcydAB strains with complementing plasmids were diluted to OD 600 = 0.2, and Δndh and ΔcydAB bearing the vector to OD 600 = 1.0.200 µL from each dilution were loaded onto an OxoPlate in triplicate, and oxygen consumption was measured as described in Materials and Methods.Bar graphs are average of three biological and three technical replicates, and error bars are standard errors.Striped bars denote strains with complementing plasmids.Note that there is only one column for ZM4 with no plasmid.The bar for ΔcydAB is too small to be seen on this scale.

FIG 4
FIG4 Thioflavin T (ThT) fluorescence in ZM4 and Δndh and ΔcydAB strains after exposure to oxygen without (A) or with (B) 2% glucose addition.Strains were grown in Zymomonas rich medium in anoxic conditions overnight.Cultures were pelleted by centrifugation, washed once in phosphate-buffered saline (PBS), and resuspended in PBS at final OD 600 of 0.5.ThT was added to 10 µM, and 0.2 mL of each culture was loaded on 96-well microplate.The plate was incubated at 30°C with shaking in the plate reader, and OD 6oo and fluorescence (460/528 nm) were measured for 2 hours.After this time, glucose was added and measurement continued for an additional 2 hours.X-axis shows 10 min before and 1 hour after addition of glucose, which was set at 0. The curves are average of three biological and three technical repeats, and the shadowed area represents the standard error.

FIG 5 7 FIG 6
FIG 5 Oxygen consumption by ΔcydABΔndh with complementing plasmids.Strains were grown as described in Fig. 3. ZM4/vector was diluted to OD 600 = 0.2, while the double mutant strains bearing the vector or a complementing plasmid were diluted to OD 600 = 1.0.O 2 uptake rate was calculated as described in Materials and Methods and in Fig. 3. Bar graphs are the average of three biological and three technical replicates, and error bars are standard errors.

FIG 7
FIG7 Oxygen consumption with heterologous expression of Lactobacillus brevis noxE.ΔcydAB/pRLnoxE strain was grown in Zymomonas rich defined medium containing spectinomycin and indicated concentrations of IPTG, at 30°C in an anaerobic chamber overnight.ZM4/vector was grown without IPTG.Oxygen partial pressure was measured as described in Materials and Methods and in Fig.3.Bar graphs show the average oxygen uptake rate after normalization.n ≥ 6 for ΔcydAB/pRLnoxE and n = 3 for ZM4/vector.

FIG 8
FIG8 Growth of ΔcydAB with heterologous expression of Lactobacillus brevis noxE.ZM4 and ΔcydAB strains, bearing the vector or pRLnoxE, were grown in Zymomonas rich defined medium, supplemented with spectinomycin and IPTG at the indicated concentrations.Two hundred microliters from each culture at the indicated IPTG level was loaded on a 96-well clear microplate in triplicate.Cells were grown on a plate shaker in 30°C at ambient oxygen pressure for 48 hours.Growth was monitored by measuring absorbance at 600 nm in a plate reader.After 48 hours, cultures from three wells were combined and the final OD 600 was measured by a spectrophotometer with a 1-cm path length.(A) Growth curve from a representative experiment measured by plate reader; average of three technical replicates.(B) OD 600 after 48 hours of incubation; average from three independent plates, with three technical replicates each.Error bars are standard errors.

FIG 9
FIG 9 Assessment of ROS in ZM4, Δndh, and ΔcydAB strains.ZM4, Δndh, and ΔcydAB strains were incubated with PBS alone, PBS with glucose, or PBS with menadione.ROS formation was evaluated by assessing CellROX Green fluorescence in singlet, live (Sytox Red ꟷ ) bacteria using flow cytometry.Representative density plots in panel (A) show the percentage of CellROX Green + bacteria for incubation with PBS alone.The relative fold changes compared to the average percentage of CellROX green + cells in ZM4/PBS samples are shown in panel (B).Fold change was averaged from three to four biological replicates for each strain in up to two separate experiments.Error bars are 95% confidence intervals.
After this time, shaking was stopped and reference (540/590) and indicator (540/650) fluorescence were measured every 3.2 min for 1 hour.Oxygen partial pressure as percentage of air saturation (pO 2 ) was calculated from the formula pO 2 = 100 × (k0 / IR − 1)/(k0 / k100 − 1), where IR = fluorescence indicator/fluorescence reference (Fi/Fr) of unknown sample; k0 = Fi/Fr of Cal 0%; and k100 = Fi/Fr of Cal 100%.The final value of O 2 concentration in mg/L was obtained by multiplying pO 2 by a conversion factor of 0.091.O 2 consumption rate was calculated from the linear part of the slope and normalized to OD 600 = 1.

TABLE 1
Efficiency of ethanol production a a Glucose and ethanol concentrations were measured by HPLC as described in Materials and Methods.Efficiency of ethanol production (%) was calculated from maximum theoretical yield produced from glucose used after 48 hours (2-mol ethanol/mol glucose).

TABLE 2
Bacterial strains and plasmids

TABLE 3 Oligonucleotides
a F, forward primer; R, reverse primer.
Wells of 96-well, flat-bottomed OxoPlate (OP96C, Precision Sensing) were filled with 200 µL of each culture in triplicate.Control wells were filled with 400 µL of oxygen-free calibration solution (Cal 0%) or 200 µL of 100% air-saturated calibration solution (Cal 100%), in quadruplicate, as recommended by the manufacturer.The wells containing Cal 0% were covered with adhesive foil.Cal 0% was prepared by dissolving 0.15 g of Na 2 SO 3 in 15-mL water in a 15-mL, tightly closed VWR vial.Cal 100% was obtained by vigorous vortexing of 20 mL of H 2 O in a 50-mL VWR conical tube for 2 min and subsequent gentle moving of liquid for 1 min to reduce air oversaturation.