Systems Analysis of NADH Dehydrogenase Mutants Reveals Flexibility and Limits of Pseudomonas taiwanensis VLB120’s Metabolism

While Pseudomonas has the capability for high metabolic activity and the provision of reduced redox cofactors important for biocatalytic applications, exploitation of this characteristic might be hindered by high, constitutive activity of and, consequently, competition with the NADH dehydrogenases of the respiratory chain. The in-depth analysis of NADH dehydrogenase mutants of Pseudomonas taiwanensis VLB120 presented here provides insight into the phenotypic and metabolic response of this strain to these redox metabolism perturbations. This high degree of metabolic flexibility needs to be taken into account for rational engineering of this promising biotechnological workhorse toward a host with a controlled and efficient supply of redox cofactors for product synthesis.

ABSTRACT

Obligate aerobic organisms rely on a functional electron transport chain for energy conservation and NADH oxidation. Because of this essential requirement, the genes of this pathway are likely constitutively and highly expressed to avoid a cofactor imbalance and energy shortage under fluctuating environmental conditions. We here investigated the essentiality of the three NADH dehydrogenases of the respiratory chain of the obligate aerobe Pseudomonas taiwanensis VLB120 and the impact of the knockouts of corresponding genes on its physiology and metabolism. While a mutant lacking all three NADH dehydrogenases seemed to be nonviable, the single or double knockout mutant strains displayed no, or only a weak, phenotype. Only the mutant deficient in both type 2 dehydrogenases showed a clear phenotype with biphasic growth behavior and a strongly reduced growth rate in the second phase. In-depth analyses of the metabolism of the generated mutants, including quantitative physiological experiments, transcript analysis, proteomics, and enzyme activity assays revealed distinct responses to type 2 and type 1 dehydrogenase deletions. An overall high metabolic flexibility enables P. taiwanensis to cope with the introduced genetic perturbations and maintain stable phenotypes, likely by rerouting of metabolic fluxes. This metabolic adaptability has implications for biotechnological applications. While the phenotypic robustness is favorable in large-scale applications with inhomogeneous conditions, the possible versatile redirecting of carbon fluxes upon genetic interventions can thwart metabolic engineering efforts.

IMPORTANCE While Pseudomonas has the capability for high metabolic activity and the provision of reduced redox cofactors important for biocatalytic applications, exploitation of this characteristic might be hindered by high, constitutive activity of and, consequently, competition with the NADH dehydrogenases of the respiratory chain. The in-depth analysis of NADH dehydrogenase mutants of Pseudomonas taiwanensis VLB120 presented here provides insight into the phenotypic and metabolic response of this strain to these redox metabolism perturbations. This high degree of metabolic flexibility needs to be taken into account for rational engineering of this promising biotechnological workhorse toward a host with a controlled and efficient supply of redox cofactors for product synthesis.

INTRODUCTION

Many industrially relevant molecules, e.g., ethanol, butanediol, or isoprene, are more reduced than the industrially used sugars glucose and sucrose, or alternative, upcoming carbon sources such as xylose or glycerol (1–3). The microbial production of those favored compounds hence is inherently redox limited, i.e., by the supply of reduced redox cofactors, generally NADH or NADPH. This bottleneck has been overcome in some cases, e.g., 1,4-butanediol and 1,3-propanediol production in Escherichia coli (4, 5) or l-lysine synthesis in Corynebacterium glutamicum (6). The strategies applied optimized the host metabolism by metabolic engineering (4, 7, 8) or adapted the process conditions by (co-)feeding reduced substrates (9), applying microaerobic conditions, or using nongrowing cells with reduced competition and cellular demand for the redox cofactor (10–13). Alternatively, microorganisms can be identified that naturally outperform the classic, industrial workhorses with respect to redox cofactor supply. Pseudomonads are outstanding in this regard, as they exhibit a driven-by-demand phenotype which allows strongly enforced metabolic activity under stress conditions with increased energy demand, reported to result in a more than 2-fold carbon uptake rate and an 8-fold increase of the NAD(P)H regeneration rate relative to standard growth conditions (12, 14, 15). This behavior holds great promise for using this species for the bioproduction of highly reduced chemicals such as phenol, (S)-styrene oxide, rhamnolipids, and methyl ketones (16–20). Yet, competition is high as the NAD⁺/NADH couple functions as a coenzyme in over 300 oxidation/reduction reactions (21). Pseudomonas strains without apparent fermentative metabolism are obligate aerobes that rely on constitutive activity of the NADH dehydrogenases to ensure adequate oxidation of NADH to NAD⁺. Hence, we argue here that a naturally high NADH oxidation activity might impair the effective fueling of production pathways with reducing equivalents. We here set out to provide an in-depth analysis of the redox metabolism of Pseudomonas taiwanensis VLB120, a strictly aerobic bacterium, focusing on the role and essentiality of the individual NADH dehydrogenases for NADH oxidation and energy conservation.

While the mammalian mitochondrial electron transport chain constitutes only NADH dehydrogenase type 1, a multisubunit enzyme referred to as Nuo or complex 1 (22), which couples the electron transfer to proton translocation and hence contributes to ATP generation (23), aerobic bacteria have developed diverse NADH oxidation capabilities linked to the respiratory chain. Besides the Nuo complex, most species possess one to two isozymes of the non-proton-translocating type 2 dehydrogenase (Ndh), also termed alternative NADH dehydrogenase, which transfers electrons from NADH to ubiquinone but does not contribute to the membrane potential (23, 24). In some species, a third sodium-pumping type 3 dehydrogenase (Nqr) can be found. As with the facultative aerobic yeast Saccharomyces cerevisiae, several bacteria lack the Nuo complex and possess only type 2 dehydrogenases or are reported to mainly rely on the activity of this enzyme for NADH reoxidation (25, 26). Likewise, the genome of P. taiwanensis VLB120 encodes two types of NADH dehydrogenases, type 1 (EC 7.1.1.2) and two isoforms of type 2 (EC 1.6.99.3). Type 1 is encoded by the genes PVLB_15600 to PVLB_15660, designated the nuo operon. The two type 2 NADH dehydrogenases are encoded by PVLB_13270 and PVLB_21880, designated ndh-1 and ndh-2, respectively. Ndh-1 and Ndh-2 both consist of a single polypeptide chain.

In the present study, NADH dehydrogenase mutants of P. taiwanensis VLB120 were generated and characterized regarding growth, respiratory activity, and transcriptional and proteomic changes to elucidate the impact of redox metabolism perturbation on the cellular physiology.

RESULTS

NADH dehydrogenase activity is vital for P. taiwanensis but single enzymes of the redundant oxidation system are dispensable.

The NADH dehydrogenase type 1 operon encoded by nuoA to nuoN (PVLB_15600 to PVLB_15660) and the two type 2 NADH dehydrogenases encoded by ndh-1 (PVLB_13270) and ndh-2 (PVLB_21880) were successfully deleted from the P. taiwanensis VLB120 genome using the I-SceI-based pEMG plasmid (27). The double knockout Δndh-1 Δndh-2 (ΔΔndh) mutants and Δnuo Δndh-1 were successfully obtained, however, several attempts failed to generate the double knockout of Δnuo and Δndh-2. All gene deletions were confirmed by Sanger sequencing. The five NADH dehydrogenase mutants demonstrated that the NADH dehydrogenases Nuo, Ndh-1, and Ndh-2 are not essential individually. While the presence of either Nuo or Ndh-2 is sufficient to sustain the viability of P. taiwanensis VLB120, Ndh-1 seems to be unable to compensate for the loss of Nuo and Ndh-2. Similarly, it has been reported that single deletions of NADH dehydrogenases in P. aeruginosa PAO1 did not result in a growth defect or decrease in NADH oxidation activity, whereas in the double knockout (ΔnuoIJ Δndh) and triple knockout (ΔnuoIJ Δndh ΔnqrABCDEF) the NADH oxidation activity was abolished (28). In conclusion, Nuo and Ndh account for the total NADH dehydrogenase activity in this Pseudomonas strain. Note that while P. aeruginosa is a facultative anaerobe able to respire on nitrate and ferment pyruvate, P. taiwanensis VLB120 does not possess the necessary enzymatic makeup for this.

A total loss of NADH dehydrogenase activity in the obligate aerobic P. taiwanensis VLB120 strain seems to be lethal, indicating that the strain relies on the presence of these dehydrogenases for NADH oxidation and that alternative, native NADH-consuming reactions do not suffice to efficiently reoxidize this vital cofactor under the tested conditions.

The ΔΔndh mutant exhibits a growth-phase-dependent growth defect.

P. taiwanensis VLB120 and the five NADH dehydrogenase deletion strains Δndh-1, Δndh-2, ΔΔndh, Δnuo, and Δnuo Δndh-1 were characterized for growth, glucose utilization, CO₂ formation, and oxygen consumption in batch shake-flask experiments. The single NADH dehydrogenase type 2 mutants, Δndh-1 and Δndh-2, showed the same growth and sugar co-utilization profile as the wild-type P. taiwanensis VLB120 (Fig. 1A to C). The loss of the megaplasmid pSTY during NADH dehydrogenase deletions resulted in a growth advantage for the generated mutants, which was determined to result in a 14% higher growth rate for P. taiwanensis VLB120 pSTY⁻ compared to the pSTY⁺ wild type (29). For a comparison of mutants and wild type, the growth rate of the wild type was corrected accordingly and is referred to as μ_recalc.

FIG 1.

Physiological characterization of P. taiwanensis VLB120 wild type (A) and the NADH dehydrogenase deficient mutants Δndh-1 (B), Δndh-2 (C), ΔΔndh (D), Δnuo (E), and Δnuo Δndh-1 (F). The strains were cultured in MSM with 25 mM glucose. The OD_600nm (black circles), glucose levels (blue squares), and gluconate levels (green triangles) were measured over time. The shadowed area in (D) indicates the first growth phase. The data shown are the means of biological triplicates; error bars show the standard deviation. μ_recalc is the growth rate of P. taiwanensis VLB120 pSTY⁻. The wild type OD_600nm values are plotted (gray, open circles) in graphs (B) to (F) for comparison.

While the single gene deletion mutants Δndh-1 and Δndh-2 showed a wild-type physiology (Fig. 1, Table 1), the type 2 double mutant ΔΔndh reproducibly showed two growth phases (Fig. 1D). After wild type-like growth in the first phase, the growth rate dropped drastically in the second growth phase. Interestingly, the strong decrease in the growth rate (∼86%) was not correlated with an equal reduction in the carbon uptake, which showed a decrease of only ∼38%.

TABLE 1.

Calculated carbon uptake, gluconate accumulation, oxygen formation rates, biomass, and the gluconate yield of wild type and NADH dehydrogenase mutants during exponential growth

Strain	Mean carbon uptake rate (mmol gcdw−1 h−1) ± SDa	Mean biomass (g liter−1) ± SDb	Mean gluconate yield (mmol gcdw−1) ± SDc	Mean gluconate accumulation (mM) ± SDd	Mean surplus O2 consumption (mM) ± SDe
Wild type	7.3 ± 0.4	0.7 ± 0.0	13.7 ± 2.1	8.9 ± 1.1	9.2 ± 1.3
Δndh-1	7.9 ± 0.2	0.7 ± 0.0	11.6 ± 1.7	8.6 ± 1.3	8.1 ± 0.1
Δndh-2	7.2 ± 0.2	0.6 ± 0.0	12.8 ± 1.2	7.7 ± 0.3	6.3 ± 0.5
ΔΔndh	8.9 ± 0.5/3.5 ± 0.2	0.4 ± 0.0	20.7 ± 1.1	8.8 ± 0.4	7.2 ± 1.6
Δnuo	7.0 ± 1.2	0.6 ± 0.1	25.3 ± 2.5	14.5 ± 0.8	12.9 ± 0.2
Δnuo Δndh-1	6.1 ± 0.2	0.8 ± 0.0	20.9 ± 1.4	16.1 ± 1.3	14.1 ± 3.9

^aFor ΔΔndh, separate growth rates were determined for phase 1 (2 to 6 h) and phase 2 (6 to 8 h). For all other mutants, growth rates were calculated for the exponential phase (between 3 to 4 h to between 6 to 7 h after inoculation). SD, standard deviation; g_cdw, grams cell dry weight.

^bThe biomass concentration at the time point of the monitored maximum gluconate concentration according to ^d; shown is the mean of duplicate or triplicate experiments and the corresponding standard deviation.

^cThe gluconate yield was calculated by dividing the monitored maximum gluconate concentration according to ^d with the corresponding biomass concentration according to ^b; shown is the mean of duplicate or triplicate experiments and the corresponding standard deviation. g_cdw, grams cell dry weight.

^dThe gluconate accumulation was determined from offline monitored measurements; shown is the mean of duplicate or triplicate experiments and the corresponding standard deviation.

^eThe surplus oxygen consumption was calculated from the sectional integrals between the OTR (mmol liter⁻¹ h⁻¹) and CTR (mmol liter⁻¹ h⁻¹) between start of the cultivation and the time point of intersection of CTR and OTR (see Fig. 2); shown is the mean of duplicate experiments and the corresponding standard deviation.

Pseudomonas can catabolize glucose either via the phosphorylative or the oxidative pathway. In the latter, a membrane-bound glucose dehydrogenase (Gcd) oxidizes periplasmic glucose to gluconate coupled with the reduction of pyrroloquinoline quinone (PQQ). The phosphorylative pathway starts in the cytoplasm with the phosphorylation of glucose to glucose-6-phosphate catalyzed by the glucokinase (Glk) (30, 31).

The mutant ΔΔndh showed a significant increase in the specific gluconate yield in the early exponential growth phase. The same behavior was observed in the Δnuo and Δnuo Δndh-1 mutants (Table 1).

Besides the characterization for growth and glucose consumption, the respiratory behavior of the wild type and NADH dehydrogenase mutants was studied (Fig. 2, Fig. S1 in the supplemental material). Again, only the ΔΔndh mutant showed a different phenotype characterized by a stagnating oxygen transfer rate (OTR) after 6 h (Fig. 2B). This change in the OTR development is an indication for substrate inhibition, here, potentially by NADH, which cannot be oxidized at the rate required for fast growth. The onset of the reduced specific oxygen uptake rate also correlated well with the change in the growth rate (Fig. 1D).

FIG 2.

Respiratory activity of P. taiwanensis VLB120 and the ΔΔndh mutant. (A) CTR and OTR rates of the wild-type strain; the highlighted area corresponds to the surplus of consumed oxygen. The area was calculated from the sectional integrals between the OTR (dashed line) and CTR (solid line). (B) Oxygen transfer rates during cultivation of P. taiwanensis VLB120 (black dashed line) and mutant ΔΔndh (green dashed line).

During growth on glucose, the respiratory quotient (RQ), defined as the ratio of OTR and CO₂ transfer rate (CTR), is generally close to one (32, 33). Due to the oxidation of glucose to gluconate in the periplasm of Pseudomonas strains, the measured OTR for all tested mutants during the first 6 h of cultivation was higher than the CTR, resulting in an RQ below 1 (Fig. 2A, Fig. S1). Indeed, the surplus of consumed oxygen, calculated from the sectional integrals between the OTR and CTR ($\int \mathit{OTR}dt-\int \mathit{CTR}dt$), correlated with the produced gluconate (Table 1, Fig. 2A). During glucose conversion, roughly half of the overall consumed oxygen was used for the oxidation of glucose to gluconate and the reoxidation of the reduced PQQ formed by the glucose dehydrogenase activity. Consequently, in the glucose phase, the cells can partially uncouple glucose oxidation and energy provision from NADH formation, relieving the dependence on NADH dehydrogenase activity. The O₂ and CO₂ transfer rates of the Δndh-2 and ΔΔndh mutants (Fig. S1) showed a double peak, which occurred in the same time frame as glucose depletion, and, hence, might be due to the diauxic shift from glucose to gluconate. We assume that the diauxic shift also occurred in the other strains but was not recorded by the measurement frequency of three measurements per hour. The respiratory coefficient on gluconate was close to one for all Pseudomonas strains, indicating that no products other than biomass and CO₂ were formed during catabolism of this substrate.

NADH dehydrogenase gene deletions affect expression levels but do not result in altered in vitro NADH oxidation activities.

To further elucidate the NADH oxidation activity in the different mutants, and hence, the importance of the three NADH dehydrogenases for oxidizing NADH and fueling the electron transport chain, we performed in vitro NADH oxidation assays. Inverted membrane vesicles were prepared at early, mid-, and late exponential growth phase, and the NADH oxidation rate was determined from the decrease in absorbance at 340 nm over time. Note that the SDS-PAGE of the membrane fraction showed up to 21 prominent protein bands (data not shown). Therefore, we cannot exclude the presence of further membrane-bound NADH-dependent enzymes, e.g., the transhydrogenase PntAB, which might have contributed to the measured NADH oxidation rate. However, there is a high probability that the NADH oxidation is very specific for NADH dehydrogenases as most NADH-dependent enzymes, e.g., alcohol or aldehyde dehydrogenase, require electron acceptors other than O₂. Additional experiments with alternative electron acceptors have not been performed. In the early exponential growth phase, in which none of the strains showed a growth defect, all single mutants possessed NADH oxidation activities at levels similar to the wild type of around 1.2 U mg protein⁻¹ (Table 2), which is in the range of in vitro rates reported for other organisms (34). Overall, the NADH oxidation rate was rather stable in all mutants, indicating high metabolic flexibility of P. taiwanensis VLB120 to maintain redox homeostasis.

TABLE 2.

Specific NADH oxidation activities of inverted membrane vesicles of P. taiwanensis VLB120 wild type and NADH dehydrogenase mutants in the early and late exponential growth phases

Strain	Mean specific NADH oxidation activitya (U mgprotein−1) ± SD
	Early exponential	Late exponential
Wild type	1.2 ± 0.2	0.7 ± 0.2
Δndh-1	1.3 ± 0.1	0.5 ± 0.1
Δndh-2	1.0 ± 0.2	0.4 ± 0.1
ΔΔndh	1.1 ± 0.1	0.5 ± 0.2
Δnuo	1.2 ± 0.1	0.5 ± 0.1
Δnuo Δndh-1	0.8 ± 0.1	0.5 ± 0.2

^aMean values and standard deviations were determined from independent, biological triplicates.

To further substantiate this hypothesis, we examined potential changes at the transcriptional level by quantitative PCR (qPCR) on samples taken in the early, mid-, and late exponential growth phase. High-pressure liquid chromatography (HPLC) analysis showed that glucose and/or gluconate were still left when sampling the late exponential growth phase, i.e., the cells were still metabolically active (data not shown). The fold changes were normalized against the wild type in the corresponding growth phase. The single and double deletions of the type 2 NADH dehydrogenase encoding genes (Fig. 3A to C) had only minor effects (fold changes of < 2) on the remaining NADH dehydrogenase gene expression, while the type 1 deletion strains Δnuo and Δnuo Δndh-1 showed a substantial upregulation of the ndh-2 gene expression (Fig. 3D and E). The expression of the ndh-1 gene in both Δnuo and Δnuo Δndh-1 was unaffected; we observed only a small increase for mutant Δnuo in the early growth phase. This finding suggests that ndh-2 is probably the only NADH dehydrogenase gene that is regulated in response to the cellular NADH/NAD⁺ ratio. The consequent essentiality would further explain why the double deletion of nuo and ndh-2 was lethal. The observation that the ΔΔndh strain is growth impaired only during mid- to late exponential growth indicates that either the Nuo complex is less active in these phases or that the PQQ-dependent glucose dehydrogenase activity during the early growth phase enables sufficient ATP synthesis independent of NADH dehydrogenase activity.

FIG 3.

Relative gene expression of the NADH dehydrogenase-encoding genes ndh-1, ndh-2, and nuoA in NADH dehydrogenase mutants Δndh-1 (A), Δndh-2 (B), ΔΔndh (C), Δnuo (D), and Δnuo Δndh-1 (E) at early, mid-, and late exponential growth phase normalized to the corresponding values of the wild type. mRNA abundance was determined by quantitative PCR. Values were normalized to the relative transcript levels of P. taiwanensis VLB120 in the corresponding growth phase. nuoA was used as a proxy for the expression of the nuo operon. Gene deletions in the respective mutants are marked with “X” and were not analyzed by qPCR. Experiments were performed in biological triplicates.

Double deletion of the type 2 NADH dehydrogenases affects intracellular redox cofactor levels.

We found that the NADH oxidation rate was not (or only slightly) compromised by the introduced gene deletions, but that the ndh-2 level was significantly upregulated, suggesting that its expression is controlled by the redox state of the cell. Moreover, despite that we performed the in vitro enzyme assay with a physiological meaningful NADH concentration of 125 μM (35) to mimic in vivo conditions, we cannot exclude differences between these and the actual in vivo NADH dehydrogenase activities of the mutants, as the redox cofactor levels might have been altered.

For this reason, we determined the intracellular abundance of NADH and NAD⁺ in the early, mid-, and late exponential growth phase. Since the two single mutants of type 2 dehydrogenases had no growth phenotypes and showed no apparent changes on the transcriptional level, we restricted the analysis to the two double mutants ΔΔndh, and Δnuo Δndh-1 and the single Δnuo deletion mutant.

The Δnuo Δndh-1 mutant showed a higher NADH/NAD⁺ ratio in the late exponential growth phase but also a high variability in the triplicate experiments, thereby curtailing the statistical significance. The double mutant ΔΔndh had a significantly increased NADH/NAD⁺ ratio in the mid- and late exponential growth phase compared to the wild type (Fig. 4). This significantly increased NADH/NAD⁺ ratio in ΔΔndh probably triggered the observed drop in the growth rate in the mid-exponential phase, e.g., because of a potential inhibition of the Nuo complex. To test this hypothesis, we overexpressed the water-forming NADH oxidase (Nox) from Streptococcus pneumoniae (36). Nox is known to be highly specific for NADH, unable to oxidize NADPH, and has been described to produce no toxic hydrogen peroxide (37). The enzyme activity hence results solely in NADH oxidation and is suitable to elucidate the effect of relief from NADH accumulation. The overexpression of nox did not restore the wild-type phenotype but we observed a higher respiratory activity in mutant ΔΔndh Nox⁺ in comparison to mutant ΔΔndh (Fig. S3). In contrast to the response of P. putida KT2440, elevated Nox activity in the ΔΔndh mutant did not lead to a decrease in growth rate or biomass yield (14, 38).

FIG 4.

Quantification of the NADH/NAD⁺ ratio in the P. taiwanensis VLB120 (black) and the NADH dehydrogenase mutants ΔΔndh (green), Δnuo (orange), and Δnuo Δndh-1 (red) in early, mid- and late exponential growth phase.

Proteomic analysis reveals rerouting of the carbon flux in the ΔΔndh mutant.

We further performed shotgun proteomics analysis to explain possible metabolic changes in early, mid-, and late exponential growth phase in P. taiwanensis VLB120 due to NADH dehydrogenase deletions. The relative quantitative data were used to categorize the detected proteins into three groups: (i) significantly upregulated or (ii) downregulated proteins (fold changes of > 2; adjusted P value < 0.05); and (iii) weak/no effect proteins (fold changes of < 2). The proteins were further grouped into functional categories according to the KEGG database classification (39), e.g., transport, carbohydrate metabolism, or amino acid metabolism (Table S2). The most strongly represented categories are summarized in Fig. 5.

FIG 5.

Significant changes at proteome level of P. taiwanensis VLB120 NADH dehydrogenase mutants in early (A), mid- (B), and late (C) exponential growth phase relative to the wild type. Proteins are clustered into functional categories according to the KEGG classification system (39). Each bar represents the number of proteins in the depicted category, the abundance of which was either increased or decreased in response to NADH dehydrogenase deficiency. Experiments were performed in biological triplicates.

Per the physiological and transcript data, we did not observe significant changes in the proteome for either NADH dehydrogenase type 2 single mutants (Δndh-1: 9 of 24 proteins significantly up/downregulated; Δndh-2: 8 of 36 proteins significantly up/downregulated) (Fig. 5, Table S1 and S2). Proteomic changes in both type 1 mutants (Δnuo: 50 of 139 proteins significantly up/downregulated; Δnuo Δndh-1: 60 of 165 proteins significantly up/downregulated) were more significant compared to the type 2 single-gene knockout mutants and very similar to each other (Fig. 5). The double deletion mutant ΔΔndh showed more alterations in the proteome in the early and late exponential phase (17 and 37 of 107 proteins significantly up/downregulated, respectively) than in the mid-exponential phase (2 significantly up/downregulated proteins) (Fig. 5; Table S1).

We next focus on changes observed in the ΔΔndh mutant for proteins related to carbon uptake, energy generation, and oxidative stress response with respect to highlighting distinct differences from the type 1 NADH dehydrogenase mutants.

The OprB-I porin (PVLB_20075), a carbohydrate selective porin, and the D-gluconate transporter GntT (PVLB_13665) located in the outer and inner membrane, respectively, showed greater increases in the ΔΔndh mutant during the early and late exponential growth phases, while the glucokinase quantity was strongly reduced in all growth phases (Table 3). These data suggest that the ΔΔndh mutant strain oxidized glucose via glucose dehydrogenase (Gcd) to gluconate to a greater extent than the wild type. In contrast, the quantity of OprB-I in the type 1 NADH dehydrogenase mutants during the later growth phase was decreased. This change might, however, be explained by the faster glucose depletion in these mutants (Fig. 1).

TABLE 3.

Protein abundance in NADH dehydrogenase mutants relative to the wild type^c

^aGene function assignment differs from Fig. 5.

^bNumbers indicate the fold change for upregulated (+) and downregulated (−) proteins relative to the wild type, and the lack of a number indicates no differential production.

^cProteins marked with a diamond (◊) are discussed in the text.

During the late exponential growth of the ΔΔndh mutant, all enzymes of the arginine deiminase (ADI) pathway were more strongly expressed, while they were significantly downregulated in the NADH dehydrogenase type 1 mutants (Table 3). This pathway catalyzes a three-step conversion of arginine to ornithine, ammonium, and carbon dioxide coupled to ATP generation (40). Likewise, isocitrate lyase (AceA), the first enzyme of the glyoxylate shunt, was upregulated in the ΔΔndh mutant but downregulated in the Δnuo mutant, which instead showed a slight upregulation of the 2-oxoglutarate dehydrogenase complex of the TCA cycle during mid- and late exponential growth. These changes indicate that the mutant ΔΔndh used the glyoxylate shunt and not exclusively the TCA cycle in the late exponential growth phase.

We further observed remarkable changes in proteins combating oxidative stress. While deletion of the nuo operon (Δnuo and Δnuo Δndh-1) resulted in a generally reduced abundance of proteins involved in the oxidative stress response, those mutants deficient in one of the two type 2 dehydrogenases displayed increased levels of peroxidases and peroxiredoxin proteins. The abundance of the catalase-peroxidase KatG was strongly decreased in both type 1 mutants, whereas it was weakly increased in the ΔΔndh mutant. Additionally, only the peroxidase encoded by PP_0235 and the quinone reductase ChrR were more highly expressed in the ΔΔndh mutant, with the latter reported to be induced by superoxide (41), while peroxiredoxin AhpC was weakly upregulated in the single-gene deletion mutants Δndh-1 and Δndh-2 (Table S2). These findings indicate that the deletion of both type 2 dehydrogenases increases oxidative stress. We determined ROS formation in P. taiwanensis VLB120 and mutant ΔΔndh using the ROS-sensitive dye 2’,7’-dichlorodihydrofluorescein diacetate (H₂-DCFA), which is oxidized by ROS to fluorescent DCF. Indeed, we observed an increase in ROS formation in ΔΔndh concomitant with the reduction in growth (Fig. S2), underlining the proteomic results and strengthening our hypothesis.

DISCUSSION

This in-depth analysis of NADH dehydrogenase mutants has revealed high metabolic robustness of P. taiwanensis VLB120 to a partial loss of the three NADH dehydrogenases, but also the essential nature of residual NADH dehydrogenase activity, as the simultaneous deficiency of Nuo and Ndh-2 was lethal, likely due to inefficient NADH oxidation or ATP provision.

In accordance with the observed phenotypic robustness of most mutants, in vitro NADH oxidation activities in the mutant strains were not reduced. While this can be explained for those mutants deficient in the nuo operon by the concordant upregulation of ndh-2, no transcriptional changes in NADH dehydrogenase-related genes were observed for the other mutants. In contrast, the mutant with Nuo as the sole NADH dehydrogenase (ΔΔndh) showed a growth phenotype in the mid-exponential growth phase. While the in vitro NADH oxidation capacity was not altered, several changes in protein and metabolite levels were observed. We have summarized our current model of the potential underlying metabolic changes in Fig. 6. The wild type-like growth of the ΔΔndh mutant on glucose is likely sustained by periplasmic glucose oxidation to gluconate instead of phosphorylation, which partially uncouples the oxidation of the carbon source from NADH formation. The increased gluconate yield of the ΔΔndh mutant supports this hypothesis. The monitored respiratory activity and the match of gluconate accumulation and surplus oxygen consumption emphasize that no side products other than gluconate were produced to sustain the wild type-like growth. In vitro studies have shown the formation of reactive oxygen species (ROS) such as superoxide (O₂˙⁻) and hydrogen peroxide (H₂O₂) by enzymes of the electron transport chain due to electron leakage to oxygen. Cells activate antioxidant defense systems to combat ROS, which can result in severe cell damage or even death. Nuo (complex I) and cytochrome bc₁ (complex III) are considered the main sites for ROS in mitochondria and P. fluorescens (42, 43) but not in E. coli, which does not possess complex III (44) and rather employs type 2 dehydrogenase. It has further been reported that an oversupply of NADH can enhance ROS production (45, 46).

FIG 6.

Proposed metabolic changes caused by type 2 NADH dehydrogenase deficiency in P. taiwanensis VLB120. An increased NADH/NAD⁺ ratio (1) might result in substrate inhibition of the Nuo complex as well as ROS production (2), which is reported for this NADH dehydrogenase (45, 46). Rerouting of the flux through the TCA cycle into the glyoxylate shunt (3) reduces redox cofactor formation (48–50) and helps to scavenge reactive oxygen species by glyoxylate (48, 52). Limited ATP provision from oxidative phosphorylation can be mitigated by upregulation of the ADI pathway, based on our proteomics data (4) (58, 72). The light representation of the Ndh dehydrogenase indicates deficiency of both isozymes. ETC, electron transport chain; ROS, reactive oxygen species; ADI, arginine deiminase pathway; Nuo, type 1 NADH dehydrogenase; Ndh, type 2 NADH dehydrogenase; Sdh, succinate dehydrogenase; bc1, cytochrome bc₁ (complex III); cbb3, cytochrome cbb₃ (complex IV); QH₂, ubiquinol; Q, ubiquinone; SUC, succinate, SUCCoA, succinyl-CoA; FUM, fumarate.

The observed ROS accumulation in ΔΔndh during mid-exponential growth phase might, hence, be either due to an increased Nuo activity relative to the wild type or to an elevated NADH/NAD⁺ ratio. In line with this hypothesis, it has been shown that Mycobacterium tuberculosis NADH dehydrogenase mutants with a similarly elevated NADH/NAD⁺ ratio were more susceptible to (additional) oxidative stress than those with a lower NADH/NAD⁺ ratio (47). ROS induces stress, but a further explanation for the reduced growth and respiratory activity may lie in the potential NADH inhibition of metabolic enzymes. However, while the latter increased upon nox expression, Nox-mediated NADH oxidation did not restore growth, indicating that further limitations remained.

The activation of the glyoxylate shunt in ΔΔndh, as indicated by the proteome data, might contribute to stress reduction in two ways. First, this shortcut of the TCA cycle bypasses NAD(P)H-producing steps (48–50). For Pseudomonas putida KT2440, the production of a surplus of NADPH during growth on glucose has been reported (51) and it was speculated that the excess NADPH is converted to NADH by the transhydrogenase PntAB, which has been shown to be expressed in this strain under similar growth conditions (38). Assuming PntAB transhydrogenase activity in P. taiwanensis, which is equipped with the respective genes, attenuated NADP⁺ reduction would, hence, result in an overall reduced NADH formation. Second, the glyoxylate formed by the isocitrate lyase AceA activity, which was upregulated in ΔΔndh, can react with hydrogen peroxide to produce formate and CO₂ (48, 52). This ROS combating strategy has been reported for Pseudomonas aeruginosa, Burkholderia cenocepacia, and Staphylococcus aureus, even though S. aureus has no functional glyoxylate shunt (48, 53–55). Note, however, that neither increased formate dehydrogenase abundance nor formate accumulation was observed in the ΔΔndh mutant.

The deletion of nuo and the accompanying higher Ndh-2 activity did not result in a similar stress response, which is in accordance with corresponding M. tuberculosis mutants (47).

A probable energy shortage due to reduced respiratory activity might have been counteracted by ATP generation via the arginine deiminase (ADI) pathway, which seems to be activated in ΔΔndh according to the proteome data. This pathway generates 1 mol ATP per mol arginine (40, 56) and has been reported to be activated upon energy depletion in lactic bacteria (57) and pseudomonads, e.g., in P. putida DOT-T1E under energy-demanding solvent stress conditions (58, 59) and in P. aeruginosa under oxygen limiting conditions (40).

In this study, we showed high metabolic flexibility of P. taiwanensis VLB120 to interventions in the redox metabolism, which confers robust phenotypic behavior by a possible rerouting of metabolic fluxes. This metabolic adaptability and phenotypic robustness can be advantageous for biocatalysis but simultaneously challenging because it impedes the prediction of mutant behavior and can lever out metabolic engineering efforts. Hence, to effectively turn this promising microbe into a controllable, biotechnological workhorse, further biological and physiological system analyses, such as ¹³C metabolic flux analysis, are needed.

MATERIALS AND METHODS

Strains, media, and culture conditions.

Bacterial strains used in this study are listed in Table 4. Strains were propagated in lysogeny broth (LB) containing 10 g liter⁻¹ peptone, 5 g liter⁻¹ sodium chloride, and 5 g liter⁻¹ yeast extract (60). Cetrimide agar (Sigma-Aldrich, St. Louis, MO, USA) was used after mating procedures to select for Pseudomonas. Growth and characterization experiments were performed using mineral salt medium (MSM) (61) containing 3.88 g liter⁻¹ K₂HPO₄, 1.63 g liter⁻¹ NaH₂PO₄, 2 g liter⁻¹ (NH₄)₂SO₄, 0.1 g liter⁻¹ MgCl₂·6H₂O, 10 mg liter⁻¹ EDTA, 2 mg liter⁻¹ ZnSO₄·7H₂O, 1 mg liter⁻¹ CaCl₂·2H₂O, 5 mg liter⁻¹ FeSO₄·7H₂O, 0.2 mg liter⁻¹ Na₂MoO₄·2H₂O, 0.2 mg liter⁻¹ CuSO₄·5 H₂O, 0.4 mg liter⁻¹ CoCl₂·6H₂O, 1 mg liter⁻¹ MnCl₂·2H₂O supplemented with 25 mM glucose. For the preparation of solid LB, 1.5% agar was added to the medium. For plasmid maintenance and in the gene deletion procedure, antibiotics were added to the medium as required. Gentamicin and kanamycin were used at concentrations of 25 mg liter⁻¹ and 50 mg liter⁻¹, respectively. Because of the leaky expression of nox from plasmid pS2311, P. taiwanensis VLB120 strains bearing plasmid pS2311 were grown without the addition of the inducer cyclohexanone.

TABLE 4.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Reference
E. coli
DH5α	fhuA2 lac(del)U169 phoA glnV44 Φ80' lacZ(del)M15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17	New England Biolabs
DH5α λpir1	F−, Δlac169, rpoS(Am), robA1, creC510, hsdR514, endA, recA1uidA(ΔMluI)::pir-116; host for oriV(R6K) vectors in high copy number	Thermo Fisher Scientific
HB101 pRK2013	SmR, hsdR-M+, proA2, leuB6, thi-1, recA; bears plasmid pRK2013	(73)
DH5α pSW-2	Gmr, DH5α bearing pSW-2	(27)
DH5α λpir1 pEMG	Kmr, DH5α λpir1 bearing plasmid pEMG	(27)
DH5α λpir1 pEMG_ko_ndh1	Kmr, PVLB_13270 deletion plasmid	This study
DH5α λpir1 pEMG_ko_ndh2	Kmr, PVLB_21880 deletion plasmid	This study
DH5α λpir1 pEMG_ko_nuo	Kmr, PVLB_15600-15660 deletion plasmid	This study
Plasmids
pS2311·Nox	Kmr; derivative of vector pSEVA2311 with the nox gene from S. pneumoniae	(36)
P. taiwanensis
VLB120	Wild type	A. Schmid (UFZ, Leipzig, DE)
VLB120 pSTY−	VLB120 devoid of megaplasmid pSTY	(29)
VLB120 Δndh-1	Δndh-1 (PVLB_13270), pSTY−	This study
VLB120 Δndh-2	Δndh-2 (PVLB_21880), pSTY−	This study
VLB120 ΔΔndh	ΔΔndh (PVLB_13270, PVLB_21880), pSTY−	This study
VLB120 Δnuo	Δnuo (PVLB_15600-15660), pSTY−	This study
VLB120 Δnuo Δndh-1	Δndh-1 (PVLB_13270), Δnuo (PVLB_15600-15660), pSTY−	This study
VLB120 pS2311·Nox	Kmr, VLB120 bearing pS2311·Nox
VLB120 ΔΔndh pS2311·Nox	Kmr, ΔΔndh (PVLB_13270, PVLB_21880), pSTY− bearing pS2311·Nox	This study

^aGm^r, gentamicin resistance; Km^r, kanamycin resistance.

Batch-flask experiments were performed in 50 ml medium in 500-ml flasks under oxic conditions on a horizontal rotary shaker with a throw of 50 mm and frequency of 300 rpm. E. coli was grown at 37°C, Pseudomonas at 30°C. The chemicals used in this work were obtained from Carl Roth (Karlsruhe, Germany), Sigma-Aldrich (St. Louis, MO, USA), or Merck (Darmstadt, Germany) unless stated otherwise. The main cultures were inoculated from liquid precultures to an approximate OD_600nm of 0.05. All experiments were performed in biological triplicates unless stated otherwise.

Plasmid cloning and generation of deletion strains.

Genomic DNA of P. taiwanensis VLB120 was isolated using the High Pure PCR template preparation kit (Hoffmann-La-Roche, Basel, Switzerland). Upstream (TS1) and downstream (TS2) regions with a length of 400 to 800 bp flanking the specific target gene were amplified using Q5 high-fidelity polymerase (New England Biolabs, Ipswich, MA, USA). Primers were ordered as unmodified DNA oligonucleotides from Eurofins Genomics (Ebersberg, Germany) and are listed in Table 5. The suicide delivery vector pEMG was isolated using the NEB Monarch plasmid miniprep kit (New England Biolabs, Ipswich, MA, USA). The isolated plasmid was digested with restriction enzymes purchased from New England Biolabs (Ipswich, MA, USA). For plasmid construction, Gibson Assembly using NEB Builder Hifi DNA assembly (New England Biolabs, Ipswich, MA, USA) was used. Plasmids were transformed into electrocompetent E. coli DH5α λpir1 via electroporation (62). Transformants and chromosomally engineered Pseudomonas were screened by colony PCR using OneTaq 2× Master Mix (New England Biolabs, Ipswich, MA, USA). The cell material was lysed in alkaline polyethylene glycol for enhanced colony PCR efficiency as described previously (63).

TABLE 5.

Primers used in this study

Primer	Sequence (5′–3′)a	Description
SN019	gataacagggtaatctgCGCAGGATGAAAGCTAAACC	TS1 ndh-1 forward
SN020	aacagccaTGAGTCGTTCGAATAACTAC	TS1 ndh-1 reverse
SN021	acgactcaTGGCTGTTCAGCAATGATGG	TS2 ndh-1 forward
SN022	cgggtaccgagctcgGCAAGGGCGAGCATGATGAC	TS2 ndh-1 reverse
SN023	gataacagggtaatctgTCGACCTCAACACGCACTTC	TS1 ndh-2 forward
SN024	ccggcaagCGATGCGATGAGTCATGG	TS1 ndh-2 reverse
SN025	tcgcatcgCTTGCCGGCGATAAAGCCAG	TS2 ndh-2 forward
SN026	cgggtaccgagctcgCGGCACTCCCAGATAACTTG	TS2 ndh-2 reverse
SN027	ATACGGGCCGTTCATCAGTC	Verification cointegration ndh-1 forward
SN028	GCGATCTTGCGAATGGTGTC	Verification cointegration ndh-1 reverse
SN029	CCGGCTGAATGACGAATG	Verification cointegration ndh-2 forward
SN030	GTTACGACCCGGTGTATG	Verification cointegration ndh-2 reverse
SN112	tcagataactTTAAGTAGAGCGGCGACTTG	TS1 nuo reverse
SN113	agggataacagggtaatctgCCTTATCGCCGCCGAATCAC	TS1 nuo forward
SN114	ctctacttaaAGTTATCTGAACGGGCTTGG	TS2 nuo forward
SN115	atccccgggtaccgagctcgGCGCTCCAGTTGGTGGATTC	TS2 nuo reverse
SN116	CTCGTCCAAGCCACCTGATG	Verification cointegration nuo forward
SN117	AGCCTCAAGGTCATGGTCTG	Verification cointegration nuo reverse
SN171	CGGACACAGACCATGCATAC	Verification cointegration, binding in nuoA
SN200	CTGCACACCTATGCCTACAA	qPCR ndh-2 forward
SN201	TACAGCGACACATAGAACATCC	qPCR ndh-2 reverse
SN214	TTGGCCCAGAGGAAATCAC	qPCR rpoB forward
SN215	GGCACCGACGTAGACAATAC	qPCR rpoB reverse
SN234	AGAACGAACCCTTCGAATCC	qPCR nuoA forward
SN235	GCATCGCGACCAGATAGAAT	qPCR nuoA reverse
SN228	CGAATACGTCGCTAGCCATAC	qPCR ndh-1 forward
SN229	ATCACTTTCAGGTGCTCGTC	qPCR ndh-1 reverse

^aUnderlined nucleotides refer to gene-specific regions, and lowercase letters indicate overhangs. The primer efficiencies of the qPCR primer pairs for ndh-2 (SN200, SN201), rpoB (SN214, SN215), nuoA (SN234, SN235), and ndh-1 (SN228 and SN229) were 104.4%, 101.8%, 104.4%, and 103.4%, respectively.

Targeted gene deletions were performed using the I-SceI-based system developed by Martinez-Garcia and de Lorenzo (27). The conjugational transfer of the mobilizable knockout plasmids from E. coli DH5α λpir1 to Pseudomonas was performed via triparental patch mating (16). After conjugation, the pSW-2 plasmid encoding the I-SceI endonuclease was conjugated into Pseudomonas cointegrates. The addition of 3-methylbenzoate for the induction of I-SceI expression was omitted, as the basal expression level was sufficient. Kanamycin-selective clones were directly isolated, positive clones were cured of pSW-2 and restreaked several times. The gene deletion was confirmed by colony PCR and Sanger sequencing. No complementation studies were performed because it has been shown for the I-SceI scarless gene deletion method we used that the double-strand breaks do not result in random mutations. Moreover, the method does not introduce foreign DNA material, making side effects very unlikely.

Analytical methods.

The optical density of cell suspensions was measured at a wavelength of 600 nm using an Ultrospec 10 spectrophotometer (GE Healthcare, Chicago, IL, USA). The cell dry weight (CDW) was calculated by multiplying OD_600nm with a gravimetrically determined correlation factor of 0.39. For HPLC analysis the samples were centrifuged at 17,000 × g for 5 min and the supernatant was stored at –20°C until further analysis.

Glucose and gluconate concentrations were measured by high-performance liquid chromatography using a Beckman System Gold 126 solvent module equipped with a System Gold 166 UV-detector (Beckman Coulter) and a Smartline RI detector 2300 (KNAUER Wissenschaftliche Geräte, Berlin, Germany). Analytes were separated on the organic resin column Metab AAC (ISERA, Düren, Germany) and eluted with 5 mM H₂SO₄ at an isocratic flow of 0.6 ml min⁻¹ at 40°C for 20 min. Glucose and gluconate were analyzed using the RI detector, whereas gluconate was determined with the UV detector at a wavelength of 210 nm.

Proteomic profiling of NADH dehydrogenase mutants.

Samples for proteome profiling were taken during early, mid-, and late exponential growth at an OD_600nm of 0.5, 2.5, and after depletion of glucose, respectively, and checked with test strips for rapid detection of glucose (Medi-Test; Macherey-Nagel, Düren, Germany). Proteins were extracted from cell biomass and subsequently prepared for shotgun proteomic experiments as described previously (64). All samples were analyzed on an Agilent 6550 iFunnel Q-TOF mass spectrometer (Agilent Technologies, Santa Clara, CA) coupled to an Agilent 1290 UHPLC system. Peptides (20 μg) were separated on a Sigma–Aldrich Ascentis Peptides ES-C18 column (2.1 mm × 100 mm, 2.7 μm particle size, operated at 60°C) at a 0.400 ml min⁻¹ flow rate and eluted with a gradient as follows. The initial condition was 95% solvent A (0.1% formic acid) and 5% solvent B (99.9% acetonitrile, 0.1% formic acid). Solvent B was increased to 35% over 120 min, and then increased to 50% over 5 min, then up to 90% over 1 min, and held for 7 min at a flow rate of 0.6 ml min⁻¹, followed by a ramp back down to 5% B over 1 min, where it was held for 6 min to reequilibrate the column to original conditions. Peptides were introduced to the mass spectrometer from the LC by using a Jet Stream source (Agilent Technologies) operating in positive-ion mode (3,500 V). Source parameters employed gas temperature (250°C), drying gas (14 liters min⁻¹), nebulizer (35 psig), sheath gas temperature (250°C), sheath gas flow (11 liters min⁻¹), VCap (3,500 V), fragmentor (180 V), and OCT 1 RF Vpp (750 V). The data were acquired with Agilent MassHunter Workstation Software, LC/MS Data Acquisition B.06.01 operating in Auto MS/MS mode, whereby the 20 most intense ions (charge states, 2 to 5) within 300 to 1,400 m/z mass range above a threshold of 1,500 counts were selected for tandem mass spectrometry (MS-MS) analysis. MS-MS spectra (100 to 1,700 m/z) were collected with the quadrupole set to “medium” resolution and were acquired until 45,000 total counts were collected or for a maximum accumulation time of 333 ms. Former parent ions were excluded for 0.1 min following MS-MS acquisition. The acquired data were exported as mgf files and searched against the pan proteome that is highly related to Pseudomonas taiwanensis VLB120 with Mascot search engine version 2.3.02 (Matrix Science). The resulting search results were filtered and analyzed by Scaffold v 4.3.0 (Proteome Software Inc.). The normalized spectral counts of each sample were exported from Scaffold, and the relative quantity changes of identified proteins in mutant samples were calculated in comparison to the wild-type sample. The statistical significance of these changes and the adjusted P values were evaluated by limma R package. (See below for data availability.)

RNA preparation and analysis.

Samples for transcription analysis were taken during early, mid- and late exponential growth at an OD_600nm of approximately 0.5, approximately 2.5, and after glucose depletion, respectively. Prior to RNA isolation, the culture sample was diluted with the DNA/RNA protection reagent of the Monarch Total RNA miniprep kit (New England Biolabs, Ipswich, MA, USA), followed by mechanical lysis with ZR BashingBead lysis tube (0.5 mm) (Zymo Research, Irvine, CA, USA) for 1 min using the Mini-Beadbeater-16 (Biospec, Bartlesville, OK, USA). After a centrifugation step at 16,000 × g for 2 min the supernatant was transferred into a new tube. An equal volume of RNA lysis buffer of the Monarch Total RNA miniprep kit was added and the RNA isolation was continued as described in the supplier’s manual. After the last elution step, an additional in-tube DNase treatment was done using RNase-free DNase I (New England Biolabs, Ipswich, MA, USA). The final RNA yield and purity were evaluated by the absorption ratio A₂₆₀/A₂₈₀ measured with a Nanodrop (Thermo Scientific, Rockford, IL, USA). The synthesis of cDNA for reverse transcription was carried out with a Protoscript II first strand cDNA synthesis kit (New England Biolabs, Ipswich, MA, USA) using 120 ng total RNA and 60 μM random hexamers. The qPCR analyses were conducted with 5 μl of the reverse transcription reaction mixture with gene-specific primers (Table 5) and the Luna Universal qPCR Master Mix (New England Biolabs, Ipswich, MA, USA) was used. Primers for qPCR were designed with the PrimerQuest Tool of IDT technologies. Gene expression levels for each individual sample were normalized relative to the internal reference gene rpoB and the wild type in the corresponding growth phase calculated by a mathematical method based on the calculated real-time PCR efficiencies (65). The qPCR was performed with the CFX96 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). All qPCRs were performed in triplicates.

Inverted membrane vesicle preparation and NADH oxidation activity.

Cultures were harvested at early exponential growth phase at an optical density (OD_600nm) of approximately 0.5, as well as in the late exponential growth phase (OD_600nm of 3 to 4). Inverted membrane vesicles were prepared as described by Borisov (66). Briefly, cells were centrifuged for 8 min at 5,000 × g and resuspended in 2 ml spheroplast buffer (200 mM Tris-HCl pH 8.0, 2 mM EDTA, 30% sucrose), centrifuged again, and resuspended in 1 ml spheroplast buffer. Spheroplasts were prepared using lysozyme (0.03 g) and incubated for 30 min at room temperature. Spheroplasts were centrifuged for 10 min at 5,000 × g and resuspended in 2 ml sonication buffer (100 mM HEPES-KOH pH 7.5, 50 mM K₂SO₄, 10 mM MgSO₄, 2 mM DTT, 0.5 mM PMSF). The vesicles were sonicated (Bioruptor, Diadenode, Belgium) in 4 cycles of 30 s at high intensity with an intermediate pause of 30 s in ice water. The inverted membrane vesicles were centrifuged twice for 10 min at 5,000 × g to remove cell debris. The supernatant was centrifuged for 30 min at 120,000 × g and the resulting pellet was resuspended in the assay buffer (25 mM HEPES, 25 mM BIS-TRIS propane pH 7, 10 mM MgSO₄).

The freshly prepared inverted membrane vesicles were immediately used for the determination of the NADH oxidation activity, as we observed a rapid activity decline when the membrane samples were stored on ice. 150 μl ml⁻¹ isolated membrane fractions were added to the assay buffer, and the reaction was initiated by the addition of 125 μM NADH. The total volume of the assay was 200 μl. The NADH oxidation was monitored over 30 min at 340 nm in a Synergy MX microplate reader (BioTek, Winooski, VT, USA). For calculating the specific enzyme activity, we used the NADH molar extinction coefficient ε_NADH = 6.22 mM⁻¹cm⁻¹, where one unit of activity was the quantity that catalyzed the oxidation of 1 μmol of NADH per min. The protein concentration was measured with the reducing agent compatible Pierce BCA Protein assay kit (Thermo Scientific, Rockford, IL, USA).

Respiration activity monitoring.

The cultivations and measurements of the oxygen transfer rate (OTR) and the carbon dioxide transfer rate (CTR) were performed in a modified RAMOS System, developed by the chair of Biochemical Engineering (RWTH Aachen University) (33, 67). The standard RAMOS for shake flasks is commercially available from the Kühner AG (Birsfelden, Switzerland) or HiTec Zang GmbH (Herzogenrath, Germany). All cultivations were performed in 250-ml Ramos flasks with 10% (vol/vol) filling volume using MSM medium supplemented with 25 mM glucose. The cultures were inoculated from liquid precultures to an approximate OD_600nm of 0.05. The OTR and CTR were measured thrice per hour. All experiments were performed in biological duplicates.

Redox cofactor quantification.

Samples for redox cofactor analysis were collected from early, mid-, and late exponential growth phase at OD_600nm of approximately 0.6, 2.2, and 4.0, respectively. The samples were rapidly transferred into 15-ml Falcon tubes containing 5 ml of quenching solution (acetonitrile:methanol:water, 40:40:20, vol/vol) with ¹³C-labeled cell extracts at –40°C. After three freeze-thaw cycles, the samples were centrifuged at 13,000 × g for 5 min and concentrated by evaporating the quenching solvent using a vacuum concentrator (SAVANT SpeedVac; Thermo Fisher Scientific, San Diego, CA, USA) for 5 h followed by lyophilization (LABCONCO; FreeZone, Kansas City, MO, USA). All dried extracts were stored at −80°C until analysis or resuspended in LC-MS grade water for LC-MS analysis. All redox cofactor metabolites were measured on an AB SCIEX Qtrap1 5500 mass spectrometer (AB SCIEX, Framingham, MA, USA) operated in negative ion and selected multiple reaction monitoring (MRM) mode. The column XSELECT HSS XP (150 mm × 2.1 mm × 2.5 μm) (Waters, Milford, MA, USA) with ion-pairing technique was used for the chromatography separation as previously described (68). Peak integration and metabolite quantification were performed using an isotope-ratio-based approach on Multi-Quant 3.0.2 (AB SCIEX) software as previously described (69, 70).

ROS assay.

The ROS-sensitive fluorescent dye 2’,7’-dichlorodihydrofluorescein diacetate (H₂-DCFA, Sigma-Aldrich, St. Louis, MO, USA) was used to monitor the ROS formation in P. taiwanensis VLB120. Cells were harvested at multiple time points during the growth experiment and centrifuged for 5 min at 8,000 × g. The pellet was washed once in the mineral salt medium described above but lacking nitrogen (MSM−N) and centrifuged again. The cells were resuspended to an optical density of OD_600nm of ∼0.4 in 1 ml MSM−N but supplemented with 25 mM glucose to allow respiratory activity. The ROS-sensitive H₂-DCFA dye was added to the cells to a final concentration of 20 μM. The cells were incubated in the dark for 30 min at 30°C at 250 rpm for aeration. After the incubation, the cells were immediately analyzed in the microplate reader Synergy MX (BioTek, Winooski, VT, USA). The oxidized fluorescent DCF was measured using Ex/Em 504/529 nm. The measurement was performed in biological and technical triplicates. As a positive control, 1 mM H₂O₂ was included in all experiments and treated in the same way as the biological samples (data not shown).

Data availability.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (71) partner repository with the data set identifier PXD013623.