Secreted Flavin Cofactors for Anaerobic Respiration of Fumarate and Urocanate by Shewanella oneidensis: Cost and Role

Shewanella species are prevalent in marine and aquatic environments, throughout stratified water columns, in mineral-rich sediments, and in association with multicellular marine and aquatic organisms. The diversity of niches shewanellae can occupy are due largely to their respiratory versatility. Shewanella oneidensis is a model organism for dissimilatory metal reduction and can respire a diverse array of organic and inorganic compounds, including dissolved and solid metal oxides. The fumarate reductase FccA is a highly abundant multifunctional periplasmic protein that acts to bridge the periplasm and temporarily store electrons in a variety of respiratory nodes, including metal, nitrate, and dimethyl sulfoxide respiration. However, maturation of this central protein, particularly flavin cofactor acquisition, is poorly understood. Here, we quantify the fitness cost of flavin secretion and describe how free flavins are acquired by FccA and a homologous periplasmic flavoprotein, UrdA.

ABSTRACT

Shewanella oneidensis strain MR-1, a facultative anaerobe and model organism for dissimilatory metal reduction, uses a periplasmic flavocytochrome, FccA, both as a terminal fumarate reductase and as a periplasmic electron transfer hub for extracellular respiration of a variety of substrates. It is currently unclear how maturation of FccA and other periplasmic flavoproteins is achieved, specifically in the context of flavin cofactor loading, and the fitness cost of flavin secretion has not been quantified. We demonstrate that deletion of the inner membrane flavin adenine dinucleotide (FAD) exporter Bfe results in a 23% slower growth rate than that of the wild type during fumarate respiration and an 80 to 90% loss in fumarate reductase activity. Exogenous flavin supplementation does not restore FccA activity in a Δbfe mutant unless the gene encoding the periplasmic FAD hydrolase UshA is also deleted. We demonstrate that the small Bfe-independent pool of FccA is sufficient for anaerobic growth with fumarate. Strains lacking Bfe were unable to grow using urocanate as the sole electron acceptor, which relies on the periplasmic flavoprotein UrdA. We show that periplasmic flavoprotein maturation occurs in careful balance with periplasmic FAD hydrolysis, and that the current model for periplasmic flavin cofactor loading must account for a Bfe-independent mechanism for flavin transport. Finally, we determine that the metabolic burden of flavin secretion is not significant during growth with flavin-independent anaerobic electron acceptors. Our work helps frame the physiological motivations that drove evolution of flavin secretion by Shewanella.

IMPORTANCE Shewanella species are prevalent in marine and aquatic environments, throughout stratified water columns, in mineral-rich sediments, and in association with multicellular marine and aquatic organisms. The diversity of niches shewanellae can occupy are due largely to their respiratory versatility. Shewanella oneidensis is a model organism for dissimilatory metal reduction and can respire a diverse array of organic and inorganic compounds, including dissolved and solid metal oxides. The fumarate reductase FccA is a highly abundant multifunctional periplasmic protein that acts to bridge the periplasm and temporarily store electrons in a variety of respiratory nodes, including metal, nitrate, and dimethyl sulfoxide respiration. However, maturation of this central protein, particularly flavin cofactor acquisition, is poorly understood. Here, we quantify the fitness cost of flavin secretion and describe how free flavins are acquired by FccA and a homologous periplasmic flavoprotein, UrdA.

INTRODUCTION

Shewanella oneidensis strain MR-1 is a gammaproteobacterium found in a variety of aquatic and marine environments, ranging from sediments to aquatic multicellular organisms (1, 2). As a model organism for dissimilatory metal-reducing bacteria, S. oneidensis is capable of respiring a wide variety of organic and inorganic compounds, including nitrate, sulfate, trimethylamine N-oxide (TMAO), dimethyl sulfoxide (DMSO), urocanate, fumarate, and both solid and dissolved metal oxides (2–5). Central in the reduction of many of these terminal electron acceptors, including nitrate, DMSO, and metals, are two periplasmic proteins: a flavocytochrome and fumarate reductase, FccA (SO_0970), and a small tetraheme cytochrome (STC) encoded by the gene cctA (SO_2727). Both STC and FccA receive electrons from the inner membrane tetraheme cytochrome CymA and donate electrons to the MtrABC metal respiration complex in the outer membrane, forming a periplasmic electron transfer hub (6, 7). FccA shares 59% amino acid sequence identity with the well-characterized flavocytochrome c fumarate reductase (Fcc₃) of S. frigidimarina NCIMB400 and is highly abundant in the periplasm of both organisms (7–10). FccA requires a noncovalent flavin adenine dinucleotide (FAD) cofactor to function (8, 10, 11), which is produced intracellularly and transported through the inner membrane by exporter Bfe (12). It is unclear whether FccA acquires its cofactor due to Bfe-driven FAD transport or if other processes lead to its flavin cofactor acquisition.

While FAD is the primary flavin secreted by S. oneidensis (13), flavin mononucleotide (FMN) and riboflavin are predominantly found in the extracellular space of Shewanella cultures, ranging in concentrations between 250 nM and 1 μM (14–16). At these concentrations, extracellular flavins act as electron shuttles to mediate reduction of solid substrates, such as metal oxides and electrodes, and additional supplementation enhances reduction rates (12, 14–19). Additionally, extracellular flavins have the potential to play a role as cofactors for outer membrane cytochromes OmcA and MtrC, accelerating reduction of mineral surfaces via direct interaction (20–24). Although FAD can also function as an electron shuttle, it is efficiently cleaved in the periplasm to FMN and AMP by the metallophosphoesterase UshA, which further enables growth of S. oneidensis with AMP as its sole carbon source through hydrolysis of AMP to adenosine and inorganic phosphate (13). Periplasmic UshA activity presents a potential problem for cofactor binding to FccA, which, as a CxxCH motif-bearing flavocytochrome (11), enters the periplasm in an unfolded state through the Sec system and is processed by the CcmABCDEFGH complex (25), presumably before acquiring noncovalently bound FAD. Another periplasmic flavoprotein, the urocanate reductase UrdA, is homologous to FccA in its FAD-binding domain (5). UrdA is predicted to be Sec secreted (26), indicating it enters the periplasm in an unfolded state, as is the case with FccA. Unlike FccA, however, UrdA does not bind heme and contains a covalent FMN-binding motif in addition to its noncovalent FAD-binding domain (5).

At present, the processes behind cofactor loading and maturation of either FccA or UrdA, especially the factors that balance flavinylation with UshA activity, are not well understood. Here, we lay a basis for the metabolic costs of flavin secretion and the cellular conditions required for free acquisition by FccA and UrdA. We probe flavin secretion requirements for both growth and enzymatic activity in fumarate and urocanate respiration. Finally, we provide insights for further elucidation of specific mechanisms behind periplasmic flavoprotein maturation in S. oneidensis.

RESULTS AND DISCUSSION

Metabolic burden of FAD secretion.

To determine the metabolic cost of flavin secretion in S. oneidensis, we performed competition assays for fitness of a flavin-nonsecreting Δbfe (ΔSO_0702) deletion mutant (12). When grown in coculture with a genomic green fluorescent protein (GFP)-bearing strain of S. oneidensis (MR-1+gfp strain), using lactate as the carbon source and fumarate as the sole terminal electron acceptor, the Δbfe mutant was outcompeted (Fig. 1A), with a cost to fitness relative to that of MR-1+gfp near 14% (Table 1). The cost to fitness associated with bfe deletion under fumarate-respiring conditions was slightly higher in an inverse experiment in which the Δbfe mutant engineered to produce GFP (Δbfe+gfp strain) was competed against wild-type (WT) MR-1 (Fig. 1B). GFP production itself manifests a fitness cost, as it does in the WT background, under fumarate-respiring conditions (Table 1; see also Fig. S1 in the supplemental material). When grown with TMAO as the sole terminal electron acceptor, the Δbfe mutant displayed near-WT fitness (Fig. 1C), while gfp expression in the Δbfe strain led to an expected fitness cost compared to that of the WT (Fig. 1D), likely due to GFP production. Finally, when grown in competition under nitrate-respiring conditions, the Δbfe mutant had an apparent fitness benefit (Fig. 1E) while the Δbfe+gfp strain had a nearly neutral relative fitness (Fig. 1F). While at first glance this appears to suggest that FAD secretion has a slight metabolic burden under nitrate respiration, we are unable to conclude whether this is the case, as the difference in fitness between the Δbfe and MR-1+gfp strains is no greater than the fitness cost associated with gfp expression under nitrate-respiring conditions (Table 1 and Fig. S1). These results show that the benefits afforded by FAD secretion under fumarate respiration greatly outweigh the potential metabolic costs. Furthermore, TMAO and nitrate competition assay results suggest that the cost of flavin secretion is minimal for S. oneidensis even under anaerobic conditions in which FAD does not perform a role as a cofactor. Similarly, there was not a clear cost or benefit to flavin secretion under aerobic conditions (Fig. S2). Altogether these results led us to the hypothesis that active secretion of FAD into the periplasm is required for maximal function of FccA.

FIG 1.

Anaerobic competition assays. Assays containing 20 mM d,l-lactate as the carbon source with S. oneidensis MR-1 expressing GFP (MR-1+gfp) competed against the Δbfe mutant (A, C, and E) or WT MR-1 competed against the Δbfe+gfp mutant (B, D, and F) on 40 mM fumarate (A and B), 20 mM TMAO (C and D), or 40 mM nitrate (E and F). Symbols and error bars indicate averages and standard errors of the means (SEM) for measured ratios; data are from n = 3 replicate cultures, except for panels A and E, where n = 6.

TABLE 1.

Relative fitness of Δbfe, gfp-expressing Δbfe, and gfp-expressing MR-1 strains versus WT strain

Strain	Relative fitnessa (w)
	Fumarate	TMAO	Nitrate
Δbfe (vs MR-1+gfp)	0.858 ± 0.011b	0.990 ± 0.016	1.039 ± 0.008b
Δbfe+gfp (vs MR-1)	0.839 ± 0.008	0.948 ± 0.016	0.994 ± 0.005
MR-1+gfp (MR-1)	0.954 ± 0.021		0.967 ± 0.011

^aValues are averages ± SEM; all data represent n = 3 replicate cultures, except where noted otherwise.

^bn = 6.

FAD secretion by bfe is required for peak FccA activity.

To quantify the contribution of Bfe-exported FAD to enzymatic function of FccA, methyl viologen (MV) oxidation assays (Fig. 2) were performed using Δbfe, ΔushA (ΔSO_2001), and Δbfe ΔushA deletion mutant strains of S. oneidensis. FAD was separately added to determine whether defects could be rescued by supplementation and whether UshA activity inhibits binding of exogenous flavin by FccA. When FAD was not provided exogenously, both Δbfe and Δbfe ΔushA mutant strains had strong defects for bulk FccA activity (Fig. 2A), with an unsupplemented Δbfe strain displaying only ∼18% FccA activity of the WT level (Table 2). Growth with FAD at 1 μM or higher fully restores FccA activity in a Δbfe ΔushA mutant (Fig. 2B), but a Δbfe mutant is only slightly rescued by exogenous FAD at 10 μM (Fig. 2C), a concentration 5 to 50 times greater than total flavin secretion by Shewanella species under aerobic and anaerobic growth conditions (14–16). These results indicate a greater effect of Bfe-driven FAD secretion on total fumarate reductase activity (82% reduction in FccA activity in Δbfe strain) than we anticipated from fitness scores under fumarate respiration (∼15% cost to fitness for Δbfe strain). Furthermore, the finding that exogenous FAD does not fully restore FccA function in a Δbfe background unless ushA is also deleted suggests that FAD must be derived intracellularly to be used as a periplasmic cofactor, pointing to uncharacterized FAD chaperone activity, which may facilitate binding of Bfe-derived FAD to periplasmic proteins and preempt UshA cleavage.

FIG 2.

Methyl viologen assay for fumarate reductase (FccA) activity. Methyl viologen oxidation coupled to fumarate reduction via FccA in whole cells was measured for WT, Δbfe, ΔushA, and Δbfe ΔushA strains and abiotic controls with no flavin supplementation (A), 1 μM FAD supplementation (B), and 10 μM FAD supplementation (C). Symbols and error bars represent averages and SEM from n = 3 technical replicate assays.

TABLE 2.

Methyl viologen oxidation/fumarate reduction rate per OD of cells for Δbfe strains^a

Strain	Oxidation/reduction rate, MV+ Abs606 (min−1) per cell OD600
	No FAD addition	1 μM FAD	10 μM FAD
WT	−71.94 ± 0.98	−69.78 ± 1.06	−58.38 ± 1.57
Δbfe	−16.58 ± 0.70	−16.14 ± 0.84	−18.90 ± 0.20
ΔushA	−67.79 ± 1.24	−69.43 ± 0.92	−66.60 ± 1.75
Δbfe ΔushA	−19.74 ± 0.31	−64.09 ± 1.35	−75.67 ± 1.25
Abiotic	−3.97 ± 0.18	−4.48 ± 0.36	−3.64 ± 0.04

^aValues are averages ± SEM from n = 3 technical replicate assays. Abs₆₀₆, absorbance at 606 nm.

Bfe is not essential for growth with fumarate.

While Δbfe mutants had a measurable fitness defect and displayed severely diminished FccA activity, Δbfe and Δbfe ΔushA mutants showed relatively minor defects in growth with fumarate as the sole terminal electron acceptor even when no exogenous flavins were added (Fig. 3A and Table 3). Subsequent transfers of strains into fresh medium in an attempt to eliminate carryover of cofactor-loaded FccA did not impact growth (Fig. S3). As with biochemical assays, growth defects were rescued by exogenous FAD supplementation in the Δbfe ΔushA but not the Δbfe strain (Fig. 3B and C and Table 3). While observed growth rate (k) differences between the Δbfe mutant and WT MR-1 (specific growth, k_Δbfe/k_WT, ≈0.77) align well with the observed fitness defect of Δbfe under fumarate respiration (w_Δbfe, 0.86 versus that for the MR-1+gfp strain), they do not match FccA activity by methyl viologen assay. These results suggest that far more FccA is produced in S. oneidensis than is strictly required to respire fumarate, a notion supported by the proposed function of FccA as a transient periplasmic electron transfer hub (6, 7, 27) in addition to its role as a terminal fumarate reductase. Together with biochemical assays, the ability of the Δbfe mutant to grow on fumarate implies that the FccA produced by Δbfe strains retains its FAD cofactor and is lower in abundance than that in the WT.

FIG 3.

Anaerobic growth assay for phenotypic assessment of Δbfe strains during fumarate respiration with flavin additions. Growth of WT, Δbfe, ΔushA, and Δbfe ΔushA strains was measured in anaerobic minimal medium containing 20 mM d,l-lactate and 40 mM fumarate without exogenous flavin supplementation (A) and with 1 μM FAD supplementation (B) or 10 μM FAD supplementation (C). Symbols and error bars represent averages and SEM from n = 3 replicate cultures. See Table 3 for growth rate measurements.

TABLE 3.

Specific growth rates of bfe mutants during fumarate respiration

Strain	Growth rate (h−1) witha:
	No exogenous FAD	1 μM FAD	10 μM FAD
WT	0.586 ± 0.024	0.570 ± 0.009	0.529 ± 0.016
Δbfe	0.451 ± 0.018	0.440 ± 0.010	0.483 ± 0.008
ΔushA	0.568 ± 0.032	0.572 ± 0.033	0.565 ± 0.027
Δbfe ΔushA	0.438 ± 0.013	0.562 ± 0.007	0.583 ± 0.038

^aValues are averages ± SEM from n = 3 replicate cultures.

FccA production and fccA expression in Δbfe strain.

To address questions of FccA production and FAD cofactor retention in the Δbfe strain, we purified FccA from Δbfe and WT backgrounds and characterized it by nuclear magnetic resonance (NMR) spectroscopy. We observed decreased abundance of FccA in the Δbfe strain compared to that of the WT strain by heme stain (Fig. 4A and Table 4 and Fig. S4A), and FccA purified from a Δbfe background (Fig. S4B) is similar to WT-derived FccA by NMR spectrum (Fig. 4B), suggesting that Δbfe strain-derived FccA retains its cofactor. However, a small percentage has a different conformation, given by the small peaks that appear in the NMR spectrum (shown with gray squares), suggesting that a small fraction of the protein is unfolded or has a different conformation state than the wild-type protein.

FIG 4.

(A) Protein gel comparison between Δbfe and WT strains. SDS-PAGE (12%) heme-stained gel loaded with 1 ml each of anaerobic (SBM) cultures of WT, Δbfe, and Δbfe strains grown with 10 μM FAD or 10 μM FMN, lysed with bacterial cell lysis buffer (NZYTech). Heme stain from ΔfccA cells grown aerobically in LB medium at 30°C displays lack of FccA. (B) One-dimensional ¹H NMR spectrum of FccA isolated from Δbfe (blue trace) and wild-type MR-1 (red trace) strains.

TABLE 4.

Densitometry results of heme-stained SDS-PAGE gel^a

Strain	Density	Relative density (%)
WT	14,337.761	100.00
Δbfe	5,397.033	37.64
Δbfe with FAD	6,666.497	46.50
Δbfe with FMN	6,304.983	43.97

^aData are derived from n = 1 gel image shown in Fig. 4A.

One possible explanation for the decreased fitness, growth rate, and fumarate reduction rates observed in the Δbfe compared to the WT strain is that bfe deletion decreases expression of fccA. Similarly, rescue of the Δbfe ΔushA strain by exogenous FAD could be explained by positive regulation of fccA by a hypothetical flavin-sensing mechanism. While we observed decreased abundance of FccA protein in a Δbfe mutant, there were no significant differences from the WT strain in fccA expression as determined by reverse transcription-quantitative PCR (RT-qPCR) in either the Δbfe or Δbfe ΔushA strain with or without exogenous flavin addition (Table 5). A likely explanation for the discrepancy between expression and protein abundance is that FccA is unstable without FAD and is thereby degraded. Notably, when the Δbfe ΔushA mutant was grown without FAD supplementation, its bulk biochemical activity was not rescued by exogenous flavins added to the assay (Fig. S5 and Table S1), altogether indicating that Δbfe mutants and Δbfe ΔushA mutants grown without flavin supplementation have a lower total abundance of FccA than WT cells or Δbfe ΔushA mutants supplemented during growth.

TABLE 5.

FccA expression by RT-qPCR

Strain	2ΔΔCTa (recA, WT)
	SBM	FAD	FMN	RF
Δbfe	1.03 ± 0.17	0.97 ± 0.22	0.75 ± 0.51	1.37 ± 1.02
Δbfe ΔushA	1.26 ± 0.62	0.79 ± 0.34	N/A	1.23 ± 0.09
WT ΔCT (recA)	−7.89 ± 0.37	−8.60 ± 0.50	−8.73 ± 0.47	−7.68 ± 0.69

^aValues for Δbfe and Δbfe ΔushA strains are reported as 2-log threshold cycles (C_T) relative to the WT and normalized to recA for n = 3. WT values are reported as the threshold cycle relative to recA. Data for Δbfe ΔushA strain with FMN supplementation are not available. n = 3. All flavins were at 1 μM concentration. P > 0.05 for all one-sample t tests against 1.0.

Bfe is required for normal growth with urocanate respiration.

To determine whether phenotypes exhibited under fumarate respiration in bfe mutants hold for other respiratory conditions requiring Sec-secreted periplasmic flavoproteins, we tested growth phenotypes under urocanate respiration. As is the case under fumarate respiration, a Δbfe mutant is able to grow with urocanate as the sole terminal electron acceptor, but unlike the case with fumarate respiration, the Δbfe strain exhibits a severe lag phase (Fig. 5) along with a severe growth rate defect (Table 6). Without exogenous FAD supplementation, the Δbfe ΔushA strain also shows a growth defect with urocanate (Fig. 5A), but it is partially rescued by exogenous FAD supplementation at a concentration of 10 μM (Fig. 5B), with a distinct lag phase (approximately 8 h) compared to those of wild-type and ΔushA strains, along with a slight growth defect. Importantly, ΔcymA and ΔurdA mutants failed to grow under these conditions (Fig. 5A and B), even after approximately 30 days of continued incubation (data not shown), indicating that UrdA receives its electrons via CymA and that growth of bfe mutants after extended lag was due to urocanate respiration. Altogether, these results indicate that Bfe is much more important for FAD cofactor acquisition by UrdA than by FccA.

FIG 5.

Anaerobic growth assay for phenotypic assessment of Δbfe, ΔfccA, and ΔcymA strains during urocanate respiration with flavin additions. Growth of WT, Δbfe, ΔushA, Δbfe ΔushA, ΔcymA, and ΔurdA strains was measured in minimal medium containing 10 mM d,l-lactate and 20 mM urocanate without exogenous flavin supplementation (A) and with 10 μM FAD or FMN supplementation (B). Measurements were also conducted with ΔfccA, ΔfccA ΔushA, ΔfccA Δbfe, and ΔfccA Δbfe ΔushA strains without exogenous flavin supplementation (C) and with 10 μM FAD or FMN supplementation (D). Dotted lines at an OD₆₀₀ of 0.025 indicate arbitrarily chosen culture density thresholds for lag-phase comparison. Symbols and error bars represent averages and SEM from n = 3 replicate cultures.

TABLE 6.

Specific growth rates of bfe and fccA mutants during urocanate respiration^a

Strain	Growth rate (h−1) with:
	No exogenous FAD	10 μM FAD
WT	0.502 ± 0.013	0.477 ± 0.020
Δbfe	0.157 ± 0.012	0.087 ± 0.001
ΔushA	0.492 ± 0.008	0.515 ± 0.020
Δbfe ΔushA	0.285 ± 0.020	0.405 ± 0.002
ΔfccA	0.509 ± 0.004	0.471 ± 0.007
Δbfe ΔfccA	0.226 ± 0.011	0.115 ± 0.009
ΔushA ΔfccA	0.507 ± 0.008	0.504 ± 0.006
Δbfe ΔushA ΔfccA	0.299 ± 0.004	0.473 ± 0.015

^aValues are averages ± SEM from n = 3 replicate cultures.

As FccA is highly abundant in the periplasm (7), it is possible that its presence inhibits rescue of the Δbfe ΔushA strain by exogenous FAD on urocanate by binding available FAD before it can be used to flavinylate newly synthesized UrdA. Indeed, a Δbfe ΔushA ΔfccA mutant has a shorter lag phase (approximately 40 h) (Fig. 5C) than a Δbfe ΔushA strain (approximately 65 h) (Fig. 5A). More strikingly, when the Δbfe ΔushA ΔfccA strain was supplied with 10 μM FAD, it did not exhibit a measurable lag (Fig. 5D), and its growth rate was 15% faster than that of a Δbfe ΔushA strain under the same conditions, matching that of the WT (Table 6). Further, with one exception (Δbfe ΔushA strain without FAD supplementation), deletion of fccA in mutant strains exhibiting lag on urocanate resulted in both decreased lag duration and significantly increased growth rate. These observations are consistent with FccA preferentially acquiring FAD under conditions of periplasmic flavin limitation (resulting from the loss of Bfe). Considering this result, it is interesting that the WT does not exhibit a measurable lag phase on urocanate and suggests a direct role for Bfe in periplasmic flavin cofactor loading. The way in which UrdA acquires FAD, either directly from Bfe or through a chaperone, is seemingly not competitively inhibited by the presence of FccA in wild-type cells.

The growth phenotype differences between fumarate and urocanate respiration of Δbfe mutants possibly are due to the additional covalent FMN cofactor requirement for UrdA. Covalent attachment of FMN to UrdA has been reported to occur in the periplasm via flavin transferase ApbE (28). ApbE purified from Vibrio harveyi used FAD as the FMN donor substrate for flavinylation of Na⁺-NQR, and the reaction did not proceed with FMN as the substrate (29), which indicates that cofactor loading of both FAD and FMN onto UrdA in S. oneidensis occurs in a manner that preempts UshA activity. Since ApbE is used for covalent FMN transfer to UrdA (28), the extra steps involved in this process may present a higher barrier for flavin cofactor acquisition than that with FccA.

Conclusions and future perspective.

The inner membrane flavin exporter Bfe is responsible for providing the FAD cofactor for FccA in S. oneidensis MR-1, but approximately 10 to 18% of FccA is able to acquire its cofactor in the absence of Bfe. Furthermore, this small pool of Bfe-independent FccA is sufficient to support robust growth with fumarate as the sole terminal electron acceptor, albeit at a 14% loss in fitness. Thus, S. oneidensis produces substantially more FccA than is needed strictly for fumarate respiration, supporting a proposed electron transfer hub model of FccA whereby it overlaps in function with STC as a temporary capacitive buffer for electrons generated by oxidative metabolism, primarily during extracellular metal reduction but also during nitrate and DMSO respiration (6).

Specific mechanisms through which secreted FAD is acquired by periplasmic flavoproteins before being cleaved by UshA remain to be elucidated. Although ApbE isolated from Vibrio cholerae has been shown to bind FAD with high affinity, it is not clear whether it plays a role in noncovalent FAD acquisition by flavoproteins. Interestingly, transposon insertions in apbE yield an apparent fitness defect in transposon sequencing experiments with fumarate as the sole electron acceptor (30). However, this defect could be due to several potential downstream effects of an apbE knockout, including NADH-dehydrogenase inactivation, and additional apbE homologs are found in the S. oneidensis genome. A model in which Bfe directly interacts with a chaperone and/or flavin transferase, such as ApbE, is a promising prospect but would be incomplete without accounting for the ability of Δbfe mutants to grow on fumarate without flavin supplementation. A secondary mechanism for FAD secretion and/or leakage through the inner membrane has not been uncovered thus far.

This work has implications for biotechnological applications of Shewanella involving respiration of solid substrates, supporting the concept that exogenous flavin supplementation to Shewanella cultures greatly benefits respiratory processes that utilize flavins as electron shuttles, such as metal oxide mineral and electrode surface reduction. Critically, these results show that exogenous flavin supplementation does not significantly aid in periplasmic flavin cofactor acquisition by MR-1. The requirement for S. oneidensis to produce its own secreted flavins as cofactors places it in stark contrast with Listeria monocytogenes, which was recently discovered to have a flavin-based extracellular electron transfer pathway despite possessing no riboflavin biosynthesis pathways, making it dependent on its environment to acquire flavin cofactors (31). Interestingly, L. monocytogenes is able to use flavins as electron shuttles for reduction of iron oxide minerals, and, as part of its extracellular electron pathway, it also possesses an FMN transferase, FmnB, with FAD substrate requirements similar to those of ApbE (31). Finally, flavin secretion does not seem to come at a significant metabolic cost during aerobic, nitrate, or TMAO respiration, while it provides a significant benefit to the reduction of fumarate and solid metal oxides such as ferrihydrite and birnessite (12). Ultimately this work suggests that S. oneidensis is evolutionarily adapted to thrive in flavin-poor environments and that its ability to provide the communal benefit of flavin secretion is borne not just out of their utility as electron shuttles but also out of necessity for filling its own cofactor requirements.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

S. oneidensis strain MR-1 was used as a wild-type control in all experiments and as the parent strain for all gene deletions. Table 7 presents a list of strains, plasmids, and primers used in this study. The MR-1+gfp strain was constructed by insertion of gfpmut3* (32), under the control of the promoter P_A1/04/03 (33), into the neutral attTn7 insertion site downstream of gene glmS in MR-1 (34, 35). Double homologous recombination was used to target the gene insertion as well as gene deletions and has been described previously (36). Inocula for all strains were routinely prepared for anaerobic growth experiments from frozen stock isolated on 1.5% lysogeny broth (LB) agar plates and then grown in successive aerobic overnight cultures in liquid LB and minimal medium. Inocula were transferred by syringe to anaerobic minimal medium after washing and concentrating cells to equal turbidities. Unless otherwise noted, minimal growth medium consisted of Shewanella basal medium (SBM) (19), containing 5 ml/liter of vitamin mix excluding riboflavin (37), 5 ml/liter of trace mineral mix (15), and 0.05% Casamino Acids (Fisher), buffered with 10 mM HEPES and adjusted to pH 7.2 using NaOH. In all growth experiments and overnight cultures in minimal medium, d,l-lactate was supplied as the sole carbon source. In anaerobic cultures, fumarate, urocanate, TMAO, or nitrate was used as the sole terminal electron acceptor where indicated.

TABLE 7.

Strains, plasmids, and primers used in this study

Strain, plasmid, or primer	Description or sequence	Cut site	Source or reference
Strains
JG274	MR-1, wild-type		3
JG686	JG274, ΔfccA (ΔSO_0970)		40
JG1079	JG274, ΔushA (ΔSO_2001)		13
JG1637	MR-1+gfp, JG274, with gfpmut3* under constitutive A1/O4/O3 phage promoter expression at neutral insertion locus downstream of glmS		This study
JG1758	JG274, Δbfe (ΔSO_0702)		12
JG1759	JG1079, ΔushA Δbfe		12
JG2761	JG1637, gfpmut3*, Δbfe		This study
JG4278	JG1758, Δbfe ΔfccA		This study
JG4279	JG1079, ΔushA ΔfccA		This study
JG4280	JG274, ΔurdA (ΔSO_4620)		This study
JG4281	JG1759, ΔushA Δbfe ΔfccA		This study
UQ950	E. coli DH5α λ(pir) used for cloning		36
WM3064	E. coli strain used for conjugation		36
Plasmids
pSMV3	Deletion vector, Kmr-only version of pSMV10, sacB		36
pΔbfe	pSMV3 backbone		12
pΔfccA	pSMV3 backbone		40
pΔurdA	573 bp upstream and 595 bp downstream of SO_4620, including the first 8 and last 10 codons, in pSMV3 backbone at SacI and SpeI restriction sites		This study
pAK1	Homologous region targeting neutral insertion site 7 bp downstream of glmS cloned into pSMV3, with a SpeI recognition sequence at targeted insertion site; see primers ALK23, ALK24, ALK25, and ALK26		This study
pURR25	R6K derivative containing promoter PA1/O4/O3 and gfpmut3*		35
pAK2	Promoter PA1/O4/O3 and gfpmut3* amplified from pURR25 using primers ALK27 and ALK28, cloned into pAK1 at SpeI site		This study
pBAD202::fccAStrep	Directional TOPO expression vector containing Strep-tagged fccA		7
pBAD202::fccA	pBAD202::fccAStrep modified by site-directed mutagenesis kit (NZYTech) to remove Strep tag, using primers Flavo_stop_forw and Flavo_stop_rev		This study
Primers
FccAQF	ACCTGCTGCAATGACTACTG		This study
FccAQR	CTTGAAGATGCAAGCGGTAAAG		This study
RecAQF	AGCTATAGCCGCTGAAATCG		41
RecAQR	CCTCGACATTGTCATCATCG		41
ΔurdA UP F	ACATGAGCTCCTGGACCCAGAACTTTATCTC	SacI	This study
ΔurdA UP R OE	TGCGATACGCGTAACAGCAATACCAATAATGG		This study
ΔurdA DN F OE	GCTGTTACGCGTATCGCAGGACAAGAAG		This study
ΔurdA DN R	TTATACTAGTGCATCACCCGCAACTTTA	SpeI	This study
ALK23	NNNNGGGCCCGGCGGCACGTTATTGGTTA	ApaI	This study
ALK24	NNNNACTAGTACTGGCGGTTTTTTATTGG	SpeI	This study
ALK25	NNNNACTAGTACCGCCAGTTAGGCGGTTT	SpeI	This study
ALK26	NNNNGAGCTCTCCTGATGTCGCGAGCTTCG	SacI	This study
ALK27	NNNNACTAGTGTTCCGCGCACATTTCCCGA	SpeI	This study
ALK28	NNNNACTAGTCGGCAACCGAGCGTTCTGAAC	SpeI	This study
Flavo_stop_forw	CTGCCGCTAAATTCGCTAAAGATAATTAAGCTTGGAGCCACCCGCAGTTCG		This study
Flavo_stop_rev	CGAACTGCGGGTGGCTCCAAGCTTAATTATCTTTAGCGAATTTAGCGGCAG		This study

Competition assays.

Triplicate LB overnight cultures of two strains (GFP labeled and unlabeled) were centrifuged at 8,000 × g and then washed and resuspended in SBM, diluting to an optical density at 600 nm (OD₆₀₀) of 0.1. Equal volumes of each strain were mixed, and then 500 μl of the 50:50 mixture was added to 4.5 ml of minimal growth medium containing 100 mM HEPES in a butyl rubber-stoppered tube to achieved a starting OD₆₀₀ of 0.01. Oxygen was removed via filtered syringe using an N₂ and CO₂ gas mixture. Following 24 h of culture, transfers were made into fresh anoxic medium at an OD₆₀₀ of 0.01. Ratios of each strain were determined by flow cytometry with a FACSCalibur (Becton, Dickinson, Franklin Lakes, NJ) equipped with 488-nm and 640-nm lasers, using the FL1 green detection channel through a 530/30 filter. GFP-producing and non-GFP-producing cells were counted using commercial FlowJo software (Ashland, OR) (see Fig. S6 for example data analysis). Relative fitness was calculated for each culture transfer as w = 1 + ($\text{ln}\frac{A_f}{B_f}-\text{ln}\frac{A_i}{B_i}$)/no. of generations, where A is the mutant and B is gfp-expressing MR-1 in forward experiments or A is the gfp-expressing mutant and B is MR-1 in reverse experiments. Variables f and i denote final strain frequencies after culture and initial strain frequencies at inoculation, respectively. Values were represented as averages across all time points and replicates, using standard errors of the means (SEM) as the measure of variance.

Methyl viologen assay.

Reduced methyl viologen was prepared by passing hydrogen gas through a 10 mM aqueous solution in the presence of a platinum wire catalyst. In a 96-well plate inside an anaerobic chamber (N₂), anaerobic cultures (20 mM lactate, 60 mM fumarate) of WT, Δbfe, ΔushA, and Δbfe ΔushA strains grown overnight with 0 μM, 1 μM, and 10 μM FAD were diluted in SBM (pH 7.2, 100 mM HEPES) to an identical turbidity (OD₆₀₀ of 0.2) and then further diluted 1/50. Reduced methyl viologen reagent was diluted to achieve an absorbance at 606 nm (Abs₆₀₆) of 3.3. Assays were prepared by mixing 100 μl each of 10 mM fumarate in SBM, diluted methyl viologen, and diluted cells. Data were immediately collected by a 96-well plate spectrophotometer. Change in Abs₆₀₆ normalized to cellular concentration in the assay was used as a proxy for MV⁺ oxidation rate by fumarate via FccA and was determined by linear regression.

RT-qPCR.

WT and Δbfe strains were grown anaerobically in minimal medium with fumarate, with and without addition of 1 mM FAD. Cell cultures were collected in mid-exponential growth phase (OD₆₀₀ of 0.260) by mixing 1:1 with RNAprotect bacterial reagent (Qiagen), centrifugation at 4,000 × g for 10 min, supernatant removal, and freezing at −80°C. RNA was purified using an RNeasy minikit (Qiagen) by following manufacturer-recommended protocols for bacterial RNA isolation, including on-column DNase I treatment. Quantitative RT-PCR was performed using an iTaq universal one-step RT-qPCR kit (Bio-Rad) by following manufacturer instructions and using 50 ng of RNA template. Data were analyzed by the 2^ΔΔCT method, with normalizing to recA and wild-type controls.

Purification of FccA.

Δbfe strain-derived FccA was purified as previously described using Δbfe strain-grown cells under aerobic conditions (38). Briefly, the soluble fraction obtained from Δbfe strain growth cells was loaded into a Q-Sepharose column previously equilibrated with 20 mM Tris buffer (pH 7.6). The fraction containing FccA eluted at approximately 150 mM NaCl was concentrated and dialyzed prior to being loaded into another Q-Sepharose column, equilibrated previously with 20 mM Tris buffer (pH 7.6). The FccA fraction was eluted at approximately 150 mM NaCl. This fraction was then loaded, after dialysis, into a hydroxyapatite column preequilibrated with 10 mM potassium phosphate buffer (pH 7.6). Pure FccA was eluted with 100 mM potassium phosphate buffer (pH 7.6). All of the chromatography fractions were analyzed by SDS-PAGE (12% gel) and UV-visible spectroscopy to select those that contain FccA. NMR experiments were performed at 25°C on a Bruker Avance II 500-MHz NMR spectrometer equipped with a 5-mm BBI probe.

Bacterial cell lysis buffer (NZYTech) was used to lyse WT and Δbfe strains in the presence and in the absence of flavins (FAD and FMN) to evaluate FccA production using SDS-PAGE. To this end, anaerobic growth was performed in SBM without flavins and in the presence of 10 μM FAD or 10 μM FMN, using as the preinoculum aerobically grown cells in 1% LB. After 24 h of culture, 1 ml of culture was used for analysis. Densitometry measurements were determined by histogram curve areas for FccA bands within a single gel image using the FIJI package of ImageJ software (39).