Bacteroides fragilis Maintains Concurrent Capability for Anaerobic and Nanaerobic Respiration

ABSTRACT

Bacteroides species can use fumarate and oxygen as terminal electron acceptors during cellular respiration. In the human gut, oxygen diffuses from intestinal epithelial cells supplying “nanaerobic” oxygen levels. Many components of the anaerobic respiratory pathway have been determined, but such analyses have not been performed for nanaerobic respiration. Here, we present genetic, biochemical, enzymatic, and mass spectrometry analyses to elucidate the nanaerobic respiratory pathway in Bacteroides fragilis. Under anaerobic conditions, the transfer of electrons from NADH to the quinone pool has been shown to be contributed by two enzymes, NQR and NDH2. We find that the activity contributed by each under nanaerobic conditions is 77 and 23%, respectively, similar to the activity levels under anaerobic conditions. Using mass spectrometry, we show that the quinone pool also does not differ under these two conditions and consists of a mixture of menaquinone-8 to menaquinone-11, with menaquinone-10 predominant under both conditions. Analysis of fumarate reductase showed that it is synthesized and active under anaerobic and nanaerobic conditions. Previous RNA sequencing data and new transcription reporter assays show that expression of the cytochrome bd oxidase gene does not change under these conditions. Under nanaerobic conditions, we find both increased CydA protein and increased cytochrome bd activity. Reduced-minus-oxidized spectra of membranes showed the presence of heme d when the bacteria were grown in the presence of protoporphyrin IX and iron under both anaerobic and nanaerobic conditions, suggesting that the active oxidase can be assembled with or without oxygen.

IMPORTANCE By performing a comprehensive analysis of nanaerobic respiration in Bacteroides fragilis, we show that this organism maintains capabilities for anaerobic respiration on fumarate and nanaerobic respiration on oxygen simultaneously. The contribution of the two NADH:quinone oxidoreductases and the composition of the quinone pool are the same under both conditions. Fumarate reductase and cytochrome bd are both present, and which of these terminal enzymes is active in electron transfer depends on the availability of the final electron acceptor: fumarate or oxygen. The synthesis of cytochrome bd and fumarate reductase under both conditions serves as an adaptation to an environment with low oxygen concentrations so that the bacteria can maximize energy conservation during fluctuating environmental conditions or occupation of different spatial niches.

INTRODUCTION

Bacteroides species are abundant in the gut microbiota of many human populations (1–5). The lumen of the gut is a largely oxygen-free environment; however, oxygen is present in low “nanaerobic” levels (1,000 to 1,500 ppm or approximately 1 to 2 μmol of O₂/L) near the epithelial cell surface, where it diffuses from host cells (6–9). Bacteroides have been shown to thrive under these nanaerobic conditions using oxygen for aerobic respiration via the cytochrome bd type oxidase (10, 11). All Bacteroides species that have been sequenced have the cyd operon that encode this enzyme complex.

The ability of Bacteroides to respire in strict anaerobic conditions is well established (10, 12). The respiratory chain responsible for anaerobic respiration (NADH to fumarate) involves two NADH dehydrogenases: the Na⁺-pumping NADH:quinone oxidoreductase (NQR) and NDH2, as well as the mobile redox carrier menaquinone, and terminates in fumarate reductase, which catalyzes the conversion of fumarate to succinate. NQR is responsible for approximately 70% of the NADH dehydrogenase activity, while the remaining 30% is contributed by NDH2 (13).

RNA sequencing (RNA-seq) analysis comparing transcript levels in bacteria grown under anaerobic or nanaerobic conditions showed that the change in O₂ concentration from 0 to 1,400 ppm leads to changes in transcription of many genes. Transcription of two genes encoding enzymes related to respiration was increased in nanaerobic conditions: nfrHA (nitrite reductase) (BF638R_0427-0430) and ccp (cytochrome c peroxidase) (BF638R_2374-2376) (11). Thus, the RNA-seq data indicate that Bacteroides fragilis is able to sense changes in oxygen concentration at these low levels and regulate expression accordingly. Transcription of genes encoding the main components of the respiratory chain is not altered over this physiologically relevant range of oxygen concentrations. Since the transcript level does not always provide an accurate assessment of enzyme activity under various conditions, we used a series of biochemical methods to directly defined the aerobic respiratory pathway.

RESULTS AND DISCUSSION

In order to grow large liquid cultures of B. fragilis with a consistent nanaerobic oxygen concentration (1,000 to 1,500 ppm) throughout the culture, we constructed a gas-flow system to pump the chamber atmosphere through the headspace of the culture flask (Fig. 1). Using this system, we showed that the B. fragilis Δfrd mutant, which is unable to respire using fumarate, grows to a very high density under nanaerobic conditions (Fig. 1), higher than previously reported (14). These data show the utility of the system and that it is superior to growth in semiclosed systems that do not allow for rapid gas exchange. The presence of nanaerobic O₂ allows the Δfrd mutant to grow, presumably by using cytochrome bd, with O₂ as the terminal acceptor, but the rate of growth is still significantly reduced compared to the wild type under similar conditions (Fig. 1). This indicates that, even when cells respire on nanaerobic oxygen, fumarate reductase can make a significant contribution to overall respiration. This suggests that cytochrome bd and fumarate reductase are both active under these conditions (see below).

FIG 1.

Nanaerobic growth system. A schematic shows the gas-flow apparatus to supply a constant and continuous concentration of oxygen to large broth cultures. The graph shows growth curves for the B. fragilis wild-type (WT) and Δfrd mutant under anaerobic and nanaerobic conditions. Cells were cultured in BHIS medium at 37°C with constant agitation (100 rpm.) Growth under each condition (in duplicate) was monitored as the change in absorbance at 600 nm.

We then used this system to grow wild type and mutants of respiratory complex genes under nanaerobic conditions and analyzed the composition and activity of the respiratory enzymes for comparison with our earlier results from cells grown anaerobically (13).

NADH:quinone oxidoreductases.

Under anaerobic conditions, two enzymes contribute to the NADH:quinone oxidoreductase (NADH dehydrogenase) activity: NQR and NDH2. These activities can be distinguished experimentally using deamino-NADH, an artificial substrate that is consumed by NQR but not NDH2 (13). We measured coupled NADH oxidation and quinone reduction activity in membranes from cells grown in nanaerobic conditions using NADH and deamino-NADH (Fig. 2A and B). We found that NQR and NDH2 contribute 77 and 23% of the total NADH dehydrogenase activity, respectively. These are values similar to those obtained using membranes from cells grown in anaerobic conditions: 70 and 30%, respectively (13).

FIG 2.

The NADH-to-menaquinone section of the B. fragilis respiratory chain. (A) Scheme representing the NADH to menaquinone section of the B. fragilis respiratory chain. (B) NADH:quinone oxidoreductase activity in membranes of B. fragilis. The rates of NADH oxidation (NADH ox, blue) and menadione (vitamin K₃) reduction (Q red, yellow). Deamino-NADH was used to discriminate the activity of NQR. Mean values (n = 3 to 5 measurements) are presented, and error bars show the standard errors. Data were analyzed by an unpaired t test (P < 0.05) between the different growth conditions for each corresponding activity result. The table shows the relative contributions to the total NADH:quinone oxidoreductase activity of NQR and NDH2 under anaerobic and nanaerobic conditions. (C) Liquid chromatography-mass spectrometry profile of the menaquinones (MK-8 to MK-11) identified by mass spectrometry present in B. fragilis grown anaerobically and nanaerobically in BHIS. The numbers (MK-8, -9, -10, and -11) indicate the numbers of isoprenoid units attached to the naphthoquinone ring.

This result is consistent with our previous RNA-seq analysis that showed transcription of the nqr operon and the ndh2 gene are essentially unchanged between anaerobic and nanaerobic conditions (11). Together, these results strongly indicate that expression of these genes and the activities of the corresponding enzymes are not regulated by changes in oxygen concentration from 0 to 1,500 ppm.

Menaquinones.

The NADH dehydrogenases transfer electrons to the quinone pool, a population of hydrophobic redox carriers in the membrane that shuttle electrons to the downstream components of the respiratory chain. Some facultative aerobes change the composition of their quinone pool in response to changes in oxygen. For example, Escherichia coli switches between two different types of quinone in anaerobic and aerobic growth conditions. When E. coli is grown in atmospheric oxygen, the quinone pool is comprised of ubiquinone-8, which has a one-ring benzoquinone-head group and an eight-unit isoprenoid side chain. In contrast, when E. coli is grown anaerobically, the quinone changes to menaquinone-8 (MK-8) which is a two-ring naphthoquinone with an eight-unit isoprenoid side chain (15, 16). We used mass spectrometry to analyze the composition of the quinone pool in the membranes of B. fragilis cells grown anaerobically and nanaerobically (Fig. 2A and C). B. fragilis lacks the genes for ubiquinone synthesis but encodes all the genes for menaquinone synthesis. We found that unlike E. coli grown anaerobically, the menaquinones present in the membranes of B. fragilis grown anaerobically or nanaerobically are the same with a distribution of different side chain lengths, ranging from 8 to 11 isoprenoid units (MK-8 to MK-11) with MK-10 being the most abundant (Fig. 2A and C) (17). The midpoint potential of a quinone is primarily determined by the nature of the head-group ring(s) and substituents directly attached to these ring(s) with the length of the side chain having little effect on this potential (18). Therefore, the menaquinones with different side chains in B. fragilis are expected to have similar redox potentials. Menaquinone has a lower redox potential than ubiquinone, which may confer an advantage in driving the final step of respiration at low oxygen concentrations (19, 20).

Terminal respiratory enzymes.

The crucial difference between aerobic and anaerobic respiration is in the terminal electron acceptors. For B. fragilis, aerobic respiration requires cytochrome bd, which ends the respiratory chain by transferring electrons from menaquinol to oxygen, producing water. In anaerobic respiration, fumarate reductase transfers electrons from menaquinol to fumarate, producing succinate (14). To determine whether B. fragilis changes the synthesis and/or activities of cytochrome bd and fumarate reductase under nanaerobic and anaerobic conditions, we performed several analyses. First, we measured fumarate reductase activity in membranes isolated from both WT and the Δfrd mutant grown under these two conditions. The result for wild type (WT) showed that fumarate reductase is synthesized and active in both anaerobic and nanaerobic conditions, which is consistent with the observation in the growth experiment (Fig. 1). We also showed that membranes isolated from Δfrd mutant cells grown anaerobically showed significantly less fumarate-dependent NADH consumption compared to the wild type (29%). The activity in the mutant may be due to other electron acceptors in the membrane preparations. However, when the deletion strain was complemented by inserting a single copy of the frd operon into the genome, activity was restored to a level above the original wild-type value (Fig. 3A).

FIG 3.

Analysis of fumarate reductase and cytochrome bd type oxidase activity under anaerobic and nanaerobic conditions. (A) Fumarate reductase activity in membranes from B. fragilis WT, Δfrd mutant, and complemented Δfrd attB2::pfrd strains grown in BHIS anaerobically and nanaerobically. Activity was measured in an anaerobic cuvette as the fumarate-dependent NADH oxidation. The reaction was initiated by the addition of 1 mM fumarate. Nonspecific activity was subtracted from each data point (see Materials and Methods). Measurements were repeated in triplicate and the standard error is shown. Data were analyzed by Dunnett’s test (**, P < 0.01; *, P < 0.05) between the WT strain grown anaerobically and each sample. EZR software was used (44). (B) Cytochrome bd type oxidase activity measured as the oxidation of NADH in the presence of oxygen. Activity measurements were performed on extracts from WT, the Δcyd mutant, and complemented Δcyd attB2::pcyd strain cells grown in BHIS medium supplemented with 1 μM PPIX and 2 μg/mL FeSO₄·7H₂O. Milliunits, 1 unit is defined as the amount of enzyme required to oxidize 1 μmol of NADH in 1 min in the presence of 20 μM O₂. The reported activity values indicate the means ± the standard errors. Measurements were performed in triplicate. Data were analyzed by an unpaired t test (*, P < 0.05) between mutant and WT strains for anaerobic or nanaerobic conditions. Complemented mutants carry the deleted gene as a single copy in the pNBU2 vector integrated into the attB2 site in the genome.

We next performed four different assays to analyze cytochrome bd levels under these two growth conditions. We analyzed (i) promoter activity for the operon encoding cytochrome bd (cyd) (Fig. 4A and B), (ii) membrane protein synthesis using a streptavidin tag (Fig. 4C and D), (iii) oxygen-dependent NADH consumption as an indication of cytochrome bd activity (Fig. 3B), and (iv) visible spectroscopy of hemes in membrane preparations (Fig. 4E).

FIG 4.

Characterization of cytochrome bd. (A) Promoter region of the cyd operon showing the −7 region in purple and the first ATG from BF638R_1908 in blue. In the NanoLuc construct (pcyd-pMM553), the cyd promoter controls expression of the luciferase gene. (B) NanoLuc reporter activity of the cyd promoter from B. fragilis under anaerobic and nanaerobic conditions. For these experiments, cells were cultured in BHIS medium containing hemin. For each measurement, the positive control was the promoter of the housekeeping sigma factor of B. thetaiotaomicron cloned upstream the NanoLuc gene (pMM553), and the negative control was the vector without the luciferase (NanoLuc) gene. Relative luminescence units (RLU) of luciferase have been normalized to the optical density at 600 nm (OD₆₀₀). Error bars show the standard deviations of the means. Data were analyzed by an unpaired t test (*, P < 0.05) between anaerobic and nanaerobic conditions. (C) Schematic of the site of insertion of the sequence coding for the streptavidin tag at the end of the cydA gene in the genome of B. fragilis. (D) Western blot analysis showing the presence of cytochrome bd in cells grown in BHIS medium containing hemin in anaerobic and nanaerobic conditions using antibodies against the streptavidin tag. An alpha-fold model (45) of B. fragilis CydA shows the location of the streptavidin tag at the end of the C terminus. (E) Dithionite reduced-minus-air-oxidized spectra of membranes of E. coli Δcyo and B. fragilis WT and mutant strains. B. fragilis was grown in the presence of PPIX and FeSO₄, except where hemin is indicated. Final membrane protein concentrations and scaling of spectra: E. coli Δcyo (5 mg/mL, X1); WT B. fragilis, anaerobic (5 mg/mL, X10); WT B. fragilis, nanaerobic (5 mg/mL, X10); Δcyd nanaerobic (5 mg/mL, X5); Δfrd nanaerobic with hemin (5 mg/mL X 20); and WT nanaerobic with hemin (30 mg/mL, X1). Spectra were recorded at room temperature.

Our previous RNA-seq analysis comparing B. fragilis grown in anaerobic and nanaerobic conditions showed similar cyd transcript levels in these two conditions (11). To specifically quantify transcription initiation, we used the NanoLuc reporter system and made transcriptional fusions with the promoter upstream of the cyd operon (BF638R_1908) (Fig. 4A, B). Consistent with the RNA-seq results, strong promoter activity with similar levels was found in the two conditions. Next, we performed assays to determine whether the protein product of cydA is produced under both conditions. For this, we incorporated a streptavidin tag at the end of the genomic copy of the cydA gene in the B. fragilis genome. Western blotting showed that the streptavidin tag is present in cells grown anaerobically and nanaerobically, indicating the presence of the CydA subunit of cytochrome bd in both conditions (Fig. 4D).

Cytochrome bd can also be identified on the basis of its visible spectrum. Cytochrome bd has three hemes: a six-coordinate heme (b-595), a five-coordinate heme (b-558), and a five-coordinate Fe-chlorin (“heme d”), where oxygen binds and is reduced to water (21–23). Among heme enzymes, cytochrome bd has a distinctive reduced-minus-oxidized spectrum, with a characteristic peak and trough at 630 and 650 nm, respectively, assigned to heme d (22). As an example of this, we show the spectrum from the membranes from a strain of E. coli that does not have cytochrome bo₃ and constitutively expresses cytochrome bd (24). In this spectrum, the 630- and 650-nm features can be clearly seen (Fig. 4E). Bacteroides cannot synthesize heme, and it must be supplied in the growth medium for maximal growth, usually in the form of hemin, which is a heme B ferric chloride (25–27). When wild-type B. fragilis is grown in medium containing hemin, we observed cytochrome bd activity in the cell extracts, measured as oxygen-dependent consumption of NADH (see below), but we could not detect the characteristic peak and trough at 630 and 650 nm corresponding to heme d in the reduced-minus-oxidized spectra of the cell membranes, likely due to a large contribution of hemin to the spectra (Fig. 4E). However, when the growth medium is supplemented with protoporphyrin IX (PPIX) and FeSO₄ instead of hemin, the characteristic heme d features are observed, most strongly in membranes of cells grown nanaerobically but also in membranes of cells grown anaerobically, indicating that the bacteria can synthesize heme d from these precursors (Fig. 4E). The spectra also show the expected contributions of the hemes b of cytochrome bd, although it must be noted that the heme b in fumarate reductase has a peak at the same wavelength. To further investigate these findings, we analyzed spectra of membranes from a strain of B. fragilis that lacks the cyd operon, which should not contain heme d. This strain is able to grow in anaerobic and nanaerobic conditions, presumably by respiration on fumarate. In contrast to wild-type B. fragilis, this strain shows essentially no oxygen-dependent NADH consumption (see below) and, consistent with this, the reduced-minus-oxidized spectra of membranes of cells grown nanaerobically or anaerobically do not show the characteristic heme d features (Fig. 4E). Interestingly, in the Δfrd mutant the characteristic spectral features of heme d are clearly visible in the reduced-minus-oxidized spectra of membranes from cells grown nanaerobically in the presence of hemin. The contribution of hemin to these spectra Is much smaller than in the case of the wild type (above), and it seems likely that hemin in the membranes of wild-type cells overwhelms the heme d spectral features. We do not know the reason that membranes from wild-type cells retain more hemin than those from the mutants.

Although the characteristic spectral features of heme d appeared only when cells were grown with PPIX and FeSO₄, the activity of cytochrome bd could be measured as oxygen-dependent NADH consumption, independent of whether cells were grown with PPIX and FeSO₄ or hemin (Fig. 3B). We first compared activity from extracts of wild-type cells grown in anaerobic and nanaerobic conditions with PPIX and FeSO₄ and found ~5-fold-higher activity in nanaerobically grown cells. As described above, the Δcyd strain showed essentially no activity, but a complemented strain in which a single copy of the cyd operon was inserted into the genome of the deletion strain gave even higher activity than the wild type. We also measured the activity from extracts of cells grown anaerobically with hemin. The values were slightly lower (1 mU) than for cells grown with PPIX and FeSO₄ under similar conditions (1.5 mU). This compares to a previous report of 5 mU for cells grown anaerobically in the presence of hemin (10). Thus, wild-type B. fragilis grown with hemin, rather than PPIX and FeSO₄, has cytochrome bd activity, but we were unable to resolve the spectral features characteristic of heme d likely because of the presence of excess hemin in cell membrane preparations. When PPIX is added instead of hemin, the spectral signature of heme d is observed in membranes of cells grown nanaerobically or anaerobically. For this to occur, PPIX must be converted to the “heme d” chlorin. It is not known whether this conversion is dependent on enzymatic catalysis, but these results indicate that oxygen is not strictly necessary for this conversion.

While Bacteroides can grow in both anaerobic and nanaerobic environments, we find little evidence that the respiratory chain adapts in response to changes in oxygen concentrations. Cells grown with and without nanaerobic concentrations of oxygen showed no major differences in expression or activity of NADH dehydrogenases, the composition of the quinone pool, cytochrome bd or fumarate reductase. The measured activity of cytochrome bd increases slightly when cells are grown nanaerobically, but even in the presence of oxygen, the corresponding terminal enzyme for anaerobic respiration fumarate reductase is also present and active.

The anaerobic and nanaerobic respiratory pathways from NADH to oxygen and fumarate in B. fragilis are shown schematically in Fig. 5. The split blue/green background of the respiratory chain components illustrates the results of the current work, which show that all components are synthesized in anaerobic (blue) and nanaerobic (green) conditions. This indicates that when growing nanaerobically, the bacteria can maximize energy conservation through respiration by taking advantage of the increased energy gain from aerobic respiration, while continuing to respire anaerobically (see Fig. 1). The fact that more cytochrome bd activity is observed in cells grown nanaerobically (Fig 3B) is slightly puzzling in light of the fact that RNA-seq (11) and promoter activity (Fig 4B) both indicate almost equal levels of expression. One possible explanation is that the assembly of the active enzyme is more efficient in the presence of oxygen.

FIG 5.

Scheme showing the respiratory chain of B. fragilis with the transport of electrons from NADH to oxygen and fumarate. The dual color background highlights our finding that all relevant components are present under both anaerobic (blue) and nanaerobic (green) conditions.

This seeming lack of adaptation to changes in oxygen concentrations may reflect a more fundamental accommodation to the Bacteroides niche in the colon. Our earlier RNA-seq data show that B. fragilis regulates the transcription of some genes related to respiration between anaerobic and nanaerobic conditions, but that the genes of the central respiratory pathway are not under this regulatory regime. A simultaneous capability for aerobic and anaerobic respiration has been reported for other bacteria (28–33). For example, in E. coli, the synthesis of terminal oxidases is strongly regulated by oxygen, but at low oxygen concentrations, genes for both cytochrome bd and fumarate reductase are transcribed (34). Trojan et al. (35) studied the expression of terminal oxidases with different oxygen affinities and energy conservation efficiencies and proposed that in some rapidly changing environments, such as soil, bacteria are better served by not changing the composition of the respiratory chain on the basis of small changes in oxygen concentration. It has recently been recognized that different bacteria employ many different strategies to regulate their energy metabolism in response to changes in available terminal acceptors. For B. fragilis and likely other Bacteroides, the bacteria have evolved to maintain the same central respiratory chain across the range of oxygen concentrations in which they live.

MATERIALS AND METHODS

All primers are listed in Table S1 in the supplemental material.

Bacterial strains and growth conditions.

B. fragilis (TM4000 and derivatives) were grown in brain heart infusion medium supplemented with 0.5% (wt/vol) yeast extract and 5 μg/mL hemin (BHIS) or in basal medium (36). For some experiments, hemin was substituted by 1 μM protoporphyrin IX (PPIX; Sigma-Aldrich) and 2 μg/mL FeSO₄·7H₂O for some of experiments. When required, 5 μg/mL erythromycin, 200 μg/mL gentamicin, or 50 ng/mL anhydrotetracycline (aTC) was added. Cells were grown either anaerobically or nanaerobically (1,000 to 1,500 ppm oxygen [see below]) at 37°C in a vinyl controlled atmosphere glove chamber (Coy Laboratory Products) using a 5% H₂–10% CO₂–85% N₂ gas mixture. Escherichia coli strains were grown in LB-Miller medium with 100 μg/mL ampicillin.

E. coli strain CLY (C43-DE3) (24), which lacks the cyo operon coding for the terminal oxidase cytochrome bo₃, was grown with 50 μg/mL kanamycin. Cells were harvested in the early stationary phase. Membranes were prepared using a procedure similar to B. fragilis but under aerobic conditions (see below).

Nanaerobic growth.

For nanaerobic growth, the chamber atmosphere was maintained at 1,000 to 1,500 ppm oxygen. To ensure equilibration of the atmosphere with the liquid cultures, a gas-flow system was used within the chamber (Fig. 1). An aquarium-type pump was used to create a flow of chamber atmosphere through the headspace of the culture flask. To avoid evaporation of the culture medium, the gas was first humidified using a bubbler bottle with a fritted glass bubbler. Sterility of the culture was maintained using 0.2-μm-pore-size, syringe-type filters at the entrance and exit of the culture flask. The flow rate was monitored using a variable-area flow meter (Dwyer Instruments) and maintained at 0.2 L/min. The gas was passed through a desiccant before being returned to the chamber atmosphere. Cultures were agitated using an orbital shaker at 100 rpm.

Deletion of the frd and cyd operons.

Δfrd and Δcyd mutants were created using the protocol reported previously (13, 37). In each case, the complete operon was deleted. Left and right flanking regions were amplified from B. fragilis TM4000 using Phusion polymerase (NEB) and cloned into BamHI-digested pLGB36 using NEBuilder (NEB) and transformed into E. coli S17λpir. The sequence of the plasmid(s) was verified by whole-plasmid sequencing and transferred to B. fragilis TM4000 by conjugation. Cointegrates were selected on BHIS plates containing gentamicin and erythromycin. Cointegrates were passed without antibiotics and plated to BHIS containing 50 ng/mL aTC to select for double cross-outs. Δfrd and Δcyd mutants were selected by PCR.

Complementation of Δfrd and Δcyd mutants.

The frd and cyd operons, including promoter regions, were cloned into the BamHI site of pNBU2 (38). The resulting constructs were verified by whole-plasmid sequencing and mated into the mutant strains where they integrated into the attB2 site in the genome.

Addition of a streptavidin-tag to the genomic copy of cydA.

A 1 kb PCR product of the C-terminal portion of cydA was cloned into the BamHI site of the pLGB36 plasmid using NEBuilder. A streptavidin tag (Trp-Ser-His-Pro-Gln-Phe-Glu-Lys) was added at the end of the cydA gene using the NEB Q5 mutagenesis kit. The final construct was transformed into the E. coli S17λpir, and the construct was verified by whole-plasmid sequencing. The plasmid was introduced into B. fragilis by conjugation, and the final strain was selected after recombination.

Preparation of inner membranes.

All procedures were carried out anaerobically or nanaerobically. Cells were grown until late exponential phase (for activity measurements) or stationary phase (for spectral analysis) and harvested by centrifugation at 3,800 × g for 30 min at 4°C. Supernatants were discarded and cell pellets were washed with deaerated buffer 1 (40 mM KPi (KH₂PO₄) 5 mM dithiothreitol [DTT; pH 7]). Cells were then resuspended in the same buffer and DNase and phenylmethylsulfonyl fluoride protease inhibitor were added after which cells were disrupting using two passes through a French pressure cell (20,000 lb/in²). Cell debris was removed by centrifugation at 3,800 × g for 30 min at 4°C, and the supernatant was centrifuged at 185,000 × g at 4°C overnight to pellet the membrane fraction. Membranes were washed, resuspended in buffer 1, placed in sealed tubes, and stored at −80°C until use.

Preparation of cell extract for cytochrome bd activity measurements.

All procedures were carried out under anaerobic or nanaerobic conditions. Cells were grown in BHIS supplemented with 1 μM PPIX and 2 μg/mL FeSO₄·7H₂O until late exponential/early stationary phase and harvested by centrifugation at 3,800 × g for 30 min at 4°C. The supernatant was discarded, and the cell pellet was washed and resuspended in deaerated buffer 2 (50 mM KPi, 5 mM MgCl₂·7H₂O, 0.1 M sucrose, 5 mM DTT). Resuspended cells were disrupted by sonication inside the anaerobic chamber using a probe-type sonicator (Misonix S3000) operated with a duty cycle of 4 s on/4 s off and a total time of 8 min. The preparation was then centrifuged at 3,800 × g for 30 min at 4°C to remove cell debris, and the supernatant was used immediately for activity measurements (see below).

Luciferase assay of promoter activity.

Promoter activity was assayed using the luciferase system (NanoLuc; Promega). The promoter region of the cyd operon (Fig. 4) was cloned into pMM553 (39), from which the promoter for the B. thetaiotaomicron housekeeping sigma factor BT1311 was removed. This construct was transferred from E. coli S17λpir to B. fragilis by conjugation, and the correct integration into the attB2 site in the genome was confirmed by PCR. pMM553 and pNBU2 plasmids were transferred into B. fragilis and used as positive and negative controls (40). For the promoter activity assays, cells were grown for 4 h in BHIS medium anaerobically or nanaerobically. Cells were harvested at midexponential phase using a microtube centrifuge at 16,100 × g for 1 min at room temperature, and the supernatant was discarded. Cell pellets were frozen briefly by placing in a −20°C freezer, thawed, and resuspended in 100 μL of Bugbuster protein extraction reagent (Millipore). Samples were incubated for 10 min at room temperature with constant agitation and then centrifuged for 10 min at 10,000 × g at 4°C. Samples were maintained on ice. The NanoLuc system (Promega) was used to measure luminescence as a readout of promoter activity. Luminescence was measured in a Tecan Infinite M1000 Pro plate reader using Thermo Fisher Scientific Nunclon Flat Black 96-well plates at room temperature with an integration time of 1,000 ms and a settle time of 0 ms. Samples were measured in triplicates.

NADH:quinone oxidoreductase activity.

The NADH oxidation and menadione (vitamin K₃) reduction were measured spectrophotometrically in membrane preparations under anaerobic conditions using an anaerobic cuvette apparatus as previously described (13). Briefly, the absorbance changes at 343 nm (ε = 6.22 mM⁻¹cm⁻¹) and 262 nm (ε = 14 mM⁻¹cm⁻¹) were monitored for the oxidation of NADH (or deamino-NADH) and the reduction of menadione, respectively. The absorbance change at 262 nm was then adjusted by subtracting the calculated contributions of the produced NAD⁺ and the consumed NADH. The absorbance of NAD⁺ and NADH at 262 nm (ε = 17.8 mM⁻¹cm⁻¹ and 14.3 mM⁻¹cm⁻¹, respectively) were calculated using the NADH absorbance at 343 nm. The reaction was initiated by the addition of 60 μg/mL membrane protein to a buffer containing 50 mM Tris-HCl, 1 mM EDTA, 100 mM NaCl, 5% (vol/vol) glycerol, 0.05% n-dodecyl β-maltoside (pH 7), 100 μM NADH, and 50 μM menadione at room temperature under an argon atmosphere. Activity measurements were repeated at least three times. Activity values were normalized to protein concentration.

Fumarate reductase activity.

The fumarate-dependent oxidation of NADH was measured spectrophotometrically in membrane preparations using the same anaerobic cuvette apparatus as in the NADH dehydrogenase activity above (30). NADH oxidation was followed at 343 nm (ε = 6.22 mM⁻¹cm⁻¹). Next, 50-μg/mL membranes were added to a deaerated buffer (50 mM KH₂PO₄, 5 mM MgCl₂ [pH 7.6]) and 200 μM NADH, followed by incubation for 2 min at room temperature under an argon atmosphere. The reaction was initiated by the addition of 1 mM fumarate. As a control, buffer 1 (40 mM KPi, 5 mM DTT [pH 7]) was added in place of membranes, and this value was subtracted from the calculated activity. All activity measurements were repeated at least three times. Activity values were normalized to protein concentration.

Cytochrome bd oxidase activity.

The oxygen-dependent NADH oxidation was used to measure cytochrome bd activity using the same anaerobic cuvette apparatus referenced above (13). The reaction was initiated by injection of 100 μL of air-saturated water, resulting in a final oxygen concentration of 20 μM into anaerobic preequilibrated sample containing cell extracts at a protein concentration of 1.65 mg/mL in buffer (50 mM KP_i, 5 mM MgCl₂·7H₂O, 0.1 M sucrose) and 200 μM K₂-NADH. NADH was measured at 343 nm (ε = 6.22 mM⁻¹cm⁻¹). All activity measurements were repeated at least three times. Activity values were normalized to protein concentration.

Menaquinone analysis.

(i) Extraction. Cultures were grown until stationary phase in BHIS medium in both anaerobic and nanaerobic conditions at 37°C. Cells were harvested by centrifugation at 3,800 × g for 30 min at 4°C. Supernatants were discarded, and pellets were frozen until extraction. Pellets were resuspended in 10 mL of di-H₂O, and 1 g of sodium bicarbonate was added. Then, 30 mL of 60:40 methanol-petroleum ether was added to the sample, which was then vortexed and incubated for 2 h with agitation at room temperature (41). Samples were centrifuged at 2,000 × g for 15 min at room temperature. The organic (upper phase) was collected and dried under a nitrogen stream. The aqueous (lower phase) from the original separation was extracted a second time using 30 mL of a methanol-petroleum ether mix for additional 30 min with agitation before centrifugation at 2,000 × g for 30 min. The upper organic phase was separated and dried under a nitrogen stream, and the products of the two extractions were combined and protected from light until analysis.

(ii) Liquid chromatography-mass spectrometry. Samples from the menaquinone extraction discussed above were redissolved in methanol-isopropanol (1:1) and injected using a Agilent 1200 nano-HPLC system (Agilent, Palo Alto, CA) and an Agilent 1200 autosampler. Agilent Microsorb-MV 100-5 C₁₈ column (150 × 4.6 mm) was used to separate components. The injection volume was 8 μL, and the flow rate of the mobile phase was 300 μL/min. The mobile phase consisted of 15 mM ammonium acetate in methanol (solvent A) and isopropanol (solvent B). An LTQ XL Orbitrap mass spectrometer (Thermo, Bremen, Germany) was used as a detector. The mass spectrometer was operated in APCI mode with the detection of negatively charged ions. The detection mass range was m/z 400 to 1,100. Tandem mass spectrometry experiments were performed to confirm component structures using data-dependent acquisition. The most intense peaks in mass spectra were fragmented using higher-energy collisional dissociation with nitrogen as the collision gas. The mass accuracy during all experiments was >4 ppm. The following solvent gradient was used: t (min), %B: 0,50; 5,50; 45,0; 50,0; 51,50; and 60,50. The results are presented (Fig. 3C) in the form of a time-based chromatogram, where only peaks identified by mass spectrometry as menaquinones are shown.

Western blotting to detect the streptavidin tag.

B. fragilis cells containing the streptavidin tag at the C terminus of CydA were grown under anaerobic and nanaerobic conditions. Membrane fractions were prepared as described above. Each lane of the gel was loaded with 30 μg of membrane protein, as measured by a Rapid BCA assay. Samples were run on a 6 M urea/4 to 16% Tricine gel (42). The gel was transferred to a polyvinylidene difluoride membrane using the glycine-methanol transfer system (43) and probed with a streptavidin tag antibody (125 ng/mL), followed by a secondary antibody conjugated with alkaline phosphatase for color development.

Reduced-minus-oxidized visible spectra of membranes.

Membranes were resuspended in 50 mM Tris-HCl (pH 7.0), 1 mM EDTA, 100 mM NaCl, 5% (vol/vol) glycerol, and 1% (vol/vol) n-dodecyl β-maltoside. Spectra were recorded using a Hitachi U-2910 UV-Vis spectrophotometer in 1-cm light path cuvettes. “Oxidized” spectra were recorded from samples in the “air-oxidized” as the prepared state. Reduced spectra were recorded after the addition of a few grains of dithionite. Spectra were normalized to protein concentration. Membranes were isolated from B. fragilis grown in BHIS (nanaerobically or anaerobically) with 1 μM protoporphyrin IX and 2 μg/mL FeSO₄·7H₂O or 5 μg/mL hemin. Membrane samples of the Δcyd and Δfrd strains were solubilized in 2% n-dodecyl β-maltoside for 1 h and centrifuged at 106,000 × g for 30 min at 4°C. The supernatant was used for measurements.

Protein concentration measurements.

Protein concentration of cell extracts and membranes were determined by using a rapid bicinchoninic acid protein determination kit (Thermo Fisher). All samples were measured in triplicate.