|Home | About | Journals | Submit | Contact Us | Français|
Cytochrome bd oxidase operons from more than 50 species of bacteria contain a short gene encoding a small protein that ranges from ~30 to 50 amino acids and is predicted to localize to the cell membrane. Although cytochrome bd oxidases have been studied for more than 70 years, little is known about the role of this small protein, denoted CydX, in oxidase activity. Here we report that Escherichia coli mutants lacking CydX exhibit phenotypes associated with reduced oxidase activity. In addition, cell membrane extracts from ΔcydX mutant strains have reduced oxidase activity in vitro. Consistent with data showing that CydX is required for cytochrome bd oxidase activity, copurification experiments indicate that CydX interacts with the CydAB cytochrome bd oxidase complex. Together, these data support the hypothesis that CydX is a subunit of the CydAB cytochrome bd oxidase complex that is required for complex activity. The results of mutation analysis of CydX suggest that few individual amino acids in the small protein are essential for function, at least in the context of protein overexpression. In addition, the results of analysis of the paralogous small transmembrane protein AppX show that the two proteins could have some overlapping functionality in the cell and that both have the potential to interact with the CydAB complex.
Cytochrome oxidases are heterologous membrane protein complexes found in all domains of life. They perform the last step in the electron transport chain during aerobic respiration and in the process promote the creation of a charge gradient across the membrane that is used to synthesize ATP (1–3). Although cytochrome oxidases share this function, they are a diverse family of proteins (4, 5). Members of the protein family are distinguished by the substrate they recognize for electron transfer (6). Cytochrome c oxidases accept electrons from cytochrome c, whereas the structurally similar cytochrome bo oxidases accept electrons from quinol and related small molecules. Although phylogenetic analysis suggests that cytochrome bo oxidases evolved from a common ancestor with the cytochrome c oxidases (6, 7), another group of quinol oxidases, the cytochrome bd oxidases, appears to be evolutionarily distinct and structurally divergent from other oxidases (8, 9).
It is common for bacterial species to contain multiple cytochrome oxidase operons encoding a combination of cytochrome c, cytochrome bo, and/or cytochrome bd oxidases (10–14). The genome of the bacterium Escherichia coli contains three oxidase operons, one encoding a cytochrome bo oxidase and two encoding cytochrome bd oxidases (15). It is becoming increasingly apparent that the expression of multiple cytochrome oxidases imparts advantages to bacteria in surviving specific growth and stress conditions (16). For example, although cytochrome bd oxidases are typically less efficient at creating the charge gradient for ATP synthesis, bd oxidases have been found to have a higher affinity for oxygen than other cytochrome oxidases (6, 13, 16, 17). It is thought, therefore, that bd oxidases are used by bacteria to continue aerobic respiration under conditions of low oxygen (18). This theory is supported by gene expression data in E. coli, which demonstrate that cytochrome bo oxidase is highly expressed during aerobic growth. In contrast, cytochrome bo oxidase expression is reduced and cytochrome bd oxidase levels increase under low-oxygen conditions (8, 14, 19).
In addition to allowing facultative anaerobes to continue aerobic respiration under low-oxygen conditions, cytochrome bd oxidase complexes also help protect bacteria from cytotoxic agents synthesized by the host immune system. Nitric oxide and reactive oxygen species are compounds released by macrophages to kill engulfed bacteria (20, 21), and cytochrome bd oxidases have been shown to be less sensitive than other cytochrome oxidases to many such compounds (18, 20). Consistent with cytochrome bd oxidase being important for bacterial survival within macrophages, cytochrome bd oxidase genes in Brucella suis, the cause of brucellosis, were shown to be induced upon macrophage engulfment, and were required for bacterial survival in the macrophage phagosomes (22). Together, these data suggest that cytochrome bd oxidases may play multiple roles in bacterial infection, helping bacteria survive periods of oxygen deprivation and exposure to cytotoxic agents released by the host immune system.
Cytochrome bd oxidase operons in E. coli and in bacteria throughout the proteobacteria contain a short gene predicted to encode an ~30- to 50-amino-acid small protein containing a single hydrophobic α-helix (see Fig. S1 in the supplemental material). Although cytochrome bd oxidase complexes have been studied for more than 70 years, little is known about these small proteins. E. coli contains two cytochrome bd oxidase complexes, the cytochrome bd-I oxidase complex encoded by the cydA-cydB-ybgT-ybgE (cydABTE) operon (Fig. 1A) and the cytochrome bd-II oxidase complex encoded by the appC-appB-yccB-appA (appCBBA) operon. The operons are paralogues, and both contain paralogous small proteins of unknown function: YbgT in the cydABTE operon and YccB in the appCBBA operon. In this report, we present data suggesting that the small protein encoded in the E. coli cydABTE operon, YbgT, is a member of the cytochrome bd-I oxidase complex and is required for oxidase activity. In deference to the recent renaming of the Brucella abortus YbgT homologue as CydX, we will use the same nomenclature for the E. coli small protein (23). We also present evidence for potential functional overlap of CydX and its paralogous small protein, YccB, here renamed AppX, and investigate the contributions of specific amino acids to CydX functionality.
All strains, plasmids, and oligonucleotides used in this study are listed in Table 1 and Tables S1 and S2 in the supplemental material, respectively. All strains used were derivatives of the E. coli K-12 strain MG1655. Except where noted, mutant strains were constructed by amplifying a kanamycin cassette from the plasmid pKD4, transforming the recombinase-positive strain NM400, and transducing by P1 into E. coli MG1655 as previously described (24). To make the ΔCydX strain, the ΔybgE::kan strain was used as a template for PCR using mutagenic primers in which the cydX ATG codon had been replaced with the CAA codon, flanked by upstream DNA sequence. For the cydB-His-cydX-cat strain, a chloramphenicol resistance cassette was inserted through homologous recombination behind the cydX gene. This cydB-cydX-cat-ybgE strain was used as a template for mutagenic PCR in which the DNA sequence for a hexahistidine tag was inserted at the 3′ end of the cydB open reading frame (ORF). Using the same strategy, the cydB-His-ΔcydX::kan strain was created by performing mutagenic PCR using the ΔcydX::kan strain as a template, and the cydB-His-cydX-SPA-kan strain (SPA stands for sequential peptide affinity) was created using the cydX-SPA-kan DNA for mutagenic PCR. The ΔlacZ::PcydAB-cydA-6×His-kan strain was constructed by using a modification of the epitope tagging protocol described previously (25). Briefly, a PCR product containing the hexahistidine tag and the kan cassette was generated using the plasmid pSUB7 as the template along with a pair of chimeric primers. A recombinant was confirmed to have cydA tagged with the hexahistidine tag and was subsequently used to generate a PCR product containing cydA-6×His-kan under its native promoter(s) which is flanked by sequences homologous to the upstream and downstream of the lacZ gene respectively. Upon recombineering, the lacZ gene was replaced by the ΔlacZ::PcydAB-cydA-6×His-kan amplicon such that the His-tagged cydA is oriented in the same direction as the lac operon.
To make the CydX and AppX overexpression plasmids, the short genes were amplified by PCR from E. coli MG1655 genomic DNA, and the cydX and appX genes were then cloned into the pBAD24 plasmid. Positive transformants were screened by flanking primers, and plasmids with inserts were confirmed by sequencing. Plasmids expressing mutated CydX proteins were created by amplifying the pBAD24/cydX plasmid by PCR using mutagenic primers in which the mutated bases were flanked on either side by 15 nucleotides (nt) of wild-type sequence. The PCR products were purified and then digested with DpnI (New England Biologicals) to eliminate the wild-type plasmid template DNA. The plasmids in ampicillin-resistant colonies were screened by PCR, and positive plasmids were confirmed by sequencing.
Since ΔcydX strains show a slow-growth phenotype in oxygenated liquid culture and mixed-colony formation on plates grown aerobically, all ΔcydX transgenic strains were created using a modified version of the transformation and storage solution (TSS) protocol of Chung et al. (26). In brief, cells were grown up to an optical density at 600 nm (OD600) of 0.1 to 0.3 and then chilled on ice. An equal volume of cells and 2× TSS stock (LB with 10% [wt/vol] polyethylene glycol, 5% [vol/vol] dimethyl sulfoxide, and 50 mM Mg2+) were then mixed and incubated on ice for 30 min. After incubation, the cells were mixed with plasmid DNA and allowed to chill on ice for at least 30 min. The cells were then plated on LB plus ampicillin (100 μg/ml) and glucose (0.2%) (LB-ampicillin-glucose) plates. Plates were incubated at 30°C under anaerobic conditions to limit the possibility of the ΔcydX mutant strains developing suppressor mutations. Ampicillin-resistant colonies were then streaked to single colonies on LB-ampicillin-glucose plates and grown anaerobically. In all cases, the ΔcydX wild-type and transgenic strains grown under these conditions did not show the mixed-colony phenotype.
Zone of inhibition assays were conducted to test the sensitivity of the strains to β-mercaptoethanol, hydrogen peroxide, and hydroxylamine (Sigma-Aldrich). Before the zone of inhibition assay was conducted, the cultures were diluted in order to normalize all samples to an equivalent OD600 per ml. Assays using strains that did not contain a plasmid were plated on LB, while those strains containing a plasmid were plated on LB-ampicillin plates supplemented with arabinose (0.2%) (LB-ampicillin-arabinose) to induce protein expression from the pBAD24 plasmid. To plate the cells, the cultures were diluted into 3 ml of top agar at 55°C. After the plates had cooled, sterilized filter paper disks were placed in the center of the plate, and a 10-μl aliquot of hydrogen peroxide (30%), hydroxylamine (50% [wt/vol]), or β-mercaptoethanol (12 M) was added to the disk. The bacterial strains on the plates were then allowed to grow overnight at 30°C under aerobic conditions before the zone of inhibition diameters were scored.
The growth rate of ΔcydX and other mutant strains was assayed growing 30-ml cultures in 250-ml Erlenmeyer flasks at 37°C. All cultures were inoculated with an aliquot of overnight culture equivalent to a 1:500 dilution of E. coli MG1655 as determined by OD600. Nontransgenic strains were grown in either LB or LB-arabinose, while transformed strains were grown in LB-ampicillin-arabinose. Culture growth was measured by assaying the OD600 of each strain every 30 min.
For copurification experiments testing the potential interaction between hexahistidine-tagged CydA (CydA-His) or CydB (CydB-His) and CydX-SPA (CydX with a C-terminal SPA tag) or SPA-tagged AcrZ (AcrZ-SPA), cells were aerobically grown at 30°C. For experiments using the AppX-SPA strain, cultures were grown at 30°C under low-oxygen conditions in order to induce AppX-SPA expression (27). For the YneM-SPA strain, cultures were grown aerobically at 30°C in LB, and EDTA and SDS were added to a final concentration of 1 mM and 0.025%, respectively, in order to induce YneM expression (27). After centrifugation, cell pellets were resuspended in lysis buffer (10 mM Tris-HCl, 10 mM MgCl2, 0.5% dodecyl-β-d-maltoside, 10 mM imidazole) and lysed by sonication. Crude lysates were cleared by centrifugation, and protein concentrations were taken of the cleared lysates. Since protein concentrations varied between the cydX-SPA, acrZ-SPA, yneM-SPA, and appX-SPA strains, the protein concentrations of the sample containing CydA-His and the negative control sample lacking CydA-His for each SPA-tagged strain were normalized together for each purification. CydA-His was purified by incubating the lysate with nickel affinity gel (Sigma-Aldrich) overnight at 4°C on an orbital shaker. The gel was then washed multiple times with wash buffer (10 mM Tris-HCl, 10 mM MgCl2, 0.5% dodecyl-β-d-maltoside, 20 mM imidazole). CydA was eluted from the gel by incubating with 1× protein loading buffer (15 mM Tris [pH 6.7], 6.25% glycerol, 0.5% SDS, 0.025% bromophenol blue, 1.25% β-mercaptoethanol) and heating for 10 min at 95°C.
To purify the SPA-tagged CydX, cultures were grown under low-oxygen conditions at 37°C. Cells were harvested at exponential phase and suspended in lysis buffer (10 mM Tris-HCl [pH 8.0], 250 mM NaCl, 10% glycerol, 0.5% β-d-maltoside) prior to being lysed in a cell breaker (27,000 × g). Crude lysates were cleared by centrifugation, and the cleared lysate was incubated in anti-FLAG M2 beads for ~16 to 18 h at 4°C. After the overnight incubation, the samples were centrifuged and washed multiple times with buffer, and protein was eluted from the beads by incubating with 0.5-mg/ml 3× FLAG peptide (Sigma-Aldrich) in M2 buffer (10 mM Tris-Cl [pH 8], 100 mM NaCl, 10% glycerol, 0.1% Triton X-100). The 3× FLAG column eluate was mixed with tobacco etch virus (TEV) protease buffer (50 mM Tris-HCl [pH 8.0], 100 mM NaCl, 0.2 mM EDTA, and 0.1% Triton X-100) and 2 mM CaCl2. This protein solution was incubated with calmodulin Sepharose 4B beads (GE Healthcare Life Sciences) for ~16 h at 4°C. The calmodulin column was then washed multiple times with calmodulin binding protein (CBP) buffer (10 mM Tris-HCl [pH 8.0], 100 mM NaCl, 1 mM magnesium acetate [MgOAc], 2 mM CaCl2, 0.1% Triton X-100). Finally, the proteins bound to the calmodulin column were eluted with 1× protein loading buffer (15 mM Tris [pH 6.7], 6.25% glycerol, 0.5% SDS, 0.025% bromophenol blue, 1.25% β-mercaptoethanol) after incubating at 95°C for 10 min.
To purify the SPA-tagged strains for use in the N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD) assay, CydX-SPA, AcrZ-SPA, and E. coli MG1655 cells were aerobically grown at 30°C. Once harvested, cells were resuspended in buffer (100 mM Tris-HCl [pH 8], 150 mM NaCl, 10 mM EDTA, 0.2% β-d-maltoside, EDTA-free protease inhibitor [Roche]) and were lysed by sonication. Lysates were cleared by centrifugation and then purified on an anti-FLAG M2 (Sigma-Aldrich) column. The columns were washed twice with buffer, and the SPA-tagged proteins were eluted by incubating with buffer containing 0.5-mg/ml 3× FLAG peptide (Sigma-Aldrich).
Samples from protein copurification were run on either 16% Tricine gels or 4 to 20% Tris-glycine precast gels (Invitrogen) as described previously (24). The amount of protein loaded per lane was normalized between all samples in each gel, except for the eluate, which was not diluted before loading on a gel. The amount of protein loaded per lane for each purification and the corresponding control purification was kept constant, enabling direct comparisons between blots. Immunoblot analysis of protein levels of SPA-tagged proteins were conducted as described previously (24). To detect the hexahistidine-tagged CydA and CydB, blots were probed with monoclonal anti-6×His (GenScript) in 2% milk in phosphate-buffered saline with 0.2% Tween 20 (PBS-T) and visualized using SuperSignal West Femto substrate maximum sensitivity substrate (Thermo Scientific). Results were visualized using CL-XPosure film (Thermo Scientific).
To isolate membrane fractions for in vitro oxidase activity assays, 150-ml cultures were grown at 37°C until they reached an OD600 of 1.5 to 2. The cells were harvested by centrifugation, and the cell pellets were stored at −80°C. Prior to cell lysis, samples were resuspended in 0.02 M potassium phosphate buffer (pH 7.5) and transferred to 15-ml centrifuge tubes. The cells were lysed using sonication, and the crude lysates were cleared by centrifugation. Membrane fractions were isolated from the cleared lysate by high-speed centrifugation (120 min at 70,000 × g). Membrane pellets from the ultracentrifugation were homogenized in 0.02 M potassium phosphate (pH 7.5). Protein concentrations of the homogenized membrane samples were determined using a Coomassie protein assay (Thermo Scientific).
Oxidase assays were performed by adding homogenized membrane extracts or purified protein eluate to a solution containing N,N,N′,N′-tetramethyl-p-phenylenediamine and ascorbate (1% TMPD, 0.16 mM ascorbate). Each strain/extract was tested in quadruplicate, and in almost all cases, the protein content in each sample of an individual assay was normalized. The one exception was the eluate from control purification of extract from E. coli MG1655 on an anti-FLAG column. In this case, the protein concentration of the purified eluate was so low that a maximum volume of sample was used instead. Blank tubes receiving only buffer acted as a control to monitor oxidation of the TMPD by air. Oxidation of TMPD was measured spectrophotometrically on a NanoDrop 2000 (Thermo-Fisher Scientific) spectrophotometer. The amount of oxidized TMPD in each sample was assayed using the increase of absorbance at 611 nm (28).
E. coli cells that lack the cytochrome bd-I oxidase genes cydA and/or cydB show a number of cell growth and stress resistance phenotypes (20, 29–31). To test whether CydX is required for cytochrome bd-I oxidase activity, a ΔcydX mutant was assayed for phenotypes observed for the cydA and cydB mutants. We previously showed that while CydX is easily detected during aerobic growth, expression is induced ~4- to 8-fold during growth under low-oxygen conditions (27). In contrast, its paralogue, AppX, is expressed at very low levels during aerobic growth but is induced ~30-fold under low-oxygen conditions (27). Due to the coexpression of both cytochrome bd oxidase complexes under low-oxygen conditions, we focused on investigating potential ΔcydX mutant phenotypes under aerobic conditions. Thus, any phenotypes should be primarily due to a loss of CydAB activity. A ΔcydX mutant was tested for previously documented phenotypes of cydA and cydB mutants, including slow growth in aerobic liquid cultures, colony formation on plates grown under aerobic conditions, and sensitivity to hydrogen peroxide, hydroxylamine, and β-mercaptoethanol. In all phenotypic experiments conducted, the phenotype of ΔcydX cells was identical to the phenotypes of both a ΔcydAB strain and a strain lacking the entire operon (ΔcydABXE). The ΔcydX mutant showed slow growth in liquid culture (Fig. 2A), mixed-colony formation when grown in aerobic conditions (Fig. 2C), and sensitivity to the reductant β-mercaptoethanol (Fig. 2B). None of the strains, including the ΔcydAB mutant and ΔcydABXE mutant, however, showed significant sensitivity to hydrogen peroxide, and only a slight increase in the diameter of the zone of inhibition when exposed to hydroxylamine (data not shown). In addition, a ΔybgE mutant showed wild-type phenotypes in all assays (Fig. 2B and data not shown), suggesting that YbgE may not be required for cytochrome bd-I oxidase activity under the conditions tested. Together with the fact that the ybgE gene shows only limited synteny with the other genes in the operon (data not shown), these data suggest that YbgE may have become a member of the operon relatively recently and may not play a role in oxidase activity in E. coli.
Since the cydX gene is contained within the cydAB operon, there is the possibility that the mutant phenotypes observed for the ΔcydX mutant are due to a polar effect of deleting the cydX gene, and not a result of a lack of CydX. To test this possibility, ΔcydX mutant cells were transformed with a plasmid expressing the CydX protein. CydX expressed in trans was able to restore wild-type growth in liquid culture (Fig. 2A) and resistance to β-mercaptoethanol (Fig. 2B). In addition, a strain containing the cydX gene, but with a mutated start codon (ATG to CAA), also showed sensitivity to β-mercaptoethanol (Fig. 2B). Finally, integration of a sequential peptide affinity (SPA) epitope tag, along with a downstream kanamycin resistance cassette into the cydABXE locus, did not result in β-mercaptoethanol sensitivity (Fig. 2B) or a growth phenotype (data not shown), suggesting that gross changes to the 3′ region of the cydABXE transcript do not affect expression of CydA and CydB. Together, these data suggest that the ΔcydX phenotypes are not due to polar effects of altering the operon transcript and support the hypothesis that the CydX protein itself is required for wild-type levels of cytochrome bd-I oxidase activity.
The observation that all phenotypes of the ΔcydX mutant paralleled those of ΔcydAB and ΔcydABXE mutants suggested cytochrome bd-I oxidase activity is reduced when the small protein is absent. To test this directly, membrane fractions were isolated from aerobically grown cells of E. coli MG1655, ΔcydABXE mutant, and ΔcydX mutant and were assayed for cytochrome bd-I oxidase activity in vitro. Oxidase activity in these membrane fractions was determined by measuring the oxidation of N,N,N′,N′-tetramethyl-p-phenylenediamine (TMPD), a known substrate of cytochrome bd oxidases that shows little oxidation by other oxidases (16). Oxidation of TMPD by membrane fractions was measured visually by observation of the blue color formed when TMPD is oxidized and spectrophotometrically by measuring the change in absorbance at 611 nm. Consistent with the in vivo ΔcydX phenotypes, the level of TMPD oxidation was reduced ~2- to 3-fold in samples of ΔcydX mutant membrane extracts compared to the levels in samples of wild-type extracts (Fig. 3). In addition, expression of the CydX protein in trans was able to restore TMPD oxidation activity in the ΔcydX mutant extracts. Some change in color was observed in the ΔcydABXE mutant; however, the level of oxidation was similar to that seen in a blank sample containing only TMPD, suggesting that this signal reflected autooxidation of the TMPD.
Given the small size of the CydX protein and the location of the cydX gene in the cydABXE operon, we hypothesized that the CydX protein may be an unidentified member of the CydAB complex. To test whether CydX exists as a monomer or interacts with other proteins in vivo, we used a strain in which the chromosomal cydX gene was tagged with the SPA epitope tag at the 3′ end, allowing for synthesis of CydX with a C-terminal SPA tag (CydX-SPA). This construct had been used previously to examine growth- and stress-specific expression of the CydX protein and to confirm that CydX is localized to the E. coli cell membrane (24). Phenotypic analysis of the aerobic growth and β-mercaptoethanol sensitivity of the CydX-SPA strain showed that the tagged strain was phenotypically wild type under all conditions measured (Fig. 2B and data not shown), suggesting that the tagged protein is functional.
Analysis of extracts from cells expressing CydX-SPA using native gel electrophoresis showed that both the CydX-SPA protein and its paralogue AppX-SPA resolved as a higher-molecular-weight band than expected if the proteins exist as monomers (Fig. 4). Heating of the CydX-SPA extract led to a shift in protein migration to a lower band, suggesting that the heating had dissolved a protein complex, allowing the CydX-SPA protein to migrate as a monomer. These data suggested that CydX-SPA exists as part of a high-molecular-weight complex in vivo and that this complex can be disrupted by incubating the cell extract at 95°C (Fig. 4).
To identify proteins that potentially interact with CydX, endogenously expressed CydX-SPA was harvested from cells grown in low oxygen and then sequentially purified on a calmodulin column and by coimmunoprecipitation using anti-3×FLAG M2 beads. Eluates from this purification contained multiple proteins that were not detected in extracts from wild-type cells. Mass spectroscopic analysis of two prominent bands (~64 kDa and ~40 kDa) suggested that these proteins could be the CydA and CydB proteins (data not shown).
To directly test the hypothesis that CydX-SPA interacts with CydA, a strain was created in which the lacZ locus was replaced with a copy of the cydA gene containing a 3′ hexahistidine tag expressed under the control of the cydAB promoter. Phenotypic analysis showed that this strain had wild-type growth in liquid culture, normal growth on plates, and wild-type β-mercaptoethanol resistance (see Fig. S3 in the supplemental material), suggesting that CydAB activity is not perturbed in the strain. A potential interaction between CydA and CydX was tested by purifying CydA-His on a nickel-nitrilotriacetic acid (Ni-NTA) column and assaying for the presence of CydX-SPA in the purification eluate. For a control, extracts from a strain expressing only CydX-SPA were also loaded on a Ni-NTA column. Consistent with an interaction between CydX and the CydA complex, only eluates purified from a strain expressing both CydX-SPA and CydX-His contained detectable levels of CydX-SPA (Fig. 5A), while those isolated from a strain expressing only CydX-SPA did not (Fig. 5A). To confirm that this interaction was not due to either nonspecific hydrophobic interactions between two hydrophobic membrane proteins or to general interactions of CydA-His with the SPA tag, copurification of CydA-His and two other SPA-tagged hydrophobic transmembrane proteins, AcrZ and YneM, was also assayed. In contrast to the copurification results using CydX-SPA, there was no detectable purification of YneM-SPA on the column, independent of the presence of CydA-His in the extract (see Fig. S2 in the supplemental material). For AcrZ-SPA, a slight amount of AcrZ-SPA was detected in the eluate from the Ni-NTA column (Fig. 5B); however, this was also observed in extracts lacking CydA-His, suggesting that the AcrZ-SPA protein has some small level of affinity for the Ni-NTA column (32). Together, these data suggest that the interaction between CydX and CydA may be specific. It is important to note that probing for CydA-His in purified samples using an anti-His antibody showed that the protein was present in multiple low-molecular-weight fragments in these samples. This could reflect the labile nature of CydA, especially when expressed independent of the cydAB operon. It is clear, however, that CydX-SPA does copurify with either the full-length CydA-His protein or with fragments of the CydA-His protein, consistent with an interaction between the endogenous CydA and CydX proteins.
To confirm the potential interaction of CydA and CydX, a reciprocal purification was also performed in which CydX-SPA was tandemly purified on anti-3×FLAG and calmodulin binding protein (CBP) columns, and the purification eluates were assayed for the presence of CydA-His. Consistent with the previous purification results, CydA-His was detected in eluates from the 3× FLAG and CBP columns (Fig. 6) and was not detected in the purification of a control strain lacking CydX-SPA (Fig. 6). These data support a specific interaction between CydX and CydA.
To determine whether CydX also interacts with the CydB protein, a strain was created in which the endogenous copy of cydB is tagged at the 3′ end with a hexahistidine epitope tag at the same time that the endogenous cydX gene is tagged at the 3′ end with the SPA tag. Thus, both epitope-tagged proteins are expressed from their endogenous loci. The CydB-His strain, both with or without the CydX-SPA allele, showed normal growth in aerobic liquid cultures and plates (see Fig. S3A in the supplemental material; also data not shown), suggesting that the CydB-His allele has normal function under these conditions. The cydB-His and cydB-His/cydX-SPA strains showed sensitivity to β-mercaptoethanol (Fig. S3B), suggesting that the addition of the hexahistidine tag has compromised CydB function when cells are under reductive stress. The fact that the CydB-His strain, both with or without the CydX-SPA allele, showed normal growth in aerobic liquid cultures and plates, however, suggests that the CydB-His allele has normal function under these conditions, and we reasoned that any potential interaction between the two proteins should be maintained. CydB-His was purified on a Ni-NTA column, and the eluate from this purification was assayed for the presence of CydX-SPA. The results from the purification experiments showed that CydX-SPA was detected in eluates from the Ni-NTA column (Fig. 7) and was not detected in a control purification from a strain lacking the cydB-His allele (Fig. 7). These data support a specific interaction between CydX and CydB.
As an additional test to determine whether CydX-SPA interacts with a functional cytochrome bd oxidase, CydX-SPA was purified on an anti-FLAG column, and the eluate was assayed for oxidase activity toward TMPD. Compared to eluates from control purifications using E. coli MG1655 and AcrZ-SPA cells, TMPD oxidase activity was detected only in purified CydX-SPA eluate (Fig. 8). Taken together, these results are consistent with the hypothesis that CydX interacts with a functional cytochrome oxidase and that the oxidase is CydAB.
To begin to characterize the sequence specificity for CydX activity, we changed individual amino acids in the CydX sequence to alanine, serine, or glycine and examined the activity of each mutated protein by testing the ability of the mutated protein to complement the β-mercaptoethanol sensitivity of the ΔcydX mutant in trans. In total, 15 amino acids were individually mutated on the CydX overexpression plasmid. The residues that were mutated were selected based on an alignment of 58 CydX homologues that had either been previously identified or were identified by our own bioinformatic search for homologues (see Fig. S1 in the supplemental material). Of the 15 residues mutated, four were completely conserved (Y3, W6, A13, and E25), eight (W2, F4, I7, L8, G9, L12, F16, and A21) were conserved in a majority of homologues, and two (L11 and V18) were not widely conserved. Since it has been hypothesized that the C14 residue in the CydX protein could act to potentially coordinate a heme in the CydAB complex (23), we also generated a C14S mutant.
Of the 15 mutants tested, only four (F4A, I7A, L12A, and A21G) were unable to complement a ΔcydX mutant (Fig. 9). Surprisingly, none of these amino acids were those that are absolutely conserved, and two of the mutations that had the most effect were substitutions of a hydrophobic residue with alanine, which a priori might have been predicted to have only a minor effect. It is possible that the effects of some mutations could be suppressed by the protein overexpression, which could help stabilize an interaction between CydX and the CydAB complex. It is interesting to note that all four of the residues that are required for CydX activity are predicted to be located on the same side of the hydrophobic α-helix (data not shown), suggesting that this may be the side of the protein that interacts with the complex. The fact that the central cysteine can be replaced with serine without affecting small-protein activity, however, does not provide support for the hypothesis that the cysteine helps coordinate a heme in the CydAB complex. Altogether, these results suggest a surprising degree of sequence flexibility for the small protein, at least in the context of the overexpressed protein.
CydX and AppX show ~30% identity and ~60% similarity at the amino acid level (Fig. 1C). To determine whether CydX and AppX have overlapping functions, a plasmid overexpressing AppX was transformed into the ΔcydX mutant and tested for its ability to complement the mutant's sensitivity to β-mercaptoethanol. Overexpression of AppX did complement the mutant (Fig. 2B), suggesting that AppX could replace CydX in the CydAB complex, at least when expressed at high levels. We then assayed for a potential interaction between the CydAB complex and AppX by performing a copurification experiment of endogenously expressed AppX-SPA and CydA-His. To do this, cells containing chromosomally tagged AppX-SPA and CydA-His (at the lacZ locus) were grown under low-oxygen conditions to induce AppX-SPA expression. The cell lysates from these cultures were purified on a Ni-NTA column and assayed for the presence of AppX-SPA in the eluate. Although the level of copurification was much less than that observed for CydX-SPA, some AppX-SPA did copurify with CydA-His (Fig. 10A), consistent with the possibility that a small percentage of AppX molecules can interact with the CydAB complex instead of its more likely partner, the AppBC complex.
We also tested for an interaction between AppX and CydB by assaying for copurification from a strain expressing both CydB-His and AppX-SPA from their endogenous loci. Cell lysates from this strain and a negative-control strain grown under low-oxygen conditions were passed over a Ni-NTA column, and eluates were probed for the presence of AppX-SPA. Similar to the copurification results using CydA-His and AppX-SPA, AppX-SPA was detected in the eluate of the CydB-His strain and not in the negative control (Fig. 10B). Together, these data are consistent with the idea that CydX and AppX may have some functional overlap in the cell and that a fraction of these molecules could potentially interact with either of the two cytochrome bd oxidase complexes. Ultimately, further studies will be necessary to more fully understand the overlapping roles of the two small proteins.
Our knowledge about the function and prevalence of small proteins, specifically those containing 50 or fewer amino acids, in any organism is limited. These proteins are difficult to isolate and characterize biochemically (33), are challenging to reliably identify using bioinformatics (33, 34), and are often overlooked in genetic screens (34). Recently, there has been an increased focus on identifying the number and function of small proteins in bacteria, particularly Escherichia coli (24, 35–39). In addition, the biological functions of an increasing number of small proteins are being characterized, and it is becoming clear that many of these proteins have important roles in microbial physiology, including acting as signaling molecules, being involved in mechanisms to recognize membrane curvature, and acting as subunits of large protein complexes (32, 40–45).
The results presented here show that the 37-amino-acid protein CydX is essential for normal activity of the CydAB cytochrome bd oxidase complex in E. coli. These results are consistent with a recent report showing that CydX is required for CydAB activity in B. abortus (23). In the previous study, researchers showed that B. abortus ΔcydX mutants exhibited phenotypes similar to those of ΔcydB mutants, including reduced survival after engulfment by macrophages (23). These results, together with the data presented here, show that CydX involvement in cytochrome bd oxidase activity is not limited to a single species of bacteria and is likely a shared characteristic of many cytochrome bd oxidases throughout proteobacteria.
We also present biochemical data strongly suggesting that CydX is a subunit of the CydAB cytochrome bd oxidase complex. The cytochrome bd oxidase complex from E. coli and other species of bacteria has been purified many times in the past 70 years (16, 46), raising the question as to why the CydX protein had not been identified earlier. One potential explanation could be due to the difficulty inherent in visualizing and identifying small proteins using standard biochemical techniques. For example, although a number of publications show SDS-polyacrylamide gels of the purified protein complex, few of these figures show a region of the gel in which a 4-kDa protein such as CydX could be resolved. In at least one study, however, the presence of a lower-molecular-weight species in a sample of purified cytochrome bd oxidase from E. coli is reported (47). This observation is followed up in another study in which it is suggested that the low-molecular-weight species present in the sample is lipopolysaccharide (48). Interestingly, one of the characteristics that suggested the low-molecular-weight molecule was a lipopolysaccharide, namely, that it is extractable with methanol-chloroform, is a quality that has been shown to be shared by hydrophobic small proteins as well (49).
Cytochrome bd oxidases have a number of characterized functions in bacterial cells, including maintaining electron transport during low-oxygen conditions (50), acting as electron donors for the disulfide bond regulatory system (51), providing protection for oxygen-labile proteins (52), and acting as a cyanide- and nitrous oxide-resistant oxidase used by pathogenic bacteria to survive the host immune response (18, 20). In addition, these oxidases have multiple catalytic sites that accept a number of different endogenous and artificial substrates (16). Given this diversity, there are many possible functions for a small-protein subunit of cytochrome bd oxidase complex. First, it is possible that CydX is required for proper folding and/or stability of the complex. However, preliminary experiments looking at CydB-His levels in a ΔcydX mutant background suggest that this may not be the case, since CydB-His levels do not appear to be reduced in a ΔcydX mutant (C. E. VanOrsdel and M. R. Hemm, unpublished data). A second possibility is that CydX could bind proteins in addition to CydAB, potentially enabling the coordinated activation or regulation of cytochrome oxidase activity. One potential target for interaction could be the CydDC complex, a transmembrane complex that functions as a glutathione transporter and that is known to be required for cytochrome bd oxidase activity (53). This would be potentially analogous to the proposed activity of the 56-amino-acid CcoQ subunit of cytochrome oxidase of Neisseria gonorrhoeae, which may bind multiple oxidase-related proteins (54). A third, potentially more intriguing, possibility is that CydX acts as a regulatory subunit of the CydAB complex that has evolved to modify cytochrome bd oxidase activity in certain species of bacteria. Given the plethora of biological functions for this complex, it is tempting to hypothesize that the CydX protein, when bound to the CydAB complex, could alter the relative levels of activity of the complex toward different substrates. It is interesting to note that overexpression of CydX led to higher than wild-type levels of TMPD oxidation (Fig. 3), reminiscent of the phenotype of certain CydA mutants lacking a highly conserved arginine residue in the protein (55). In a study by Zhang et al., E. coli cydA R391 mutants showed increased TMPD oxidation but significantly reduced activity toward ubiquinol, suggesting that the CydAB complex had been altered in such a way to increase specificity of one substrate over another (55). Although ubiquinol activity was not measured in our current study, in future studies it will be interesting to determine whether altering CydX levels in the cell and/or its affinity for the CydAB complex may affect oxidation of other substrates besides TMPD.
In addition, it remains to be determined whether CydX is required for CydAB activity in vitro. It is possible that CydX is not essential when using a highly purified sample of CydAB, but could, if present in the purified sample, modulate the activity of the complex. If this were the case, the presence or absence of CydX in different purifications of CydAB could help account for some of the variation in enzymatic activities observed for oxygen affinity and substrate oxidation rates of the purified complex (16). This would be analogous to the findings that KdpF, a small protein encoded in a K+-translocating complex in E. coli, was required for the activity of a large protein complex (KdpABC) in vitro but was dispensable under high-lipid conditions (40). Ultimately, further enzymatic analysis of purified cytochrome bd oxidase from organisms that contain a cydX gene in the cytochrome bd oxidase operon, such as Escherichia coli and Azotobacter vinelandii, will be required to determine the exact role of the CydX protein in oxidase activity.
The sensitivity of E. coli mutants with reduced cytochrome bd-I oxidase activity to β-mercaptoethanol is striking. One of the main effects of reductant stress is the anomalous reduction of disulfide bonds in periplasmic proteins. In contrast to the E. coli cytosol, which is a reducing environment, the periplasm between the inner and outer membranes is oxidizing, and many periplasmic proteins have disulfide bonds that are essential for function. E. coli and other bacteria have evolved a well-regulated system for maintaining correct disulfide bond formation in these proteins, called the Dsb pathway, and the mechanisms responsible for this process have been well characterized (56). In short, the DsbA protein directly acts as a disulfide bond donor for thiol-containing periplasmic proteins, facilitating the formation of disulfide bonds in these substrates. After this occurs, the DsbB protein then reoxidizes the reduced DsbA protein and transfers the electron to ubiquinones, converting them to ubiquinol (56). Bader et al. have shown in vitro that both the E. coli cytochrome bo and cytochrome bd oxidases can then oxidize the ubiquinol created by the Dsb pathway, transferring the electrons to oxygen to generate water (51). On the basis of these in vitro assays, they proposed that either one of the two oxidases present during aerobic growth, Cyo or CydAB, could compensate for a loss of the other in vivo in providing electrons to the Dsb pathway (51). In contrast to the in vitro results, however, our experiments with β-mercaptoethanol suggest either that the CydAB pathway may have a greater role in supporting Dsb-related oxidation or that both complexes are required for wild-type levels of resistance to reductants.
The results presented here show that the small-protein CydX is required for the function of a large protein complex, similar to the KdpF (40) and AcrZ (32) proteins. Considering that the complexes bound by these proteins, CydAB, KdpABC, and AcrAB-TolC, have been studied extensively in the absence of knowledge about the small proteins, these data illustrate the challenge in identifying small-protein subunits of protein complexes. In addition, it suggests that other large protein complexes may also have unidentified small-protein components. The identification of these small-protein subunits will either require the characterization of the small proteins themselves, as was conducted through the characterization of KdpF, CydX, or AcrZ, or through the purification and biochemical analysis of protein complexes with a focus on isolating and identifying small-protein subunits in particular.
We thank Gisela Storz for supporting experiments that were conducted at the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development. We also thank Gisela Storz and Matthew Howard for helpful comments and discussions about the project and manuscript. We also thank Barry Margulies for technical assistance and Jacqueline Thompson for substantial logistical assistance throughout the project.
This research was supported by an R15 grant (1R15AI094548-01) from the National Institute of Allergies and Infectious Disease, National Institutes of Health. It was also supported by funds to the Jess and Mildred Fisher Endowed Chair of Biological Sciences at Towson University, by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, and by postdoctoral fellowships from the Life Sciences Foundation (M.R.H.). Students were supported by Undergraduate Research Grants from the Fisher College of Science and Mathematics and from Towson University.
Published ahead of print 7 June 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.00324-13.