The copper-linked Escherichia coli AZY operon: Structure, metal binding, and a possible physiological role in copper delivery.

The Escherichia coli yobA-yebZ-yebY (AZY) operon encodes the proteins YobA, YebZ, and YebY. YobA and YebZ are homologs of the CopC periplasmic copper-binding protein and the CopD putative copper importer, respectively, whereas YebY belongs to the uncharacterized Domain of Unknown Function 2511 family. Despite numerous studies of E. coli copper homeostasis and the existence of the AZY operon in a range of bacteria, the operon's proteins and their functional roles have not been explored. In this study, we present the first biochemical and functional studies of the AZY proteins. Biochemical characterization and structural modeling indicate that YobA binds a single Cu2+ ion with high affinity. Bioinformatics analysis shows that YebY is widespread and encoded either in AZY operons or in other genetic contexts unrelated to copper homeostasis. We also determined the 1.8 Å resolution crystal structure of E. coli YebY, which closely resembles that of the lantibiotic self-resistance protein MlbQ. Two strictly conserved cysteine residues form a disulfide bond, consistent with the observed periplasmic localization of YebY. Upon treatment with reductants, YebY binds Cu+ and Cu2+ with low affinity, as demonstrated by metal-binding analysis and tryptophan fluorescence. Finally, genetic manipulations show that the AZY operon is not involved in copper tolerance or antioxidant defense. Instead, YebY and YobA are required for the activity of the copper-related NADH dehydrogenase II. These results are consistent with a potential role of the AZY operon in copper delivery to membrane proteins.

Copper is an essential prokaryotic micronutrient that serves as a cofactor for key enzymes involved in respiration and redox defense such as cytochrome c oxidase, NADH dehydrogenase II (NDH-2), and superoxide dismutase (1, 2). However, copper is also toxic if present in high concentrations and/or in the wrong cellular compartments, partly because of copper-mediated generation of free radicals via the Fenton reaction and disruption of iron–sulfur clusters (3, 4, 5, 6). Therefore, bacteria employ a variety of resistance and regulation mechanisms that allow copper delivery to designated target proteins, while also affording protection from copper overload (2, 6).

Copper resistance mechanisms primarily involve two chromosomally encoded multicomponent systems, Cue (Cu efflux) and Cus (Cu sensing). The Cue system consists of the periplasmic multicopper oxidase CueO, the membrane-embedded ATP-driven copper efflux pump CopA, and the transcriptional regulator CueR, which upregulates the expression of CueO and CopA in response to elevated copper concentrations (7, 8, 9). The Cus system includes the tripartite CusABC complex, which spans the inner membrane, periplasm, and outer membrane. CusABC functions as a proton gradient–driven efflux pump, exporting Cu+ either directly from the cytosol or from the periplasm with the aid of the periplasmic protein CusF (10, 11). The Cus system responds to high copper concentrations via regulation by the CusRS two-component system (12).

The plasmid-borne Cop/Pco (copper resistance or plasmid-borne copper resistance) genes represent a third major bacterial copper handling system that includes at least six proteins, CopABCDRS (13, 14). Unlike the Cue and Cus systems, the roles of the individual Cop proteins are not well understood. CopB and CopD have been proposed to import copper into the periplasm and cytoplasm, respectively (15), whereas CopA is a periplasmic multicopper oxidase (16) and CopC is a periplasmic copper-binding protein (17) suggested to work in concert with CopD (15). Copper-inducible expression of these proteins is regulated by the CopRS two-component system. While gene disruption studies have linked these operons to copper resistance (14, 18), expression of the CopC and CopD proteins alone confers copper hypersensitivity (15), suggesting an import function. Genes encoding CopC and CopD are found together or as fusions in the genomes of a range of bacteria in the absence of the other cop genes present in the Cop/Pco plasmid-encoded resistance operons (19, 20, 21, 22). One CopCD fusion protein, Bacillus subtilis YcnJ, has been implicated in import by genetic disruption studies (20). While pathways of bacterial copper import to the cytoplasm have not been investigated extensively because of the extracytoplasmic localization of most cuproenzymes, recent data suggest that some extracytoplasmic cuproenzymes, such as cytochrome c oxidase and periplasmic copper, zinc superoxide dismutase, are metallated by copper that is first imported to the cytoplasm and then exported back to the periplasm (23, 24, 25, 26).

Genes encoding CopC and CopD homologs are also found on the Escherichia coli chromosome. These proteins are designated YobA (CopC homolog) and YebZ (CopD homolog), and along with a third protein, YebY (Domain of Unknown Function 2511 [DUF2511]) are encoded by the AZY operon (27). Two transcriptional initiation sites homologous to the copper-responsive “copper boxes” of the pco resistance operon (28) are present at positions -34 and -22. Notably, unlike the pco, cop, and cus gene clusters, the AZY operon is also regulated by the small noncoding RNA, FnrS. Under anaerobic conditions, FnrS negatively regulates the AZY operon and numerous additional genes mainly involved in respiration and antioxidant defense (29, 30). Given that E. coli has been the focus of numerous copper homeostasis studies, it is surprising that the AZY operon proteins have rarely been mentioned in the literature (17, 31). Despite the presence of this operon in many strains of E. coli and in other proteobacteria, firmicutes, and actinobacteria, its physiological role and the proteins involved have not been investigated. Here, we present an initial characterization of the YobA–YebZ–YebY system, including bioinformatics analysis of the YebY family, determination of the YobA and YebY metal-binding properties, the high-resolution crystal structure of YebY, and a possible physiological role of the AZY operon. The combined findings provide new insight into the function of this operon.

Results

Model structure and metal-binding properties of YobA

YobA is a member of the CopC family (PF04234) of periplasmic copper-binding proteins. All structurally characterized CopC proteins exhibit a seven-stranded β-barrel fold (17, 32, 33, 34, 35, 36). While a small subset of CopC proteins contain distinct Cu+-binding and Cu2+-binding sites, most family members bind a single Cu2+ ion. Based on its sequence, YobA belongs to this majority class, in which the Cu2+ ion is coordinated by an amino-terminal nitrogen, two histidine side-chain nitrogens, and an aspartic acid (17). In homology models of YobA generated using five different NMR and crystal structures of CopC from four different organisms as templates, residues His27, Asp111, and His113 are well positioned to form the predicted Cu2+-binding site (Fig. S1).

To study metal binding by YobA, we amplified the complete yobA gene (including its signal peptide) from the genome of the WT E. coli strain BW25113 and inserted it into a pET21 expression plasmid. A stop codon was inserted before the His6 sequence to avoid possible complications from metal binding to the His6 tag. Mature YobA (∼11 kDa) was purified to homogeneity (Fig. S2) from osmotic shock extracts of cells overexpressing YobA by a combination of ion exchange and gel filtration chromatography, and its identity was verified by MS. We then measured Cu2+ binding by isothermal titration calorimetry, which gave a K value of ∼3.3 × 108 M−1 (K ∼ 3 × 10−9 M) (Fig. 1A). Consistent with the lack of a methionine-rich Cu+-binding motif (17), YobA displayed no measurable Cu+ binding (Fig. 1B). We also measured the binding affinity of YobA for several additional divalent metal ions. We could not detect binding of Mn2+, Cd2+, Fe2+, Pb2+, or Mg2+ (Table S1). Binding of Zn2+, Ni2+, Co2+, and Hg2+ was observed, albeit with affinities that are two to three orders of magnitude lower than that of YobA for Cu2+ (Fig. 1C and Table S1). Similar results were obtained in metal-mediated tryptophan fluorescence quenching experiments. The most prominent quench was observed in the presence of Cu2+ (Fig. 2A), whereas no quench was observed upon addition of Cu2+ in the presence of the reducing agent DTT (Fig. 2B). Collectively, these results indicate that YobA is a typical Cu2+-binding CopC protein.

Figure 1

Isothermal titration calorimetry measurements of metal binding by YobA. Shown are the serial injections of 2 μl of (A) 0.12 mM CuSO4, (B) 0.12 mM CuSO4 + 1 mM TCEP, and (C) 0.2 mM ZnSO4 into a 200 μl solution of 12 μM YobA. Results are representative of experiments repeated at least three times. The upper panels show the calorimetric titration, and the lower panels display the integrated injection heat derived from the titrations, for which the best-fit curve (red trace) of a simple 1:1 interaction model was used to calculate the K. The K values are mean ± STD (standard deviation) of three independent experiments. Also shown are the goodness of the fit (Chi2/DoF) and the changes in enthalpy (ΔH) and entropy (ΔS) upon binding. TCEP, Tris(2-carboxyethyl)phosphine.

Figure 2

Fluorescence quenching measurements of metal binding by YobA.A, the fluorescence (excitation of 280 nm and emission of 325 nm) of 2 μM YobA was measured in the presence of 10 μM CuSO4, NiSO4, CoSO4, and ZnSO4. The fractional quench of fluorescence (ΔF/F) relative to metal-free YobA is shown. The results are the mean ± standard deviation of the mean of triplicates. B, the fluorescence intensity (excitation of 280 nm and emission of 325 nm) of 2 μM YobA in the absence or the presence of 10 μM of CuSO4, 1 mM DTT, or 10 μM CuSO4 + 1 mM DTT, as indicated. Normal distribution of the data was verified by the Shapiro–Wilk test (α = 0.05), and statistics were calculated using one-way ANOVA. ∗∗∗p < 0.005. ns, not significant.

Bioinformatics analysis and cellular localization of YebY

Unlike YobA and YebZ, which belong to the known CopC and CopD protein families, respectively (17, 22), YebY belongs to the uncharacterized Protein family 10709 (pfam10709)/DUF2511 protein family (17, 31), and its connection to copper homeostasis is unknown. As such, we conducted a comprehensive bioinformatics analysis of sequences associated with pfam10709 available in the Joint Genome Institute (JGI)/Integrated Microbial Genomes genome database. The majority of YebY sequences are found in Gram-negative bacteria, and specifically in γ-proteobacteria (Fig. 3A), most commonly within the Enterobacteriaceae family. Other well-represented families include Erwiniaceae, Yersiniaceae, Morganellaceae, and Pectobacteriaceae. The next most represented class is Actinobacteria, including Bifidobacteriaceae, Microbacteriaceae. Mycobacteriaceae, Nocardiaceae, and Streptosporangiaceae. The E. coli K12 YebY investigated here belongs to the largest cluster of sequences, suggesting that it is an appropriate representative of γ-proteobacterial YebY homologs and the broader Pfam10709/DUF2511 family. The only previously studied family member is the putative lantibiotic self-resistance protein MlbQ from the Gram-positive actinomycete Microbispora ATCC PTA-5024 (37). Comparisons of YebY and MlbQ will therefore provide useful insight into the similarities and differences between γ-proteobacterial and actinobacterial Pfam10709 members (Fig. 3A).

Figure 3

Sequence similarity network of YebY homologs constructed using the EFI/EST sequence similarity network generation tool at an alignment score threshold of 48. Colored by (A) bacterial family and (B) genomic proximity to CopC and CopD. EFI/EST, Enzyme Function Initiative/Enzyme Similarity Tool.

We assessed the cellular localization of the YebY family by identifying predicted signal sequences using the LipoP 1.0 server (38, 39). The vast majority of analyzed sequences contain a predicted signal peptide, with a minority predicted to harbor lipoprotein signal peptides (Fig. S3). E. coli YebY contains a predicted signal peptide, whereas MlbQ contains a lipoprotein signal peptide that may facilitate lipid attachment to Cys36, the first residue after the predicted cleavage site, and subsequent membrane anchoring. The widespread presence of signal peptides indicates that YebY homologs are extracytoplasmic in both Gram-positive and Gram-negative bacteria. To verify the cellular localization of YebY, we cloned the full sequence, including the signal peptide, from E. coli BW25113 and inserted it into a pET-derived expression vector harboring a C-terminal His6 tag. YebY-His6 was then expressed in the E. coli BL21-Gold (DE3) strain, and the cells were fractionated into cytosolic, membrane, and periplasmic compartments. Immunoblotting of SDS-PAGE indicates that YebY localizes exclusively to the periplasm (Fig. S4A).

In addition, we performed a genomic neighborhood analysis using representative node sequences to determine how frequently genes encoding YebY family members are encoded in conjunction with copC or copD-like genes. Approximately 52% of representative genes encoding YebY homologs are found in proximity to both copC and copD in AZY-like genomic contexts (Fig. 3B). Of those genes, >97% exhibit the canonical gene ordering of yebY adjacent to copD and two genes away from copC. Two E. coli species harbor YebZ–YebY fusions that are present next to YobA. However, a significant number of genes encoding YebY homologs exist in different genomic contexts. Many homologs are not part of an operon, including members of the Pectobacteriaceae, Micrococcales, Mycobacteriaceae, Erwinia, and Chryseobacteria. In other cases, yebY is present in an operon unrelated to copper homeostasis. For example, YebY is encoded between the site-specific recombinase XerD (COG4974) and the bacteriophage CI repressor helix–turn–helix domain-containing protein (PF07022) in a number of Enterobacteriaceae. In some Mycobacteriaceae, yebY is downstream of a transcriptional regulator (contains XRE-family HTH domain, COG1396). By contrast, the homolog MlbQ is present along with genes encoding ABC transporters in an operon associated with lantibiotic resistance (37).

Biochemistry and structure of YebY

To facilitate biochemical analysis of YebY, we purified it from osmotic shock extracts. To avoid possible complications from metal binding to the tag, we inserted a stop codon before the His6 sequence. Untagged mature YebY (∼11 kDa) was purified to homogeneity from the periplasmic compartment using size-exclusion chromatography (SEC; Fig. S4B, inset). Inductively coupled plasma MS (ICP–MS) analysis indicated that YebY purifies in the absence of metal ions (Table S2). SEC with multiangle light scattering (SEC–MALS) analysis of purified YebY showed a homogeneous monomer with a molar mass of 11 kDa (Fig. S4B). Given the presence of two highly conserved cysteine residues and the periplasmic localization of YebY, we investigated the possibility of a disulfide bond. On SDS-PAGE, the oxidized form of YebY migrates faster than the reduced form, suggesting a more compact conformation of the former (Fig. S5A). In addition, reduction of YebY decreased the midpoint temperature for thermal denaturation by ∼5 °C, consistent with the presence of a stabilizing disulfide bond (Fig. S5B). Moreover, replacement of either Cys48 or Cys113 by site-directed mutagenesis to Ala or Ser completely abrogated protein expression. Collectively, these results indicate that a Cys48–Cys113 disulfide bond is indeed formed.

To gain a molecular-level understanding of YebY, we took advantage of the two methionine residues in YebY and prepared a selenomethionine-labeled analog (SeMet-YebY) for crystallography. SeMet-YebY crystallized readily, and the 1.8 Å resolution structure was solved using Se single-wavelength anomalous dispersion phasing (Table 1). The space group is P21 with 12 monomers in the asymmetric unit arranged as dimers. The monomer structure comprises an antiparallel four-stranded β sheet surrounded by four α helices. As anticipated, the C-terminal cysteine (Cys113) forms a disulfide bond with a cysteine at the C terminus of the second β-strand (Cys48), stabilizing a compact fold (Fig. 4, A and B). Alignment of the structure with the NMR structure of the only previously characterized family member, MlbQ (26.9% amino acid identity) (37, 40), reveals a strong similarity with an alignment RMSD of 1.84 Å for 294 atoms and a disulfide bond in the same location (Fig. 4C). While there are slight differences in some of the secondary structure elements, the only major difference between the two proteins is the length. The first 58 residues of MlbQ, which include a 34-residue amino-terminal lipoprotein signal peptide, are unstructured (not shown in Fig. 4C). The unstructured residues following the signal peptide are presumably a linker between the membrane anchor and the folded domain. In contrast, the first secondary structure element begins at position 25 in the full-length YebY sequence. Thus, the first sequentially and structurally conserved residues are Trp37 (YebY) and Trp71 (MlbQ). The similarity between these two homologs, which are relatively distant within the Pf10709/DUF2511 family, suggests a high degree of structural conservation within the family.

Table 1

Data collection and refinement statistics

SeMet-YebY
Data collection
Space group	P21
Cell dimensions
a, b, c (Å)	96.1, 53.2, 132.4
α, β, γ (°)	90, 102.3, 90
Resolution (Å)	32.79–1.88 (1.95–1.88)
Rsym	0.06082 (1.14)
I/σI	9.96 (0.61)
Completeness (%)	90.84 (55.27)
Multiplicity	2.0 (2.0)
No. of reflections	193,327 (11,618)
Refinement
Resolution (Å)	32.79–1.88 (1.95–1.88)
Reflections used in refinement	97,246
Rwork/Rfree (%)	23.5/27.0
No. of atoms
Protein	8659
Water	451
B-factors (Å2)
Protein	40.7
Water	43.3
RMSDs
Bond lengths (Å)	0.031
Bond angles (°)	1.95

Values in parentheses represent the highest-resolution shell.

Figure 4

Crystal structure of YebY.A, cartoon diagram of YebY. B, topology diagram of YebY indicating the location of the disulfide bond. C, superposition of YebY (green) and MlbQ (Protein Data Bank accession code: 2MVO; purple).

YebY displays low affinity for copper and other metal ions

Given that YebY is encoded in an operon with putative (YebZ) and confirmed (YobA) copper-binding proteins and contains two highly conserved cysteine residues that could bind copper in their reduced state, we investigated its metal-binding properties. YebY lacking the His6 tag in its native state or treated with the reductant tris(2-carboxyethyl)phosphine (TCEP) was incubated with up to 10 equivalents of Cu+ or Cu2+ in an anaerobic chamber. CD spectra confirm that reduction of the disulfide does not cause a significant change in secondary structure (Fig. S6). Following incubation for ∼2 h, unbound copper was removed by desalting. The copper and protein concentrations were then measured by ICP–MS and absorbance at 280 nm, respectively. Without treating YebY with TCEP, very little binding was observed for either Cu+ or Cu2+ (Fig. 5A). However, the TCEP-treated (reduced) protein showed measurable binding of both Cu+ and Cu2+, with a seeming preference toward the former: at a molar ratio of 2:1, ∼30% of YebY was occupied by Cu+, whereas five equivalents of Cu2+ were required to reach the same occupancy (Fig. 5A). To estimate the binding affinity of reduced YebY for Cu+, we measured the quenching of intrinsic tryptophan fluorescence and observed a dose-dependent response with a K of 50 to 100 μM (Fig. 5B). These ICP–MS and tryptophan fluorescence results suggest that the two cysteine residues involved in the disulfide bond can bind Cu+ with low affinity.

Figure 5

Metal-binding properties of YebY.A, YebY (32–56 μM) with (+) or without (−) pretreatment with reductant was incubated with the indicated molar equivalents of Cu+ or Cu2+ under anaerobic conditions. Unbound material was removed by desalting, and the amount of YebY-bound copper was measured by ICP–MS. B, YebY (5 μM) was incubated in the presence of 1 mM DTT with the indicated concentrations of Cu+. The change in fluorescence emission at 340 nm was measured following excitation in 280 nm. C, YebY (2.5 μM) was incubated in the presence of 1 mM DTT and 250 μM of the indicated metals. The change in fluorescence emission at 340 nm was measured following excitation at 280 nm. Results shown in (A–C) are means of at least three repeats ± standard deviation of the mean. ICP–MS, inductively coupled plasma MS.

Quenching of intrinsic tryptophan fluorescence was also used to evaluate the possibility that YebY interacts with other metal ions. A strong quench of fluorescence was observed in the presence of Co2+, Ni2+, and Ag+, whereas a negligible quench was observed in the presence of either Zn2+ or Mn2+ (Fig. 5C). Therefore, unlike YobA, which is a highly specific high-affinity Cu2+-binding protein, YebY seems to bind metal with low affinity and specificity, bringing into question the physiological relevance of YebY as a copper-binding protein, and suggesting that it may play a different role in the AZY operon.

A possible physiological role for the AZY operon

To probe the physiological role of the AZY operon, we constructed E. coli strains harboring single deletions of yobA, yebZ, or yebY or a triple AZY knockout and compared the susceptibility of WT and deletion strains to various stressors. First, we tested for a role in copper tolerance, since periplasmic copper-binding proteins have been proposed to contribute to copper tolerance (41, 42). However, the WT and ΔAZY strains exhibited similar growth over a broad range of copper concentrations (0–2 mM) (Figs. 6A and S7). Because of the redundancy of bacterial copper resistance mechanisms, the distinction between tolerant and sensitive phenotypes often requires the use of hypersensitive strains in combination with overexpression of the rescuing protein (43). We therefore overexpressed the individual genes or the entire operon in the background of the copper-sensitive E. coli strain GG44, which lacks the copper P1B-ATPase efflux pump CopA (43, 44). As shown in Figure 6B, the overexpressed AYZ proteins could not rescue this copper-sensitive strain. These combined results argue against a role for the AZY operon in copper tolerance.

Figure 6

The AYZ operon is not involved in copper tolerance or antioxidant defense.A, cultures of WT (black), ΔyobA (red), ΔyebY (green), ΔyebZ (blue), or ΔAYZ (purple) Escherichia coli were grown for 12 h in Davis minimal media in the presence of 0 to 2 mM CuSO4, as indicated. B, cultures of WT E. coli (black) or the copper-sensitive GG44 strain (all other curves) were grown for 12 h LB media in the presence of 0.1 mM IPTG and 0 to 2 mM CuSO4, as indicated. The GG44 cells were transformed with an empty control vector (orange) or with plasmids for the overexpression of YobA (red), YebY (green), YebZ (blue), or AYZ (purple). C, cultures of WT (black) or ΔAYZ (gray) E. coli were grown in LB media for the indicated times in the absence (solid traces) or the presence (dashed traces) of 0.0025% hydrogen peroxide. D, cultures of WT (black) or ΔAYZ (gray) E. coli were grown in LB media for the indicated times in the absence (solid traces) or the presence (dashed traces) of 0.25 mM paraquat. Results shown in (A–D) are means of at least three repeats, and error bars (shown unless smaller than the icons) represent ± standard deviation of the mean.

In addition to copper, the AZY operon is regulated by the small noncoding RNA, FnrS, which is known to regulate genes involved in antioxidant defense (29, 30). We therefore investigated the possibility that the operon is involved in antioxidant defense. When challenged with the oxidizing agents, hydrogen peroxide (Fig. 6C) or paraquat (Fig. 6D), the ΔAZY strain grew very similarly to the WT strain, suggesting that the AZY operon does not play a major role in antioxidant defense.

Taking into account the copper-binding capacity of YobA and the periplasmic localization of YobA and YebY, another possibility is that the operon functions in copper delivery to membrane-embedded copper proteins. Such a role has been demonstrated for two periplasmic copper chaperones of the purple photosynthetic bacteria Rhodobacter capsulatus (45). To probe for copper content of membrane proteins, we isolated membranes from early exponential-phase cultures of WT and ΔAZY cells and monitored the activity of NADH-NDH-2. This membrane-embedded enzyme catalyzes the transfer of electrons from NADH to its copper cofactor or to copper cofactors of other membrane proteins (46, 47, 48). The activity of NDH-2 therefore provides a proxy for the amount of membrane-bound copper.

In buffered solution (in the absence of membranes), NADH is stable, and its oxidation is not observed on the timescale of the experiment (Fig. 7, black trace). Addition of the WT E. coli membrane fraction greatly accelerated the rate of NADH oxidation (Fig. 7, red trace). The oxidation rate of NADH in the presence of membranes prepared from cells lacking NDH-2 (Δndh) was nearly identical to that observed in buffer (Fig. 7, green trace), corroborating previous reports that in such early exponential-phase cultures, NDH-2 is the main NADH oxidase (49, 50). Relative to WT membranes, membranes prepared from the ΔAZY cells displayed significantly reduced NADH oxidation activity (Fig. 7, blue trace), despite containing WT-like levels of NDH-2 as determined by LC–MS2 proteomic comparison of WT and ΔAZY cells (Fig. S8). These results imply that the AZY operon contributes to copper delivery to NDH-2 and/or to other membrane proteins. We then prepared membranes from strains carrying single deletions of yobA, yebZ, or yebY. The ΔyebZ membranes displayed the same NADH oxidation activity as the WT membranes (Fig. 7, cyan trace). Of the two remaining genes, the deletion of yobA more strongly affected the NADH oxidation activity compared with the deletion of yebY (Fig. 7, compare purple and orange curves), and the combined effects of the single deletions ΔyobA and ΔyebY accounted for the effect observed upon deletion of the entire operon. Importantly, transformation of the ΔyobA and ΔAZY strains with plasmids encoding the deleted genes restored the NADH oxidation activity to WT levels (Fig. S9). Collectively, these results suggest that YebZ is dispensable for NDH-2 activity, whereas YobA and YebY contribute to it.

Figure 7

YobA and YebY contribute to the activity of NDH-2. At the time indicated by the arrow (∼3 min), 1 mM NADH was injected into buffered solution in the absence (black trace) or the presence (all other traces) of membrane fractions (1 mg/ml) prepared from early exponential phase cultures of WT Escherichia coli (red), Δndh cells (green), ΔAZY (blue), ΔyebZ (cyan), ΔyobA (purple), or ΔyebY (orange). The rate of NADH oxidation was monitored by measuring the absorbance at 340 nm. Results shown are means of triplicates ± standard deviation of the mean. NDH-2, NADH dehydrogenase II.

Discussion

The E. coli AZY operon encodes two periplasmic proteins, YobA and YebY, and a third uncharacterized membrane protein, YebZ, which belongs to the CopD family of putative copper importers. While genes encoding YobA and YebZ homologs are typically found together or as fusions (19, 20, 21, 22), our bioinformatics analysis shows that YebY is frequently found in other genomic contexts, perhaps suggesting distinct roles within and beyond the AZY operon. YobA contains the canonical His–Asp–His C0–2 Cu2+-binding site (17), and as expected, displays high specificity and affinity for Cu2+ (K ∼ 3 nM). In contrast, YebY is more promiscuous and binds several metals (including Cu2+ and Cu+) with low affinity (K = 50–100 μM). Metal binding by YebY is enhanced following reduction of the protein, suggesting that the two highly conserved cysteines of YebY can participate in metal coordination. However, the low affinity of copper binding by YebY along with the stabilization imparted by the disulfide suggests that this metal binding is likely not physiologically relevant.

Genetic disruptions indicate that the AZY operon is not involved in copper tolerance or antioxidant defense but instead point to a possible role in copper delivery to membrane targets. It is intriguing that despite the fourfold to fivefold difference in copper-binding affinities between YobA and YebY, both seem to be required for full activity of the copper-dependent NDH-2. Given the striking structural similarity between YebY and MlbQ, the function of YebY may also be related to that of MlbQ. Resistance to the self-produced lantibiotic NAI-107 in Microbispora ATCC PTA-5024 is proposed to derive from direct binding of NAI-107 to MlbQ (37). Notably, copper has been previously linked to antibiotic resistance in E. coli (51). Further investigation of the role of the AZY operon in copper delivery and potentially antibiotic resistance represents an important future direction.

Despite their broad distribution and extensive structural and biochemical characterization, the roles of YobA family members have remained surprisingly elusive (13, 17, 34, 45, 52). Even less is known about YebZ homologs (CopD family), and YebY proteins were characterized here for the first time, despite the presence of the AZY operon in E. coli. By contrast, the roles of the copper-detoxifying systems, Cue and Cus, are fairly well established (53, 54). One possible explanation for this dichotomy in knowledge may lie in the universality of copper toxicity and copper tolerance mechanisms versus the divergence of copper utilization. The challenge of copper toxicity is faced by all bacteria, and the main defense mechanisms appear to have been highly conserved throughout evolution. Copper utilization, on the other hand, is less universal: only a handful of copper-dependent/containing proteins are present in each bacterium, and this subset varies between different bacteria (2). Moreover, in some bacteria, copper is needed in compartments or metabolic pathways that do not exist in other bacteria (e.g., thylakoids of cyanobacteria (55), intracytoplasmic membranes of methanotrophic bacteria (56), reduction of N2O by denitrifying bacteria (57)). These specialized needs likely led to divergence of the utilization/delivery pathways, rendering the assignment of physiological roles difficult. In E. coli, and perhaps in other enterobacteria, we suggest that the AZY system is not involved in copper tolerance or import but rather in its delivery to specific membranal targets.

Experimental procedures

Homology modeling of YobA

Template structures were collected using the Hidden Markov model (HMM)-based homology detection algorithm HHpred (58). Five CopC NMR and crystal structures from four different organisms were identified as potential templates for modeling. These included the crystal structures of CopC from E. coli (Protein Data Bank [PDB]: 1LYQ; 46% sequence identity), Pseudomonas syringae (PDB: 2C9Q; 39% sequence identity), Methylosinus trichosporium OB3b (PDB: 5ICU; 29% sequence identity), and Pseudomonas fluorescens (PDB: 6NFR; 37% sequence identity), and an NMR structure of CopC from P. fluorescens (PDB: 6TPB). For YobA and each of the templates, sequences of homologs were collected using HMMER (59) against the Clean-UniProt database with a maximal sequence identity of 95% and a minimal sequence identity of 35%. These sequences were then aligned using MAFFT (60) to produce multiple sequence alignments (MSAs). Next, each of the template MSAs was aligned to the MSA of YobA using HHalign (61). Pairwise alignments between the sequences of YobAs and each of the templates were deduced from these profile-to-profile MSAs.

For each of the five templates, MODELLER-9.18 (62) was used with default parameters to produce 100 models of YobA. A short steepest descent energy minimization was then carried out for each model using GROMACS-2019 (63) and the AMBER99SB-ILDN force field (64). The model with the lowest predicted energy was selected for further analysis. Finally, the MSA of YobA was used with the ConSurf Web server (65) to calculate the conservation scores of amino acids that were mapped onto the 3D model. Overall, the YobA model is consistent with the expected evolutionary pattern, in which the core of the protein is highly conserved and the periphery more variable. One exception is a highly conserved patch on the protein's surface, consisting, among others, of amino acids His 27, Asp 111, and His 113. Importantly, the equivalent residues in the P. fluorescens CopC crystal structure bind Cu2+ ions (36).

Cloning, expression, and protein purification

The yobA and yebY genes were PCR amplified from the parental strain of the Keio collection (66) (E. coli strain BW25113) and inserted into a pET-21b (Novagen) expression vector carrying a His6 C-terminal affinity tag. The His6 tag was used for preparation of YebY for crystallization. For all metal-binding assays, tag-free variants of YebY and YobA were prepared by inserting a stop codon before the His6 coding region. YebY and YobA were expressed in E. coli BL21-Gold (DE3) cells grown at 37 °C in either LB or terrific broth glycerol supplemented with 100 μg/ml ampicillin. To induce protein expression, 1 mM IPTG was added to midexponential phase cultures for 2 to 4 h. Cells were then harvested by centrifugation for 20 min at 8000g and stored at −80 °C. Periplasmic extracts were prepared by osmotic shock: cell pellets were resuspended (10 ml/1 g cells) in 40% sucrose, 1 mM EDTA, 10 mM Tris–HCl, pH 7.5, and agitated gently at 4 °C for 1 to 3 h, followed by 5 to 15 min agitation at room temperature. The cells were then squirted into 100-fold excess volume milliliters of ice-cold 0.5 to 250 mM EDTA, pH 8, and stirred vigorously for 5 min. Prior to centrifugation at 8000g for 30 min, 50 mM Tris–HCl, pH 7.5, 250 mM NaCl, 30 μg/ml DNase (Worthington), 1 EDTA-free protease inhibitor cocktail tablet (Roche) (or 1 mM PMSF), and 1 to 2 mM MgCl2 were added to the cell suspension. Because of the high periplasmic content of YobA or YebY, this step alone yielded 50 to 80% purity. Tag-free YobA and YebY were further purified by SEC using a HiLoad 16/600 Superdex 75 or 200 column (GE Healthcare) using 50 mM Tris–HCl, pH 7.5, 250 mM NaCl. Fractions were analyzed by SDS-PAGE, pooled, and concentrated to ∼10 to 20 mg/ml using Amicon Ultra concentrators (Millipore) with a molecular cutoff of 8 to 10 kDa.

Selenomethionine-labeled His6-tagged YebY was purified as described for native YebY, with the following modifications. Osmotic shock was performed using 40% sucrose, 1 mM EDTA, 10 mM Tris, pH 8.0 followed by addition of 2 mM MgCl2, 20 mM imidazole, 1 mM PMSF, 20 mM Tris–HCl, pH 8.0, and 250 mM NaCl. The protein was purified using a 5 ml Ni-loaded HisTrap column by elution with imidazole. Imidazole was then removed using a HiPrep 26/10 column equilibrated with 50 mM Tris–HCl, pH 7.5, 250 mM NaCl, and the protein was incubated with 10 mM EDTA for 1 h followed by two cycles of dialysis against 50 mM Tris–HCl, pH 7.5, 250 mM NaCl, 1 mM TCEP, once for 3 h and again overnight.

Isothermal titration calorimetry of YobA

Calorimetric measurements were performed with a MicroCal iTC200 System (GE Healthcare). Prior to the experiment, YobA was dialyzed extensively against 25 mM Tris–HCl, pH 8, 150 mM NaCl. To reduce buffer incompatibility, metal stocks were prepared in this dialysis buffer. All measurements were carried out at 25 °C. Aliquots (2 μl) of metal in the indicated concentrations were added by a rotating syringe to the reaction chamber containing 200 μl of 12 μM YobA. Cu2+ was added as CuSO4, and Cu+ was added as CuSO4 + 1 mM TCEP. Data fitting was performed with ORIGIN 7.0 assuming a simple 1:1 binding model in which the metal-free form of the protein is in equilibrium with the metal-bound species.

Bioinformatics of YebY

YebY belongs to the DUF2511 family and is part of Pfam10709. All sequences corresponding to Pfam10709 (12,376) were downloaded from JGI/Integrated Microbial Genomes on April 13, 2020. A sequence similarity network was created using the EFI-EST web tool (67) with an E-value cutoff of 5. Representative nodes were created using a 100% sequence identity cutoff, resulting in 1224 unique amino acid sequences. These sequences were then aligned against the HMM for Pfam10709 using HMMalign (68) and visualized in Jalview (69). The vast majority of the downloaded sequences contained two highly conserved cysteine residues, so truncated sequences lacking one or both cysteine residues were removed from the dataset. A sequence similarity network was then created using the EFI-EST web tool (67) with an E-value cutoff of 5. Representative nodes were created using a 95% sequence identity cutoff, resulting in 539 unique amino acid sequences. The resulting network was visualized using Cytoscape 3.7.2 (70). Metadata from JGI were added for each representative node gene to visualize on the network, including the family of each representative node sequence. The LipoP 1.0 server was used to predict whether each representative sequence contains a signal peptide or lipoprotein signal peptide (38). In addition, the genomic neighborhoods and the proximity of PF10709 to CopC (PF04234) and CopD (PF05425) genes were assessed by evaluating five genes upstream and downstream of YebY.

SEC–MALS

SEC–MALS was used to determine the oligomeric state of untagged YebY. Aliquots of YebY were thawed and buffer exchanged using 10 kDa MWCO (molecular weight cutoff) concentrators (Millipore) into 20 mM Hepes, pH 7.0, 100 mM NaCl. Samples (250 μl, ∼300 μM) were prepared at room temperature immediately before use and analyzed using an Agilent 1260 series HPLC system equipped with diode array detection absorbance in-line with a DAWN HELEOS II multiangle static light scattering detector (Wyatt Technology), a QELS dynamic light scattering detector (Wyatt Technology), and a T-rEx differential refractive index detector (Wyatt Technology). The samples were injected onto a Superdex 75 Increase 10/300 GL column (GE Healthcare) that had been pre-equilibrated in running buffer (20 mM Hepes, pH 7.0, 100 mM NaCl). The buffer was stored at room temperature, and the column was kept at 8 °C. Each sample was run at 0.4 ml min−1 for 60 min. Data processing and analysis were performed using Astra software, version 5.3.4 (Wyatt Technology).

Crystallization and structure determination

Selenomethionine-labeled His6-tagged YebY was prepared to facilitate structure determination. YebY was transformed into the methionine-auxotrophic strain B834(DE3). About 5 ml of starter cultures were grown overnight at 37 °C with shaking in total M9 medium supplemented with 0.4% glucose, 2 mM magnesium sulfate, 0.1 mM calcium chloride, 25 μM ferric sulfate (or 40 μM ferric citrate), 100 μg/ml ampicillin, and 50 μg/ml methionine. The starter culture was used to inoculate 1 l of the same media to a starting absorbance of 0.05 to 0.1 at 600 nm. The culture was grown to an absorbance of 1 at 600 nm before it was centrifuged (10 min, 5500g, 25 °C). The supernatant was decanted, and the pellet was resuspended in the same media but with omission of methionine. Following 4 to 8 h shaking at 37 °C, selenomethionine was added to a concentration of 50 μg/ml. After 30 min, expression was induced with 0.5 mM IPTG, and the culture was grown overnight at 25 °C. The cells were pelleted by centrifugation (8500g, 30 min, 4 °C), flash frozen in liquid nitrogen, and stored at −80 °C until use.

Selenomethionine-labeled His6-tagged YebY was crystallized by sitting drop vapor diffusion at room temperature by mixing 0.5 μl of 14.2 mg/ml SeMet-YebY with 0.5 μl of well solution containing 0.2 M ammonium sulfate, 0.1 M bis–Tris–HCl (pH 6.5), and 25% (w/v) PEG 3350. Rod-shaped crystals grew within 2 weeks. Crystals were cryoprotected in a solution containing 40% PEG 400 and flash frozen in liquid nitrogen. Diffraction data were collected at LS-CAT Sector 21 of the Advanced Photon Source at Argonne National Laboratory at a wavelength of 0.9786 Å. XDS was used to index, integrate, and scale the data (71).The crystals belong to space group P21, and there are 12 monomers in the asymmetric unit. The structure was solved by single-wavelength anomalous dispersion using phenix.autosol (72). Phenix.autobuild (73) was used to obtain an initial model with an R/Rfree of 27.6%/31.8%. This model was improved through iterative rounds of model building using Coot (74) and refinement using phenix.refine (75) resulting in a final R/Rfree value of 23.5%/28.2% (Table 1). The final model includes 1112 residues in 12 chains and 451 water molecules. The Ramachandran plot indicates that 97.1% of residues are in favored regions with 2.57% in allowed regions and the remaining 0.09% are outliers. The Molprobity (76) score is 1.74 (85th percentile).

Metal-binding analysis of YebY

The copper-binding assays were performed under inert atmosphere in a Coy anaerobic chamber. The buffer (20 mM Hepes, pH 7.0, 100 mM NaCl) was degassed on a vacuum line with three rounds of pumping followed by purging with nitrogen with continuous stirring. The buffer was then closed to the atmosphere and brought into the glove box, where the cap was removed and it was stirred for 2 days. Aliquots (50 μl) of YebY were thawed, and ∼2.6 molar equivalents of TCEP were added to YebY from a freezer stock (1 M TCEP dissolved in MilliQ water). The samples were brought into the glove box and incubated with TCEP for ∼1 h. The samples were then applied to a PD-10 desalting column and eluted using 20 mM Hepes, pH 7.0, 100 mM NaCl according to the manufacturer’s instructions. The eluted YebY was aliquoted into microcentrifuge tubes; each aliquot contained 4 × 10−8 mols. Tetrakis(acetonitrile)copper(I) hexafluorophosphate (stored in the anaerobic chamber) was freshly dissolved in anhydrous dimethyl sulfoxide (stored in the anaerobic chamber) to ∼50 mM copper concentration, and Cu+ was added to YebY (0, 1, 2, 5, or 10 molar equivalent; <0.6% dimethyl sulfoxide per sample). Alternatively, Cu2+ was added from a 50 mM stock of CuCl2. The samples were gently inverted to mix thoroughly and incubated for ∼2 h before YebY was separated from unbound Cu+ using a PD-10 column following the manufacturer’s instructions. YebY was eluted with 3 ml of buffer (instead of 3.5 ml to avoid coelution with unbound copper) and collected in 15-ml metal-free polypropylene tubes. The samples were gently mixed by inversion, and the protein concentration was measured using absrbance at 280 nm (εoxidized = 8605 M−1 cm−1, εreduced = 8480 M−1 cm−1 used for YebY treated with TCEP). The absorbance of each sample was ∼0.1 a.u. in a pathlength cuvette of 1 cm. The metal content of the samples was measured by ICP–MS using a Thermo iCAP Q Inductively Coupled Plasma Mass Spectrometer equipped with a CETAC ASX260 autosampler in the Quantitative Bio-element Imaging Center core facility at Northwestern University. ICP–MS samples were prepared by combining 0.5 ml of YebY sample, 9 ml of MilliQ water, and 0.5 ml of concentrated nitric acid (TraceSelect). Standard curves were prepared from a dilution series of a multielement standard (Inorganic Ventures).

CD spectroscopy

A Jasco J-815 CD instrument housed in the Keck Biophysics Facility at Northwestern University was used for all measurements. Proteins were buffer-exchanged into CD buffer (1 mM Tris–HCl, ±TCEP, pH 7.5). Each sample (20 μM protein, 300 μl) was transferred to a quartz cuvette (1 mm path length). Spectra were recorded from 190 to 280 nm using continuous scan mode (50 nm/min) and 2 nm bandwidth. All data represent averages of two replicate baseline-subtracted scans, where the baseline was obtained from a sample of CD buffer.

Tryptophan fluorescence quenching experiments

Purified untagged YebY or YobA was dialyzed against 20 mM Tris, pH 7.5, 150 mM NaCl, and 10 mM EDTA for 4 h to remove any metals remaining in the preparation, and then against three rounds of 20 mM Hepes, pH 7.0, and 150 mM NaCl to wash away the EDTA. The proteins were diluted to 2.5 μM in 20 mM Pipes, pH 7.5, 150 mM NaCl, and metals were added as indicated. Cu2+ was added as CuSO4, and Cu+ was added as CuSO4 + 1 mM DTT. Where specified, 1 mM DTT was added to both metal and YebY before mixing. Triplicates of 100 μl were measured using a monochromator-based Tecan M200 plate reader (excitation of 280 nm and emission of 320 nm).

Growth experiments

Overnight cultures were diluted to an absorbance of 0.05 at 600 nm, and 150 μl aliquots were grown in 96-well plates in a Tecan Infinite Pro microplate reader. Cultures were grown in LB or in Davis minimal media (as indicated) in the absence or the presence of the indicated concentrations of CuSO4, H2O2, or paraquat. Cells were grown with intermittent agitation (30 s on, 2 min off) for 16 h.

Construction of knockout strains

The single-gene deletion strains ΔyobA, ΔyebZ, and ΔyebY were obtained from the E. coli genetic stock center at Yale University. The triple knockout (ΔyobA/ΔyebZ/ΔyebY, i.e., ΔAZY) was generated using the lambda Red recombinase system as described previously (77).

NDH-2 activity assay

Early exponential phase cultures (absorbance at 600 nm ≤0.1) grown in LB at 37 °C were harvested by centrifugation (20 min, 8000g, 4 °C) and resuspended in 50 mM Tris–HCl, pH 8, 150 mM NaCl, and 1 mM PMSF. Cells were disrupted using a tip sonicator, and debris was removed by centrifugation at 17,000g. Membranes were collected by ultracentrifugation at 150,000g for 1 h, washed once with 50 mM Tris–HCl, pH 7.5, resuspended in the same buffer + 10% glycerol (w/v), frozen in liquid nitrogen, and stored in −80 °C. Membranes were thawed on ice and diluted to 1 mg/ml with 50 mM Tris–HCl, pH 7.5, and 90 μl were dispensed in triplicates to a 96-well plate, and the assay was initiated by injection of NADH to a final concentration of 1 mM.

MS analysis

Three colonies each from E. coli WT strain BW25113 and the ΔAZY strain were grown overnight in M9 minimal media supplemented with 0.5% glucose, 1% thiamine, 1 mM MgSO4, 10 μM CaCl2, and 0.025% arginine and lysine. Cultures were then diluted and grown to early log phase (absorbance at 600 nm ∼0.15). Cells were washed with PBS, flash frozen, and stored at −80 °C. Samples were dissolved in 10 mM DTT, 100 mM Tris–HCl, pH 7.5, and 5% SDS, sonicated and boiled at 95 °C for 5 min, and precipitated in 80% acetone. The protein pellets were dissolved in 9 M urea and 400 mM ammonium bicarbonate, reduced with 3 mM DTT (60 °C for 30 min), modified with 10 mM iodoacetamide in 100 mM ammonium bicarbonate (room temperature, 30 min in the dark), and digested in 2 M urea, 25 mM ammonium bicarbonate with modified trypsin (Promega) overnight at 37 °C in a 1:50 (M/M) enzyme-to-substrate ratio. The tryptic peptides were desalted using C18 tips (Top tip, Glygen), dried, and resuspended in 0.1% formic acid.

The peptides were resolved by reverse-phase chromatography on 0.075 × 180 mm fused silica capillaries (J&W) packed with ReproSil reversed phase material (Dr Maisch GmbH). The peptides were eluted with a linear 180 min gradient of 5 to 28%, a 15 min gradient of 28 to 95%, and 25 min gradient at 95% acetonitrile with 0.1% formic acid in water at flow rates of 0.15 μl/min. MS was performed by Q Exactive HFX mass spectrometer (Thermo) in a positive mode (m/z 300–1800, resolution 120,000 for MS1 and 15,000 for MS2) using repetitively full MS scan followed by collision-induced dissociation (higher energy collisional dissociation at 27 normalized collision energy) of the 30 most dominant ions (>1 charges) selected from the first MS scan. The automatic gain control settings were 3 × 106 for the full MS and 1 × 105 for the MS/MS scans. The intensity threshold for triggering MS/MS analysis was 1 × 104. A dynamic exclusion list was enabled with exclusion duration of 20 s.

The MS data were analyzed with the MaxQuant software, version 1.5.2.8 for peak picking and identification using the Andromeda search engine, searching against the E. coli proteome from the UniProt database with mass tolerance of 6 ppm for the precursor masses and 20 ppm for the fragment ions. Oxidation of methionine and protein N terminus acetylation were accepted as variable modifications, and carbamidomethyl on cysteine was accepted as static modifications. The minimal peptide length was set to six amino acids, and a maximum of two miscleavages was allowed. The data were quantified by label-free analysis using the same software. Peptide- and protein-level false discovery rates were filtered to 1% using the target-decoy strategy. Protein tables were filtered to eliminate the identifications from the reverse database, common contaminants, and single-peptide identifications.

Data availability

The coordinates and structure factors for E. coli YebY have been deposited in the PDB with accession code 7N0J. All other data are contained within the article and supporting information. Raw data are available from the corresponding authors upon request.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.