Probing a Role of Subunit IV of the Escherichia coli bo -type Ubiquinol Oxidase by Deletion and Cross-linking Analyses*

Subunit IV of the Escherichia coli bo -type ubiquinol oxidase is a 12-kDa membrane protein encoded by the cyoD gene and is conserved in the bacterial heme-copper terminal oxidases. To probe the functional role of subunit IV, we carried out deletion analysis and chemical cross-linking experiments with a homobifunctional and cleavable reagent. Spectroscopic properties of the mu- tant oxidases suggest that the C-terminal two-third (Val 45 to His 109 ) containing helices II and III is essential for the functional expression of the oxidase complex and for the Cu B binding to the heme-copper binuclear center in subunit I. Cross-linking studies indicate that subunit IV is in close vicinity to subunit III. Based on these observations, we propose that subunit IV is present in a cleft formed by subunits I and III and assists the Cu B binding to subunit I during biosynthesis or assembly of the oxidase complex.

Subunit IV of the Escherichia coli bo-type ubiquinol oxidase is a 12-kDa membrane protein encoded by the cyoD gene and is conserved in the bacterial heme-copper terminal oxidases. To probe the functional role of subunit IV, we carried out deletion analysis and chemical cross-linking experiments with a homobifunctional and cleavable reagent. Spectroscopic properties of the mutant oxidases suggest that the C-terminal two-third (Val 45 to His 109 ) containing helices II and III is essential for the functional expression of the oxidase complex and for the Cu B binding to the heme-copper binuclear center in subunit I. Cross-linking studies indicate that subunit IV is in close vicinity to subunit III. Based on these observations, we propose that subunit IV is present in a cleft formed by subunits I and III and assists the Cu B binding to subunit I during biosynthesis or assembly of the oxidase complex.
Bacterial heme-copper terminal quinol (10,13,14) and cytochrome c (15)(16)(17) oxidases generally consist of four different subunits, I to IV. In some preparations, the fifth polypeptide with a molecular mass of 28 kDa was associated with the oxidase complex (18 -20). Subunits I, II, and III of UQO are encoded by the cyoB, cyoA, and cyoC genes, respectively, of the cyoABCDE operon (19,21,22) and are homologous to the counterparts of mitochondrial cytochrome c oxidase (5). Subunit IV (12 kDa) is conserved in bacterial heme-copper terminal oxidases (13,17,20,21,(23)(24)(25) and is likely to be the cyoD gene product, as demonstrated for the counterpart of ba-type ubiquinol oxidase from Acetobacter aceti (13), aa 3 -type menaquinol oxidase from Bacillus subtilis W23 (14), and caa 3 -type cyto-chrome c oxidase from thermophilic Bacillus PS3 (17). Subunit IV of UQO, which consists of three putative transmembrane helices (26), however, is unlikely to be a homologue of either nuclear-encoded small subunits of mitochondrial cytochrome c oxidase (21) or subunit IV of aa 3 -type cytochrome c oxidase from Paracoccus denitrificans that has a single transmembrane helix and resides in a cleft between subunits I and III (15,16 (18,19) is not encoded by the cyoE gene (29).
Subunit I serves as the catalytic center for dioxygen reduction and proton pumping and binds all the redox metal centers in the oxidase complex, low spin heme b (cytochrome b 563.5 ), high spin heme o (cytochrome o), and a copper ion (Cu B ) (see Refs. 30 -32 for reviews). Heme o and Cu B are antiferromagnetically coupled and form the heme-copper binuclear center (18,33). The axial ligands of the metal centers have been identified by site-directed mutagenesis studies (34 -38) and were recently confirmed by crystallographic studies on cytochrome c oxidases from P. denitrificans (15) and bovine heart (39). Based on photoaffinity cross-linking studies, a quinonebinding site(s) was assigned to reside in subunit II (40). Therefore, electrons seem to travel from the quinol oxidation site in subunit II to low spin heme b in subunit I (41,42), similar to electron transfer in cytochrome c oxidase (i.e., from the Cu A site in subunit II to low spin heme a in subunit I). In contrast, the functional role of subunits III and IV remains unknown, although preliminary deletion studies suggested that they are indispensable for functional expression of UQO (43).
In the present study, we examined the functional role of subunit IV of UQO by deletion analysis and found that helix I and its neighboring sequences are dispensable for the catalytic function of UQO. Our deletion studies strongly suggested that subunit IV is adjacent to the Cu B -binding site of subunit I, and cross-linking studies favored that subunit IV is in close proximity to subunit III. Based on these results, we propose the location and functional role of subunit IV in the oxidase complex.

MATERIALS AND METHODS
Bacterial Strains, Plasmids, Growth Media, and DNA Manipulations-E. coli strains, plasmids, and growth media used in this study and DNA manipulations were as described previously (27,29,34).
Construction of Other Deletion Mutants-⌬D1 was prepared by blunt-end ligation at the NcoI site in the cyoD gene using T4 DNA polymerase. In a resultant construct, a frameshift occurred due to a termination codon TAA after the artificial Pro 14 (see Fig. 2). To eliminate Gly 13 to Ala 44 , ⌬D3 was constructed by blunt-end ligation at the NcoI and EagI sites using T4 DNA polymerase and mung bean nuclease, respectively.
Subsequently, all the junction sites of these pCYO6 derivatives were confirmed by direct plasmid DNA sequencing. Then the EcoRI-SphI DNA fragment of the mutant pCYO6 was introduced into the corresponding sites of a mini-F plasmid, pMFO21 (29). The mutant cyo operons were expressed in either ST4676 (⌬cyo cyd ϩ ) or ST2592 (⌬cyo ⌬cyd) as described previously (29).
Cross-linking Conditions-Cross-linking experiments with dithiobis(succinimidyl propionate) (DSP) (Pierce), which can covalently link two amino groups within 12 Å, were performed according to the method described by Jarausch and Kadenbach (44) with slight modifications. Purified UQO was diluted with 0.2 M triethanol amine/HCl (pH 8.5) containing 0.1% (w/v) sucrose monolaurate (Mitsubishi-Kasei Food Co., Tokyo) at a final concentration of 1 mg/ml and was incubated on ice for 2.5 h with DSP in dimethyl sulfoxide at final concentrations of 0.25, 1.0, and 2.5 mM (50-, 200-, and 500-fold molar excess, respectively). Reaction was terminated by the addition of ammonium acetate to a final concentration of 50 mM, and then the incubation was continued for 0.5 h on ice. Then the reaction mixture was diluted 10-fold with 0.1% (w/v) sucrose monolaurate and centrifuged at 20,000 ϫ g and at 4°C for 5 min. The supernatant was removed carefully to eliminate excess DSP in the precipitate. The enzyme was recovered by precipitation with 40% (v/v) ethanol and washed twice with distilled water.
Two-dimensional SDS-Polyacrylamide Gel Electrophoresis-The enzyme was dissolved in 65 mM Tris-HCl (pH 6.8) containing 8% (w/v) SDS and 10% (v/v) glycerol and then incubated at room temperature for 1 h followed by incubation at 37°C for 0.5 h. Immediately after the incubation, 0.5 g of the enzyme was subjected to two-dimensional SDS-polyacrylamide gel electrophoresis (45). The first dimension was run in a disc gel (5-mm internal diameter ϫ 9 cm) with 7 cm of 12.5% (w/v) acrylamide separation gel and 1 cm of stacking gel. After the first dimension electrophoresis, gels were taken off from a glass cylinder and incubated at 4°C for 40 min in stacking gel buffer (125 mM Tris-HCl (pH 6.8) containing 0.125% (w/v) SDS and 2% (v/v) 2-mercaptoethanol). Then the gel was placed in a upper space of the second dimension slub gel and embeded in 1.5% (w/v) agarose gel containing stacking gel buffer. The second dimension was run in a mini-slab gel containing 16% (w/v) acrylamide in the separation gel. After the electrophoresis, separation gels were fixed and stained with silver as described previously (46).
Miscellaneous-Aerobic complementation test, preparation of cytoplasmic membrane vesicles, purification of UQO, Western blotting analysis, and spectroscopic measurements were as described previously (29,34). Restriction endonucleases and other enzymes for DNA manip-ulations were purchased from Takara Shuzo Co. (Kyoto, Japan) or New England BioLabs, Inc. Other chemicals were commercial products of analytical grade.

Construction of the CyoD Deletion Mutants and Their
Catalytic Activities in Vivo-Subunit IV (CyoD) of UQO is composed of 109 amino acid residues and has three putative transmembrane helices I-III with the N-terminal end in the cytoplasm (21, 26) (Fig. 1). CyoD (21) is homologous to subunit IV of ba-type ubiquinol oxidase from A. aceti (13) and P. denitrificans (20), aa 3 -type menaquinol oxidase from B. subtilis (23), and caa 3 -type cytochrome c oxidase from thermophilic Bacillus PS3 (17), B. subtilis (24), and alkalophilic B. firmus OF4 (25). Among these seven subunit IV sequences, 17-53% amino acid residues (32% in average) are conserved, and Phe 22 , Leu 24 , Leu 28 , and Thr 29 in helix I and Ala 54 , Gln 57 , Phe 65 , His 67 , and Met 68 in cytoplasmic loop II/III are found to be invariant (Fig.  1). To probe the topological domain(s) of subunit IV essential for its functional or structural role(s), we constructed the nine CyoD deletion mutants (⌬D1-⌬D9) by gene engineering (Fig. 2). In ⌬D1, ⌬D8, and ⌬D9 stop codons were introduced at the respective positions to terminate translation, whereas in other mutants the corresponding DNA fragments were deleted from the cyoD gene (Fig. 2).
The catalytic activities of the mutant oxidases were examined by genetic complementation test (Table I). The mutant cyoD genes were subcloned into a single copy expression vector pMFO21 (cyoABCDE ϩ ) and expressed in a terminal oxidase double deletion strain ST2592 (⌬cyo ⌬cyd), which cannot grow aerobically on nonfermentable carbon source via oxidative phosphorylation. The wild-type strain ST2592/pMFO1 grew aerobically on minimal 0.5% glycerol plates, whereas the vector control strain ST2592/pHNF2 (34,43) and the CyoD deletion mutants except ⌬D2 and ⌬D3 failed to grow aerobically (Table  I). This result indicates that CyoD is required for functional expression of UQO and that the N-terminal cytoplasmic domain and transmembrane helix I are dispensable for the function of subunit IV. These mutations did not affect their temperature sensitivity for growth (23,30,37, and 42°C) and were not suppressed by supplementation of copper ions at final concentrations of 0.1 and 1 mM. In addition, we found that the expression levels of subunit I in the deletion mutant membranes were not affected at all (Table I), indicating that the defective CyoD deletions did not alter the stability of the mutant enzymes.

Effects of the CyoD Deletions on the Redox Metal Centers of UQO-
The effects of the CyoD deletions on low spin heme b and high spin heme o were examined by UV-visible spectroscopy using cytoplasmic membrane vesicles isolated from the UQO deletion strain ST4676 (⌬cyo cyd ϩ ) harboring the pMFO1 derivatives. Low spin heme b (cytochrome b 563.5 ), which exhibits a split ␣ peak at 556 and 563.5 nm in dithionite-reduced minus air-oxidized difference spectra at 77 K (18,34), was quantitated by an amplitude of 563.5 nm peak in their second order finite difference spectra ( Fig. 3a and Table I). The content of high spin heme o (cytochrome o) was estimated from reduced CO-bound minus dithionite-reduced difference spectra at room temperature ( Fig. 3b and Table I). The copper content (i.e., Cu B of the binuclear center) was determined by atomic absorption spectroscopy (Table I).
As expected from the genetic complementation test, only ⌬D2 and ⌬D3 mutant oxidases retained all the redox metal centers comparable with the wild-type levels. In contrast, all the functionless deletion mutant oxidases, ⌬D1 and ⌬D4 to ⌬D9, reduced the amplitudes of low spin heme and high spin heme signals to two-thirds of the wild-type control and were completely or largely devoid of bound copper ions. It should be noted that the defective mutant oxidases show 1-3 nm blue shifts of the Soret peak in CO-binding difference spectra (Fig.  3b). However, these blue shifts were smaller than those of the defective cyoE mutant oxidases and were not associated with replacement of high spin heme o by heme B (27)(28)(29). 2 These results indicate that the C-terminal two-third of subunit IV (Val 45 to His 109 ) including transmembrane helices II and III (see Fig. 1) is required for copper binding to the binuclear center in subunit I during biosynthesis or assembly of the oxidase complex. This suggests that subunit IV is located in close proximity to subunit I. Cross-linking Analysis of Subunit Organization in the UQO Complex-To probe the proximity of subunits in the UQO complex, the purified enzyme (18) was subjected to chemical crosslinking with DSP, an amino group-directed homobifunctional and thiol cleavable reagent. Fig. 4 shows a separation profile of

TABLE I Characterizations of the CyoD deletion mutant oxidases
The in vivo catalytic activity of the mutant oxidases was evaluated by the genetic complementation test. The amounts of cytochrome o and low spin heme b (cytochrome b 563.5 ) were determined by CO-binding difference spectra and 77 K redox difference spectra in Fig. 3, respectively. Specific content of copper was estimated by atomic absorption spectroscopy. Expression level of the mutant UQOs was examined by Western blotting analysis using the anti-subunit I antiserum, followed by densitometric analysis of the subunit I band using a Shimadzu CS-9000 double-wavelength flying spot scanner. Strain ST4676 harboring mini-F plasmid pMF01 (cyo ϩ ) and pHNF2 with no insert was used as the wild-type control (wild type) and the negative control (control), respectively.  3. Second order finite difference spectra of dithionitereduced minus air-oxidized difference spectra (a) and CO-reduced minus reduced difference spectra (b) of cytoplasmic membranes isolated from the subunit IV deletion mutants. a, spectra were recorded at 77 K with a Shimadzu UV-3000 double-wavelength spectrophotometer at a protein concentration of 3 mg/ml in 120 mM Tris-HCl (pH 7.4). Measurements were done with a bandwidth of 1 nm, a light path of 2 mm, and a scan speed of 50 nm/min. b, determinations were carried out at room temperature at a protein concentrations of 0.5 mg/ml with a light path of 10 mm. Strain ST4676 (⌬cyo cyd ϩ ) carrying mini-F plasmid pMFO1 (cyo ϩ ) or pHNF2 was used as the wild-type control (WT) and a vector control (control), respectively. the cross-linked products on two-dimensional SDS-polyacrylamide gel electrophoresis. The cross-linking between subunits III and IV was reproducibly detected in the presence of DSP at final concentrations of 0.5 and 1.0 mM (data not shown) and 2.5 mM (Fig. 4). In addition, weak cross-linking signals between subunits I and II and between subunits I and V (28-kDa polypeptide) were also evident. These biochemical evidence support a close association of subunit IV with subunit III in the UQO complex.

DISCUSSION
Functional Role of Subunit IV-There are three papers reporting a role of subunit IV in bacterial heme-copper terminal oxidases. Very recently, Prochaska et al. (47) showed that 75% removal of subunit IV by proteolytic digestion did not affect proton pumping activity of caa 3 -type cytochrome c oxidase from thermophilic Bacillus PS3. It is consistent with the previous observations that the two-subunit preparations of the E. coli UQO (3) and of cytochrome c oxidase from P. denitrificans (48,49) have been reported to be fully functional; therefore, subunit IV is dispensable for the catalytic function of the bacterial heme-copper terminal oxidases. On the contrary, Papa and colleagues (50,51) found that a deletion in the B. subtilis qoxD gene encoding subunit IV of aa 3 -type menaquinol oxidase significantly depressed respiration and proton pumping but did not affect the content of aa 3 -type cytochrome in the membranes. Thus, they concluded that subunit IV is essential for the function of aa 3 -type menaquinol oxidase in B. subtilis. Present deletion studies on subunit IV of UQO revealed that only the C-terminal two-thirds is required for functional expression of the enzyme and that subunit IV assists the Cu B binding to the binuclear center in subunit I during biosynthesis or assembly of the oxidase complex (Table I). In conclusion, subunit IV of bacterial heme-copper terminal oxidases is indispensable for the functional expression of the oxidase complex, although it can be removed in vitro without a loss of the enzymatic activity (3,48,49).
Subunit Organization of the UQO Complex-Recent crystallographic studies on cytochrome c oxidase revealed that threedimensional organization of subunits I, II, and III is nearly the same in 4-subunit enzyme from P. denitrificans (15) and 13-subunit enzyme from bovine heart (39,52). Because of homology of these subunits, UQO must have a similar molecular structure though helices I and II of subunit III are transferred at the C-tereminus of subunit I on the gene level (21). By gene fusion experiments, Ma et al. (53) demonstrated that transmembrane helices of subunits II, I, and III (i.e., products of the cyoA, cyoB, and cyoC genes) are consecutively arranged in the UQO complex in a tail-to-head manner, as found in cytochrome c oxidases (15,52).
Subunit IV of the P. denitrificans enzyme consists of a single transmembrane helix with the N terminus in the cytoplasm and is not homologous to subunit IV of other bacterial oxidases and any nuclear-encoded subunits of eukaryotic oxidases (15,16). In the P. denitrificans enzyme, the upper part of the transmembrane helix of subunit IV has indirect contacts with subunit III mediated by a bound phospholipid, whereas the nine C-terminal residues touch residues of transmembrane helix III of subunit III (i.e., helix I of UQO's subunit III) and helices VII and VIII of subunit I (15). Therefore, subunits I and III of UQO seem to form a similar binding pocket for subunit IV (Fig. 5).
Cross-linking studies of UQO with DSP as a chemical probe showed that the amino group(s) of subunit IV (i.e., the N terminus, Lys 16 of the N-terminal tail and Lys 71 of loop II/III), which all are expected to be in cytoplasmic domains (26) (Fig.  1), is in close proximity (within 12 Å) to one of the amino groups of subunit III. When subunits II, I, and III of UQO were fused at the gene level in this order, the resultant four-subunit oxidase complex was shown to be functional (53). Similarly, we fused subunits III and IV by linking the C terminus (helix V) of subunit III at the periplasm with periplasmic loop I/II of subunit IV and found that the fused UQO was inactive, 3 suggesting that helix V of subunit III is distal to subunit IV. Our deletion studies also suggest that subunit IV of UQO is close to the Cu B binding site (Table I), which is provided by helices VI to VIII of subunit I (15, 30 -32, 34 -39). Cu B is ligated by His284 in helix VI and His 333 and His 334 in helix VII of UQO. In addition, both the N-terminal tail and helix I are shown to be dispensable for the function (i.e., ⌬D2 and ⌬D3). Perturbations of high spin heme o in the subunit IV mutants support possible interaction of subunit IV with the binuclear center (Fig. 3b), because a complete loss of the Cu B center in the His 333 mutant of subunit I did not qualitatively alter a CO-binding difference spectrum (54). This can be interpreted that proper arrangement of amino acid side chains seems to be strictly required at the binuclear center (55)

Role of Subunit IV in bo-type Quinol Oxidase
Based on these findings, we propose that subunit IV of UQO is present in a cleft between subunits I and III (Fig. 5), as is the case in cytochrome c oxidase from P. denitrificans (15). Because helices II and III of subunit IV are indispensable for the function, these helices may have direct contacts with either subunit I or III. Current alanine-scanning mutagenesis and suppressor mutation studies indicate that helix III of subunit IV interacts directly or indirectly with helices VII and VIII of subunit I. 3 In conclusion, we present here the possible functional role of subunit IV and its structural domains needed for subunit interactions in the UQO complex. Through these intersubunit interactions, subunit IV seems to assist for the Cu B binding to subunit I during biosynthesis or assembly of UQO. Further molecular biological studies on subunit IV will provide a clue for understanding of the folding process of UQO and the functional roles of the nuclear-encoded subunits in eukaryotic cytochrome c oxidases.