(Received for publication, March 4, 1997, and in revised form, March 27, 1997)
From the Department of Biological Sciences, Cytochrome bo-type ubiquinol oxidase
is a four-subunit heme-copper terminal oxidase and functions as a
redox-coupled proton pump in the aerobic respiratory chain of
Escherichia coli. On the basis of deletion and chemical
cross-linking analyses on subunit IV, we proposed that subunit IV is
essential for CuB binding to subunit I and that it is
present in a cleft between subunits I and III (Saiki, K., Nakamura, H.,
Mogi, T., and Anraku, Y. (1996) J. Biol. Chem. 271, 15336-15340).
To extend previous studies, we carried out alanine-scanning mutagenesis
for selected 16-amino acid residues in subunit IV to explore
subunit-subunit interactions in bo-type ubiquinol oxidase. We found that only the replacement of Phe83 in helix III
resulted in the reduction of the catalytic activity but that this did
not significantly affect the UV-visible spectroscopic properties and
the copper content. This suggests that individual amino acid
substitutions, including the six invariant residues, are not enough to
alter such properties of the metal centers. Extragenic suppressor
mutations were isolated for the Phe83 In the aerobic respiratory chain of Escherichia coli,
cytochrome bo-type ubiquinol oxidase
(UQO)1 is predominantly expressed under
highly aerobic growth conditions (1) and functions as a redox-coupled
proton pump (2). UQO consists of four different subunits (3, 4) that
are encoded by the cyoABCDE operon (5). Subunits I, II, and
III (CyoB, CyoA, and CyoC, respectively) of UQO are homologous to the
counterparts of mitochondrial and bacterial cytochrome c
oxidases, and thus UQO belongs to the heme-copper terminal oxidase
superfamily (5, 6). Subunit I binds all three redox metal centers, low
spin heme b (cytochrome b563.5), high
spin heme o (cytochrome o), and a copper ion
(CuB) (7-11). The latter two metal centers form the heme-copper binuclear center that serves as the catalytic center for
dioxygen reduction and proton pumping. Site-directed mutagenesis studies demonstrated that six invariant histidines in subunit I serve
as the axial ligand of the metal centers and that a bundle of helices
VI-VIII and X form a binding pocket for the binuclear center. Subunit
II does not have any metal center but seems to provide the ubiquinol
oxidation site (12). Subunits III and IV can be removed from the UQO
complex without loss of the catalytic activity (13); however, they are
essential for the functional expression of the UQO complex
(14).2 Subunit IV (CyoD), whose homologues
can be found only in the bacterial terminal oxidases (5, 14), is
proposed to be located in a cleft between subunits I and III and to
assist the CuB binding to subunit I (14). Deletion analysis
of CyoD revealed that the functional domain is located in the
C-terminal two-thirds (Val45-His109)
containing helices II and III (14) (see Fig. 1). The cyoE gene, which presents at the 3
To extend previous studies on subunit IV (14), we carried out
alanine-scanning mutagenesis to identify the functional amino acid
residues that are involved in interactions between subunits I and IV.
Subsequently, we isolated and characterized extragenic suppressor
mutations for the defective subunit IV mutation, namely D-F83A.3 We found that extragenic mutations
are located in transmembrane helices VII and VIII of subunit I. Thus,
we conclude that subunit IV is in close vicinity to the CuB
binding site of subunit I and functions as a domain-specific molecular
chaperon in the UQO complex.
E. coli strains, plasmids, and growth
media used in this study and DNA manipulations were as described
previously (8, 14, 17). Ampicillin was supplemented to growth media at
a final concentration of 15 µg/ml for mini-F plasmids or 50 µg/ml
for multicopy plasmids.
Alanine-scanning
mutagenesis of CyoD was carried out using phagemid pCYOF6 and mutagenic
primers in the range of 21-30 nucleotides (14). The 0.4-kb
EcoRI-EagI or the 0.4-kb
EagI-EcoRV fragment containing the
cyoD mutations was isolated from phagemid DNAs and replaced
with the counterpart in the wild-type multicopy plasmid pCYO6 (14). The
nucleotide sequence of the corresponding region in the recombinant
plasmid was confirmed by direct plasmid sequencing (8). For expression
of the mutant enzymes, the EcoRI-SphI fragment of
the mutant pCYO6 was introduced into the corresponding sites of mini-F
plasmid, pMFO21 (14). To facilitate the subcloning of the entire
cyo coding region, a single copy expression vector pHNFO11P,
which carried the gene-engineered unique ClaI and
BamHI sites before and after the cyo promoter,
respectively, was constructed from pMFO1 (8). Similarly, a multicopy
expression vector pBR4 was constructed from pHN3795-1 (18).
pHNFO11P-O9B and pBR4-O9B are the derivatives of pHNFO11P and pBR4,
respectively, and contain the whole cyo operon.
ST2592 ( pCYO64-DHisCt was
constructed by cloning of a synthetic His5 tag linker at
the 3 Cytoplasmic
membranes were isolated from ST4676 ( The membranes were diluted to a protein
concentration of 1.5 mg/ml with 50 mM Tris-HCl (pH 7.4)
containing 1 mM phenylmethylsulfonyl fluoride, 0.5 M NaCl, and 1.5% sucrose monolaurate SM-1200 (SM) (Mitsubishi-kasei Food Co., Tokyo) and solubilized with stirring at
4 °C for 1 h. The mixture was centrifuged at 180,000 × g for 1 h, and all (mutants) or half (wild type) of the
supernatant was passed twice through 12 ml of Ni-nitrilotriacetic acid
(NTA)-agarose (Qiagen Inc.) at a flow rate of ~6 ml/min. After
washing Ni-NTA-agarose resins with 150 ml of wash buffer (50 mM Tris-HCl (pH 7.4) containing 0.1 mM
phenylmethylsulfonyl fluoride, 0.5 M NaCl, and 0.1% SM), UQO was eluted by 30 ml of wash buffer containing 0.1 M
imidazole. Peak fractions were desalted and concentrated by
ultrafiltration using a Macrosep 100 (Filtron Technology). Contaminated
phospholipids were removed by anion-exchange chromatography (20) after
a 20-fold dilution with 50 mM Tris-HCl (pH 7.4) containing
0.1 mM phenylmethylsulfonyl fluoride and 1% SM.
Other analytical procedures were as described
previously (8, 13, 17). Restriction endonucleases and other enzymes for DNA manipulations were purchased from Takara Shuzo Co. (Kyoto, Japan)
or New England BioLabs, Inc. Other chemicals were commercial products
of analytical grade.
Subunit IV (CyoD) of the
UQO complex is composed of 109 amino acid residues and has three
transmembrane helices (I-III) with the N terminus in the cytoplasm
(Refs. 5 and 15; Fig. 1). Homologues of CyoD are found
only in bacterial quinol and cytochrome c oxidases (5, 14,
21). Among eight known subunit IV sequences, 17-51% of the amino acid
residues are conserved and Phe22, Leu24,
Leu28, and Thr29 in helix I, Ala54
and Gln57 in helix II, and Phe65,
His67, and Met68 in the cytoplasmic loop
II-III are found to be invariant and appear to be essential for the
UQO function. However, our recent deletion analysis demonstrated that
the N-terminal cytoplasmic domain, helix I, and the loop I-II are
dispensable in CyoD (14). Subunit IV seems to be present in a cleft
between subunits I and III of the UQO complex (14), as found in a
crystal structure for bacterial aa3-type
cytochrome c oxidase (22). To explore subunit-subunit
interactions in the heme-copper terminal oxidases, we extended the
previous studies on subunit IV and carried out alanine-scanning
mutagenesis on 12 conserved amino acid residues and 4 aromatic or
charged amino acid residues within or at the boundary of transmembrane
helices of subunit IV (Fig. 1, Table I).
Table I.
Characterizations of the CyoD alanine mutants in cytoplasmic membranes
Department of Life Science,
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
Ala mutation of
subunit IV and identified as missense mutations in helices VII and VIII
in subunit I. These observations provide further support for specific
interactions of subunit IV with helix VII and/or VIII, the
CuB binding domain, of subunit I and suggest that subunit
IV functions as a domain-specific molecular chaperon in the oxidase
complex.
-end of the cyo operon,
encodes heme O synthase, which supplies heme O specifically to the
binuclear center of UQO (16).
Fig. 1.
Topological model of subunit IV (CyoD) of
bo-type ubiquinol oxidase from E. coli
(modified from Chepuri and Gennis (15)). Amino acid residues
conserved in six out of eight subunit IV sequences of the bacterial
heme-copper terminal oxidases (5, 14, 21) are shown in the
circles. Phe83, whose substitution by analine
reduced the catalytic activity, is indicated by the
diamond.
[View Larger Version of this Image (28K GIF file)]
Bacterial Strains, Plasmids, Growth Media, and DNA
Manipulations
cyo
cyd)/pBR4-O9B-DF83A, which had been anaerobically grown
overnight in 2 × YT medium supplemented with 0.5%
NaNO3, was inoculated into 5 ml of minimal medium A (19)
containing 0.5% glycerol, 0.5% glycerol plus 0.05% glucose, or 0.5%
glucose and then cultured aerobically at 37 °C for 3 days. Cultures
with significant turbidity were streaked on minimal/0.5% glycerol
plates, and revertants were isolated after incubation at 37 °C for 2 days. Assignment of the DNA segments carrying the suppressor mutations was carried out by subcloning of the restriction fragments of plasmid
DNA isolated from revertant strains into pBR4-O9B-DF83A. Recombinant
plasmids were then introduced into ST2592 to confirm the ability of
transformants to grow aerobically on minimal/glycerol plates. The
identified DNA fragments were subcloned into pUC119 for DNA sequencing
analysis with a model 373A DNA sequencer (Applied Biosystems Inc.).
-end of the cyoD gene using the genetically engineered
unique SplI site in pCYO64 (17). Then, the 1.7-kb EcoRI-SphI fragment of pCYO64-DHisCt was
subcloned into pHNF11OP-O9B to give pHNFO11P-O9B-DHisCt.
cyo
cyd+) harboring the pMFO11P-O9B or pBR4-O9B
derivatives (14, 17). A large scale isolation of the membranes was
performed as described previously (20).
Alanine-scanning Mutagenesis of CyoD
Mutant
Catalytic activity
Cyt o
Cu
Cyt
b563.5
Subunit I
nmol/mg
protein
nmol/mg protein
Wild
type
+
0.39
0.33
+++
+++
Control
0.05
0.01
CyoD
(
D5)a
0.26
0.01
++
+++
D-Y18A
+
0.32
0.30
+++
+++
D-F22A
+
0.32
0.31
+++
++
D-T29A
+
0.30
0.38
+++
+++
D-F33A
+
0.41
0.44
+++
+++
D-W34A
+
0.30
0.30
+++
++
D-Q57A
+
0.32
0.29
+++
+++
D-H61A
+
0.27
0.30
+++
+++
D-F65A
+
0.35
0.38
+++
+++
D-H67A
+
0.29
0.31
+++
+++
D-M68A
+
0.30
0.35
+++
+++
D-K71A
+
0.37
0.48
+++
+++
D-E74A
+
0.33
0.33
+++
+++
D-F81A
+
0.31
0.30
+++
+++
D-F83A
0.29
0.28
++
+++
D-W97A
+
0.35
0.22
+++
+++
D-W100A
+
0.29
0.41
+++
+++
a
Data taken from Ref. 14.
Catalytic activities of the mutated oxidases were examined by genetic
complementation test (Table I). Using a single copy vector pMFO21, the
cyoD mutant operons were expressed in a terminal oxidase-deficient strain ST2592 that cannot grow aerobically on nonfermentable carbon sources via oxidative phosphorylation. The wild-type and all the CyoD alanine mutants except D-F83A grew aerobically on minimal/0.5% glycerol plates, whereas the D-F83A mutant
and D5 mutant, which carries the complete cyoD deletion (14), as well as the vector control strain ST2592/pHNF2 (8) failed to
grow aerobically (Table I). This result indicates that alanine can be
substituted for the 15 amino acid residues examined in the present
study without a significant loss of the UQO functions. To test the gene
dosage effect, we overexpressed the D-F83A mutant UQO in
ST2592/pHN3795-1-DF83A and found that it can grow slowly on
minimal/glycerol plates (Table II), indicating that the
D-F83A mutant oxidase is partially functional.
|
Effects of
the alanine substitutions in CyoD on the redox metal centers were
examined using cytoplasmic membrane vesicles isolated from strain
ST4676 (cyo cyd+) harboring the pMFO1
derivatives. Low spin heme b was quantitated as cytochrome
b563.5 (8, 20) by an amplitude of the 563.5-nm peak in the second order finite difference spectra of redox difference spectra at 77 K. The content of high spin heme o was
estimated as cytochrome o (20) from CO binding difference
spectra at room temperature. The CuB content was estimated
as the amount of bound copper ions in the membranes (8, 20). As shown
in Table I, all of the mutations in CyoD did not largely alter the
amounts of high spin and low spin heme signals and CuB. The
amounts of the three metal centers in the D-F83A mutant were reduced to
80% that of the wild-type levels, whereas the defective CyoD deletions completely lost the CuB center (14).
To identify a defect(s) caused by the D-F83A mutation,
we isolated extragenic suppressor mutants based on their ability to grow aerobically on minimal/glycerol medium. To facilitate genetic manipulations of the large cyo operon (i.e. 5 kb), we constructed a multicopy expression vector pBR4-O9B where the
2.2-kb 5- and the 2.5-kb 3
-noncoding regions of the cyo
operon in pHN3795-1 (18) have been deleted, but it contained five
genetically engineered restriction sites (NheI,
ApaI, XhoI, MluI, and
Bsu36I)4 in the structure genes
and ClaI and BamHI sites around the
cyo promoter (Fig. 2). Spectroscopic analysis
of the cytoplasmic membranes showed that the expression level of UQO in
strain ST4676 by multicopy vector pBR4-O9B was slightly less than those
by single copy vectors pMFO1 and pHNFO11P-O9B (~85%; Tables I and
II) although the generation time of ST2592/pBR4-O9B in minimal/glycerol
medium was nearly identical to that of ST2592/pHNFO11P-O9B (data not
shown). Accordingly, the ST2592/pBR4-O9B-DF83A mutant was also unable
to grow aerobically in minimal/glycerol medium (Table II).
We isolated four spontaneous revertants independently from strain ST2592/pBR4-O9B-DF83A that had been grown in minimal medium containing 0.5% glycerol plus 0.05% glucose (R1, R3, and R4) or minimal/0.5% glycerol medium (R2) (Table II). Plasmid DNAs isolated from the revertants were digested with convenient restriction enzymes as indicated in Fig. 2a, and the resultant fragments were subcloned into the corresponding sites of pBR4-O9B-DF83A to test the ability of recombinant plasmids to support the aerobic growth of ST2592 on minimal/glycerol plates. After several rounds of screening, the suppressor mutation of the R1 revertant was mapped in the 0.14-kb XhoI-MluI fragment and those of the R2, R3, and R4 revertants were mapped in the 0.29-kb MluI-Bsu36I fragment (Fig. 2a). Sequencing analysis showed that all the suppressor mutations were extragenic missense mutations in the subunit I (cyoB) gene as follows. A codon for Leu326 (CTG) was changed to Pro (CCG) in R1 (B-L326P), that for Ile357 (ATC) to Thr (ACC) in R2 (B-I357T), and that for Val361 (GTG) to Met (ATG) in R3 and R4 (B-V361M) (Fig. 2b). Leu326 is located in transmembrane helix VII where His333 and His334 ligate CuB at the periplasmic end (9-12). Ile357 and Val361 are present in helix VIII, which also consists of a binding pocket for the heme-copper binuclear center (8-11). These results indicate the presence of specific interactions of subunit IV with transmembrane helices VII and/or VIII of subunit I in the UQO complex.
Effects of Suppressor Mutations on the Metal CentersThe suppressor mutations were subcloned into pHNFO11P-O9B, and their effects on the high spin and low spin hemes were examined in the presence and the absence of the original D-F83A mutation. Western blotting analysis of the cytoplasmic membranes indicated that all of the mutations did not affect the stability of the mutant UQOs (Table II). UV-visible spectroscopic analysis of the membranes showed that reductions in the amounts of cytochromes o and b563.5 by the D-F83A mutation were partially suppressed by the B-V361M whereas all three individual suppressor mutations significantly reduced heme signals although they did not affect the aerobic growth of the mutant strains (Table II).
Characterizations of the Purified Mutant UQOsThe mutant UQOs
were purified by Ni-NTA-agarose chromatography using the
His6 tag including His109 at the C terminus of
CyoD (D-HisCt). The purity of the isolated D-HisCt UQO from the small
scale culture was estimated to be ~90%. Unexpectedly, we found that
the wild-type and mutant UQOs without the His tag can be purified
similarly to the His-tagged UQO (Fig. 3). Upon the
addition of the His tag, the apparent molecular mass of subunit IV
increased from 13.5 to 14.5 kDa in 14% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (lanes 1 and
2). Alternatively, the complete CyoD deletion eliminated a
13.5-kDa polypeptide (lane 4), providing further support for
identification of subunit IV as CyoD. It should be noted that a 7-kDa
polypeptide seems to be associated with the UQO complex even after
DEAE-5PW HPLC (lanes 5-9) and size exclusion HPLC (data not
shown). It can be seen in the UQO preparation obtained by
anion-exchange HPLC (20) when the gel was carefully fixed (data not
shown). In contrast, subunit V (20) is almost absent in the UQO
preparation with the small scale protocol (lanes 1-4) as
compared with the large scale protocol (lanes 5, 7, and
8). These results indicate that subunit IV is not required
for the assembly of the UQO complex and that the mutations in subunits
I (B-L326P, B-I357T, and B-L361M) and IV (D-F83A) do not disrupt the
quaternary structure of the UQO complex.
The effects of the D-F83A and suppressor mutations on the metal centers
and the catalytic activity were further studied using the purified
enzyme obtained with the large scale protocol. The isolated UQOs were
estimated to be ~70% pure based on their heme contents (Fig.
4, Table III). As expected from the
complementation test (Table II), the D-F83A mutation reduced
ubiquinol-1 oxidase activity to 55% that of the wild-type control
(Table III). The B-I357T and B-V361M subunit I mutations completely
suppressed the defects caused by the D-F83A mutation in subunit IV,
whereas the B-L326P mutation reduced CO binding activity of high spin heme o and quinol oxidase activity to 79 and 41%, respectively, that
of the wild-type levels (Fig. 4, Table III). The peak of pyridine
ferrohemochrome from the mutant UQOs was found at 554.5 nm, the same as
the wild-type enzyme that binds 1 mol each of hemes B and O (20),
suggesting that these mutations did not alter the specificity of the
heme binding sites. In addition, the amount of cytochrome
b563.5 remained unaltered (Fig. 4, Table III),
indicating that a defect is distal to the low spin heme binding site.
These results suggest that the defect of the D-F83A mutation cannot be
attributed to a loss of any metal centers.
|
Recently, the crystal structures of aa3-type cytochrome c oxidases have been solved at atomic resolution for the 4-subunit enzyme from Paracoccus denitrificans (22) and for the 13-subunit enzyme from bovine heart (23). Coordination chemistry of the redox metal centers revealed by molecular biological studies on bacterial enzymes (8-11) was confirmed by crystallographic efforts (22, 23). This structural information provides us with new insights into the molecular mechanism of redox-coupled proton pumping. Subunits I and II are known to serve as the redox reaction centers in the oxidase complex, whereas the functional role of other small subunits remains obscured. The nuclear encoded "supernumeral" subunits of eukaryotic oxidases are unlikely to be involved in electron transfer or proton pumping (23) but may serve as regulators of the cytochrome c oxidase activity through ATP binding and/or the switching of tissue-specific isoforms (24). Subunit III of bacterial cytochrome c oxidase has been suggested to be required for assembly of subunits I and II into an active complex (25).
Comparison of the bacterial subunit IV sequences indicates that helix I
to the loop II/III is more conserved than the other portions; however,
our deletion analysis of CyoD showed that the N-terminal one-third
(i.e. the N terminus to the loop I/II) of subunit IV is
dispensable in the catalytic function of UQO (14). CyoD was suggested
to assist the binding of copper ions to the CuB site during
the folding process of subunit I or assembly of the UQO complex (14).
Alanine-scanning mutagenesis demonstrated that only a substitution of
Phe83 in helix III reduced the catalytic activity without
significant effects on the redox metal centers and the assembly of the
UQO complex. Since Phe or Tyr is conserved at position 83, aromatic side chains at the cytoplasmic end of helix III may be crucial for
folding of subunit IV or interaction with subunit I. Aromatic side
chains on intrinsic membrane proteins are frequently found within
membrane boundaries (26) and may be stabilized by the choline head
group of phospholipids through cation- interactions (27). Individual
replacements of other conserved amino acid residues in subunit IV were
not enough to alter subunit-subunit interactions; therefore, the
functional role of subunit IV seems to be structural rather than
enzymatic in contrast to CyoE (16, 17).
Recently, Ni-NTA affinity chromatography was shown to be
efficient for small scale purification to near homogeneity of the His-tagged UQO and cytochrome c oxidase from
Rhodobacter sphaeroides (28)5
and may be suitable for isolation of unstable mutant enzymes. UQO with
the His-tagged CyoD was purified to ~90 and 70% using small scale
and large scale protocols, respectively (Fig. 3, Table III). However,
it should be noted that an endogenous cation binding site in the
C-terminal periplasmic domain of subunit II (29) seems as efficient as
the His tag introduced at the C terminus of CyoD (Fig. 3). Purification
of the CyoD-deficient mutant oxidase by this method demonstrated that
subunit IV is not required for the assembly of subunits I-III (CyoD
(
D5) in Fig. 3, lane 4). Based on the deletion and
chemical cross-linking analyses on CyoD, we proposed previously that
subunit IV is located in a cleft between subunits I and III (14),
similar to the four-subunit cytochrome c oxidase of P. denitrificans (22). Specific interactions of helices II and III of
subunit IV with subunit I has been suggested from a complete loss of
the CuB center in the CyoD deletion mutants (14). In the
present study, we identified the extragenic suppressor mutations for
the D-F83A mutation in helices VII and VIII of subunit I (B-L326P,
B-I357T, and B-V361M). It has been shown previously that helices VI,
VII, and VIII of subunit I form the CuB binding pocket and
contain the axial ligands for CuB (His284,
His333, and His334) (8-11). These results
suggest the presence of direct interactions of subunit IV with subunit
I.
In conclusion, subunit IV is dispensable for enzyme activity (13) but is required for assembly of the oxidase complex into a catalytically active form through interactions with the CuB binding domain of subunit I.
We thank H. Nakamura of Riken for valuable comments and R. B. Gennis of the University of Illinois for unpublished observations.