Cloning of the cta operon from alkaliphilic Bacillus firmus OF4 and characterization of the pH-regulated cytochrome caa3 oxidase it encodes.

We have cloned and sequenced the DNA of alkaliphilic Bacillus firmus OF4 encompassing the cta operon that encodes a pH-regulated cytochrome caa3 oxidase. The gene organization is identical with that of the homologous Bacillus subtilis caa3 oxidase locus (van der Oost, J., von Wachenfeld, C., Hederstedt, L. & Saraste, M. (1991) Mol. Microbiol. 5, 2063-2072). The deduced amino acid sequences of the four putative structural subunits (CtaC-F) indicate substantial similarity to caa3-type oxidases from other Bacillus species and to other members of the family of mitochondrial-type aa3 oxidases. A marked paucity of basic residues was noted in the cytochrome c-containing domain of CtaC, which faces the highly alkaline external milieu. We have also purified the enzyme as a three-subunit complex, with possible trace amounts of a fourth subunit. N-terminal sequence analysis of the two largest subunits confirmed them to be encoded by the cloned cta genes. An additional, minor caa3 component with distinctive chromatographic properties was noted during purification. Analysis of mRNA with a ctaD probe revealed an abundant 4-kilobase message of the right size to encode CtaC-F. The cellular content of this message varied with growth pH. Cells grown at pH 10.5 contained 2 to 2.5 times more message than those grown at pH 7.5, in good correspondence with the relative amounts of caa3 oxidase found in the cells. The ctaB gene, immediately upstream from the ctaC-F genes, was found to be transcribed onto a low abundance 5-kilobase message, which is likely also to encode CtaC-F. Levels of this message were not affected by growth pH.

a 2-3-fold higher concentration in the membranes of cells grown at pH 10.5 uersus pH 7.5, suggesting some mechanism of pH-dependent regulation (3). Increases in cytochrome 0 , albeit more modest on a percentage basis of this more abundant species, are also observed during growth at high pH. Elevated levels of caa3 oxidase were additionally found in cells grown at pH 7.5 in the presence of sublethal amounts of the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP)' (3). Thus, increased amounts of cytochrome caa3 are associated with two distinct growth conditions where Ap is low: pH 10.5 or pH 7.5 in the presence of CCCP. In addition, a mutant, CCCP-resistant strain was found to contain constitutively elevated levels of cytochrome caag when grown at pH 7.5 (3).
ATP synthesis by alkaliphilic bacilli such as B. firmus OF4 poses a bioenergetic problem. Oxidative phosphorylation at alkaline pH has been shown to deviate from the chemiosmotic model of Mitchell (4) in two respects (5,6). Firstly, the protonmotive driving force ( A p ) is very low at pH 10.5 and above and is insufficient to account for the observed phosphorylation potential unless a high and variable H+/ATP stoichiometry is posited. Secondly, artificially .imposed diffusion potentials fail to energize ATP synthesis at pH values above 9.5. To explain these discrepancies, it has been hypothesized that oxidative phosphorylation at very alkaline pH values may involve the direct, intramembranal transfer of protons from respiratory chain complexes to the Fo sector of the FIFO ATP-synthase (5). In addition, one or more of the respiratory chain complexes whose concentration increases during growth at very alkaline pH is presumably responsible for the increased transmembrane electrical potential observed in cells grown at such pH values (1). By virtue of its elevated expression at high pH, the cytochrome caa3 oxidase is an attractive candidate for mediating one of these roles. The involvement of a pH-regulated complex in the putative direct proton pathway is specifically supported by physiological data. For example, when cells grown at pH 7.5 were starved and then re-energized at pH 10.5, they synthesized much less ATP than similarly treated cells that had been grown at pH 10.5 (6), although the Ap generated was the same. In order to better understand oxidative phosphorylation at the molecular level, we recently cloned and sequenced the atp genes encoding the FIFO ATP-synthase (7). Several nonconservative substitutions at otherwise highly conserved residues were noted in the predicted sequences of the a-and c-subunits of the Fo moiety, and equivalent substitutions were found in the genes of a second, distinct alkaliphilic Bacillus (8). TO complement these efforts, a parallel project was initiated on the cytochrome caag oxidase, and we will pursue a comparable The abbreviations used are: CCCP, carbonyl cyanide m-chlorophenylhydrazone; PCR, polymerase chain reaction; PMSF, phenylmethylsulfonyl fluoride; aa, amino acid; kb, kilobase(s). study of cytochrome 0. Here, we report on the cloning and sequencing of the cta genes encoding the caa3 oxidase and the further purification and characterization of the enzyme.

MATERIALS AND METHODS
Bacterial Strains and Growth Conditions-B. firmus OF4 and its derivative strain 811M (met-, strR) were grown as previously described (3). During the course of this work, it was noted that cells grew better at pH 10.5 when the ammonium sulfate was omitted from the growth medium, and this modification was then adopted routinely. Escherichia coli strains DH5aMCR (Gibco BRL) and JM109, used as plasmid hosts, were grown in LB medium, with the addition of ampicillin (100 pg/ml) where appropriate.
Cloning Procedures-Standard protocols were employed for all cloning procedures (9,10). .Radioactive probes were prepared by random priming of appropriate DNA templates, using a commercial kit (New England Biolabs). Plasmid DNA for sequencing was prepared by large scale alkaline lysis and purified by centrifugation through a CsCl gradient. Both strands of clones were sequenced using an Applied Biosystems 373A Sequencer. Appropriate oligonucleotide primers were synthesized using an Applied Biosystems 380B DNA synthesizer or were obtained commercially (Genset, Paris). Computer analysis employed the GCG suite of programs (11) running on a VAX 4000-300 computer. A Perkin-Elmer Cetus Instruments thermal cycler was used to perform polymerase chain reactions (PCR). The entire DNA sequence has been deposited in the GenBank library; throughout this paper, the numbering of bases follows that of the deposited sequence.
Initial attempts to generate a probe for the B. firmus OF4 cm3 oxidase genes utilized a DNA fragment from the Bacillus subtilis cta operon, but this DNA failed to generate a signal in hybridization screens. Instead, PCR was performed on purified B. firmus OF4 chromosomal DNA using two oligonucleotides, D l and D2, designed to hybridize to the DNA encoding two highly conserved regions of subunit I about 150 bases apart. The probe sequences were: Dl, GGICACCCIGAGGTITACAT (I denotes inosine); D2, ACATGTGGTGIACCCAIACCAT (Fig. 20). After 25 cycles of amplification (initial 5 annealed at 30 "C, remainder at 55 T ) , a 150base pair product was found and ligated into pGEM5Zf(+) (Promega) by the method of Marchuk et al. (12). The resulting construct was designated pCOXlOll (Fig. 1). Sequencing revealed the insert to be highly homologous to the corresponding regions of the cytochrome caa3 oxidase genes cloned from Bacillus subtilis (13) and Bacillus species PS3 (14). The PCR product was used to probe a library of B. firmus OF4 DNA (library 3 of Ref. 15). Three overlapping clones were isolated (pCOX17, pCOX43, pCOX163), which encompassed most of the ctaB-cta0 region (Fig. 1).
To obtain the 5' end of the ctaB gene and upstream sequence, a second probe, comprising an already sequenced portion of the gene, was prepared by PCR (primers: B1, GATCGTGACATAGAT-CATCTTATGG, bases 2601-2625; B2, TAACCCGATAA-CAGCTGCAGTCGG, complement. of bases 2736-2759), and further screening of the library led to the isolation of pCOX1711 (Fig. 1). To obtain the 3' end of the ctaD gene and downstream sequence, a further probe was prepared by PCR, using a forward primer (D3, CTGGGTATTCTACATT GGTT, bases 5699-5718) based on a known sequence from the pCOX43 insert, and a reverse primer (El, CCAIACIACATCIATAAAATGCCA) designed to recognize a portion of the ctaE gene, which was expected to lie immediately downstream from cta0. Subsequent screening of the library allowed the isolation of pCOX31 (Fig. 1). AS a precaution against the possibility of deletions or rearrangements of DNA (13), PCR was performed on B. firmus OF4 chromosomal DNA and the pCOX clones, using several combinations of sequencing primers. Products of the expected size were obtained in all cases, indicating that no rearrangements had occurred during the cloning procedures.
Purification of the Cytochrome caa3 Complex-The initial stages of purification followed precisely those employed in the FIFo-ATPase isolation procedure (17). Briefly, washed membranes were extracted with 50 mM octyl glucoside in a buffer containing 20% glycerol, 50 mM Tricine-KOH, pH 8, 5 mM MgC12,5 mM p-aminobenzamidine, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol, and 3 mg/ml soybean asolectin. The cytochrome caa3-containing supernatant after high speed centrifugation was subjected to ammonium sulfate fractionation. The cytochromes remained soluble at 60% saturation; the salt concentration was lowered in this fraction by several steps of ultrafiltration and dilution during which Triton X-100 was added to a concentration of 0.1%. The sample was then diluted with 2 volumes of Buffer A (10% glycerol, 25 mM Tris-SO4, pH 7, 1 mM EDTA, 0.1 mM PMSF, and 0.5% Triton X-100) and loaded onto a DEAE-Sepharose CL-GB column (2.5 X 18 cm) equilibrated with Buffer A. After washing the column with several bed volumes of Buffer A, a linear gradient between 0 and 0.6 M (NH4),S04 (800 ml) was applied to the column, and 5.2-ml fractions were collected. The major peak of cytochrome c oxidase activity, eluting as a broad peak between about 300 and 450 mM (NH&S04, was concentrated by ultrafiltration using a PM-30 membrane and dialyzed overnight against 10 mM potassium phosphate, pH 7.5,l mM EDTA, and 0.1 mM PMSF. The dialyzedpreparation was split into 2 volumes, and each portion was applied to a 1.5 X 9.5-cm hydroxyapatite column. The column was washed successively with 2 bed volumes each of 10, 25, and 50 mM potassium phosphate, pH 7.5 (buffers also contained 1 mM EDTA, 0.1 mM PMSF, and 0.5% Triton X-100). Cytochrome caa3 was eluted with 100 mM potassium phosphate. The enzyme eluted rather sharply off this column and was collected in a small volume which was further concentrated by ultrafiltration. The preparation was divided into aliquots, frozen in liquid Nz, and stored at -70 "C.
SDS-Polyacrylamide Gel Electrophoresis and Related Procedures-The polypeptide composition of cytochrome caa3 was examined on SDS-10% T polyacrylamide gels (12 X 13 X 0.15 cm, with 1.5-cm stacking gel) according to the Tricine buffer formulation of Schagger and von Jagow (18). Samples were incubated in sample buffer for 15 min at room temperature prior to electrophoresis. The gels were run overnight, fixed for 1 h in 9% acetic acid, 45% methanol and stained with Coomassie G (19) or with silver (20). Heme staining was performed as described in Refs. 21 and 22. For N-terminal amino acid sequence analysis of subunits I and 11, the purified complex was applied to a mini-gel of identical composition. Following electrophoresis, the gel was transferred, without pre-equilibration, to a polyvinylidene difluoride membrane (Applied Biosystems), using a buffer containing 20% methanol, 192 mM glycine, 25 mM Tris free base, pH 8.3. To identify the subunits, the membrane was briefly stained with Coomassie Brilliant Blue R (0.25% in 9% acetic acid, 45% methanol), destained, and equilibrated with 50% methanol (high performance liquid chromatography grade) before drying. Amino acid sequencing was performed on a Porton Instruments Model 2090E automated gas phase Sequencer, using modified Edman degradative chemistry.
Assays-Cytochrome c oxidase activity was measured by the change in absorbance at 550 nm of reduced horse heart cytochrome c. The reaction mixture contained, in a 1-ml volume, 50 mM MES-NaOH, pH 6.0,30 pM cytochrome c, 0.1% Triton X-100, and sample, The reaction was carried out at 20 "C in a Shimadzu UV-160 spectrophotometer. Protein content of samples was estimated by the method of Lowry et al. (23); Triton X-100-containing samples were compared to bovine serum albumin standards also containing Triton X-100 and were centrifuged in a table top model following color development. Spectra were taken in a Perkin-Elmer model 550 spectrophotometer at room temperature with the slit width fixed at 2 nm. Scanning speed was 120 nm/min. Appropriate baseline spectra were taken and subtracted from the sample spectra. Spectroscopic analysis of the cytochrome content of cell membranes was performed as described previously (31, except that sucrose was omitted from the bacterial suspension medium, and a sample concentration of 2 mg of protein/ ml was used.  Open reading frames are boxed, with the direction of transcription indicated by the arrows. The cloned fragments are shown below; the initial PCR product, cloned as pCOX1011, is indicated by the *.

DNA Sequence Analysis
phr-An incomplete open reading frame upstream of ctaA encoding 339 amino acids (aa) was found by the FASTA program to be 42% identical with the C-terminal335 aa of E. coli deoxyribodipyrimidine photolyase (photoreactivating enzyme) (24). In contrast, the pycA pyruvate carboxylase gene is found in the equivalent position in B. subtilis (25). Interestingly, B. subtilis, along with other naturally competent euhacteria, is reported to lack a photoreactivation mechanism (26,27); to date, we have been unable to induce a competent state in B. firmus OF4.
ctaA-The ctaA gene is predicted to encode a 297-aa protein with a PI of 10.3 and a molecular weight of 32,369. The most likely initiation codon is a GUG (complement of bases 2017-2019). As in B. subtilis, the gene is transcribed in the opposite direction to the rest of the cta genes (28). A region of dyad symmetry lies midway between the stop codons of the phr and ctaA genes and probably acts as a transcriptional terminator, perhaps for both genes. Overall, the alkaliphile ctaA gene product is 35% identical with that of B. subtilis, the only other species in which the sequence of this gene has been reported (25), with most identity in the N-terminal 200 aa (Fig. 2 A ) .
ctaB-The ctaB gene is predicted to encode a 312-aa protein with a PI of 9.8 and a molecular weight of 34,325. B. firmus OF4 CtaB is 52% identical with the B. subtilis protein and 61% identical with the Bacillus PS3 partial sequence deduced from the published DNA sequence (14) (Fig. 2B). All of the 25 residues noted as invariant in B. subtilis CtaB, E. coli CyoE, Paracoccus denitrificans ORF1, and yeast COX10 (13) are present in the predicted alkaliphile CtaB. A hydropathy plot (not shown) suggests a topology similar to those suggested for the Paracoccus and E. coli homologues (29,30), with seven transmembrane segments. A conserved DNA sequence exists upstream of the B. subtilis qoxA, ctaB, and men genes (31), but no equivalent sequence is apparent in B. firmus OF4.
ctaC-The ctaC gene for subunit I1 of the cytochrome oxidase complex is predicted to encode a 342-aa protein with a PI of 4.3 and a molecular weight of 38,141. However, Nterminal amino acid sequencing yielded the sequence (X)-L-( X denotes unidentifiable residue), corresponding to residues 22-48 and suggested that the mature protein begins with Cys-22; its molecular weight would be 35,588. A similar cleavage of a signal sequence has been demonstrated in Bacillus PS3 (14) and postulated in B. subtilis, where Cys residues are predicted in the corresponding regions of both the ctaC and qoxA gene products (13,31). Subunits I1 of these oxidases may be lipoproteins, as the sequences surrounding the cleavage site cysteines match the consensus sequence of Gramnegative lipoprotein precursors (31); however, that of B. firmus OF4, L-T-G-C-L-G, fits the consensus before the cysteine, but not after.  ing the cytochrome c domain, which is probably external to the cytoplasmic membrane, is extremely acidic in the alkaliphile. This is due in part to the deletion of otherwise conserved residues, e.g  (14). Interestingly, the start codon is UUG (confirmed by sequencing in both pCOX163 and pCOX17), which allows efficient translation initiation in B. subtilis (33).

G-E-E-N-L-T-A-L-D-P-K-G-P-Q-A-Q-(X)-I-Y-D-N-M-I-L
The predicted amino acid sequence is 64% identical with that of B. subtilis CtaD, 63% identical with PS3 CaaD, and 47% identical with B. subtilis QoxB (Fig. 2 0 ) . Subunit I contains the binding sites for heme a3 and CUB (the binuclear center) and heme a. All the residues likely to be involved in these centers (32)

A Y W E V M T M H C T
? I ?
Irm,"ii"lLr M I ? L $ basic residues, or by their replacement with neutral or acidic nine is cleaved off (14). The predicted amino acid sequence is originally thought to play a role in proton translocation, but such a role appears doubtful, since the activity of the Paracoccus enzyme was apparently unimpaired when its equivalent and disruption of the Paracoccus gene resulted instead in ctaF-The ctaF gene for oxidase subunit IV is predicted to weight of 13,017. The alkaliphile sequence is 37% identical with B. subtilis CtaF, 31% identical with PS3 CaaF, and 25% identical with the first 100 residues of B. subtilis QoxD (Fig.  2F). Immediately downstream from the stop codon is a 30-"ald-ar"~tIII.

-I L ? A T ? ? C T ? C~R C C T A D C P T S A D L V A L D L V ? I M L L n v L r A L r A t r L a A s n s A c e e T T z W ? n v T L v . x T m L L L m aup VL.
impaired assembly of the oxidase complex (36).

m p ? A D M W~~~
A m ? U 117   The alkaliphile CtaC protein exhibits 39% sequence iden-N~ significant homology to any sequence in the PIR database titY with E. subtilis CtaC, 43% identity with p s 3 CaaC, and was found. Finally, orfc, reading from the AUG codon (bases 22% identity with B. subtilis QoxA (Fig. 2C). All Of the 9381-9383) may encode a hydrophilic 153-aa protein with a residues implicated in the formation of the CUA and heme c PI of 4.5 and a molecular weight of 17,748. This product may binding Sites are conserved in the B. firmus OF4 subunit 11 also be lipid-anchored, as residues 18-23 (L-A-A-c-G-s) (13,14,32). One change of note concerns the conserved match exactly the Gram-negative consensus sequence; its ammatic sequence F-W-W-Q-F-D-Y-E (residues 139-146), acidic PI is consistent with an exterior location. However, no where Glu-146 replaces the Pro-146 of PS3 and both B. subtilis significant homology to any deposited sequence was found. subunits 11. In addition, the large C-terminal region contain-While this is the first description of the genes distal to a Bacillus cytochrome oxidase operon, none of them seems likely to be involved with expression of the oxidase.

M Q T W S C W K R D L A R R R K W A -K K W R H M I ? V L M I L L T aip M A N K S A K H S H P X H V C I I L S I V L T
mRNA Abundance and Size Results from a typical Northern blot are shown in Fig. 3. The ctuD probe hybridized to a message of approximately 4 kb. Such a transcript would be just the right length to encode the four structural subunits of the cytochrome oxidase (CtaC-F), but too small to encode CtaB in addition. Considerably more of this message was found in pH 10.5-grown cells than in pH 7.5-grown cells. Densitometric analysis of the autoradiographs gave a relative abundance of 2.4 +. 0.251 ( n = 4), in good agreement with the relative abundance of cytochrome caa3 oxidase in the cell membranes (3). Although data are not shown, the ctuD probe was also used to determine mRNA levels in cells grown at pH 7.5 in the presence of 1 PM CCCP. The levels of ctu mRNA were generally higher in such cells than in the pH 7.5-grown controls, but were quite variable. One preparation contained as much message as pH 10.5grown cells, while two others showed little or no increase over pH 7.5-grown controls. As cytochrome caa3 levels were elevated in all three batches of cells, as observed previously (3), the timing of sampling for the mRNA determinations may be critical.
Although the coding region of the ctuB gene occupies only 1 kb, Northern analysis revealed a 5-kb low abundance message, with no sign of any shorter products (Fig. 4). As the c t d gene, immediately upstream of ctuB, is transcribed in the opposite direction, it is likely that the ctuB message spans from ctuB to ctuF. This message would not have been detected with the ctuD probe, due to the close proximity of the high abundance 4-kb message. Levels of the 5-kb message were very similar in cells grown at pH 7.5 or at pH 10.5. Northern analysis using a ctaA probe revealed a low abundance 1.2-kb message, consistent with the expected size of the gene and unaffected by growth pH (data not shown).

Protein Purification
Purification data are summarized in . Table I. We had observed previously that most of the cytochromes extracted from B. firmus OF4 membranes remained soluble at 60% ammonium sulfate saturation (17); in other work, we noted that cytochrome caa3 bound rather tightly to DEAE resins and eluted from these resins at relatively high ionic strength (2). By combining ammonium sulfate fractionation with anion exchange chromatography, we could obtain a cytochrome caa3 preparation that was almost spectrally and electrophoretically pure. with just a trace of cytochrome b contamination. As shown in Fig. 5, the major fraction of cytochrome c oxidase activity eluted from the DEAE-Sepharose CL-GB column as a broad peak between 300 and 450 mM ammonium sulfate. A minor peak representing less than 10% of the total activity was found at approximately 150 mM ammonium sulfate. The identity and characterization of the minor peak is currently under investigation. The major peak was further purified on a hydroxyapatite column which removed the cytochrome b contaminant. The reduced versus oxidized difference spectrum of the purified complex showed absorption maxima at 600,551,521,443, and 418 nm, indicative of a cytochrome uand c-containing complex.
The polypeptide composition of the purified cytochrome caa3 complex was determined on SDS-10% T polyacrylamide gels (18). The complex was resolved into three bands with approximate sizes of 44,000, 37,500, and 22,500 (Fig. 6). As observed for other very hydrophobic polypeptides, subunit I migrates anomalously. Both Coomassie staining (Fig. 6, lane 2) and silver staining of the complex (Fig. 6, lane 3 ) indicated that the preparation was quite pure. Subunit I was better stained by silver than by Coomassie. Heme staining of the gel showed that the cytochrome c of the complex is in subunit I1 (Fig. 6, lane 4 ) , consistent with the predicted amino acid sequence of subunit 11. Importantly, N-terminal amino acid sequencing confirmed that subunits I and I1 were indeed the products of the cloned ctuD and ctuC genes.
B. firmus OF4 cytochrome caa3 oxidized cytochrome c from both horse heart and Saccharomyces cerevisiae, the latter being a slightly better substrate. The pH optimum of the reaction was acidic (pH 6). These results are in agreement with the those observed with the purified complex from alkaliphilic B. firmus RAB (38).

DISCUSSION
The organization of the cytochrome caa3 oxidase genes in alkaliphilic B. firmus OF4 is identical with the corresponding loci in B. subtilis (13,28) and Bacillus PS3 (14). Four genes, ctuC-F, correspond to the four putative structural subunits of the enzyme. Among these three species, however, a foursubunit complex has been isolated to date only from PS3 (39). The purified complex from the alkaliphile clearly has three subunits (Fig. 6). A faint silver-stained band at around 10 kDa (Fig. 6, lane 3 ) might correspond to subunit IV, although this would be present, at least in this preparation, in substoichiometric amounts. The presence of the putative fourth subunit has not yet been demonstrated in the B. subtilis a a 3 oxidase (40), although the qox operon has a homologue of ctuF.
The functions of the two other ctu genes, c t d and ctuB, remain to be elucidated. In view of the relatively low abundance of ctuB mRNA in B. firmus OF4, it would be surprising if the protein were present in stoichiometric amounts in the alkaliphile oxidase complex, although it could be one of the minor species. A nonsense mutation in B. subtilis c t d and a spontaneous deletion encompassing the entire c t d and ctuB genes both led to a total lack of detectable heme u (25, 28), Membranes equivalent to about 400 mg of protein were extracted with octyl glucoside; measurement of the enzymatic activity of the * Micromoles of cytochrome c oxidized min".
suggesting that neither cm3-nor m3-type oxidase was present; there is no gene equivalent to ctuB upstream of the qox operon (31). Recent data of Svensson and Hederstedt (41) suggest that c t d and ctuB are required, together, for the synthesis of heme u, and that one of them encodes a b-type cytochrome that functions in the synthetic reactions. The hydropathy profiles of the two CtaA proteins whose deduced sequences are now available (Fig. 2 A ) are almost identical. Taking into consideration the fact that exterior connecting loops of bacterial plasma membrane proteins contain a paucity of positively charged residues relative to cytosolic loops (42), a speculative model would include eight transmembrane helices and would place the two pairs of conserved cysteines facing the exterior, where they could be involved in metal binding.
No equivalent to CtaA has yet been identified in E. coli, which is unable to synthesize heme a.
The products of all four alkaliphile structural genes show substantial identity to their counterparts in other cytochrome oxidases from Bacillus species (Fig. 2). Subunit I in particular shows significant identity to subunits I from other prokaryotes and eukaryotes (32). Ultimately, we will focus on the possibility of specific sequence deviations that may relate to the bioenergetics of alkaliphiles (6); such an analysis will require a better understanding of the native structure, its subunits and their roles, and will entail the examination of mutants that are compromised in oxidative phosphorylation specifically a t high pH, but generate the same Ap pattern as the wild-type. An observation in the current study that is relevant to alkaliphilicity, although not to the bioenergetics of the organism per se, is the striking deletion of basic residues in the external, hydrophilic cytochrome c-binding domain of CtaC, or their replacement by neutral or acidic residues; in Fig. 7, a model of the gene product is shown, with an x indicating residues a t which such a replacement has occurred. In general, adaptation to growth at extremely alkaline pH features the avoidance of basic residues, even more than is usually found (42) in those enzymes (43), parts of enzymes or structural proteins (44) that must function in the highly alkaline exterior. With respect to polytypic membrane proteins, the extreme avoidance of basic residues in the external hydrophilic loops of alkaliphile proteins may offer assistance in making topological predictions.
In addition to other, alternate terminal oxidases, B. subtilis and with models of other subunits I1 (14,32). type quinol oxidase (40,45) encoded by the qox genes (31). B. cereus also has both a cuu3 and an a a 3 oxidase (46). Might a similar situation exist in B. firmus OF4? To date, there is no evidence for more than one uu3/caa3-type oxidase in the alkaliphile, unless the minor component observed during the current purification work is such a species (Fig. 5). During extensive screening of the B. firmus OF4 DNA library, only clones of the ctu region were obtained; however, Saraste et ul. (13) also failed to detect hybridization of their B. subtilis ctu probe to the chromosomal qox DNA, despite the similarity of the two sets of protein products. At the protein level, we have consistently observed only a single a-peak at 600 nm in reduced-minus-oxidized spectra. By contrast, the B. subtilis uu3 oxidase shows a peak at 600 nm, and the cuu3 oxidase a peak at 605 nm. Furthermore, the B. subtilis uu3 enzyme, unlike the c m 3 , is not significantly reduced by ascorbate/ TMPD (45). We have never seen any increased signal from dithionite-reduced versus ascorbate/TMPD-reduced samples, nor was any additional signal apparent when ascorbate/ TMPD-reduced samples were further reduced with dithionite. No qualitative differences were observed in the 600 nm region when malate-grown and glucose-grown B. firmus OF4 were compared; by contrast, expression of the uu3 oxidase in B. subtilis was higher in rich or glucose minimal media than in succinate minimal media (31). The minor component eluted from the Sepharose column may indeed represent an additional uu3-type oxidase species in the alkaliphile, but is probably more likely to be a cuus variant than a counterpart to the B. subtilis uu3. P. denitrificuns possesses two very closely related genes for subunit I, with the protein products differing significantly only in the N-terminal 15 aa (47). The second gene was not detected until a deletion of the first gene was found to be without any phenotypic effect. It is notable that under conditions of slight oxygen limitation, Bacillus PS3 synthesizes an alternative form of cytochrome cuu3 in which the heme u3 is replaced by an o-type heme (48) to produce an uco-type oxidase (49). In contrast, the facultative alkaliphile, Bacillus YN-2000, synthesizes an uco-type oxidase under all conditions of aeration and appears not to produce any c a stype oxidase (50).
The cytochrome content of bacterial species is sensitive, both qualitatively and quantitatively, to growth conditions (e.g. Refs, 2, 3, 31, 46, and 51), and lesions in one oxidase may be compensated for by elevated levels of another (52). The up-regulation of the cuu3-oxidase of B. firmus OF4 at high pH (perhaps, really, low Ap) is particularly striking and is of likely importance in the special physiological adaptations of the alkaliphile to its environment. The increase in the cuu3 oxidase during growth at pH 10.5 could be accounted,for by an increase in mRNA for the ctu structural genes, presumably reflecting some combination of increased transcription and message stability. The detection of the 5-kb transcript encompassing ctuB-F was unexpected. The most likely explanation of our data is the existence of two promoters, one upstream of ctuB and unaffected by pH, and a second between ctuB and ctuC which is pH-regulated. Alternatively, one could postulate that processing of the 5-kb transcript yields the 4-kb transcript, which is more stable under all conditions, but especially so during growth at high pH. An interesting finding in this context is an extensive region of potential secondary structure in the mRNA between the ctuB and ctuC genes (bases 3265-3324, Fig. 8), detected using the FOLD program of the GCG suite (53). Similar structures exist in the PS3 and B. subtilis ctu sequences, although the estimated folding energies of these structures are higher than that of the alkaliphile. The B. firmus OF4 structure does not possess the hairpin structure of a typical transcriptional terminator (e.g. that following ctuF) and may, instead, be part of a promoter region. It will be of interest to further delineate the mechanism(s) of the pH-dependent increase in ctu operon expression and to study transcription in the CCCP-resistant mutant strain, CC2, in which cytochrome oxidase levels are elevated in pH 7.5-grown cells (3).