Soluble CuA-binding domain from the Paracoccus cytochrome c oxidase.

In cytochrome c oxidase the C-terminal part of subunit II is outside the membrane and contains a copper center called CuA. We have expressed this domain of the Paracoccus denitrificans oxidase in a soluble form. Data obtained by quantitative copper-to-protein measurements, electrospray mass spectrometry, and electron paramagnetic resonance spectroscopy show that the center contains two copper atoms probably in a mixed valence configuration. Its absorbance spectrum is similar to that of the copper center A in nitrous oxide reductase. The EPR spectrum suggests that the center in the soluble protein is closely related to the native CuA site in the cytochrome oxidase complex. However, it seems likely that the copper center in the soluble domain is more exposed to the aqueous milieu than in the intact complex because its absorbance and EPR spectra are sensitive to pH. At alkaline pH one of the coppers in the site acquires type-2 character, indicating that it may be coordinated to a new ligand. The pK of this reversible change is about 8.2. The CuA-binding fragment is able to oxidize cytochrome c.

CUA is a redox center in cytochrome c oxidase that is involved in electron transfer from cytochrome c to the active site. It is distinct from types 1, 2, and 3 copper sites (for a recent review, see Ref. 1) but is similar to the center A of a multicopper enzyme, nitrous oxide reductase. Recent evidence suggests that both sites have two copper atoms in a mixed valence [Cu(I)-Cu(II)] S = '/z configuration (2)(3)(4)(5)(6)(7).
Cytochrome c oxidase is a membrane protein complex which catalyses the reduction of dioxygen to water and functions as a redox-linked proton pump. Its functional core is made of two subunits, which bind four metal centers. The membraneembedded subunit I contains the active site where oxygen is reduced. This is a bimetallic center formed by the iron of a high-spin heme and a copper called CUB. Electrons from cytochrome c enter the active site via CUA and a low-spin heme (2,(8)(9)(10)(11). The latter is also bound to the major subunit whereas CuA is located in subunit 11.
The CuA-binding domain is outside the membrane bilayer. It is involved in cytochrome c binding, and CuA may be the primary electron acceptor in cytochrome oxidase (4,12,13).
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$.Supported by a predoctoral fellowship from the Academy of Finland.
T To whom correspondence should be addressed EMBL, Meyerhofstrasse 1, Postfach 10.2209, D-69012 Heidelberg, Germany. Tel.: 49-6221-387365;Fax: 49-6221-387306. We have recently studied an engineered CuA-like center built into the CyoA subunit of the Escherichia coli cytochrome bo quinol oxidase complex. These studies on the isolated soluble protein fragment and on its mutants have shown that this purple copper center is indeed binuclear. The main ligands of the two coppers appear to be 2 cysteines, 2 histidines, and a methionine (7). Here we continue this study with a homologous domain isolated from a genuine cytochrome c oxidase. It has been expressed from the ctaC gene of Paracoccus denitrificans. The results obtained with this native CUA-binding domain confirm the binuclear nature of the site.

EXPERIMENTAL PROCEDURES
Expression of the Subunit II Fragment-The region of ctaC coding for amino acid residues 128-280 (14) was amplified by polymerase chain reaction. The primers introduced the upstream NcoI and downstream Hind11 restriction sites to the DNA fragment. The NcoI site affects the residue following N-terminal methionine which is valine in our construct but leucine in the native CtaC sequence (Tables I and I1 (14,15)). The polymerase chain reaction fragment was inserted into a modified pET3d vector described in (4); the resulting plasmid is called pET.PD1. This construct was used to express the subunit I1 domain in E. coli BL21(DE3) cells. For protein purification, freshly streaked LB plates containing ampicillin (0.1 mg/ml) were used to inoculate 100-ml cultures in 250-ml Erlenmeyer bottles. After shaking (200 revolutions/min) for 3 h at 37 "C, these cultures were used to inoculate 1-liter cultures in twelve 2-liter Erlenmeyer bottles. The cells were grown 3-4 h to the optical density of 0.6 at 600 nm, and the expression was induced with isopropyl-thio-P-galactoside (0.2 mM). The cells were harvested 2.5 h after induction, washed with 20 mM Tris (pH 8.2), and stored at -80 "C.
Protein Refolding and Purification-About 15 g of E. coli cells (the yield from 6 liters of bacterial culture) were suspended in 40 ml of our "standard Tris buffer": 20 mM Tris (pH 8.2) containing proteolytic inhibitors phenylmethylsulfonyl fluoride (0.15 mM) and benzamidine (5 mM), and broken with a French press at 4 "C. Viscosity of the suspension was reduced by adding deoxyribonuclease (2 mg) and MgSO, (1 mM). The suspension was centrifuged for 40 min at 40,000 revolutions/min in a Beckman Ti-45 rotor. Almost all of the subunit I1 fragment was in inclusion bodies under these expression conditions. The supernatant was therefore discarded, and the pellet was resuspended into 50 ml of standard Tris buffer containing 3% (w/v) Triton X-100. After incubation for 2 h on ice, the suspension was briefly centrifuged to separate the inclusion body pellet from the solubilized membrane proteins. The pellet was washed with 50 ml of standard Tris buffer containing 1% of Triton X-100, and stored at -20 "C.
The pellet obtained above was dissolved in 40 ml of 6 M urea, 20 mM Tris (pH 8.2). The dissolved protein was first dialyzed for 4 h against 2 M urea, 20 mM Tris (pH 8.  The amino acid sequence of subunit I1 is taken from Ref. 14. N-terminal residues methionine and valine which are added to the native sequence, are shown in bold. C-terminal extension of the protein (15) is indicated with small letters. The copper-binding ligands (7) are underlined. Residue numbers are on the right. in a 3.5-kDa cutoff tubing. A part of the protein precipitates during dialysis. This was removed by centrifugation for 40 min at 45,000 revolutions/min in a Beckmann Ti-60 rotor.

MVFRS QEMPNDPDLV IKAIGHQWWr
The supernatant was applied to a Q-Sepharose fast flow anionexchange column (50-ml bed volume) equilibrated with 20 mM Tris (pH 8.2) containing 200 PM CuC12. The flow rate was 6 ml/min, and a linear NaCl gradient from 0 to 1 M was developed during 120 min. The peak fractions containing the subunit I1 fragment, which elutes a t 0.45 M NaC1, were pooled, concentrated in Microsep concentrators (Filtron Co., MA) to approximately 1 ml and loaded to a Superdex 75 HiLoad gel filtration column (120 ml, Pharmacia LKB Biotechnology Inc.) which had been equilibrated with 20 mM BisTris (pH 7.0), 50 PM CuC12. The flow rate was 1.0 ml/min. The subunit I1 fragment elutes at 65-70 ml. All the columns were run with a Pharmacia fast protein liquid chromatography instrument at room temperature.
Spectroscopy-Optical spectra were recorded with a Perkin Elmer Lambda 2 Spectrophotometer a t room temperature. Electrospray mass spectra were recorded with a Sciex API I11 instrument as described in Ref. 7. Apoprotein samples for mass spectra were produced by incubating the subunit I1 fragment with 10 mM EDTA for 24 h. The protein samples were desalted by gel filtration on Sephadex G-25 columns (PD-10, Pharmacia) equilibrated with distilled water. 0.05% formic acid was added to the sample just before spraying into  7.0 1.9 53 9.0 1.9 57 mass spectrometer. EPR spectra were recorded with a Bruker ER 2OOD-SRC X-band spectrometer equipped with a standard TE102 rectangular cavity and an Oxford Instruments ESR-9 helium flow cryostat. Temperatures above 100 K were obtained with a nitrogen gas flow system. Quantifications of the EPR spectra were performed under non-saturating conditions as described earlier (16). For EPR, the samples were exchanged to buffers with desired pH by gel filtration in PD-10 columns. 50 mM MES was used at pH 6.0, and 50 mM HEPES was used at pH 7.0-9.0.
Miscellaneous-Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate was carried out using the buffer system of Laemmli (17). Analytical ultracentrifugation was kindly carried out with a Beckman XLA Optima centrifuge by Ariel Lustig (Biocenter, University of Basel). Copper was determined according to Ref. 18. The millimolar extinction coefficient of the purified protein is 48.7 mM" cm" at 278 nm; this value is based on protein concentration determined by quantitative amino acid analysis. The reaction between cytochrome c and the CuA-binding fragment was measured spectrophotometrically. Horse heart cytochrome c (Sigma) was reduced with solid sodium dithionite which was subsequently removed by gel filtration in a PDlO column. The CuA-binding domain was crystallized by  The two forms (a and b) of the apoprotein correspond to the fulllength sequence ( Table I)

RESULTS
Expression of the CuA-binding Domain-Several constructs were made in order to express the CuA-binding domain from subunit I1 of the P . denitrificans cytochrome c oxidase in E.
coli. The highest amount of protein was expressed with a construct which has its N terminus immediately after the second predicted transmembrane helix (20) and its C terminus 17 residues before ctaC translation stop codon (Tables I and  11). These last 17 amino acids make a C-terminal extension of the precursor which is removed post-translationally (15). The protein was expressed at a high level in the cytoplasm of E. coli, but it was mostly in inclusion bodies (Fig. 1). Only 10-20% of the fragment was found in soluble fraction. We could not find spectroscopically any purple copper in this minor soluble fraction after purification and addition of copper. This indicates that the soluble form subunit I1 fragment may be modified or incorrectly folded in the bacterial cytoplasm.
In contrast, we were able to produce a homogeneous protein preparation with a characteristic purple copper spectrum by refolding the denatured protein of inclusion bodies by the removal of urea (see "Experimental Procedures"). It appeared to be necessary to refold the protein in the presence of DTT in order to obtain a preparation which binds CuA quantitatively. It is also necessary to introduce the copper after the refolding of the protein. The addition of copper during the folding process led to the formation of a colorless product.
The renatured protein was purified using two chromatographic steps: first anion-exchange on a Q-Sepharose column and then gel filtration on a Superdex 75 column (Fig. 1). In the anion-exchange chromatography, the subunit 11 fragment was separated from /3-lactamase, the other major protein in the inclusion bodies (upper bund in lane 2 in Fig. 1). The protein was more than 95% pure after chromatography on Q-Sepharose, and the purification was finalized with the gel filtration step. The hydrodynamic molecular mass of the subunit I1 fragment was measured to be 17 kDa by sedimentation equilibrium. This agrees with the calculated molecular weight (Table IV) and shows that the protein is a monomer. Fig. 2 shows needle-like crystals of the CuA-binding subunit I1 fragment. These have been obtained at pH 4.25 using polyethylene glycol as the precipitant. The crystals have a purple color. The fact that the protein crystallizes suggests that it is homogeneous and properly folded.
Binuclear Nature of Copper Center-The results of colorimetric copper measurements (Table 111) show that the mon- omeric protein binds almost exactly two coppers. We also carried out electrospray mass spectroscopy to determine the number of copper atoms bound in the COII domain. The mass spectrometric data indicate that the apoprotein has two different molecular species with a mass difference of 131 daltons (Fig. 3A). The mass of the larger apoprotein fragment is 17,146 daltons which is equivalent to the calculated molecular mass without a formyl group in the N-terminal methionine. The 17,014 dalton mass of the smaller species probably cor- responds to the protein which has lost the N-terminal methionine (131 daltons). Both protein species can apparently bind copper (Fig. 3B); note, however, that the smaller form with two bound copper atoms (17,142 daltons) cannot be resolved from the larger apoprotein (17,146 daltons). The mass differences between each apoprotein species and their copper saturated forms are 126 and 128 daltons ( Fig. 3 and Table IV). This result indicates that there are indeed two coppers (127 daltons) bound to this fragment. Only about 50% of this copper is visible in EPR which is consistent with the proposed mixed valence structure (Table 111).
Spectroscopical Properties of the Cua Center-The CuAbinding domain has a characteristic purple color. At pH 7.0 it has a strong absorbance maximum at 480 nm with a shoulder at 530 nm. Two additional maxima are present at 363 and 808 nm (Fig. 4). Only the latter flat peak can be seen in the intact cytochrome oxidase complex. However, the absorbance spectrum is similar to the spectra of the center A in NzO reductase (22-24) and of the engineered CuA-like center in purple CyoA (4). The extinction coefficients for these maxima are given in Table V.
The EPR spectrum of the Cu-binding domain (Fig. 5) at the neutral or acidic pH is similar to the spectrum of the native CUA site in cytochrome oxidase. The g values and hyperfine coupling constants are the same as those of the intact oxidase and the engineered purple CyoA (see Refs. 2,4,9).
The Isolated CUA Center Is Exposed to Solvent-The absorbance spectrum of the CuA domain is dependent on pH. At pH 6-7 the strongest absorbance is in the 450-550 nm region, whereas the absorbance around 360 nm is weak. With increasing pH the absorbance in the 360 nm region increases and is shifted to 370 nm. Simultaneously, there is a decrease of  (Fig. 4). An isosbestic point for the spectral change is present at 420 nm. Titration of the absorbance changes at 370 and 480 nm gives a pK which is close to 8.2 (Fig. 6). The absorbance changes caused by pH are reversible.
Also the EPR spectrum of the CUA domain changes at alkaline pH (Fig. 5). The CUA spectrum becomes merged to another signal which is similar to the spectrum of a type 2 copper center. The EPR spectrum of the samples at low pH begins to broaden at temperatures above 130 K which is typical of the EPR behavior of CUA center in cytochrome c oxidase (25). This broadening is not seen for the type 2 signal at pH 9. No change in the EPR spectrum of the CuA site in the intact Paracoccus cytochrome c oxidase could be detected when the pH was increased to 9. The monomer binds two coppers at pH 7 and 9 (Table 111), and in both cases only half of the copper is visible in EPR.
The sensitivity of the isolated purple center to pH suggests that it is more exposed to the aqueous milieu than the site in the intact oxidase complex. This could also have an effect on the redox potential of the center. We have not been able to measure the midpoint potential accurately because the titration cannot be carried out reversibly.
Reaction with Cytochrome c- Fig. 7 shows that reduced cytochrome c is oxidized when it is mixed with oxidized CuAbinding fragment. The equilibrium constant for the redox reaction between the CuA-binding domain and cytochrome c is 0.7 as estimated from the titration data similar to those shown in the figure. This would correspond to a reduction potential of 240 mV for the CuA site in the soluble domain.

DISCUSSION
This study on a native CuA-binding domain confirms the earlier conclusion on the binuclear structure of the center which was reached by studying an engineered copper center (4, 7). The Paracoccus subunit I1 fragment has potential for further structural and functional investigations on the CUA site. It may also be useful for future experiments on the interaction and electron transfer between cytochrome c and the copper site. However, the isolation of the CuA site from the complex has altered the center. It has become pH sensitive and somewhat labile; this could be explained by a more open structure, which may allow water to enter. It seems to be difficult to oxidize and reduce it reversibly; this has hampered our efforts to measure the redox potential of the isolated CUA site.
The copper site in the intact cytochrome oxidase complex is not sensitive to the slightly alkaline pH. This sensitivity seems to be an artifact of the isolated domain. The phenomenon is, however, interesting. The EPR spectra at high pH suggest a change in the ligation of the copper center. The pK of this change (8.2) would fit to the deprotonation of a thiol group. A cysteine with a thiolate should be a stronger metal ligand than a cysteine with a protonated sulfhydryl group. Therefore, we think that another amino acid is responsible for the spectral change. One possibility is that a tyrosine residue is close to the copper center and coordinates to one of the coppers when it loses a proton and becomes a phenolate. Oxygen ligation would fit to the appearance of a type 2 copper center.
An aromatic residue in the vicinity of copper site could be involved in the electron transfer pathway from CUA to the other metal centers in cytochrome c oxidase. It has also been proposed that a tyrosine could be a ligand of the reduced CUA center (26).
The optical spectrum of CUA cannot be recorded in the intact oxidase complex because it is mostly covered by the absorbance of hemes. However, magnetic circular dichroic spectrum of the center in cytochrome oxidase has suggested that the optical spectrum has a strong signal in the 500 nm region (27). The spectrum at pH 6 and 7 (Fig. 4) has the same features as the optical spectra of nitrous oxide reductase and the engineered purple copper site in CyoA (4,7,(22)(23)(24). They all have an absorbance maximum in the 480-540 region. However, the three spectra have also subtle differences which suggest that the geometry and ligation of binuclear sites are not exactly the same.