Eur. J. Biochem. 210, 133-138 (1992) 0FEBS 1992
EPR studies of cytochrome aa3 from Sulfolobus acidocaldarius Evidence for a binuclear center in archaebacterial terminal oxidase Stefan ANEMULLER
’, Eckhard BILL’,
Giinter SCHAFER3, Alfred X. TRAUTWEIN
’ and Miguel TEIXEIRA’
Centro de Tecnologia Quimica e Biologica and Universidade Nova de Lisboa, Oeiras, Portugal Institut fur Biochemie, Medizinische Universitat zu Liibeck, Federal Republic of Germany
’ Institut fur Physik and
(Received April 14/August 31, 1992) - EJB 92 0527
The purified cytochrome aa3-type oxidase from Suljolobus acidocaldarius (DSM 639) consists of a single subunit, containing one low-spin and one high-spin A-type hemes and copper [Anemiiller, S. and Schafer, G. (1990) Eur. J. Biochem. 191, 297-3051. The enzyme metal centers were investigated by electron paramagnetic resonance spectroscopy (EPR), coupled to redox potentiometry. The lowspin heme EPR signal has the following g-values: g, = 3.02, g, = 2.23 and g, = 1.45 and the highspin heme exhibits an almost axial spectrum (gy = 6.03 and g, = 5.97, E/D< 0.002). In the enzyme as isolated the low-spin resonance corresponds to 95 10% of the enzyme concentration, while the high-spin signal accounts for only 40 f 5%. However, taking into account the redox potential dependence of the high-spin heme signal, this value also rises to 95 & 10%. The high-spin heme signal of the Suljolobus enzyme shows spectral characteristics distinct from those of the Paracoccus denitrijiicuns one: it shows a smaller rhombicity (8, = 6.1 and g, = 5.9, E/D = 0.004 for the P. denitrijiicuns enzyme) and it is easier to saturate, having a half saturation power of 148 mW compared to 360 mW for the P. denitrificans protein, both at 10 K. The EPR spectrum of an extensively dialyzed and active enzyme sample containing only one copper atom/enzyme molecule does not display CuAlike resonances, indicating that this enzyme contains only a CUB-typecenter. The EPR-redox titration of the high-spin heme signal, which is assigned to cytochrome a3, gives a bell shaped curve, which was simulated by a non-interactive two step redox process, with reduction potentials of 200 f 10 mV and 370 f 10 mV at pH = 7.4. The decrease of the signal amplitude at high redox potentials is proposed to be due to oxidation of a CuB(1) center, which in the Cu,(II) state is tightly spin-coupled to the heme u3 center. The reduction potential of the low-spin resonance was determined using the same model as 305 & 10 mV at pH = 7.4 by EPR redox titration. Addition of azide to the enzyme affects only the high-spin heme signal, consistent with the assignment of this resonance to heme a3. The results are discussed in the context of the redox center composition of quinol and cytochrome c oxidases.
The thermoacidophilic archaebacterium Suljolobus acidocaldarius grows aerobically at temperatures up to 85 “C and pH values of 2- 3 [l]. This phylogenetically very early organism is likely to use a chemiosmotic mechanism for energy conservation [2, 31. Although the unequivocal identity of the individual proton pumps has not yet been established, a possible candidate has been purified as a terminal oxidase of the cytochrome aa3 type [4, 51. This enzyme consists of a single subunit and was found to be a novel species of cytochrome ua3, since caldariellaquinone [6], the ubiquinone analogue in the membrane of Suljolobus, is used as reductant instead of reduced cytochrome c.
Optical spectra clearly indicated the presence of two different heme centers with apparent redox potentials of 220 mV and f370 mV at pH = 7.4 [5]. One of the hemes binds CO and therefore was identified as heme a3. From resonance Raman spectra, it could be concluded that in both the oxidized and the fully reduced state hemes a and u3 are in the sixcoordinated low-spin and six-coordinated high-spin heme configuration, respectively [7]. The resonance Raman spectra of the oxidized enzyme are similar to those of the mammalian and plant cytochrome aa3, but the spectrum of the reduced form differs remarkably: in contrast to the beef heart enzyme, the formyl vibrations of heme a ruled out significant hydrogen bonding interactions. Preliminary EPR studies confirmed the presence of highCorrespondence to S . Anemiiller, Centro de Tecnologia Quimica e spin and low-spin heme centers [5]. However, both heme Biologica, Apt. 127, P-2780 Oeiras, Portugal centers were not quantitated and their redox behaviour was Fax: +351 1 4428766. Abbreviations. Ph(NMeZ)’, N,N,N’,N’-tetramethyl-l,4-phenyl- not studied. In this paper we report data on spin quantification of both heme centers, as well as on the redox potentials of enediamine dihydrochloride. Enzyme. Cytochrome uu3 (EC.1.9.3.1) the high-spin and low-spin hemes, obtained by EPR redox
+
134 titrations. We propose that archaebacterial quinol oxidase is another example of cytochrome aa3 with a binuclear center in which the high-spin heme and CuB are magnetically coupled. The saturation behaviour of heme a3 has also been studied and compared to that of the Paracoccus denitrificans enzyme. Studies on the ligand binding effect of azide corroborate our conclusions.
g-values
20.10.
6.0
4.0
3.0
2.0
MATERIALS AND METHODS Cultivation of Sulfolobus acidocaldarius (DSM 639), membrane preparation and purification of cytochrome aa3 were done as described previously [5].Membrane protein was determined as in [XI and detergent-solubilized protein was determined by the Lowry method in the presence of SDS [9]. Paracoccus denitrficans cytochrome ua3 was a kind gift from Prof. B. Ludwig (Medical University of Liibeck, Federal Republic of Germany). Spectroscopic methods
100
200
300
400
500
B [mTI
, Fig. 1. EPR spectrum of the purified Sulfolobus cytochrome uu3 (upper trace) and theoretical simulations of the low-spin (middle trace) and of the high-spin (lower trace) heme signals. Protein concentration, 4.9 mg/ ml in 0.8 M KH2P04, 0.05% sarcosyl, pH = 7.4; heme iron content, 61.8 pM; copper content, 1.8- 3.4 Cu/cytochrome-aa3 (determined from analogous preparations). Microwave frequency, 9.453 GHz; microwave power, 2 mW; modulation frequency, 100 kHz; modulation amplitude, 1 mT; time constant: 0.041 s; gain: 1 . 6 ~ sweep width, 0.01 -0.51 T; temperature, 20 K. The low-spin heme simulation was generated by using the following g-values and linewidths: g, = 3.02, g, = 2.23 and g, = 1.45; 5 mT, 5 mT and 20 mT. The high-spin heme signal was simulated with g, = 6.03, g, = 5.97 and g, = 2.0 and the following linewidths: 8 mT, 15 mT and 20 mT.
Visible spectra of detergent-solubilized cytochrome aa3 were measured in a Hewlett Packard HP 8450 A diode array spectrophotometer. N,N,N',N'-Tetramethyl-l,4-phenylenediamine dihydrochloride [Ph(NMeJ2] oxidation was followed as described previously at 546 nm [5]. Heme A content was determined using E~~~~~ (reduced - oxidized) = 12 mM-' cm- [lo]. Atomic absorption spectroscopy measurements of copper content were performed with a Hitachi 180-80 polarized Zeeman atomic absorption spectrophotometer at 324.8 nm. EPR spectra were recorded with an X-band Bruker ER 200 D and an X-band Bruker ESP 380 spectrometer EPR redox titration equipped with an ESR 900 continuous-flow helium cryostat Potentiometric redox titrations were performed in a buffer from Oxford Instruments. containing 0.8 M KH2P04, 0.05% sarcosyl, pH 7.4 under anaerobic conditions as described by Dutton [13]. The redox mediators used, at concentrations between 40 - 200 pM Preparation of a l-Cu/cytochrome-aa3 sample were: phenazine ethosulfate (+ 55 mV), phenazine methoCytochrome aa3 in a buffer containing 0.8 M KH2P04, sulfate (+ 80 mV), 1,2-naphthoquinone (+ 180 mV), 1,20.05% sarcosyl, pH 7.4 was dialyzed at 4°C for 60 h against naphthoquinone-Csulfonic acid ( + 215 mV), 2,6-dichloro100-fold volume excess of the same buffer, supplemented with phenol indophenol (+ 217 mV), 2,3,5,6-tetramethyl-1,410 mM EDTA. Reduced-minus-oxidized difference spectra, phenylenediamine (+ 250 mV), N,N,N,N-tetramethyl-l,4as well as measurements of the Ph(NMe2), oxidation rate, phenylenediamine (+ 260 mV). The reduction potentials of were performed on samples before and after dialysis. Typical the mediators at pH = 7.0 are indicated in brackets. The oxidation rates were 2.9 & 0.3 pmol Ph(NMez), oxidized . ambient redox potentials were adjusted by using solutions of mg-l . min-', before and after dialysis. sodium ascorbate, sodium dithionite and K,[Fe(CN),], and were measured with a platinum electrode (P 1312) and a reference (Ag/AgCl) electrode (K 8040), both purchased from Spin quantitation Radiometer Electronics (Copenhagen, Denmark). The vessel Quantitation of the low-spin heme signal from cytochrome was continuously flushed with deoxygenated argon. The reaa3 was performed in relation to a metmyoglobin azide stan- duction potentials are quoted relative to the standard hydrodard (0.2 mM), prepared according to Bolard and Garnier gen electrode. [I 11. To avoid interference by other resonances, the intensity of the enzyme low-spin signals was determined indirectly, Chemicals through double-integration of a simulated spectrum, whose Oxidation/reduction mediators were obtained from intensity was adjusted to the corresponding experimental Aldrich Chemicals. Other reagents were from Merck, Sigma spectrum, obtained under non-saturating conditions. Both and Serva, all of the highest purity commercially available. integrals, from the enzyme signal and the standard, were then normalized using the Aasa and Viinngard correction for different g-factors [12]. The integral intensity of the high-spin heme RESULTS resonance was determined by the same procedure. Its value was then compared with that of the low-spin heme, after EPR signals of the isolated enzyme correction for the g-values and the Boltzmann factor of the resonance doublet.
The EPR spectrum at 20 K and 2.0 mW microwave power of purified Sulfolobus cytochrome aa3 is shown in Fig. 1 (up-
135 per trace). The resonances clearly indicate the presence of a high-spin heme component at a g-value of about 6 and a lowspin heme with features around g = 3 and g = 2.2.A small signal at g = 4.3is also detected, showing the presence of lowsymmetry non-heme adventitious iron. In the g = 2 region a resonance at g = 2.08,presumably due to copper, as well as a radical signal are present. These signals were observed in variable amounts, depending on the preparation. The highspin heme resonance could be simulated with g, = 6.03,g, = 5.97 and g, = 2.00 (Fig. 1, lower trace), indicating a very small rhombicity E/D < 0.002.The low-spin heme signal was simulated with g, = 3.02,g, = 2.23 and g, = 1.45 (Fig. 1, middle trace). In some preparations, a second low-spin heme component was present with g-values at g, = 2.93 and g, = 2.26,which accounted for not more than 10% of the total low-spin heme concentration.
g-values 6.0
20.10.
100
4.0
2.0
3.0
400
300
200
500
B [mTI
Spin quantitation The intensities of the low-spin and high-spin heme resonances of the as isolated enzyme were determined as indicated in Materials and Methods. For the low-spin heme signal a value of 0.95 f 0.1 spin/enzyme molecule was obtained. The intensity of the high-spin heme resonance, without correction for Boltzmann factors, was found to correspond to approximately 34% of the concentration of isolated cytochrome uu3. This value, however, is an underestimation, because high-spin ferric heme has a resonating I & lj2 > Kramers doublet with EPR silent I & 3/2 > and I 5/2 > excited doublets, usually about 10 - 30 cm- higher in energy [14],being partially occupied at 20 K. An upper limit for the true concentration of the EPR-detectable high-spin iron at the ambient redox potential was calculated using the lower limit estimate of the D-value, 10 cm-', obtained by the method described in [15]. Applying the corresponding Boltzmann distribution, a final value of 40 & 5% was calculated for the enzyme as isolated. Since the resting potential of the isolated enzyme was determined to be consistently about + 370 mV, correction for the EPR-silent fraction according to the bell-shaped titration behaviour of the high-spin heme (see below) raises the value determined to 95 10% of the concentration of cytochrome uu3.
EPR spectrum of the l-Cu/cytochrome-aa3 sample To investigate further whether the copper signal belongs to an intrinsic constituent of the enzyme, an extensive dialysis against 10 mM EDTA was performed yielding a product with only one copper atom/enzyme molecule. This sample retained, within experimental error, the same catalytic activity with Ph(NMe,), (see Methods) and displayed no changes of its optical spectrum as compared to the original preparation. The EPR spectrum of the 1-Cu/cytochrome-uu3 sample is shown in Fig. 2.The resonances of the high-spin (g = 6)and the lowspin (gz = 3.03 and g, = 2.21) hemes are clearly displayed and by the same procedure as described above, they quantitate to 0.85 0.1 spin/heme u and heme u3 (after correction for the bell-shaped titration curve). On the other hand, the g = 4.3 signal is more intense than in the enzyme preparation as isolated and appears to dominate the experimental spectrum due to an increased amount of adventitious iron(II1). However, the signal does not account for a correspondingly high integral intensity because it extends only in a narrow field range of about 100 mT; therefore, spin quantitation by double-integration, assuming equal population of the three Kramers doublets, reveals only 0.15 spin/enzyme molecule.
Fig. 2. EPR spectrum of the l-Cu/cytochrome-aa3 sample. The enzyme was prepared as described in Materials and Methods. Protein concentration, 0.43 mg/ml in 0.8 M KH,PO,, 0.05% sarcosyl, pH = 7.4; heme iron content, 20.8 FM; copper content, 0.9 Cu/cytochrome-aa3 (determined from two analogous preparations). Microwave power, 2.4 mW; temperature, 9.3 K. Other conditions as in Fig. 1.
2.0
10 power
100
lmWl
Fig. 3. Power saturation of the high-spin heme signals from cytochrome au3 of S. acidocaldurius (A) and P. denitvijicans (0) at 10 K. The Paracoccus enzyme (protein concentration 8 mg/ml) was partially reduced by ascorbate (1.8 mM) prior to the experiment to generate the high-spin heme signal. The Suljolohus enzyme (protein concentration 3.6 mg/ml) was used as prepared. The logarithm of the relative amplitudes (ampl) of the g = 6 derivative peaks of both species, divided by the square root of the microwave power, is presented as a function of the microwave power.
Like the quantitations of the heme signals, this indicates a slight degree of protein damage due to the extensive dialysis. However, the extension of protein degradation is low, since both the catalytic activity and the optical spectrum of the enzyme are essentially unaltered after the dialysis. The copper signal as observed in the isolated enzyme disappeared suggesting that it wasbdueto adventitious copper not involved in the catalytic process. Relaxation behaviour of the high-spin heme signal The microwave-power dependence of the high-spin heme signal is displayed in Fig. 3. For comparison purposes, the corresponding curve of cytochrome uu3 from Purucoccus
136
o m
I""'
'
4 0.4
0.24
I
o i
100
150
200
250
300
350
400
450
E , rnV _____. ~~
1 Fig. 5. Effect of azide on cytochrome aa3 of Sulfolobus. The enzyme (3.6 mg/ml) was incubated for 1 h in the presence of 0.1 M azide. Upper trace, enzyme as prepared; lower trace, enzyme plus azide; heme iron concentration, 61.6 pM; copper content, 1.8-3.4 Cu/ cytochrome-aa3 (determined from analogous preparations). Microwave power: 20 mW. Other conditions as in Fig. 1.
._ ? 0.64
Z i 0.41
w
i
100
150
200
250
300 E , rnV
350
400
450
Fig. 4. Plot of redox titration data of high-spin and low-spin heme centers of cytochrome aa3. The redox titrations were performed at pH 7.4 as described in Materials and Methods. The signal amplitudes of the g = 6 resonance (upper part, two titrations) and the g = 3.03 resonance (lower part, one titration) are displayed against Eh. For the simple model without interacting potentials (solid lines), the high-spin data were fitted using El = 200 mV and E2 = 370 (n = 1) and the low-spin data were fitted with E = 305 mV (n = 1). The curves for the model with interacting centers (dashed lines), were obtained using E(heme a3) = 195 mV, E(heme a) = 255 mV, E(Cu,) = 370 mV and a heme - heme interacting potential of - 55 mV. Protein concentrations were in the range of 1.7 mg/ml.
denitrificans is also shown. The saturation behaviour of both proteins is clearly different: the high-spin heme signal from the Suljolobus enzyme saturated at a significantly lower power (half-saturation at 148 mW at 10 K) than the signal from the Paracoccus enzyme (half-saturation at 360 mW at 10 K). In order to ensure that the obviously lower value of the Suljolobus enzyme is not due to protein modification during the preparation procedure, the corresponding high-spin heme signal in the membranes from Suljolobus was investigated. The halfsaturation value in the membrane-bound state was determined consistently as 150 mW at 10 K (not shown), suggesting the integrity of the enzyme after the isolation. Redox potentiometry of the heme components
Titration curves of the amplitudes of the EPR derivative signals (measured at g = 6 and g = 3.02 for the high-spin and low-spin hemes, respectively) against the solution redox potentials, at pH 7.4, are shown in Fig. 4. Besides the changes in intensity, no other alterations of the heme signals were
observed during the titration. The bell-shaped curve of the high-spin heme (upper part) was fitted by a Nernst equation for two consecutive one-electron processes (n = 1). Starting from the low-potential side, a large increase in signal height is observed due to the oxidation of the high-spin heme center with a reduction potential of 200 f 10 mV. With further increase of the solution redox potential the intensity of the resonance passes through a maximum and finally declines with a reduction potential of 370 10 mV. The decline of the intensity at high redox potentials is proposed to be due to exchange coupling with increasing amounts of oxidized CuB. The titration curve and its simulation readily explain that in samples of the enzyme as isolated (resting potential = 370 mV, as judged by the intensity of the low-spin and high-spin heme resonances) only about 40% of the high-spin heme is EPR detectable. We note that a small amount of high-spin heme signal (10-15% of the maximum signal intensity), which is probably due to damaged protein, could not be reduced down to 0 mV. In the lower part of Fig. 4 the titration curve for the lowspin heme is shown. Within the accuracy of the available data, the curve was described by a one-electron Nernst equation ( n = 1) and a reduction potential of 305 f 10 mV. A preliminary analysis of the redox data for both heme centers from Sulfolobus cytochrome aa3 was done through an interactive model for a three-redox-center protein, using a set of equations as described in [16]. The result of this analysis is depicted in Fig. 4 (dashed curves). The intrinsic reduction potentials for each center (defined as the reduction potential of the center when all the other enzyme centers are fully reduced) and the interacting potential between the hemes that best reproduced the experimental data yielded E(heme a 3 ) = 195 f 10 mV, E(heme a) = 255 & 10 mV and E(CuB) = 370 10 mV and a negative heme-heme interacting potential of -55 10 mV. The fitting did not improve by assuming an interacting potential between heme a and CuB. From the quality of the fits a decision for one of the models can not be made at present (see Discussion).
137 Ligand-binding behaviour
The effect of azide on the EPR spectrum of cytochrome aa3 from Sulfolobus was examined. Upon addition of 0.1 M
azide to the enzyme solution, the high-spin heme signal broadened, and showed a shoulder due to rhombic splitting, whereas the low-spin heme resonance amplitude was only slightly affected. The radical signal totally disappeared after the incubation with azide (Fig. 5). Formation of a new lowspin heme signal at g, = 2.9, which is typical for the low-spin heme component of cytochrome a3 with azide in the bovine heart enzyme [17], could not be detected.
DISCUSSION The spectroscopic data presented in this paper allowed a detailed characterization of the metal centers from the Sulfolobus acidocaldarius aa3-type terminal oxidase, a caldariella-quinol oxidase. Significant biochemical and biophysical similarities to the terminal oxidases of the bo-type from Escherichia coli [18] and of the aa3-type from mammalian or bacterial sources [17] were found. Both types of oxidases contain copper and high-spin and low-spin hemes. In the mammalian enzyme, both hemes are of the A-type, while in the E. coli enzyme they were recently shown to be of a new type, called heme-0, structurally related to heme A [19], but with a blue-shifted pyridine hemochrome spectrum. From DNA sequence data, large similarities between subunit I of mammalian cytochrome aa3 and cytochrome bo of E. coli have been demonstrated [20]. In functional terms, both enzymes have been shown to act as redox-driven proton pumps [21,22], but they differ in the naturally used reductant. Cytochrome uu3 generally oxidizes reduced cytochrome c, whereas cytochrome bo functions as an ubiquinol oxidase. Meanwhile, an alternative terminal oxidase from Bacillus subtilis, cytochrome aa3-600, could also be shown to act as an aa3-type quinol oxidase [23]. Thus, the capability of cytochrome aa3 to use quinols as substrates seems to be a common feature in the eubacterial and archaebacterial urkingdoms. The EPR spectra of the Sulfolobus enzyme exhibit a highspin and a low-spin heme signal, accounting for about 100% of the enzyme concentration. The g-values (g, = 3.02, g, = 2.23 and g, = 1.45) and the crystal field parameters (tetragonality, AIL = 3.35 and rhombicity, V/A = 1.725) of the low-spin heme component are very similar to those of the bovine heart cytochrome aa3 and E. coli cytochrome bo [17, 181, suggesting a comparable bis-histidine axial ligation of heme a. The high-spin heme A exhibits a single spectrum, with g, = 6.03, g, = 5.97, g, = 2.0, and E/D< 0.002. The rhombicity of this signal is smaller than the rhombicity of the resonance from the P. denitrijiicans cytochrome aa3, with g, = 6.1 and g, = 5.9 (E/D = 0.004). In the membraneous as well as in the purified state the high-spin heme signal from the Sulfolohus enzyme shows a power saturation behaviour different from that of the heme a3 center from the P. denitrificans enzyme. Adopting an Orbach mechanism [24] for the spin- relaxation process, this feature reflects stronger zerofield interaction of the high-spin heme iron in the Sulfolobus enzyme as compared to the Paracoccus enzyme. The ligandbinding behaviour of cytochrome aa3 from Sulfolobus resembles that of cytochrome bo from E. coli [18], as demonstrated by the azide-binding properties of both enzymes. After addition of the ligand, the high-spin heme resonances of both enzymes are broadened, whereas no high-spin to low-spin
transition occurs, as detected in cytochrome c oxidases [17]. This may be a newly detected property characteristic for quinol oxidases as a whole. In accordance with the reactivity of azide with this archaebacterial terminal oxidase, the highspin heme species, which is sensitive to azide, is attributable to cytochrome a3, while the low-spin heme component is due to cytochrome a. The EPR spectrum of the extensively dialyzed enzyme, which essentially retained its activity and contains only one copper atom/protein molecule, does not show any copper EPR signal. This observation proves the absence of a center analogous to CuA of mammalian cytochrome 4u3, as has been also shown for the E. coli cytochrome bo [25] and the B. subtilis cytochrome aa3-600 [23]. This type of copper center in cytochrome c oxidases is a ligand of subunit I1 of the terminal oxidase complexes. Therefore it might have been missed in the present preparation consisting of only a single polypeptide species. In fact, the recently described operon coding for the Sulfolobus complex [26] shows that more than a single subunit is necessary to form the functional complex in vivo ; however, the subunit I1 equivalent also does not contain any detectable Cu- binding sites as concluded from the deduced amino acid sequence. This data implies that Sulfolobus cytochrome aa3 contains only a CUB-typecenter, which is EPR-silent either due to the diamagnetic 3d1' configuration in its monovalent Cu(1) state or due to exchange coupling with high-spin ferric ion in its divalent Cu(I1) state. The reduction potentials of the heme centers of the Sulfolobus enzyme were determined to be 305 & 10 mV for the low-spin heme and 200 & 10 mV and 370 & 10 mV for the high-spin heme, at pH = 7.4, using a simple model without considering heme - heme interacting potentials. These values are similar to those determined for the mammalian cytochrome au3 [17]. The data for the Sulfolobus enzyme can also be explained by a more complex model, including hemeheme interaction. The set of parameters that best fitted the experimental data points yielded an interacting potential between heme a and heme a3 of - 55 & 10 mV, indicative of negative cooperativity, and intrinsic reduction potentials for each heme of 195 & 10 mV (heme a3)and 255 & 10 mV (heme a). The interaction potential obtained is similar to that found for the E. coli cytochrome bo [25]. However, the intrinsic heme reduction potentials are different, reflecting either the different heme structures or the different reduction potentials of the natural reductants: menaquinone has Eo = 0 mV, while caldariella quinone has Eo = 100 mV [5]. The reduction potentials obtained through the EPR redox titration are different from those previously presented, obtained from a visible redox titration [4]. However it is not possible to directly compare the data obtained by these two different spectroscopic methods until a full set of thermodynamic parameters (intrinsic reduction potentials and interacting potentials between each metal center) describing the redox pattern of Sulfolobus cytochrome 4u3 is unequicocally determined [17]. Also, the experimental data available at the present moment do not allow a clear decision between the two redox models. Further studies are in progress to clarify this point which is of fundamental importance for the understanding of the enzyme function. The bell-shaped titration curve for the high-spin heme resonance of the Sulfolobus cytochrome oxidase indicates that the catalytic site is a strongly coupled binuclear center composed of heme a3 and CU,, similar to the corresponding centers in mammalian cytochrome aa3 [17] and cytochrome bo from E. coli [25]. The process with Eo = 370 & 10 mV is
138 assigned to the oxidation of the Cu, center, yielding a ferriheme a,-Cu(Ii) binuclear center with a net integer spin system and consequent loss of the high-spin heme EPR spectrum. Other evidence for the presence of a tightly coupled heme -copper pair is given by the results obtained with the extensively dialysed enzyme, containing only one copper atom/enzyme molecule, as the copper center involved in the binuclear center is not EPR-detectable under normal conditions. We note that under no experimental conditions were we able to detect an integer-spin spectrum from the spincoupled pair heme u3(S = 5/2)/Cu(II),(S = 1/2). We presume that a large zero-field splitting of the S,,,,, = 2 ground-state prohibits the otherwise typical 'g = 14' resonance [27]. i n conclusion, the data presented show that the metal centers in the cytochrome aa3 from Sulfolobus consist of a low-spin heme a, a high-spin heme a3 and a CuB-typecenter. The heme u3 and the Cu, form a binuclear center, similar to those observed in mammalian cytochrome aa3 and E. coli cytochrome bo. This implies that all physiological functions in terminal electron transport are unified in the single-subunit cytochrome uu3 from Sulfolobus. Furthermore, the absence of CuA seems to be characteristic for quinol oxidases as also shown for E. coli cytochrome bo [25] and B. subtilis cytochrome au3-600 [23], while its presence is a prerequisite for cytochrome c to replace quinol as an electron donor. Parts of these data were presented in poster form at the 5th International Conference on BioInorganic Chemistry, 4- 10 August 1991, Oxford, UK. The authors would like to thank Mrs Antje Lassen for skilful technical assistance, and Prof. A.V. Xavier for critical discussions. This work was supported in part by Junta Nacional de Investigap?o Cientijica e Tecnoldgica, Portugal, grant PC-CEN-649 (MT) and by a grant from the Deutsche Forschungsgemeinschaft, Federal Republic of Germany (S.A.).
REFERENCES 1. Rrock, T.D., Brock, K.M., Belly, R.T. & Weiss, R.L. (1972) Arch. Microhiol. 84, 54 - 68.
2. Moll, R. & Schafer, G. (1988) FEBSLett. 232, 359-363. 3. Liibben, M. & SchCfer, G. (1989) J . Bacteriol. 171,6101 -6116. 4. Anemiiller, S. & Schafer, G. (1989) FEBS Lett. 244, 451 -455. 5. Anemiiller, S. & Schafer, G. (1990) Eur. J. Biochem. 191, 297305. 6. Langworthy, T.A., Tornabene, T.G. & Holzer, G. (1982) Zentralbl. Bacteriol. Hyg. I Aht. Orig. C3, 228-244. 7. Hildebrandt, P., Heibel, G., Anemiiller, S. & Schafer, G. (1991) FEBS Lett. 283, 131 - 134. 8. Watters, C. (1978) Anal. Biochem. 88, 695-698. 9. Peterson, G.L. (1977) Anal. Biochem. 83, 346-356. 10. Kuboyama, M., Yong, F.C. & King, T.E. (1972) J. Biol. Chem. 247,6375-6383. 11. Bolard, J. &Gamier, A. (1972) Biochim. Biophys. Acta263, 535549. 12. Aasa, R. &Vanngard, T. (1975) J. Mugn. Res. 19,308-315. 13. Dutton, P.L. (1978) Methods Enzymol. 54, 43 1-435. 14. Aasa, R., Albracht, S.P.J., Falk, K.-E., Laane, B. & Vanngard, T. (1976) Biochim. Biophys. -4cta 422,260-272. 15. Slappendel, S., Veldink, G.A., Vliegenthard, J.F.G., Aasa, R. & Malmstrom, B.G. (1980) Biochim. Biophys. Acta 642, 30-39. 16. Coletta, M., Catarino, T., LeCall, J. & Xavier, A.V. (1991) Eur. J. Biochem. 202, 1101 -1106. 17. Wikstrom, M., Krab, K . & Saraste, M. (1981) Cytochrome oxidase: a synthesis, pp. 55 - 1 16, Academic Press, London. 18. Hata, A., Kirino, Y., Matsuura, K., Itoh, S., Hiyama, T., Konishi,K. & Anraku,Y. (1985) Biochim. Biophys. Acta 810, 62-72. 19. Puustinen, A. & Wikstrom, M. (1991) Proc. Nut1 Acad. Sci. USA 88,6122-6126. 20. Chepuri, V., Lemieux, L., Au, D.C.T. & Gennis, R.B. (1990) J . Biol. Chem. 265,11185 - 11 192. 21. Puustinen, A,, Finel, M., Virkki, M. & Wikstrom, M. (1989) FEBS Lett. 249, 163- 167. 22. Wikstrom, M. (1977) Nature 266,271 -273. 23. Lauraeus, M., Haltia, T., Saraste, M. & Wikstrom, M. (1991) Eur. J. Biochem. 197, 699 - 705. 24. Finn, C.B.P., Orbach, R. &Wolf, W.P. (1961) Proc. Phys. Soc. Lond. 77, 261 -268. 25. Salerno, J.C., Bolgiano, B., Poole, R.K., Gennis, R.B. & Ingledew, W.J. (1990) J. Biol. Chem. 265, 4364-4368. 26. Saraste, M., Holm, L., Lemieux, L., Liibben, M. & van der Oost (1991) Biochem. Soc. Trans. 19,608-612. 27. Hagen, W.R., (1982) Biochim. Biophys. Acta 708, 82-98.
EPR studies of cytochrome aa3 from Sulfolobus acidocaldarius Evidence for a binuclear center in archaebacterial terminal oxidase Stefan ANEMULLER
’, Eckhard BILL’,
Giinter SCHAFER3, Alfred X. TRAUTWEIN
’ and Miguel TEIXEIRA’
Centro de Tecnologia Quimica e Biologica and Universidade Nova de Lisboa, Oeiras, Portugal Institut fur Biochemie, Medizinische Universitat zu Liibeck, Federal Republic of Germany
’ Institut fur Physik and
(Received April 14/August 31, 1992) - EJB 92 0527
The purified cytochrome aa3-type oxidase from Suljolobus acidocaldarius (DSM 639) consists of a single subunit, containing one low-spin and one high-spin A-type hemes and copper [Anemiiller, S. and Schafer, G. (1990) Eur. J. Biochem. 191, 297-3051. The enzyme metal centers were investigated by electron paramagnetic resonance spectroscopy (EPR), coupled to redox potentiometry. The lowspin heme EPR signal has the following g-values: g, = 3.02, g, = 2.23 and g, = 1.45 and the highspin heme exhibits an almost axial spectrum (gy = 6.03 and g, = 5.97, E/D< 0.002). In the enzyme as isolated the low-spin resonance corresponds to 95 10% of the enzyme concentration, while the high-spin signal accounts for only 40 f 5%. However, taking into account the redox potential dependence of the high-spin heme signal, this value also rises to 95 & 10%. The high-spin heme signal of the Suljolobus enzyme shows spectral characteristics distinct from those of the Paracoccus denitrijiicuns one: it shows a smaller rhombicity (8, = 6.1 and g, = 5.9, E/D = 0.004 for the P. denitrijiicuns enzyme) and it is easier to saturate, having a half saturation power of 148 mW compared to 360 mW for the P. denitrificans protein, both at 10 K. The EPR spectrum of an extensively dialyzed and active enzyme sample containing only one copper atom/enzyme molecule does not display CuAlike resonances, indicating that this enzyme contains only a CUB-typecenter. The EPR-redox titration of the high-spin heme signal, which is assigned to cytochrome a3, gives a bell shaped curve, which was simulated by a non-interactive two step redox process, with reduction potentials of 200 f 10 mV and 370 f 10 mV at pH = 7.4. The decrease of the signal amplitude at high redox potentials is proposed to be due to oxidation of a CuB(1) center, which in the Cu,(II) state is tightly spin-coupled to the heme u3 center. The reduction potential of the low-spin resonance was determined using the same model as 305 & 10 mV at pH = 7.4 by EPR redox titration. Addition of azide to the enzyme affects only the high-spin heme signal, consistent with the assignment of this resonance to heme a3. The results are discussed in the context of the redox center composition of quinol and cytochrome c oxidases.
The thermoacidophilic archaebacterium Suljolobus acidocaldarius grows aerobically at temperatures up to 85 “C and pH values of 2- 3 [l]. This phylogenetically very early organism is likely to use a chemiosmotic mechanism for energy conservation [2, 31. Although the unequivocal identity of the individual proton pumps has not yet been established, a possible candidate has been purified as a terminal oxidase of the cytochrome aa3 type [4, 51. This enzyme consists of a single subunit and was found to be a novel species of cytochrome ua3, since caldariellaquinone [6], the ubiquinone analogue in the membrane of Suljolobus, is used as reductant instead of reduced cytochrome c.
Optical spectra clearly indicated the presence of two different heme centers with apparent redox potentials of 220 mV and f370 mV at pH = 7.4 [5]. One of the hemes binds CO and therefore was identified as heme a3. From resonance Raman spectra, it could be concluded that in both the oxidized and the fully reduced state hemes a and u3 are in the sixcoordinated low-spin and six-coordinated high-spin heme configuration, respectively [7]. The resonance Raman spectra of the oxidized enzyme are similar to those of the mammalian and plant cytochrome aa3, but the spectrum of the reduced form differs remarkably: in contrast to the beef heart enzyme, the formyl vibrations of heme a ruled out significant hydrogen bonding interactions. Preliminary EPR studies confirmed the presence of highCorrespondence to S . Anemiiller, Centro de Tecnologia Quimica e spin and low-spin heme centers [5]. However, both heme Biologica, Apt. 127, P-2780 Oeiras, Portugal centers were not quantitated and their redox behaviour was Fax: +351 1 4428766. Abbreviations. Ph(NMeZ)’, N,N,N’,N’-tetramethyl-l,4-phenyl- not studied. In this paper we report data on spin quantification of both heme centers, as well as on the redox potentials of enediamine dihydrochloride. Enzyme. Cytochrome uu3 (EC.1.9.3.1) the high-spin and low-spin hemes, obtained by EPR redox
+
134 titrations. We propose that archaebacterial quinol oxidase is another example of cytochrome aa3 with a binuclear center in which the high-spin heme and CuB are magnetically coupled. The saturation behaviour of heme a3 has also been studied and compared to that of the Paracoccus denitrificans enzyme. Studies on the ligand binding effect of azide corroborate our conclusions.
g-values
20.10.
6.0
4.0
3.0
2.0
MATERIALS AND METHODS Cultivation of Sulfolobus acidocaldarius (DSM 639), membrane preparation and purification of cytochrome aa3 were done as described previously [5].Membrane protein was determined as in [XI and detergent-solubilized protein was determined by the Lowry method in the presence of SDS [9]. Paracoccus denitrficans cytochrome ua3 was a kind gift from Prof. B. Ludwig (Medical University of Liibeck, Federal Republic of Germany). Spectroscopic methods
100
200
300
400
500
B [mTI
, Fig. 1. EPR spectrum of the purified Sulfolobus cytochrome uu3 (upper trace) and theoretical simulations of the low-spin (middle trace) and of the high-spin (lower trace) heme signals. Protein concentration, 4.9 mg/ ml in 0.8 M KH2P04, 0.05% sarcosyl, pH = 7.4; heme iron content, 61.8 pM; copper content, 1.8- 3.4 Cu/cytochrome-aa3 (determined from analogous preparations). Microwave frequency, 9.453 GHz; microwave power, 2 mW; modulation frequency, 100 kHz; modulation amplitude, 1 mT; time constant: 0.041 s; gain: 1 . 6 ~ sweep width, 0.01 -0.51 T; temperature, 20 K. The low-spin heme simulation was generated by using the following g-values and linewidths: g, = 3.02, g, = 2.23 and g, = 1.45; 5 mT, 5 mT and 20 mT. The high-spin heme signal was simulated with g, = 6.03, g, = 5.97 and g, = 2.0 and the following linewidths: 8 mT, 15 mT and 20 mT.
Visible spectra of detergent-solubilized cytochrome aa3 were measured in a Hewlett Packard HP 8450 A diode array spectrophotometer. N,N,N',N'-Tetramethyl-l,4-phenylenediamine dihydrochloride [Ph(NMeJ2] oxidation was followed as described previously at 546 nm [5]. Heme A content was determined using E~~~~~ (reduced - oxidized) = 12 mM-' cm- [lo]. Atomic absorption spectroscopy measurements of copper content were performed with a Hitachi 180-80 polarized Zeeman atomic absorption spectrophotometer at 324.8 nm. EPR spectra were recorded with an X-band Bruker ER 200 D and an X-band Bruker ESP 380 spectrometer EPR redox titration equipped with an ESR 900 continuous-flow helium cryostat Potentiometric redox titrations were performed in a buffer from Oxford Instruments. containing 0.8 M KH2P04, 0.05% sarcosyl, pH 7.4 under anaerobic conditions as described by Dutton [13]. The redox mediators used, at concentrations between 40 - 200 pM Preparation of a l-Cu/cytochrome-aa3 sample were: phenazine ethosulfate (+ 55 mV), phenazine methoCytochrome aa3 in a buffer containing 0.8 M KH2P04, sulfate (+ 80 mV), 1,2-naphthoquinone (+ 180 mV), 1,20.05% sarcosyl, pH 7.4 was dialyzed at 4°C for 60 h against naphthoquinone-Csulfonic acid ( + 215 mV), 2,6-dichloro100-fold volume excess of the same buffer, supplemented with phenol indophenol (+ 217 mV), 2,3,5,6-tetramethyl-1,410 mM EDTA. Reduced-minus-oxidized difference spectra, phenylenediamine (+ 250 mV), N,N,N,N-tetramethyl-l,4as well as measurements of the Ph(NMe2), oxidation rate, phenylenediamine (+ 260 mV). The reduction potentials of were performed on samples before and after dialysis. Typical the mediators at pH = 7.0 are indicated in brackets. The oxidation rates were 2.9 & 0.3 pmol Ph(NMez), oxidized . ambient redox potentials were adjusted by using solutions of mg-l . min-', before and after dialysis. sodium ascorbate, sodium dithionite and K,[Fe(CN),], and were measured with a platinum electrode (P 1312) and a reference (Ag/AgCl) electrode (K 8040), both purchased from Spin quantitation Radiometer Electronics (Copenhagen, Denmark). The vessel Quantitation of the low-spin heme signal from cytochrome was continuously flushed with deoxygenated argon. The reaa3 was performed in relation to a metmyoglobin azide stan- duction potentials are quoted relative to the standard hydrodard (0.2 mM), prepared according to Bolard and Garnier gen electrode. [I 11. To avoid interference by other resonances, the intensity of the enzyme low-spin signals was determined indirectly, Chemicals through double-integration of a simulated spectrum, whose Oxidation/reduction mediators were obtained from intensity was adjusted to the corresponding experimental Aldrich Chemicals. Other reagents were from Merck, Sigma spectrum, obtained under non-saturating conditions. Both and Serva, all of the highest purity commercially available. integrals, from the enzyme signal and the standard, were then normalized using the Aasa and Viinngard correction for different g-factors [12]. The integral intensity of the high-spin heme RESULTS resonance was determined by the same procedure. Its value was then compared with that of the low-spin heme, after EPR signals of the isolated enzyme correction for the g-values and the Boltzmann factor of the resonance doublet.
The EPR spectrum at 20 K and 2.0 mW microwave power of purified Sulfolobus cytochrome aa3 is shown in Fig. 1 (up-
135 per trace). The resonances clearly indicate the presence of a high-spin heme component at a g-value of about 6 and a lowspin heme with features around g = 3 and g = 2.2.A small signal at g = 4.3is also detected, showing the presence of lowsymmetry non-heme adventitious iron. In the g = 2 region a resonance at g = 2.08,presumably due to copper, as well as a radical signal are present. These signals were observed in variable amounts, depending on the preparation. The highspin heme resonance could be simulated with g, = 6.03,g, = 5.97 and g, = 2.00 (Fig. 1, lower trace), indicating a very small rhombicity E/D < 0.002.The low-spin heme signal was simulated with g, = 3.02,g, = 2.23 and g, = 1.45 (Fig. 1, middle trace). In some preparations, a second low-spin heme component was present with g-values at g, = 2.93 and g, = 2.26,which accounted for not more than 10% of the total low-spin heme concentration.
g-values 6.0
20.10.
100
4.0
2.0
3.0
400
300
200
500
B [mTI
Spin quantitation The intensities of the low-spin and high-spin heme resonances of the as isolated enzyme were determined as indicated in Materials and Methods. For the low-spin heme signal a value of 0.95 f 0.1 spin/enzyme molecule was obtained. The intensity of the high-spin heme resonance, without correction for Boltzmann factors, was found to correspond to approximately 34% of the concentration of isolated cytochrome uu3. This value, however, is an underestimation, because high-spin ferric heme has a resonating I & lj2 > Kramers doublet with EPR silent I & 3/2 > and I 5/2 > excited doublets, usually about 10 - 30 cm- higher in energy [14],being partially occupied at 20 K. An upper limit for the true concentration of the EPR-detectable high-spin iron at the ambient redox potential was calculated using the lower limit estimate of the D-value, 10 cm-', obtained by the method described in [15]. Applying the corresponding Boltzmann distribution, a final value of 40 & 5% was calculated for the enzyme as isolated. Since the resting potential of the isolated enzyme was determined to be consistently about + 370 mV, correction for the EPR-silent fraction according to the bell-shaped titration behaviour of the high-spin heme (see below) raises the value determined to 95 10% of the concentration of cytochrome uu3.
EPR spectrum of the l-Cu/cytochrome-aa3 sample To investigate further whether the copper signal belongs to an intrinsic constituent of the enzyme, an extensive dialysis against 10 mM EDTA was performed yielding a product with only one copper atom/enzyme molecule. This sample retained, within experimental error, the same catalytic activity with Ph(NMe,), (see Methods) and displayed no changes of its optical spectrum as compared to the original preparation. The EPR spectrum of the 1-Cu/cytochrome-uu3 sample is shown in Fig. 2.The resonances of the high-spin (g = 6)and the lowspin (gz = 3.03 and g, = 2.21) hemes are clearly displayed and by the same procedure as described above, they quantitate to 0.85 0.1 spin/heme u and heme u3 (after correction for the bell-shaped titration curve). On the other hand, the g = 4.3 signal is more intense than in the enzyme preparation as isolated and appears to dominate the experimental spectrum due to an increased amount of adventitious iron(II1). However, the signal does not account for a correspondingly high integral intensity because it extends only in a narrow field range of about 100 mT; therefore, spin quantitation by double-integration, assuming equal population of the three Kramers doublets, reveals only 0.15 spin/enzyme molecule.
Fig. 2. EPR spectrum of the l-Cu/cytochrome-aa3 sample. The enzyme was prepared as described in Materials and Methods. Protein concentration, 0.43 mg/ml in 0.8 M KH,PO,, 0.05% sarcosyl, pH = 7.4; heme iron content, 20.8 FM; copper content, 0.9 Cu/cytochrome-aa3 (determined from two analogous preparations). Microwave power, 2.4 mW; temperature, 9.3 K. Other conditions as in Fig. 1.
2.0
10 power
100
lmWl
Fig. 3. Power saturation of the high-spin heme signals from cytochrome au3 of S. acidocaldurius (A) and P. denitvijicans (0) at 10 K. The Paracoccus enzyme (protein concentration 8 mg/ml) was partially reduced by ascorbate (1.8 mM) prior to the experiment to generate the high-spin heme signal. The Suljolohus enzyme (protein concentration 3.6 mg/ml) was used as prepared. The logarithm of the relative amplitudes (ampl) of the g = 6 derivative peaks of both species, divided by the square root of the microwave power, is presented as a function of the microwave power.
Like the quantitations of the heme signals, this indicates a slight degree of protein damage due to the extensive dialysis. However, the extension of protein degradation is low, since both the catalytic activity and the optical spectrum of the enzyme are essentially unaltered after the dialysis. The copper signal as observed in the isolated enzyme disappeared suggesting that it wasbdueto adventitious copper not involved in the catalytic process. Relaxation behaviour of the high-spin heme signal The microwave-power dependence of the high-spin heme signal is displayed in Fig. 3. For comparison purposes, the corresponding curve of cytochrome uu3 from Purucoccus
136
o m
I""'
'
4 0.4
0.24
I
o i
100
150
200
250
300
350
400
450
E , rnV _____. ~~
1 Fig. 5. Effect of azide on cytochrome aa3 of Sulfolobus. The enzyme (3.6 mg/ml) was incubated for 1 h in the presence of 0.1 M azide. Upper trace, enzyme as prepared; lower trace, enzyme plus azide; heme iron concentration, 61.6 pM; copper content, 1.8-3.4 Cu/ cytochrome-aa3 (determined from analogous preparations). Microwave power: 20 mW. Other conditions as in Fig. 1.
._ ? 0.64
Z i 0.41
w
i
100
150
200
250
300 E , rnV
350
400
450
Fig. 4. Plot of redox titration data of high-spin and low-spin heme centers of cytochrome aa3. The redox titrations were performed at pH 7.4 as described in Materials and Methods. The signal amplitudes of the g = 6 resonance (upper part, two titrations) and the g = 3.03 resonance (lower part, one titration) are displayed against Eh. For the simple model without interacting potentials (solid lines), the high-spin data were fitted using El = 200 mV and E2 = 370 (n = 1) and the low-spin data were fitted with E = 305 mV (n = 1). The curves for the model with interacting centers (dashed lines), were obtained using E(heme a3) = 195 mV, E(heme a) = 255 mV, E(Cu,) = 370 mV and a heme - heme interacting potential of - 55 mV. Protein concentrations were in the range of 1.7 mg/ml.
denitrificans is also shown. The saturation behaviour of both proteins is clearly different: the high-spin heme signal from the Suljolobus enzyme saturated at a significantly lower power (half-saturation at 148 mW at 10 K) than the signal from the Paracoccus enzyme (half-saturation at 360 mW at 10 K). In order to ensure that the obviously lower value of the Suljolobus enzyme is not due to protein modification during the preparation procedure, the corresponding high-spin heme signal in the membranes from Suljolobus was investigated. The halfsaturation value in the membrane-bound state was determined consistently as 150 mW at 10 K (not shown), suggesting the integrity of the enzyme after the isolation. Redox potentiometry of the heme components
Titration curves of the amplitudes of the EPR derivative signals (measured at g = 6 and g = 3.02 for the high-spin and low-spin hemes, respectively) against the solution redox potentials, at pH 7.4, are shown in Fig. 4. Besides the changes in intensity, no other alterations of the heme signals were
observed during the titration. The bell-shaped curve of the high-spin heme (upper part) was fitted by a Nernst equation for two consecutive one-electron processes (n = 1). Starting from the low-potential side, a large increase in signal height is observed due to the oxidation of the high-spin heme center with a reduction potential of 200 f 10 mV. With further increase of the solution redox potential the intensity of the resonance passes through a maximum and finally declines with a reduction potential of 370 10 mV. The decline of the intensity at high redox potentials is proposed to be due to exchange coupling with increasing amounts of oxidized CuB. The titration curve and its simulation readily explain that in samples of the enzyme as isolated (resting potential = 370 mV, as judged by the intensity of the low-spin and high-spin heme resonances) only about 40% of the high-spin heme is EPR detectable. We note that a small amount of high-spin heme signal (10-15% of the maximum signal intensity), which is probably due to damaged protein, could not be reduced down to 0 mV. In the lower part of Fig. 4 the titration curve for the lowspin heme is shown. Within the accuracy of the available data, the curve was described by a one-electron Nernst equation ( n = 1) and a reduction potential of 305 f 10 mV. A preliminary analysis of the redox data for both heme centers from Sulfolobus cytochrome aa3 was done through an interactive model for a three-redox-center protein, using a set of equations as described in [16]. The result of this analysis is depicted in Fig. 4 (dashed curves). The intrinsic reduction potentials for each center (defined as the reduction potential of the center when all the other enzyme centers are fully reduced) and the interacting potential between the hemes that best reproduced the experimental data yielded E(heme a 3 ) = 195 f 10 mV, E(heme a) = 255 & 10 mV and E(CuB) = 370 10 mV and a negative heme-heme interacting potential of -55 10 mV. The fitting did not improve by assuming an interacting potential between heme a and CuB. From the quality of the fits a decision for one of the models can not be made at present (see Discussion).
137 Ligand-binding behaviour
The effect of azide on the EPR spectrum of cytochrome aa3 from Sulfolobus was examined. Upon addition of 0.1 M
azide to the enzyme solution, the high-spin heme signal broadened, and showed a shoulder due to rhombic splitting, whereas the low-spin heme resonance amplitude was only slightly affected. The radical signal totally disappeared after the incubation with azide (Fig. 5). Formation of a new lowspin heme signal at g, = 2.9, which is typical for the low-spin heme component of cytochrome a3 with azide in the bovine heart enzyme [17], could not be detected.
DISCUSSION The spectroscopic data presented in this paper allowed a detailed characterization of the metal centers from the Sulfolobus acidocaldarius aa3-type terminal oxidase, a caldariella-quinol oxidase. Significant biochemical and biophysical similarities to the terminal oxidases of the bo-type from Escherichia coli [18] and of the aa3-type from mammalian or bacterial sources [17] were found. Both types of oxidases contain copper and high-spin and low-spin hemes. In the mammalian enzyme, both hemes are of the A-type, while in the E. coli enzyme they were recently shown to be of a new type, called heme-0, structurally related to heme A [19], but with a blue-shifted pyridine hemochrome spectrum. From DNA sequence data, large similarities between subunit I of mammalian cytochrome aa3 and cytochrome bo of E. coli have been demonstrated [20]. In functional terms, both enzymes have been shown to act as redox-driven proton pumps [21,22], but they differ in the naturally used reductant. Cytochrome uu3 generally oxidizes reduced cytochrome c, whereas cytochrome bo functions as an ubiquinol oxidase. Meanwhile, an alternative terminal oxidase from Bacillus subtilis, cytochrome aa3-600, could also be shown to act as an aa3-type quinol oxidase [23]. Thus, the capability of cytochrome aa3 to use quinols as substrates seems to be a common feature in the eubacterial and archaebacterial urkingdoms. The EPR spectra of the Sulfolobus enzyme exhibit a highspin and a low-spin heme signal, accounting for about 100% of the enzyme concentration. The g-values (g, = 3.02, g, = 2.23 and g, = 1.45) and the crystal field parameters (tetragonality, AIL = 3.35 and rhombicity, V/A = 1.725) of the low-spin heme component are very similar to those of the bovine heart cytochrome aa3 and E. coli cytochrome bo [17, 181, suggesting a comparable bis-histidine axial ligation of heme a. The high-spin heme A exhibits a single spectrum, with g, = 6.03, g, = 5.97, g, = 2.0, and E/D< 0.002. The rhombicity of this signal is smaller than the rhombicity of the resonance from the P. denitrijiicans cytochrome aa3, with g, = 6.1 and g, = 5.9 (E/D = 0.004). In the membraneous as well as in the purified state the high-spin heme signal from the Sulfolohus enzyme shows a power saturation behaviour different from that of the heme a3 center from the P. denitrificans enzyme. Adopting an Orbach mechanism [24] for the spin- relaxation process, this feature reflects stronger zerofield interaction of the high-spin heme iron in the Sulfolobus enzyme as compared to the Paracoccus enzyme. The ligandbinding behaviour of cytochrome aa3 from Sulfolobus resembles that of cytochrome bo from E. coli [18], as demonstrated by the azide-binding properties of both enzymes. After addition of the ligand, the high-spin heme resonances of both enzymes are broadened, whereas no high-spin to low-spin
transition occurs, as detected in cytochrome c oxidases [17]. This may be a newly detected property characteristic for quinol oxidases as a whole. In accordance with the reactivity of azide with this archaebacterial terminal oxidase, the highspin heme species, which is sensitive to azide, is attributable to cytochrome a3, while the low-spin heme component is due to cytochrome a. The EPR spectrum of the extensively dialyzed enzyme, which essentially retained its activity and contains only one copper atom/protein molecule, does not show any copper EPR signal. This observation proves the absence of a center analogous to CuA of mammalian cytochrome 4u3, as has been also shown for the E. coli cytochrome bo [25] and the B. subtilis cytochrome aa3-600 [23]. This type of copper center in cytochrome c oxidases is a ligand of subunit I1 of the terminal oxidase complexes. Therefore it might have been missed in the present preparation consisting of only a single polypeptide species. In fact, the recently described operon coding for the Sulfolobus complex [26] shows that more than a single subunit is necessary to form the functional complex in vivo ; however, the subunit I1 equivalent also does not contain any detectable Cu- binding sites as concluded from the deduced amino acid sequence. This data implies that Sulfolobus cytochrome aa3 contains only a CUB-typecenter, which is EPR-silent either due to the diamagnetic 3d1' configuration in its monovalent Cu(1) state or due to exchange coupling with high-spin ferric ion in its divalent Cu(I1) state. The reduction potentials of the heme centers of the Sulfolobus enzyme were determined to be 305 & 10 mV for the low-spin heme and 200 & 10 mV and 370 & 10 mV for the high-spin heme, at pH = 7.4, using a simple model without considering heme - heme interacting potentials. These values are similar to those determined for the mammalian cytochrome au3 [17]. The data for the Sulfolobus enzyme can also be explained by a more complex model, including hemeheme interaction. The set of parameters that best fitted the experimental data points yielded an interacting potential between heme a and heme a3 of - 55 & 10 mV, indicative of negative cooperativity, and intrinsic reduction potentials for each heme of 195 & 10 mV (heme a3)and 255 & 10 mV (heme a). The interaction potential obtained is similar to that found for the E. coli cytochrome bo [25]. However, the intrinsic heme reduction potentials are different, reflecting either the different heme structures or the different reduction potentials of the natural reductants: menaquinone has Eo = 0 mV, while caldariella quinone has Eo = 100 mV [5]. The reduction potentials obtained through the EPR redox titration are different from those previously presented, obtained from a visible redox titration [4]. However it is not possible to directly compare the data obtained by these two different spectroscopic methods until a full set of thermodynamic parameters (intrinsic reduction potentials and interacting potentials between each metal center) describing the redox pattern of Sulfolobus cytochrome 4u3 is unequicocally determined [17]. Also, the experimental data available at the present moment do not allow a clear decision between the two redox models. Further studies are in progress to clarify this point which is of fundamental importance for the understanding of the enzyme function. The bell-shaped titration curve for the high-spin heme resonance of the Sulfolobus cytochrome oxidase indicates that the catalytic site is a strongly coupled binuclear center composed of heme a3 and CU,, similar to the corresponding centers in mammalian cytochrome aa3 [17] and cytochrome bo from E. coli [25]. The process with Eo = 370 & 10 mV is
138 assigned to the oxidation of the Cu, center, yielding a ferriheme a,-Cu(Ii) binuclear center with a net integer spin system and consequent loss of the high-spin heme EPR spectrum. Other evidence for the presence of a tightly coupled heme -copper pair is given by the results obtained with the extensively dialysed enzyme, containing only one copper atom/enzyme molecule, as the copper center involved in the binuclear center is not EPR-detectable under normal conditions. We note that under no experimental conditions were we able to detect an integer-spin spectrum from the spincoupled pair heme u3(S = 5/2)/Cu(II),(S = 1/2). We presume that a large zero-field splitting of the S,,,,, = 2 ground-state prohibits the otherwise typical 'g = 14' resonance [27]. i n conclusion, the data presented show that the metal centers in the cytochrome aa3 from Sulfolobus consist of a low-spin heme a, a high-spin heme a3 and a CuB-typecenter. The heme u3 and the Cu, form a binuclear center, similar to those observed in mammalian cytochrome aa3 and E. coli cytochrome bo. This implies that all physiological functions in terminal electron transport are unified in the single-subunit cytochrome uu3 from Sulfolobus. Furthermore, the absence of CuA seems to be characteristic for quinol oxidases as also shown for E. coli cytochrome bo [25] and B. subtilis cytochrome au3-600 [23], while its presence is a prerequisite for cytochrome c to replace quinol as an electron donor. Parts of these data were presented in poster form at the 5th International Conference on BioInorganic Chemistry, 4- 10 August 1991, Oxford, UK. The authors would like to thank Mrs Antje Lassen for skilful technical assistance, and Prof. A.V. Xavier for critical discussions. This work was supported in part by Junta Nacional de Investigap?o Cientijica e Tecnoldgica, Portugal, grant PC-CEN-649 (MT) and by a grant from the Deutsche Forschungsgemeinschaft, Federal Republic of Germany (S.A.).
REFERENCES 1. Rrock, T.D., Brock, K.M., Belly, R.T. & Weiss, R.L. (1972) Arch. Microhiol. 84, 54 - 68.
2. Moll, R. & Schafer, G. (1988) FEBSLett. 232, 359-363. 3. Liibben, M. & SchCfer, G. (1989) J . Bacteriol. 171,6101 -6116. 4. Anemiiller, S. & Schafer, G. (1989) FEBS Lett. 244, 451 -455. 5. Anemiiller, S. & Schafer, G. (1990) Eur. J. Biochem. 191, 297305. 6. Langworthy, T.A., Tornabene, T.G. & Holzer, G. (1982) Zentralbl. Bacteriol. Hyg. I Aht. Orig. C3, 228-244. 7. Hildebrandt, P., Heibel, G., Anemiiller, S. & Schafer, G. (1991) FEBS Lett. 283, 131 - 134. 8. Watters, C. (1978) Anal. Biochem. 88, 695-698. 9. Peterson, G.L. (1977) Anal. Biochem. 83, 346-356. 10. Kuboyama, M., Yong, F.C. & King, T.E. (1972) J. Biol. Chem. 247,6375-6383. 11. Bolard, J. &Gamier, A. (1972) Biochim. Biophys. Acta263, 535549. 12. Aasa, R. &Vanngard, T. (1975) J. Mugn. Res. 19,308-315. 13. Dutton, P.L. (1978) Methods Enzymol. 54, 43 1-435. 14. Aasa, R., Albracht, S.P.J., Falk, K.-E., Laane, B. & Vanngard, T. (1976) Biochim. Biophys. -4cta 422,260-272. 15. Slappendel, S., Veldink, G.A., Vliegenthard, J.F.G., Aasa, R. & Malmstrom, B.G. (1980) Biochim. Biophys. Acta 642, 30-39. 16. Coletta, M., Catarino, T., LeCall, J. & Xavier, A.V. (1991) Eur. J. Biochem. 202, 1101 -1106. 17. Wikstrom, M., Krab, K . & Saraste, M. (1981) Cytochrome oxidase: a synthesis, pp. 55 - 1 16, Academic Press, London. 18. Hata, A., Kirino, Y., Matsuura, K., Itoh, S., Hiyama, T., Konishi,K. & Anraku,Y. (1985) Biochim. Biophys. Acta 810, 62-72. 19. Puustinen, A. & Wikstrom, M. (1991) Proc. Nut1 Acad. Sci. USA 88,6122-6126. 20. Chepuri, V., Lemieux, L., Au, D.C.T. & Gennis, R.B. (1990) J . Biol. Chem. 265,11185 - 11 192. 21. Puustinen, A,, Finel, M., Virkki, M. & Wikstrom, M. (1989) FEBS Lett. 249, 163- 167. 22. Wikstrom, M. (1977) Nature 266,271 -273. 23. Lauraeus, M., Haltia, T., Saraste, M. & Wikstrom, M. (1991) Eur. J. Biochem. 197, 699 - 705. 24. Finn, C.B.P., Orbach, R. &Wolf, W.P. (1961) Proc. Phys. Soc. Lond. 77, 261 -268. 25. Salerno, J.C., Bolgiano, B., Poole, R.K., Gennis, R.B. & Ingledew, W.J. (1990) J. Biol. Chem. 265, 4364-4368. 26. Saraste, M., Holm, L., Lemieux, L., Liibben, M. & van der Oost (1991) Biochem. Soc. Trans. 19,608-612. 27. Hagen, W.R., (1982) Biochim. Biophys. Acta 708, 82-98.
