Binding of ligands originates small perturbations on the microscopic thermodynamic properties of a multicentre redox protein Carlos A. Salgueiro 1,2 , Leonor Morgado 1,2 , Bruno Fonseca 1,2 , Pedro Lamosa 1 , Teresa Catarino 1,2 , David L. Turner 3 and Ricardo O. Louro 1 1 Instituto de Tecnologia Quimica e Biolo ´ gica, Universidade Nova de Lisboa, Portugal 2 Departamento de Quimica da Faculdade de Cie ˆ ncias e Tecnologia da Universidade Nova de Lisboa, Portugal 3 School of Chemistry, University of Southampton, UK The structural aspects of protein complexes have received considerable attention and several experimen- tal and computational methods for the structural determination of complexes exist [1]. Redox proteins usually form transient complexes that can be studied using NMR methods, which, in addition to the struc- tural characterization, also provide information on the lifetime and dynamics of the bound forms [2,3]. Trans- fer of electrons between redox proteins at rates com- patible with metabolic processes requires the proper orientation of the partners for close approximation of the redox centres of the donor and acceptor, and that the reduction potentials ensure a favourable driving force, which is one of the main determinants of the rate of electron transfer [4]. Experimental measure- ments of the reduction potentials of proteins involved in complexes have been reported [5–7], but the effect of partner binding on the microscopic properties of the redox centres in proteins with multiple centres has not been addressed in detail yet. Cytochromes c 3 from sulfate-reducing bacteria are small soluble proteins containing four haems, and have been assigned a fundamental role in the bioenergetic metabolism of these organisms, mediating the flow of electrons from periplasmic hydrogenases to respiratory transmembrane electron transfer complexes coupled to the transfer of protons [8–11]. Several cytochromes c 3 have been isolated and characterized in great detail with respect to structure (for a recent revision of struc- tural work see [12]), equilibrium thermodynamic prop- erties [9,13–17] and transient kinetic properties [17–19]. These studies have shown that cytochromes c 3 have the required thermodynamic properties to perform a coordinated transfer of two electrons coupled to the transfer of protons in agreement with their proposed physiological role as partners of hydrogenase [8,20,21]. Keywords cytochrome c 3 ; electron transfer; NMR; protein docking; thermodynamic properties Correspondence R. O. Louro, Instituto de Tecnologia Quimica e Biolo ´ gica, Universidade Nova de Lisboa, Rua da Quinta Grande 6, 2780-156 Oeiras, Portugal Fax: 351-21-4428766 Tel: 351-21-4469848 E-mail: louro@itqb.unl.pt (Received 15 December 2004, revised 15 February 2005, accepted 7 March 2005) doi:10.1111/j.1742-4658.2005.04649.x NMR and visible spectroscopy coupled to redox measurements were used to determine the equilibrium thermodynamic properties of the four haems in cytochrome c 3 under conditions in which the protein was bound to lig- ands, the small anion phosphate and the protein rubredoxin with the iron in the active site replaced by zinc. Comparison of these results with data for the isolated cytochrome shows that binding of ligands causes only small changes in the reduction potentials of the haems and their pairwise inter- actions, and also that the redox-sensitive acid–base centre responsible for the redox–Bohr effect is essentially unaffected. Although neither of the lig- ands tested is a physiological partner of cytochrome c 3 , the small changes observed for the thermodynamic properties of cytochrome c 3 bound to these ligands vs. the unbound state, indicate that the thermodynamic prop- erties measured for the isolated protein are relevant for a physiological interpretation of the role of this cytochrome in the bioenergetic metabolism of Desulfovibrio. Abbreviations DvHc 3 , Desulfovibrio vulgaris (Hildenborough) cytochrome c 3 ; DvHc 3 :Pi, Desulfovibrio vulgaris cytochrome c 3 with phosphate; DvHc 3 :ZnRb, Desulfovibrio vulgaris cytochrome c 3 with zinc rubredoxin; EXSY, exchange spectroscopy. FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 2251 However, no experimental data exist on the effect of binding small ligands or proteins on these properties. For cytochromes c 3 these effects have only been inves- tigated in a theoretical study where cytochrome c 3 and a redox partner were docked in silico [22]. Experimental data reported in the literature argue in favour of a specific binding of phosphate to cyto- chrome c 3 [23] instead of a simple electrostatic effect of increased ionic strength at least up to 0.2 m concen- tration. The region of positively charged amino acid residues at the surface of the cytochrome surrounding haem IV, provides ample opportunity for binding a small anion such as phosphate, as was found for chro- mate for the homologous trihaem cytochrome c 7 [24]. Also, the analysis of one-dimensional NMR experi- ments showed that cytochrome c 3 and rubredoxin form a complex with a binding constant > 10 4 m )1 , and that the most downfield shifted signal in the NMR spec- trum of the ferricytochrome displays the most obvious modification upon binding [25]. This signal was assigned to the methyl 18 2 of haem IV [26] (methyl nomenclature according to IUPAC-IUB recommenda- tions [27] and Roman numerals designate the order of attachment of the haem to the polypeptide chain), which confirms the extensive work of molecular dock- ing models for cytochrome c 3 with physiological and nonphysiological protein partners [6,22,28–30] showing always the positively charged region around haem IV as the most favoured docking site. The complex between cytochrome c 3 and rubredoxin is not physiological because the two proteins are located in different cellular compartments, but it pro- vides a convenient model for studying the effect of partner binding on the thermodynamic properties of the haems of cytochrome c 3 . Because the rubredoxin is a very acidic protein and binds to the cytochrome close to haem IV, it has the electrostatic characteristics that mimic the physiological partners such as the Fe-hydrogenase and the membrane associated multi- haem cytochromes [6,22,31,32]. This work reports the first determination of the equi- librium thermodynamic properties of a cytochrome c 3 when bound to phosphate and to an engineered form of rubredoxin where the iron was replaced by zinc. Results Figure 1 shows the comparison of representative two- dimensional exchange spectroscopy (EXSY) NMR spectra of Desulfovibrio vulgaris cytochrome c 3 with phosphate (DvHc 3 :Pi) and Desulfovibrio vulgaris cyto- chrome c 3 with zinc rubredoxin (DvHc 3 :ZnRb). It is Fig. 1. Two-dimensional EXSY NMR spectra of DvHc 3 :Pi (above the diagonal) and DvHc 3 :ZnRb (below the diagonal) at pH 7.6 showing the pattern of reoxidation in both cases. The spectrum for DvHc 3 :Pi is slightly more oxidized and therefore does not have signals for stage 1. The lines connect sig- nals of one particular methyl group (2 1 CH I 3 , 18 1 CH II 3 ,12 1 CH III 3 or 18 1 CH IV 3 ) in different oxidation stages for DvHc 3 :Pi (solid lines) and DvHc 3 :ZnRb (dashed lines). Some signals are not easily visible at the level of cut-off used to prepare the figure and were boxed for clarity. Roman and Arabic num- bers indicate the haem groups and the oxidation stages, respectively. Thermodynamic parameters in ligated proteins C. A. Salgueiro et al. 2252 FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS apparent that the spectra are very similar with respect to chemical shifts of the signals in intermediate stages of oxidation, and that formation of the complex does not lead to a marked decrease of the spectral quality in the experimental conditions used, where most of the cyto- chrome is bound to the Zn-rubredoxin (Discussion). The pH dependence of the paramagnetic chemical shifts of each haem methyl group and the data obtained for redox titrations followed by visible spectroscopy at pH 7.0 and 8.1, were used to monitor the thermo- dynamic properties of DvHc 3 :Pi. The fittings of both NMR and visible spectroscopy data are reported in Figs 2 and 3, respectively. The thermodynamic parame- ters obtained for DvHc 3 :Pi are listed in Table 1, together with the macroscopic pK a values for the five stages of oxidation. The pH dependence of the chemical shifts of the haem methyl groups 2 1 CH I 3 ,18 1 CH II 3 ,12 1 CH III 3 and 18 1 CH IV 3 both Desulfovibrio vulgaris cytochrome c 3 (DvHc 3 ) and DvHc 3 :Pi are reported in Fig. 2 by dashed and solid lines, respectively. Figure 2 shows that the major differences in chemical shifts of the sig- nals relative to the data obtained in the absence of phosphate occur for the intermediate oxidation stages of haems III and IV. However, these differences are small and give rise to only a small modification on the calculated thermodynamic properties of DvHc 3 :Pi as indicated in Table 1 with all differences < 12 meV. Also, Fig. 2 and Table 1 both show that the acid–base centre and the redox–Bohr interactions are almost undisturbed by the presence of phosphate and the resulting macroscopic pK a values are within 0.2 units Fig. 2. The pH dependence of the chemical shift of haem methyl resonances 2 1 CH I 3 ,18 1 CH II 3 ,12 1 CH III 3 and 18 1 CH IV 3 ,ofDvHc 3 :Pi at 297.3 K. Squares correspond to stage 1 of oxidation, circles to stage 2, downward pointing triangles to stage 3, and upward pointing trian- gles to stage 4. The chemical shifts of the haem methyl groups in the fully reduced stage 0 are not plotted because they are unaffected by the pH. The solid lines represent the best fit of the shifts for DvHc 3 :Pi to the model of five interacting centres using the parameters listed in Table 1. Dashed lines represent the best fit for the DvHc 3 and the nearest label (1–4) indicates the oxidation stage represented by the line. Fig. 3. Reduced fraction of DvHc 3 in the presence of 100 mM phos- phate determined from redox titrations followed by visible spectros- copy performed at pH 7.0 and 8.1. Continuous lines are the fit of the model to the data. C. A. Salgueiro et al. Thermodynamic parameters in ligated proteins FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 2253 of those measured for the isolated cytochrome. As pre- viously reported for cytochromes c 3 from Desulfovibrio gigas, Desulfomicrobium norvegicum and Desulfomicro- bium baculatum [23], the presence of phosphate induced a generalized narrowing of the line widths of the DvHc 3 haem methyl signals at intermediate redox stages when compared with the experiments performed in the absence of phosphate (data not shown). These obser- vations show that the intermolecular electron exchange is slower, which allowed the data from DvHc 3 :Pi to be collected in a NMR spectrometer operating at 300 MHz, and establishing that the intermolecular elec- tron exchange is < 340 s )1 at 1 mm and 297.3 K. The paramagnetic chemical shifts of each haem methyl group (2 1 CH I 3 ,18 1 CH II 3 ,12 1 CH III 3 and 18 1 CH IV 3 ), of DvHc 3 :ZnRb are plotted in Fig. 4 and the relative thermodynamic parameters together with the macroscopic pK a values for the five stages of oxi- dation determined from the fitting are listed in Table 2. Absolute potentials and interactions are not reported for these experiments because it is not poss- ible to perform redox titrations followed by visible spectroscopy under conditions that ensure a similar proportion of bound vs. unbound state of the cyto- chrome to those obtained in the NMR tube. This is a consequence of the very strong absorption bands of the cytochrome requiring very dilute solutions to per- form visible absorption measurements, and the possi- bility of interference from the redox mediators on complex formation. Therefore, haem I and the inter- action between haems I and IV were chosen as refer- ences because these are the most distant pair of haems in the structure and are therefore expected to have the weakest interaction [33]. The pH dependence of the chemical shifts of the NMR signals of the haem methyls obtained for DvHc 3 :ZnRb and DvHc 3 is also reported in Fig. 4 and indicated by continuous and dashed lines, respectively. The figure shows that the effect of binding Zn-rubre- doxin on the NMR signals of the haem methyls vs. the results obtained for the isolated cytochrome is very small. As observed for the case of phosphate binding, the signals for intermediate redox stages 2 and 3 of haems III and IV are the more affected. This suggests that the binding of phosphate and Zn-rubredoxin occur in a similar location on the surface of the cytochrome, which is in agreement with the fact that both are neg- atively charged molecules despite the dramatic differ- ence in size, and in agreement with previous comparative work of binding inorganic and protein partners to cytochromes c 3 [6]. Table 2 shows that as for the case of phosphate binding, the association with Table 1. Thermodynamic parameters determined for DvHc 3 [13] and DvHc 3 :Pi. (Top) Diagonal terms (in bold) represent the oxidation ener- gies of the four haems and the energies for deprotonating the acid–base centre for the fully reduced and protonated state of the protein and have standard errors < 5 meV. The off-diagonal elements represent the redox- and redox–Bohr interactions between the centres. All para- meters are reported in units of meV, making them numerically equal to the values of redox potentials and interactions reported in units of mV [DG (meV) ¼ nE (mV)]. (Bottom) Macroscopic pK a values for the five stages of oxidation, from the fully reduced protein (stage 0) to the fully oxidized (stage 4) measured from the data. Haem I Haem II Haem III Haem IV Ionizable centre DvHc 3 Haem I ) 245 ) 43 20 ) 4 ) 70 Haem II – 267 ) 88) 30 Haem III ) 334 32 ) 18 Haem IV – 284 –6 Ionizable centre 439 DvHc 3 :Pi Haem I – 247 ) 38 26 6 ) 67 Haem II – 275 316) 25 Haem III – 335 35 ) 15 Haem IV – 293 –6 Ionizable centre 428 Macroscopic pK a values Oxidation stage 01234 DvHc 3 7.4 7.1 6.4 5.6 5.3 DvHc 3 :Pi 7.2 6.9 6.3 5.6 5.3 Thermodynamic parameters in ligated proteins C. A. Salgueiro et al. 2254 FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS Table 2. Thermodynamic parameters determined for DvHc 3 [13] and DvHc 3 :ZnRb. (Top) Relative thermodynamic parameters. (Bottom) Mac- roscopic pK a values for the five stages of oxidation, from the fully reduced protein (stage 0) to the fully oxidized (stage 4). The table was pre- pared as Table 1 using the energy of oxidation of haem I and the interaction between haems I and IV as reference values. Haem I Haem II Haem III Haem IV Ionizable centre DvHc 3 Haem I 0 ) 47 24 0 ) 70 Haem II ) 22 ) 412) 30 Haem III ) 89 36 ) 18 Haem IV ) 39 ) 6 Ionizable centre 439 DvHc 3 :ZnRb Haem I 0 ) 34 18 0 ) 69 Haem II ) 30 27) 29 Haem III ) 93 37 ) 13 Haem IV ) 48 2 Ionizable centre 422 Macroscopic pK a values Oxidation stage 01234 DvHc 3 7.4 7.1 6.4 5.6 5.3 DvHc 3 :ZnRb 7.2 6.9 6.2 5.5 5.3 Fig. 4. The pH dependence of the chemical shift of haem methyl resonances 2 1 CH I 3 ,18 1 CH II 3 ,12 1 CH III 3 and 18 1 CH IV 3 ,ofDvHc 3 :ZnRb at 297.3 K. Squares correspond to stage 1 of oxidation, circles to stage 2, downward pointing triangles to stage 3, and upward pointing trian- gles to stage 4. The chemical shifts of the haem methyl groups in the fully reduced stage 0 are not plotted since they are unaffected by the pH. The solid lines represent the best fit of the shifts for DvHc 3 :ZnRb to the model of five interacting centres using the parameters listed in Table 2. Dashed lines represent the best fit for the DvHc 3 and the nearest label (1–4) indicates the oxidation stage represented by the line. C. A. Salgueiro et al. Thermodynamic parameters in ligated proteins FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 2255 Zn-rubredoxin gives rise to a small perturbation of the relative reduction potentials and redox interactions among the various centres. Furthermore, because the pH of the samples could be measured inside the NMR tube the values for the redox–Bohr interactions and the macroscopic pK a values are absolute, and therefore the macroscopic pK a values of the various redox stages show only very small modifications relative to the data obtained for the isolated cytochrome [13]. This obser- vation is in agreement with the experimentally observed binding of rubredoxin close to haem IV because the acid–base centre has been assigned to propionate D of haem I [26] which is on the opposite pole of the cytochrome and therefore should be only weakly affected by the docking. Discussion Our results demonstrate that at 100 mm phosphate binds to DvHc 3 causing narrower NMR signals and perturbing the chemical shifts of the haem methyl sig- nals in intermediate redox stages. The contraction of the line widths of the NMR signals in intermediate redox stages shows that the intermolecular electron exchange is slower than in the absence of phosphate. Given that DvHc 3 is a very basic protein, with its iso- electric point above 10, this result is contrary to the expected increase in intermolecular electron exchange rate for proteins of equal charge as the ionic strength is increased, and indicates a specific binding of phos- phate to the cytochrome [34,35]. Moreover, an increase in reduction potentials of the centres with ionic strength would be expected on electrostatic grounds for a negatively charged protein [36]. This is not observed in the current case and was also not observed for some haems in the highly homologous cyto- chrome c 3 from D. vulgaris (Miyazaki) in the presence of increased phosphate concentration [37]. The results presented in Table 1 show that the interactions among the centres in the protein are subject to small modifica- tions in the presence of phosphate, in agreement with arguments in the literature that for nonsurface resi- dues, such as the haems, the presence of counter ions should have a small effect on pair-wise charge–charge interactions between redox centres in a protein [33]. To measure the detailed equilibrium thermodynamic parameters of the redox and redox–Bohr interactions of the haems of cytochrome c 3 when forming a com- plex with a partner protein, several experimental requirements had to be met in addition to maintaining slow intermolecular- and fast intramolecular-electron transfer among the cytochrome c 3 molecules: (a) The complex had to be sufficiently small so that line broad- ening from a slower tumbling complex would not pre- vent observation of the signals at the various redox stages; (b) Bound and unbound states had to be in fast exchange in the NMR time scale so that a single signal is observed at a position that is weighted by the relat- ive proportions of these states; (c) The partner should not contain a paramagnetic centre to avoid excessive broadening of the lines, as observed for the complex between the native iron rubredoxin and cytochrome c 3 [25]; (d) The redox state of the partner should not change under the various experimental conditions probed to avoid distorting the results of the param- eters measured for the cytochrome with varying elec- trostatic interactions caused by diferent redox states of the partner. The use of Zn-rubredoxin as docking partner ful- filled all these criteria: (i) the complex has a combined mass of  20 kDa; (ii) the number of signals observed in the intermediate stages of oxidation shows that the exchange between the bound and unbound form is fast; (iii) Zn(II) is diamagnetic; and (iv) Zn(II) has a d10 electronic configuration and therefore does not present redox chemistry in the range explored in this work. The binding constant of rubredoxin to DvHc 3 is >10 4 m )1 [25], which is assumed to be essentially undisturbed by the replacement of Fe by Zn in the rubredoxin given the identical structures of the two protein forms [38,39]. Therefore, at the concentrations used in the NMR experiments over 90% of the cytochrome is bound to the rubredoxin and the effect observed in the signals of the haem methyls of DvHc 3 :ZnRb complex vs. the results in the isolated cytochrome is close to the full effect of complex forma- tion. Because the thermodynamic parameters for the haems calculated from the NMR data are relative, it could be argued that all haem potentials had been modified to a similar degree but to an unknown extent making their absolute values completely different from those measured for the isolated cytochrome. However, this scenario is unlikely because the rubredoxin is a smaller protein than the cytochrome, it binds to a spe- cific location close to haem IV, and the phenomena giving rise to such a widespread modification of the redox properties should also affect the acid–base centre for which absolute thermodynamic parameters were measured and that shows very small modifications caused by the binding of Zn-rubredoxin. Overall, the two sets of results reported show that phosphate binding and docking with the Zn-rubredox- in has a very limited effect on the redox properties of the haem groups in the cytochrome. In fact, the haem reduction potentials, redox interactions, redox– Bohr interactions and macroscopic pK a values remain Thermodynamic parameters in ligated proteins C. A. Salgueiro et al. 2256 FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS essentially unaffected (Tables 1 and 2). These results are in agreement with theoretical expectations that sur- face charges interact with redox centres with a very high effective dielectric constant and therefore their contri- bution to the reduction potential of the centres is small [40,41], and also with the experimental observation for monohaem cytochromes that the effect of anion binding or complex formation on the reduction potential is small [7,42]. However, this is the first time that the net- work of pairwise interactions in a multicentre protein has been explored when forming a complex, and these interactions also show small modifications relative to the isolated protein, indicating that the intramolecular dielectric environment is essentially undisturbed by lig- and binding or complex formation. Conclusions The presence of small negatively charged ligands such as phosphate, causes little perturbation on the equilib- rium thermodynamic properties of the haems in DvHc 3 , despite the evidence that electrostatic forces are the main drive for complex formation, and the haems are very exposed to the solvent. Also, mimics of the physiological partners such as the Zn-rubredoxin cause only small modifications in the relative reduction potentials and redox interactions among the haems in DvHc 3 . Our results show that the equilibrium thermo- dynamic data obtained for the isolated cytochrome c 3 are similar to those measured when the protein is bound to a small anion or to a mimic of physiological partners and may be discussed within the framework of their functional relevance for the role of cyto- chrome c 3 in maintaining efficiency in the bioenergetic metabolism of Desulfovibrio bacteria. Experimental procedures Bacterial growth and protein purification Zn-rubredoxin Cells of Escherichia coli BL21(DE; Universidade Nova de Lisboa, Portugal) were transformed with the plasmid pMSPL1 [43] to produce the rubredoxin from Desulfovibrio gigas. Twenty millilitres of an overnight culture were ino- culated in 1 L of the medium described in the literature [43] that was allowed to grow up to an absorbance of 0.5. The cells were induced with isopropyl thio-b-d-galactoside (25 mgÆL )1 ) and supplemented with 4 mLÆL )1 of glycerol 87% and ZnCl 2 (5 mgÆL )1 ; final concentration) to increase the amount of the zinc form of rubredoxin present in the cell cultures. After 6–10 h the cells were centrifuged and collected in 50 mm Tris with 1 mm phenylmethanesulfonyl fluoride and the Zn-rubredoxin purified as described in [44]. Purity was checked by SDS ⁄ PAGE and UV-visible spectroscopy. Cytochrome c 3 Cells of D. vulgaris (Hildenborough; Universidade Nova de Lisboa) were grown and the tetrahaem cytochrome c 3 was purified as previously described [13]. Redox titrations followed by visible spectroscopy Anaerobic redox titrations followed by visible spectroscopy were performed as described previously [45] with  2 lm pro- tein solutions in 100 mm phosphate buffer at pH 7.0 and 8.1. For each pH value the redox titrations were repeated at least twice, both in the oxidative and reductive directions to check for hysteresis. Reproducibility between the runs was typically better than 5 mV. To ensure a good equilibrium between the redox centres and the working electrode [46], a mixture of the following redox mediators was added to the protein solution at pH 7.0: indigo tetrasulfonate, indigo trisulfonate, indigo disulfonate, anthraquinone-2-7-disulfonate, anthraquinone- 2-sulfonate, safranine O, diquat, neutral red, phenosafranine, and methylviologen, all at a ratio of 100 : 8 of cytochrome vs. mediator to avoid interference caused by specific binding of mediators to the protein. For the redox titrations per- formed at pH 8.1 the mediators gallocyanine and methylene blue were added to the previous mixture. The solution potential was measured using a combined Pt|Ag ⁄ AgCl electrode (Crison, Barcelona, Spain), calibra- ted against saturated quinhydrone ⁄ hydroquinone solutions at pH 4 and 7, and the visible spectra were recorded at 297 ± 1 K in a Shimadzu UV-1203 spectrophotometer, placed inside an anaerobic glove box (Mbraun MB 150 I). The reduced fraction of the cytochrome c 3 from D. vulgaris (Hildenborough) (DvHc 3 ) was determined using the a band peak at 552 nm. The optical contribution of the mediators was subtracted by measuring the height of the peak at 552 nm relative to the straight line connecting the two isosbestic points (542 and 560 nm) flanking the a band according to the method described in the literature [45]. NMR sample preparation DvHc 3 in the presence of 100 mm phosphate The protein was lyophilized twice with 2 H 2 O (99.96% atom) and then dissolved in  500 lL 2 H 2 O (99.96% atom) 100 mm K 3 PO 4 .7H 2 O solution to a final concentration of 1mm (this sample will be referred hereafter as DvHc 3 :Pi). Identical NMR spectra of the DvHc 3 (data not shown) were obtained before and after the lyophilization, showing that the protein structure was not affected. The pH of the samples was adjusted using small volumes of NaO 2 Hor 2 HCl solutions. In the reduced and intermediate C. A. Salgueiro et al. Thermodynamic parameters in ligated proteins FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 2257 stages of oxidation the pH was adjusted inside the an- aerobic glove box with argon circulation to avoid the reoxidation of the sample. The pH values reported are direct meter readings without correction for the isotope effect [47,48]. Complete reduction of the sample was achieved by the reaction with gaseous hydrogen in the presence of catalytic amounts of the enzyme Fe-hydroge- nase isolated from D. vulgaris (Hildenborough). Partially oxidized samples were obtained by first flushing out the hydrogen from the reduced sample with argon and then adding controlled amounts of air in the microlitre range into the NMR tube with a syringe through serum caps. DvHc 3 with Zn-rubredoxin The cytochrome and the Zn-rubredoxin were lyophilized separately and a sample was prepared in  500 lL 2 H 2 O (99.96% atom) at a final concentration of 0.5 mm cyto- chrome and 0.8 mm Zn-rubredoxin (this sample will be referred hereafter as DvHc 3 :ZnRb). The sample was mani- pulated as described above for the sample containing 100 mm phosphate. NMR spectroscopy of partially oxidized samples 1 H NMR spectra were obtained either in a Bruker AMX300 for DvHc 3 :Pi or in a Bruker DRX500 Avance spectrometer for DvHc 3 :ZnRb equipped with 5 mm inverse detection probe heads. To establish the complete pattern of oxidation for each haem methyl group at each pH, several two-dimensional EXSY NMR experiments, with 25 ms mixing time, were collected at various degrees of oxidation. The spectra were recorded at 297.3 K in the pH range 5.0–8.2, measuring 4 k (t 2 ) · 1k (t 1 ) data points, and water presaturation was achieved by selective, low-power pulses of 500–800 ms. Chemical shifts values are reported in p.p.m. relative to trimethylsilyl and the spectra were calibrated using the residual water signal as internal reference [49]. Modelling the thermodynamic parameters The model used for the thermodynamic characterization of DvHc 3 [13] was applied to the data obtained for DvHc 3 :Pi and Dv Hc 3 :ZnRb. This model considers five interacting cen- tres: four haems and one acid–base centre. As shown in the Results, DvHc 3 exhibits fast intramolecular and slow inter- molecular electron exchange on the NMR time scale, in the presence of both Zn-rubredoxin and 100 mm phosphate. Therefore, each haem substituent displays five discrete NMR signals corresponding to each of the five possible macro- scopic oxidation stages of the cytochrome, connected by four steps of one-electron uptake or release. The unpaired electron in the oxidized haem causes a paramagnetic shift on its signals that is directly proportional to the fractional oxida- tion in the absence of extrinsic paramagnetic contributions. As shown previously [26], the haem methyl groups 2 1 CH I 3 , 18 1 CH II 3 ,12 1 CH III 3 and 18 1 CH IV 3 have large paramagnetic shifts and negligible extrinsic dipolar contributions, hence they are suitable for monitoring the thermodynamic proper- ties of the cytochrome. The paramagnetic chemical shift of the haem methyl resonances is a very sensitive probe of the haem environment. The fact that the shifts in the fully oxi- dized state of the protein are essentially identical to those measured for the isolated protein indicates that the binding of phosphate or the Zn-rubredoxin does not disturb the structure of the haem core. Therefore, the same methyl groups were followed here. The NMR data provide only relative values for the reduction potentials and interactions [13,50,51] and deter- mination of absolute values require the use of data from redox titrations followed by visible spectroscopy. A compu- ter program was written to fit the thermodynamic model to the NMR data (for the case of DvHc 3 :ZnRb), or to the NMR and UV-visible data sets simultaneously (for the case of DvHc 3 :Pi) using the Marquardt method for parameter optimization. The half-height widths of the NMR signals were used as a measure of the uncertainty of each NMR data point and an experimental uncertainty of 2% was assumed for the experimental points of the redox titrations. Acknowledgements The authors are grateful to Professor A.V. Xavier, for many fruitful suggestions and discussions, and to Professor Helena Santos for kindly providing the Zn-rubredoxin. The assistance of Isabel Pacheco dur- ing protein purification is gratefully acknowledged. Financial support was provided by FCT-POCTI, Co- financed by FEDER (POCTI ⁄ 42902 ⁄ QUI ⁄ 2001 to CAS, POCTI ⁄ 43435 ⁄ QUI ⁄ 2001 to TC, and BPD ⁄ 11511 ⁄ 2002 to PL). References 1 Russell RB, Alber F, Aloy P, Davies FP, Korkin D, Pichaud M, Topf M & Sali A (2004) A structural per- spective on protein–protein interactions. Curr Opin Struct Biol 14, 313–324. 2 Prudeˆ ncio M & Ubbink M (2004) Transient complexes of redox proteins: structural and dynamic details from NMR studies. J Mol Recogn 17, 524–539. 3 Hansen DF, Hass MAS, Christensen HM, Ulstrup J & Led JJ (2003) Detection of short-lived transient protein– protein interactions by intermolecular nuclear paramag- netic relaxation: plastocyanin from Anabaena variabilis. J Am Chem Soc 125, 6858–6859. Thermodynamic parameters in ligated proteins C. A. Salgueiro et al. 2258 FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 4 Marcus RA & Sutin N (1985) Electron transfers in chemistry and biology. Biochim Biophys Acta 811, 265– 322. 5 Drepper F, Hippler M, Nitschke W & Haehnel W (1996) Binding dynamics and electron transfer between plastocyanin and photosystem I. Biochemistry 35, 1282– 1295. 6 Mus-Veteau I, Chottard G, Lexa D, Guerlesquin F & Bruschi M (1992) Cytochrome c 3 –heteropolytungstate complex: a model for the interaction of the tetraheme cytochrome with its redox partners, ferredoxin and rubredoxin. Biochim Biophys Acta 1102, 353–359. 7 Burrows AL, Guo LH, Hill AO, McLendon G & Sher- man F (1991) Direct electrochemistry of proteins: inves- tigations of yeast cytochrome c mutants and their complex with cytochrome b 5 . Eur J Biochem 202, 543– 549. 8 Louro RO, Catarino T, LeGall J & Xavier AV (1997) Redox–Bohr effect in electron ⁄ proton energy transduc- tion: cytochrome c 3 coupled to hydrogenase works as a ‘proton thruster’ in Desulfovibrio vulgaris. J Biol Inorg Chem 2, 488–491. 9 Louro RO (1998) Proton-thrusters: Energy transduction performed by tetrahaem cytochromes c 3 and its physiolo- gical relevance. PhD Thesis, Universidade Nova de Lisboa, Portugal. 10 Pereira IAC, Roma ˜ o CV, Xavier AV, LeGall J & Teix- eira M (1998) Electron transfer between hydrogenases and mono- and multiheme cytochromes in Desulfovibrio sp. J Biol Inorg Chem 3, 494–498. 11 Valente FM, Saraiva LM, LeGall J, Xavier AV, Teixe- ira M & Pereira IA (2001) A membrane-bound cytochrome c 3 : a type II cytochrome c 3 from Desulfovi- brio vulgaris Hildenborough. Chembiochem 2, 895–905. 12 Araga ˜ o D, Fraza ˜ o C, Sieker L, Sheldrick GM, LeGall J & Carrondo MA (2003) Structure of dimeric cyto- chrome c 3 from Desulfovibrio gigas at 1.2 A ˚ resolution. Acta Crystallogr D Biol Crystallogr 59, 644–653. 13 Turner DL, Salgueiro CA, Catarino T, LeGall J & Xavier AV (1996) NMR studies of cooperativity in the tetrahaem cytochrome c 3 from Desulfovibrio vulgaris . Eur J Biochem 241, 723–731. 14 Park JS, Ohmura T, Kano K, Sagara T, Niki K, Kyogoku Y & Akutsu H (1996) Regulation of the redox order of four hemes by pH in cytochrome c 3 from D. vulgaris Miyazaki F. Biochim Biophys Acta 1293, 45–54. 15 Salgueiro CA, Turner DL, LeGall J & Xavier AV (1997) Reevaluation of the redox and redox–Bohr coop- erativity in tetrahaem Desulfovibrio vulgaris (Miyazaki F) cytochrome c 3 . J Biol Inorg Chem 2, 343–349. 16 Louro RO, Bento I, Matias PM, Catarino T, Baptista AM, Soares CM, Carrondo MA, Turner DL & Xavier AV (2001) Conformational component in the coupled transfer of multiple electrons and protons in a monomeric tetraheme cytochrome. J Biol Chem 276, 44044–44051. 17 Correia IJ, Paquete CM, Coelho A, Almeida CC, Catarino T, Louro RO, Fraza ˜ o C, Saraiva LM, Carrondo MA, Turner DL et al. (2004) Proton-assisted two-electron transfer in natural variants of tetraheme cytochromes from Desulfomicrobium sp. J Biol Chem 279, 52227–52237. 18 Catarino T & Turner DL (2001) Thermodynamic con- trol of electron transfer rates in multicenter redox pro- teins. Chembiochem 2, 416–424. 19 Correia IJ, Paquete CM, Louro RO, Catarino T, Turner DL & Xavier AV (2002) Thermodynamic and kinetic characterization of trihaem cytochrome c 3 from Desulfuromonas acetoxidans. Eur J Biochem 269, 5722– 5730. 20 Yagi T, Honya M & Tamiya N (1968) Purification and properties of hydrogenases of different origins. Biochim Biophys Acta 153, 699–705. 21 Aubert C, Brugna M, Dolla A, Bruschi M & Giudici- Orticoni MT (2000) A sequential electron transfer from hydrogenases to cytochromes in sulfate-reducing bac- teria. Biochim Biophys Acta 1476, 85–92. 22 Teixeira VH, Baptista AM & Soares CM (2004) Model- ing electron transfer thermodynamics in protein com- plexes: interaction between two cytochromes c 3 . Biophys J 86, 2773–2785. 23 Moura JJG, Xavier AV, Cookson DJ, Moore GR & Williams RJ (1977) Redox states of cytochrome c 3 in the absence and presence of ferredoxin. FEBS Lett 81, 275–280. 24 Assfalg M, Bertini I, Bruschi M, Michel C & Turano P (2002) The metal reductase activity of some multiheme cytochromes c: NMR structural characterization of the reduction of chromium (VI) to chromium (III) by cytochrome c (7). Proc Natl Acad Sci 99, 9750–9754. 25 Stewart DE, LeGall J, Moura I, Moura JJG, Peck HD Jr, Xavier AV, Weiner PK & Wampler JE (1989) Electron transport in sulfate-reducing bacteria. Mole- cular modeling and NMR studies of the rubredoxin– tetraheme–cytochrome c 3 complex. Eur J Biochem 185, 695–700. 26 Salgueiro CA, Turner DL & Xavier AV (1997) Use of paramagnetic NMR probes for structural analysis in cytochrome c 3 from Desulfovibrio vulgaris . Eur J Biochem 244, 721–734. 27 Moss GP (1988) IUPAC–IUB joint commission on biochemical nomenclature. Nomenclature of tetrapyr- roles. Tentative rules (1969). Eur J Biochem 178, 277–328. 28 Stewart DE, LeGall J, Moura I, Moura JJG, Peck HD, Xavier AV, Weiner PK & Wampler JE (1988) A hypothetical model of the flavodoxin–tetraheme cyto- chrome c 3 complex of sulfate-reducing bacteria. Biochemistry 27, 2444–2450. 29 Morais J, Palma PN, Fraza ˜ o C, Caldeira J, LeGall J, Moura I, Moura JJ & Carrondo M-A (1995) Structure of the tetraheme cytochrome from Desulfovibrio desul- C. A. Salgueiro et al. Thermodynamic parameters in ligated proteins FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS 2259 furicans ATCC 27774: X-ray diffraction and electron paramagnetic resonance studies. Biochemistry 34, 12830–12841. 30 El Antak L, Morelli X, Bornet O, Hatchikian C, Czjzek M, Dolla A & Guerlesquin F (2003) The cytochrome c 3 – [Fe]-hydrogenase electron-transfer complex: struc- tural model by NMR restrained docking. FEBS Lett 548, 1–4. 31 Matias PM, Saraiva LM, Soares CM, Coelho AV, LeGall J & Carrondo MA (1999) Nine-haem cyto- chrome c from Desulfovibrio desulfuricans ATCC 27774: primary sequence determination, crystallographic refinement at 1.8 and modelling studies of its interaction with the tetrahaem cytochrome c 3 . J Biol Inorg Chem 4, 478–494. 32 Matias PM, Coelho AV, Valente FM, Placido D, Le- Gall J, Xavier AV, Pereira IA & Carrondo MA (2002) Sulfate respiration in Desulfovibrio vulgaris Hildenbor- ough. structure of the 16-heme cytochrome c HmcA AT 2.5-A resolution and a view of its role in transmem- brane electron transfer. J Biol Chem 277, 47907–47916. 33 Louro RO, Catarino T, Paquete CM & Turner DL (2004) Distance dependence of interactions between charged centres in proteins with common structural fea- tures. FEBS Lett 576, 77–80. 34 Santos H (1984) Relac¸a ˜ o estrutura-func¸a ˜ o em citocromos multihe ´ micos: caracterizac¸a ˜ o por RMN dos mecanismos de transfere ˆncia electro ´ nica em citocromos c 3 de bacte ´ rias reductoras de sulfato. PhD Thesis, Universidade Nova Lisboa, Portugal. 35 Cudd A & Fridovich I (1982) Electrostatic interactions in the reaction mechanism of bovine erythrocyte super- oxide dismutase. J Biol Chem 257, 11443–11447. 36 Reid LS, Mauk MR & Mauk AG (1984) Role of heme propionate groups in cytochrome b 5 electron transfer. J Am Chem Soc 106, 2182–2185. 37 Ohmura T, Nakamura H, Niki K, Cusanovich MA & Akutsu H (1998) Ionic strength-dependent physicochem- ical factors in cytochrome c 3 regulating the electron transfer rate. Biophys J 5, 1483–1490. 38 Dauter Z, Wilson KS, Sieker LC, Moulis JM & Meyer J (1996) Zinc- and iron-rubredoxins from Clostridium pasteurianum at atomic resolution: a high-precision model of a ZnS 4 coordination unit in a protein. Proc Natl Acad Sci 93, 8836–8840. 39 Lamosa P, Brennan L, Vis H, Turner D & Santos H (2001) NMR Structure of Desulfovibrio gigas rubredoxin – A model for studying protein stabilisation by compat- ible solutes. Extremophiles 5, 301–311. 40 Cutler RL, Davies AM, Creighton S, Warshel A, Moore GR, Smith M & Mauk AG (1989) Role of arginine-38 in regulation of the cytochrome c oxidation–reduction equilibrium. Biochemistry 28, 3188–3196. 41 Zhou H-X (1994) Effects of mutations and complex for- mation on the reduction potentials of cytochrome c and cytochrome c peroxidase. J Am Chem Soc 116, 10362– 10375. 42 Battistuzzi G, Borsari M, Dallari D, Lancellotti I & Sola M (1996) Anion binding to mitochondrial cytochromes c studied through electrochemistry. Effects of the neutralization of surface charges on the redox potential. Eur J Biochem 241, 208–214. 43 Lamosa P, Turner D & Santos H (2003) Protein stabilization by compatible solutes: effect of diglycerol phosphate in the backbone dynamics of Desulfovibrio gigas rubredoxin. Eur J Biochem 270, 4606–4614. 44 Lamosa P, Burke A, Peist R, Huber R, Liu M-Y, Silva G, Rodrigues-Pousada C, LeGall J, Maycock C & San- tos H (2000) Thermostabilisation of proteins by digly- cerol phosphate. A new compatible solute from the hyperthermophile Archaeoglobus fulgidus. Appl Environ Microbiol 66, 1974–1979. 45 Louro RO, Catarino T, LeGall J & Xavier AV (2001) Cooperativity between electrons and protons in a mono- meric cytochrome c 3 : the importance of mechano-chemi- cal coupling for energy transduction. Chembiochem 2, 831–837. 46 Dutton PL (1978) Redox potentiometry: determination of midpoint potentials of oxidation–reduction compo- nents of biological electron-transfer systems. Methods Emzymol 54, 411–435. 47 Glasoe PK & Long FA (1960) Use of glass electrodes to measure acidities in deuterium oxide. J Phys Chem 64, 188–191. 48 Delgado R, Frausto da Silva JJR, Amorim MTS, Cabral MF, Chaves S & Costa J (1991) Dissociation constants of Bronsted acids in D 2 O and H 2 O: studies on polyaza and polyoxa–polyaza macrocycles and a general correlation. Anal Chim Acta 245, 271–282. 49 Pierattelli R, Banci L & Turner DL (1996) Indirect determination of magnetic susceptibility tensors in per- oxidases: a novel approach to structure elucidation by NMR. J Biol Inorg Chem 1, 320–329. 50 Coletta M, Catarino T, LeGall J & Xavier AV (1991) A thermodynamic model for the cooperative functional properties of the tetraheme cytochrome c 3 from Desulfo- vibrio gigas. Eur J Biochem 202, 1101–1106. 51 Santos H, Moura JJ, Moura I, LeGall J & Xavier AV (1984) NMR studies of electron transfer mechanisms in a protein with interacting redox centres: Desulfovibrio gigas cytochrome c 3 . Eur J Biochem 141, 283–296. Thermodynamic parameters in ligated proteins C. A. Salgueiro et al. 2260 FEBS Journal 272 (2005) 2251–2260 ª 2005 FEBS . Binding of ligands originates small perturbations on the microscopic thermodynamic properties of a multicentre redox protein Carlos A. Salgueiro 1,2 ,. Comparison of these results with data for the isolated cytochrome shows that binding of ligands causes only small changes in the reduction potentials of the