Search

We present the first solid-state NMR study of an iron-sulfur protein. The combined use of very fast (60 kHz) magic-angle spinning and tailored radiofrequency irradiation schemes allows the detection and the assignment of most of the 1 H and 13 C resonances of the oxidized high-potential iron-sulfur protein I from Ectothiorhodospira halophila (EhHiPIP I), including those in residues coordinating the Fe 4 S 4 cluster. For these residues, contact shifts as large as 100 and 400 ppm for 1 H and 13 C resonances, respectively, were observed, which represent the most shifted solid-state NMR signals ever measured in metalloproteins. Interestingly, by targeting EhHiPIP I in a crystalline environment, we were able to capture distinct paramagnetic signatures from the two conformations present in the asymmetric unit. The magnetic properties of the system were verified by following the temperature dependence of the contact-shifted cysteine resonances.

The electron paramagnetic resonance (EPR) spectrum from the [4Fe-4S](3+) cluster in several high-potential iron-sulfur proteins (HiPIPs) is complex: it is not the pattern of a single, isolated S=1/2 system. Multifrequency EPR from 9 to 130 GHz reveals that the apparent peak positions (g values) are frequency-independent: the spectrum is dominated by the Zeeman interaction plus g-strain broadening. The spectra taken at frequencies above the X-band are increasingly sensitive to rapid-passage effects; therefore, the X-band data, which are slightly additionally broadened by dipolar interaction, were used for quantitative spectral analysis. For a single geometrical [4Fe-4S](3+) structure the (Fe-Fe)(5+) mixed-valence dimer can be assigned in six different ways to a pair of iron ions, and this defines six valence isomers. Systematic multicomponent g-strain simulation shows that the [4Fe-4S](3+) paramagnets in seven HiPIPs from different bacteria each consist of three to four discernible species, and these are assigned to valence isomers of the clusters. This interpretation builds on previous EPR analyzes of [4Fe-4S](3+) model compounds, and it constitutes a high-resolution extension of the current literature model, proposed from paramagnetic NMR studies.

The temperature dependence of the mean square displacement of the (57)Fe nuclei due to motion faster than 100 ns are measured by temperature-dependent Mössbauer spectroscopy for oxidized and reduced HiPIPs from Ectothiorhodospira halophila, Chromatium vinosum WT and a Cys77Ser mutant. The behaviour is interpretable in the frame of the general model of protein dynamics distinguishing two temperature intervals. The character of harmonic and quasi-diffusional modes in HiPIPs is discussed. Dynamic information obtained from Mössbauer spectroscopy and Fe K-edge EXAFS are compared. Structure dynamics of the iron-sulfur cluster in the partially unfolded reduced HiPIP from C. vinosum was investigated by Mössbauer spectroscopy and EXAFS, indicating an intact metal centre and a protein backbone with a largely collapsed secondary structure. The role of the cofactor during protein folding is discussed. Differences in the dynamics between the native protein and the molten globule are found at physiological temperatures only. The structure and dynamic behaviour of the [Fe(4)S(4)]Cys(3)Ser cluster in the Cys77Ser mutant of the HiPIP from C. vinosum are analysed. The temperature dependence of electron relaxation in oxidized HiPIPs is investigated by Mössbauer spectroscopy and analysed theoretically, considering spin-spin and spin-lattice relaxation. The latter consists of contributions from direct phonon bottleneck and Orbach mechanisms. The data agree with former pulsed EPR results. Orbach relaxation is interpreted as due to transitions between electronic isomers of oxidized HiPIPs. With this interpretation, the energetic difference between both isomers equals the energy gap estimated from the temperature dependence of the Orbach relaxation.

This work presents the three-dimensional NMR solution structure of recombinant, oxidized, unbound PsaC from Synechococcus sp. PCC 7002. Constraints are derived from homo- and heteronuclear one-, two- and three-dimensional (1)H and (15)N NMR data. Significant differences are outlined between the unbound PsaC structure presented here and the available X-ray structure of bound PsaC as an integral part of the whole cyanobacterial PS I complex. These differences mainly concern the arrangement of the N- and C-termini with respect to the [4Fe-4S] core domain. In the NMR solution structure of PsaC the C-terminal region assumes a disordered helical conformation, and is clearly different from the extended coil conformation, which is one of the structural elements required to anchor PsaC to the PS I core heterodimer. In solution the N-terminus of PsaC is in contact with the pre-C-terminal region but slides in between the latter and the iron-sulfur core region of the protein. Together, these features result in a concerted movement of the N-terminus and pre-C-terminal region away from the F(A) binding site, accompanied by a bending of the N-terminus. In comparison, the same terminal regions are positioned much closer to F(A) and take up an anti-parallel beta-sheet arrangement in PsaC bound to PS I. The conformational changes between bound and unbound PsaC correlate with the differences reported earlier for the EPR spectra of reduced F(A) and F(B) in bound versus unbound PsaC. The observed different structural features in solution are highly relevant for unraveling the stepwise assembly process of the stromal PsaC, PsaD and PsaE subunits to the PS I core heterodimer. Electronic supplementary material to this paper can be obtained by using the Springer Link server located at http://dx.doi.org/10.1007/s00775-001-0321-3.

Mössbauer, 57Fe ENDOR, CW and pulsed EPR experiments were performed on the reduced and the oxidized high-potential iron proteins (HiPIPs) of the wild type (WT) and the C77S mutant from Chromatium vinosum. The EPR spectra of the oxidized WT and mutant show three species respectively having nearly the same g-values but strongly changed spectral contributions. Relaxation times were estimated for oxidized WT and mutant at T = 5 K with pulsed EPR. A-tensor components of both iron pairs were obtained by 57Fe ENDOR, proving a similar magnetic structure for the WT and the mutant. Electronic relaxation has to be taken into account at T = 5 K in native and mutated oxidized HiPIPs to achieve agreement between Mössbauer and 57Fe ENDOR spectroscopies. The Mössbauer spectroscopy shows that the oxidized cluster contains a pure ferric and a mixed-valence iron pair coupled antiparallel. While all cluster irons from reduced C. vinosum WT are indistinguishable in the Mössbauer spectrum, the reduced C77S mutant shows a non-equivalence between the serine-bound and the three cysteine-ligated iron ions. The Mössbauer parameters confirm a loss of the covalent character of the iron bond when S is replaced by O and indicate a shift of the cluster's electron cloud towards the serine. Mössbauer spectra of the oxidized mutant can be simulated with two models: model I introduces a single electronic isomer with the serine always ligated to a ferric iron. Model II assumes two equally populated electronic isomers with the serine ligated to a ferric iron and a mixed-valence iron, respectively. The latter model is in better agreement with EPR and NMR.

15N T(1), T(2) and (1)H-(15)N NOE were measured for the thermophilic Fe(7)S(8) protein from Bacillus schlegelii and for the Fe(4)S(4) HiPIP protein from Chromatium vinosum, which is a mesophilic protein. The investigation was performed at 276, 300, and 330 K at 11.7 T for the former, whereas only the 298 K data at 14.1 T for the latter were acquired. The data were analyzed with the model-free protocol after correcting the measured parameters for the effect of paramagnetism, because both proteins are paramagnetic. Both thermophilic and mesophilic proteins are quite rigid, with an average value of the generalized order parameter S2at room temperature of 0.92 and 0.94 for Fe(7)S(8) and Fe(4)S(4) proteins, respectively. The analyzed nitrogens for the Fe(7)S(8) protein showed a significant decrease in S2with increasing temperature, and at the highest temperature >70% of the residues had an internal correlation time. This research shows that subnanosecond rigidity is not related to thermostability and provides an estimate of the effect of increasing temperature on this time scale.

The PsaC subunit of Photosystem I (PS I) is a 9.3-kDa protein that binds two important cofactors in photosynthetic electron transfer: the [4Fe-4S] clusters FA and FB. The g-tensor orientation of FA- and FB- is believed to be correlated to the preferential localization of the mixed-valence and equal-valence (ferrous) iron pairs in each [4Fe-4S]+ cluster. The preferential position of the mixed-valence and equal-valence pairs, in turn. can be inferred from the study of the temperature dependence of contact-shifted resonances by 1H NMR spectroscopy. For this, a sequence-specific assignment of these signals is required. The 1H NMR spectrum of reduced, unbound PsaC from Synechococcus sp. PCC 7002 at 280.4 K in 99% D2O solution shows 18 hyperfine-shifted resonances. The non-solvent-exchangeable, hyperfine-shifted resonances of reduced PsaC are clearly identified as belonging to the cysteines coordinating the clusters FA- and FB- by their downfield chemical shifts, by their temperature dependencies, and by their short T1 relaxation times. The usual fast method of assigning the 1H NMR spectra of reduced [4Fe-4S] proteins through magnetization transfer from the oxidized to the reduced state was not feasible in the case of reduced PsaC. Therefore, a de novo self-consistent sequence-specific assignment of the hyperfine-shifted resonances was obtained based on dipolar connectivities from 1D NOE difference spectra and on longitudinal relaxation times using the X-ray structure of Clostridium acidi urici 2[4Fe-4S] cluster ferredoxin at 0.94 A resolution as a model. The results clearly show the same sequence-specific distribution of Curie and anti-Curie cysteines for unbound, reduced PsaC as established for other [4Fe-4S]-containing proteins; therefore, the mixed-valence and equal-valence (ferrous) Fe-Fe pairs in FA- and FB- have the same preferential positions relative to the protein. The analysis reveals that the magnetic properties of the two [4Fe-4S] clusters are essentially indistinguishable in unbound PsaC, in contrast to the PsaC that is bound as a component of the PS I complex.

Mössbauer spectra of the oxidized [Fe4S4]3+ and the reduced [Fe4S4]2+ clusters in the high-potential iron protein I from Ectothiorhodospira halophila were measured in a temperature range from 5 K to 240 K. EPR measurements and 57Fe electron-nuclear double resonance (ENDOR) experiments were carried out with the oxidized protein. In the oxidized state the cluster has a net spin S = 1/2 and is paramagnetic. As common in [Fe4S4]3+ clusters, the Mössbauer spectrum was simulated with two species contributing equally to the absorption area: two Fe3+ atoms couple to the "ferric-ferric" pair, and one Fe2+ and one Fe3+ atom give the "ferric-ferrous pair". For the simulation of the Mössbauer spectrum, g-values were taken from EPR measurements. A-tensor components were determined by 57Fe ENDOR experiments that turned out to be a necessary source of estimating parameters independently. In order to obtain a detailed agreement of Mössbauer and ENDOR data, electronic relaxation has to be taken into account. Relaxing the symmetry condition in a way that the electric field gradient tensor does not coincide with g- and A-tensors yielded an even better agreement of experimental and theoretical Mössbauer spectra. Spin-spin and spinlattice relaxation times were estimated by pulsed EPR; the former turned out to be the dominating mechanism at T = 5 K. Relaxation times measured by pulsed EPR and obtained from the Mössbauer fit were compared and yield nearly identical values. The reduced cluster has one additional electron and has a diamagnetic (S = 0) ground state. All the four irons are indistinguishable in the Mössbauer spectrum, indicating a mixed-valence state of Fe2.5+ for each.

The crystal structure of the reduced high-potential iron protein (HiPIP) from Chromatium vinosum has been redetermined in a new orthorhombic crystal modification, and the structure of its H42Q mutant has been determined in orthorhombic (H42Q-1) and cubic (H42Q-2) modifications. The first two were solved by ab initio direct methods using data collected to atomic resolution (1.20 and 0. 93 A, respectively). The recombinant wild type (rc-WT) with two HiPIP molecules in the asymmetric unit has 1264 protein atoms and 335 solvent sites, and is the second largest structure reported so far that has been solved by pure direct methods. The solutions were obtained in a fully automated way and included more than 80% of the protein atoms. Restrained anisotropic refinement for rc-WT and H42Q-1 converged to R(1) = summation operator||F(o)| - |F(c)|| / summation operator|F(o)| of 12.0 and 13.6%, respectively [data with I > 2sigma(I)], and 12.8 and 15.5% (all data). H42Q-2 contains two molecules in the asymmetric unit and diffracted only to 2.6 A. In both molecules of rc-WT and in the single unique molecule of H42Q-1 the [Fe(4)S(4)](2+) cluster dimensions are very similar and show a characteristic tetragonal distortion with four short Fe-S bonds along four approximately parallel cube edges, and eight long Fe-S bonds. The unique protein molecules in H42Q-2 and rc-WT are also very similar in other respects, except for the hydrogen bonding around the mutated residue that is at the surface of the protein, supporting the hypothesis that the difference in redox potentials at lower pH values is caused primarily by differences in the charge distribution near the surface of the protein rather than by structural differences in the cluster region.

Heteronuclear multidimensional NMR spectroscopy was used to investigate in detail the structural and dynamical properties of a partially unfolded intermediate of the reduced high-potential iron-sulfur protein (HiPIP) from Chromatium vinosum present in 4 M guanidinium chloride solution. After an extensive assignment of 15N and 1H resonances, NOE data, proton longitudinal relaxation times, and 3JHNHalpha coupling constants as well as 15N relaxation parameters (T1, T2, T1rho, and 1H-15N NOE) were obtained and used to build a structural model of the intermediate. The Fe4S4 cluster of the HiPIP plays a decisive role in determining the resulting structure, which is random in the N-terminal half of the protein and partially organized in the loops between the cysteines bound to the cluster. Consistent with the structural data, the backbone mobility is typical of folded proteins in the regions where there are elements of structure and increases with the structural indetermination.

The solution structure of the D13C variant of the thermostable Fe7S8 ferredoxin from Bacillus schlegelii has been determined by 1H-NMR spectroscopy in its oxidized form. In a variable-temperature NMR study the D13C variant was as thermostable (up to 90 degrees C) as the wild-type protein (WT). Seventy-five out of 77 amino acid residues and 81% of all theoretically expected proton resonances in the D13C Fe8S8 protein have been assigned. Its structure was determined through torsion angle dynamics calculations with the program DYANA, using 935 meaningful NOEs (from a total of 1251), hydrogen bond constraints, and NMR-derived dihedral angle constraints for the cluster-ligating cysteines. Afterwards, restrained energy minimization and restrained molecular dynamics were applied to each conformer of the family. The final family of 20 structures has RMSD values from the mean structure of 0.055 nm for the backbone atoms and of 0.099 nm for all heavy atoms. The overall folding of the WT is maintained in the mutant, except for the immediate vicinity of the new cysteine, which becomes much more similar to native Fe8S8 proteins. The two residues at positions 11 and 12, which constitute an insertion with respect to all known Fe8S8 proteins, assume a conformation that does not prevent the preceding and following residues from folding like in native Fe8S8 proteins. Clear evidence for the existence of two conformations involving almost half of the amino acid residues was found. The two conformations are structurally indistinguishable. Temperature-dependent NMR experiments show that one of them is thermodynamically more stable than the other.

A partially unfolded state of the Fe4S4-containing high potential iron-sulfur protein from Chromatium vinosum has been detected and characterized by NMR spectroscopy following addition of a concentrated solution of guanidinium chloride to the native protein. This intermediate species (i) maintains the polymetallic center, (ii) exhibits a largely collapsed secondary structure, and (iii) undergoes fast cluster decomposition upon oxidation. This information is framed into the knowledge about this class of proteins, and the possible role of this intermediate with respect to the in vivo folding/unfolding process is discussed as well its role in the slow hydrolytic degradation characteristic of oxidized HiPIPs.

The reduced high-potential iron sulfur protein I from Ectothiorhodospira halophila which contains the [4Fe-4S]2+ polymetallic center has been fully labeled with 15N and 13C. The protein is paramagnetic, the nuclear relaxation times of nuclei close to the paramagnetic ion are drastically shortened and some strategic dipolar connectivities are lost. Notwithstanding, the solution structure has been reported [Banci, L., Bertini, I., Eltis, L. D., Felli, I. C., Kastrau, D. H. W., Luchinat, C., Piccioli, M., Pierattelli, R. & Smith, M. (1994) Eur. J. Biochem. 225, 715-725]. We have performed classical HNHA, HNCA soft-COSY, soft-HCCH E. COSY and 15N-1H correlated NOESY experiments in order to obtain a set of 3J scalar coupling constants. Some experiments have been optimized to counterbalance the effect of paramagnetism. From heteronuclear single-quantum experiments preceded by a 180 degrees pulse and variable delay times, the non-selective magnetization recovery has been followed from which the contribution to dipolar relaxation of nuclei due to the interaction with the paramagnetic metal ions (rho para) has been estimated. Finally, the intensities of NOEs have been corrected for the presence of paramagnetic metal ions and these corrected values together with 3J values and rho para data have been used to obtain a well defined solution structure. The aim is that of obtaining a structure with enough constraints to be well resolved all over the protein, including the vicinity of the paramagnetic metal cluster, which is anchored to the protein through the rho para constraints. In total, 1226 corrected NOESY crosspeaks (of which 945 were found to be meaningful), 37 one-dimensional NOEs, 39 3JHNH alpha and 37 3JHNC' (providing 45 phi dihedral angle constraints) 54 3JH alpha H beta and 31 3JNH beta (providing 26 chi 1 dihedral angle constraints), 4 chi 2 dihedral angle constraints of the coordinated cysteines, obtained from the hyperfine shifts of the beta CH protons, and 58 rho para constraints, have been used for structure calculation. Restrained molecular dynamics simulations have also been performed to provide the final family of structures. This research demonstrates that stable isotope labeling provides specific advantages for the NMR investigation of paramagnetic molecules, as the small magnetic moment of heteronuclei minimizes the paramagnetic influence of unpaired electrons.

The full 1H NMR assignment of the reduced C77S mutant of Chromatium vinosum high-potential iron-sulfur protein (HiPIP) was achieved by taking advantage of the assignment available for the wild-type protein. A total of 1565 nuclear Overhauser effect (NOE) spectroscopy cross peaks were integrated and converted into distance constraints, of which 497 were found to be irrelevant. An additional 24 dipolar constraints were obtained from one-dimensional NOE difference spectra by saturating hyperfine-shifted beta CH2 cysteine/serine protons. Forty-six 3JNH-H alpha coupling constants and eight hydrogen bonds provided further constraints. Through a distance geometry approach, a family of 15 structures was calculated, which was subsequently subjected to restrained energy minimization. The root mean square deviations of the minimized structures were 0.62 +/- 0.09 and 1.09 +/- 0.11 A for backbone and heavy atoms, respectively. The resulting solution structures are very similar to those of the reduced wild-type protein (WT). An analysis of the NOEs experienced by the protons of Ser-77 in both the reduced and oxidized forms reveals that they are very similar to those experienced by Cys-77 in WT. On the basis of the hyperfine shifts observed for the Ser-77 protons and of the present structural analysis, it is concluded that the serine O gamma atom is coordinated to the polymetallic center, thus confirming the strict analogy of the electronic structures of the polymetallic center in both proteins. Capillary electrophoresis experiments demonstrate coordination of Ser-77 as an anion. Serine versus cysteine coordination in iron-sulfur proteins is briefly discussed.

Oxidized and reduced forms of high-potential iron-sulfur protein (HiPIP) from the purple non-sulfur photosynthetic bacterium Rhodoferax fermentans have been characterized using 1H-NMR spectroscopy. Pairwise and sequence-specific assignments of hyperfine-shifted 1H-NMR signals to protons of cysteine residues bound to the [4Fe-4S]3+/2+ cluster have been performed using one-dimensional NOE and exchange spectroscopy experiments. 1H-NMR hyperfine shifts and relaxation rates of cluster-bound Cys beta-CH2 protons indicate that in the [4Fe-4S]3+ cluster one iron ion can be formally described as Fe(III), while electron density corresponding to one electron is unevenly delocalized onto the remaining three iron ions. This delocalization is effected by means of two different electronic distributions interconverting rapidly on the NMR time scale. The mechanism of paramagnetic proton relaxation, studied by analyzing longitudinal relaxation rates of Cys beta-CH2 protons in HiPIPs from six different sources as a function of the Fe-S-C beta-C alpha dihedral angle, indicate that the major contribution is due to a dipolar metal-centered mechanism, with a non-negligible contribution from a ligand-centered dipolar mechanism which involves the 3p orbital of the Cys sulfur atom. A semi-quantitative tool for extracting structural information from relaxation time measurements is proposed.

We have accounted for the effect of paramagnetism on the intensities of NOEs in a 73-residue paramagnetic metalloprotein, the reduced high-potential iron sulfur protein ISO I from Ectothiorhodospira halophila, whose solution structure had been recently solved by us. The paramagnetic effects were dealt with through a suitably modified complete relaxation matrix approach. We have then recalculated the structure through a distance geometry program by minimizing the difference between the sixth roots of the calculated and experimental NOEs. The average RMSD, calculated on residues 4-71, within the structures constituting the two families decreased from 0.67 to 0.46 angstrom for backbone atoms and from 1.23 to 1.06 angstroms for all heavy atoms. The structures in the new family are for the most part within the indetermination of the previous, less resolved, family. A few specific differences are detected and related to the presence of non-negligible paramagnetic effects, which are now properly evaluated.

The NMR solution structure of the oxidized HiPIP from Chromatium vinosum has been solved. Despite the fact that the protein is paramagnetic, 85% of the 1H and 80% of the 15N signals have been assigned. Through 1537 NOEs, out of which 1142 were found to be relevant for the structure determination, a family of structures has been obtained by distance geometry calculations. These structures have then been subjected to restrained energy minimization (REM) and restrained molecular dynamics (RMD) calculations in vacuum. Finally, the mean structure of the RMD family has been treated through RMD in water. The RMSD values for the backbone and heavy atoms within the RMD family are 0.57 +/- 0.14 and 1.08 +/- 0.16 A, respectively. These values together with other parameters indicate that the structure is of good quality and as good as the structure of the reduced protein. The RMDw structures of the reduced and oxidized proteins are different beyond the experimental indetermination. The set of constraints for the reduced and oxidized forms have been used to treat the available X-ray structure by RMD in water. The two structures generated in this way are quite similar to their respective solution structures, thus confirming that the experimental constraints are capable of yielding two different structures from the same starting structural model. This is the first time that independently determined solution structures of two redox states of a paramagnetic protein are available. Differences between them and the X-ray structure are discussed.

1D and 2D 1H NMR studies of the Fe4S4 cluster containing ferredoxin I from Desulfovibrio africanus have been carried out with the aim of determining the geometry of the cluster linkages with the 4 Cys side chains that bind the cluster. This required the Cys beta CH resonances of the oxidised protein to be sequence-specifically and stereo-specifically assigned, and this was accomplished by a combination of TOCSY and NOE measurements, allied to model building based on X-ray structures of related ferredoxins. An analysis of the estimated hyperfine shifts of the Cys beta CH resonances with a Karplus-type equation relating the shifts to iron-sulfur-beta carbon-beta proton dihedral angles, taken together with the relative relaxation rates of the two beta CH2 resonances, estimated from their linewidths, then allowed the iron-sulfur-beta-carbon-alpha-carbon dihedral angles to be determined. A novel representation of the NMR data is presented which shows that the cluster dihedral angles are uniquely determined by the NMR data. The analysis reveals that the dihedral angles for D. africanus ferredoxin I are similar to the corresponding angles of other ferredoxins even though there are differences in their 1H NMR spectra. The sequence-specific and stereospecific assignments have been extended by analogy to the related Fe4S4-containing D. gigas ferredoxin I, and the stereospecific assignments to the Fe4S4-containing Thermococcus litoralis ferredoxin.

The 1H NMR assignment of the reduced HiPIP from Chromatium vinosum available in the literature [Gaillard, J., Albrand, J.-P., Moulis, J.-M., & Wemmer, D. E. (1992) Biochemistry 31, 5632-5639] has been extended up to 85% of the total protein protons. Ninety percent of the nitrogens have been assigned. Then the solution structure has been obtained using as many as 1147 meaningful NOE connectivities. The protein is sizably paramagnetic even though the ground state is a singlet. Nevertheless, the final RMSD values are 0.62 and 1.19 A for the backbone and the heavy atoms, respectively. These values compare well with those for diamagnetic proteins of the same size. The solution structure is discussed in the light of the available structural information from X-ray data.

The three-dimensional structure in solution of reduced recombinant high-potential iron-sulfur protein iso-I from Ectothiorhodospira halophila was determined using 948 relevant interproton NOEs out of the 1246 observed NOEs. The determination was accomplished using the XEASY program for spectral analysis and the distance geometry (DG) program DIANA for generation of the structure as described by Wüthrich [Wüthrich, K. (1989) Acc. Chem. Res. 22, 36-44]. The FeS cluster was simulated using an amino acid residue constructed for the present work from a cysteinyl residue with an iron and a sulfur atom attached to the terminal thiol. The family of structures obtained from distance geometry were subjected to energy minimization and molecular dynamics simulations using previously defined force field parameters. The quality of these structures at each stage of the refinement process is discussed with respect to the dihedral angle order parameter and the root-mean-square deviation of the atomic coordinates. The latter values for the backbone atoms vary from 67 pm for the distance-geometry structures to 60 pm for the energy-minimized structures to 51 pm for the structures subjected to restrained molecular dynamics. Finally, the structure in best agreement with the NOE constraints has been further treated with extensive restrained molecular dynamics in water. The solution structure is well defined and is very similar to the available X-ray structure. We do not know of any previous determination of the structure of a paramagnetic protein in solution by NMR. The effect of paramagnetism on the quality of the structure determination is discussed.

A 1H and 15N NMR investigation through two-dimensional and three-dimensional spectroscopy has been performed on the reduced form ([Fe4S4]2+) of the recombinant high-potential iron-sulfur protein (HiPIP) I from Ectothiorhodospira halophila expressed in Escherichia coli. [Fe4S4]2+ clusters in proteins are paramagnetic with a relatively low mu eff of about 0.8 mu B/iron ion, but the paramagnetic effects on nuclear relaxation are so strong as to yield T1 values of a few milliseconds and linewidths of hundreds of hertz for the nuclei closet to the paramagnetic center. Despite these features, 71 out of 73 residues were identified, most of which were assigned completely as far as proton resonances are concerned; as many as 68 residues could be assigned without any reference to the existing X-ray structure. A total of 88% of all protein protons and 58 out of 69 peptide HN nitrogen signals were assigned. To the best of our knowledge, this is the most extensive 1H assignment of a paramagnetic protein to date. Protons sensitive to the proximity of the cluster were assigned through suitable NOE spectroscopy experiments. Three out of the four coordinated cysteines were assigned, and two residues have been identified whose peptide HN protons give rise to H bonds with coordinated sulfur atoms. The inter-residue NOE cross peaks are in qualitative agreement with the secondary and tertiary structure as obtained from the available X-ray crystallographic analysis of the wild-type protein at 250-pm resolution. It is therefore shown that the expressed protein is properly folded and that it is a reliable model for the wild-type protein. These data are meaningful for the detection of structural differences among mutants in future studies.

The pH-dependence of the reduction potential determined through differential pulse voltammetry for the high potential iron sulfur proteins (HiPIP) from R. globiformis, C. vinosum, R. gelatinosus, E. vacuolata (I and II), E. halophila (I and II) is reported. A decrease in reduction potential with pH is invariably observed in the pH range where deprotonation of the imidazolium nitrogen of histidine residue(s) occurs. No pH dependence is observed for the only protein lacking histidines. It appears that surface charges like the His imidazolium groups are capable of influencing the reduction potential despite the known quencing of the electrostatic interactions due to solvent effects.

During previous research on oxidized and reduced high-potential iron-sulfur proteins (HiPIP hereafter), qualitative different electron self-exchange rates were noticed. We have now investigated this phenomenon in detail for HiPIP I and II from Ectothiorhodospira vacuolata, which differ significantly in total charge and in which the sequence homology is the largest among all known HiPIPs. We have also characterized the electronic structure of HiPIP I through 1H NMR and EPR spectroscopies to parallel the existing characterization of HiPIP II and other HiPIPs. This investigation has allowed us to propose a model, according to which the productive collisions for electron transfer occur through a hydrophobic patch near the cluster. The effects of total charge and redox potential are considered. The possible formation of dimers through the hydrophobic patch at liquid helium temperature is discussed in light of the EPR spectra.

Within the framework of an investigation of the electronic structure of oxidized high-potential iron-sulfur proteins (HiPIP), we have studied the HiPIP II from Ectothiorhodospira vacuolata, which was known to have a peculiar temperature dependence of the 1H NMR isotropic hyperfine shifts. The signals of the cysteine ligand protons have been sequence specifically assigned through NOE, NOESY, and TOCSY experiments. Nine hyperfine-shifted signals are observed: seven in the downfield and two in the upfield region. They have been assigned to the eight beta-CH2 protons of the four coordinated cysteines and to one alpha-CH cysteine proton. The two most downfield-shifted signals belong to the beta-CH2 protons of Cys 63 (Chromatium vinosum numbering) and the two upfield protons to those of Cys 43. These two pairs of protons show a Curie-type temperature dependence of the hyperfine shifts. Among the remaining five downfield-shifted signals, three show a Curie-type temperature dependence and two have an anti-Curie temperature dependence. The former are assigned to the beta-CH2 and alpha-CH protons of Cys 77 and the latter to the beta-CH2 protons of Cys 46. The shift patterns are thus similar, in a sequence-specific sense, to those of the analogous proteins from C. vinosum and Rhodocyclus gelatinosus, whereas they differ from those of Rhodocyclus globiformis HiPIP and even more from those of Ectothiorhodospira halophila HiPIP II. Oxidized HiPIPs can be formally viewed as containing a cluster of four ferric ions plus one extra electron. We present here a model based on a chemical equilibrium, fast on the NMR time scale, between two species, both of which contain a pair of iron(III) ions and a mixed-valence pair but are differently oriented within the protein frame. The EPR data are also discussed in the light of the debate on the nature of the different species detected at low temperature. The interpretation of the whole set of data on HiPIPs in the light of the present model is compared with that based on previous models.

1H one-dimensional and two-dimensional NMR spectra have been recorded for the oxidized and reduced forms of the high-potential iron-sulfur protein (HiPIP) from Rhodopseudomonas globiformis which has the highest known reduction potential. The spectrum of the oxidized protein is similar to that of Chromatium vinosum and Rhodocyclus gelatinosus HiPIP but different from that of the HiPIP II from Ectothiorhodospira halophila. Surprisingly, site-specific assignment has shown that in the oxidized protein the distribution of oxidation numbers within the cluster is very similar to that found for E. halophila HiPIP II and different from that of the other two proteins. The spectrum of the reduced species is very similar to that of all other HiPIPs known to date, indicating very similar electronic and geometric structures for the reduced forms. These findings are discussed in terms of cluster structure in HiPIPs and of redox potentials.