Solution structure of the functional domain of Paracoccus denitrificans cytochrome c552 in the reduced state


  • Primož Pristovšek,

    1. Institute of Biophysical Chemistry, J.W. Goethe-University of Frankfurt, Germany;
    Search for more papers by this author
    • *Present address: National Institute of Chemistry, Ljubljana, Slovenia.

      Note: the coordinates of the functional domain of Paracoccus denitrificans cytochrome c552 in the reduced state have been deposited in the RCSB Protein Data Bank with the PDB ID code 1C7M.

  • Christian Lücke,

    1. Institute of Biophysical Chemistry, J.W. Goethe-University of Frankfurt, Germany;
    Search for more papers by this author
  • Britta Reincke,

    1. Institute of Biophysical Chemistry, J.W. Goethe-University of Frankfurt, Germany;
    Search for more papers by this author
  • Bernd Ludwig,

    1. Institute of Biochemistry, Molecular Genetics,
      J. W. Goethe-University of Frankfurt, Germany
    Search for more papers by this author
  • Heinz Rüterjans

    1. Institute of Biophysical Chemistry, J.W. Goethe-University of Frankfurt, Germany;
    Search for more papers by this author

H. Rüterjans, Institut für Biophysikalische Chemie, Biozentrum, N230, 1. OG, Marie-Curie-Straße 9, 60439 Frankfurt a.M., Germany. Fax: +49 69 7982 9632, Tel.: +49 69 7982 9631, E-mail:


In order to determine the solution structure of Paracoccus denitrificans cytochrome c552 by NMR, we cloned and isotopically labeled a 10.5-kDa soluble fragment (100 residues) containing the functional domain of the 18.2-kDa membrane-bound protein. Using uniformly 15N-enriched samples of cytochrome c552 in the reduced state, a variety of two-dimensional and three-dimensional heteronuclear double-resonance NMR experiments was employed to achieve complete 1H and 15N assignments. A total of 1893 distance restraints was derived from homonuclear 2D-NOESY and heteronuclear 3D-NOESY spectra; 1486 meaningful restraints were used in the structure calculations. After restrained energy minimization a family of 20 structures was obtained with rmsd values of 0.56 ± 0.10 Å and 1.09 ± 0.09 Å for the backbone and heavy atoms, respectively. The overall topology is similar to that seen in previously reported models of this class of proteins. The global fold consists of two long helices at the N-terminus and C-terminus and three shorter helices surrounding the heme moiety; the helices are connected by well-defined loops. Comparison with the X-ray structure shows some minor differences in the positions of the Trp57 and Phe65 side-chain rings as well as the heme propionate groups.


soluble fragment of cytochrome c552


electron transfer

Cytochrome c is a well-established electron mediator between the last two complexes of the mitochondrial redox chain, the cytochrome bc1 complex (complex III) and the aa3-type cytochrome c oxidase (complex IV). It contains a covalently bound heme moiety, with thioether linkages to the cysteine residues in the conserved CXYCH motif. During the redox cycle, the iron atom alternates between the diamagnetic reduced (2+) and the paramagnetic oxidized (3+) state. It is octahedrally coordinated by the four porphyrin nitrogens and two axial ligands, a histidine and a methionine, which is a common feature in this class of cytochromes [1,2].

Respiratory electron-transfer (ET) complexes are integral membrane proteins, which are not easily amenable to structural studies. The molecular basis of ET has therefore been understood in detail for only a few selected model systems [3,4]. The structures of c-type cytochromes, however, have been studied extensively [2]. NMR solution structures have been solved for horse heart cytochrome c[5–8], yeast iso-1-cytochrome c[9,10], cytochrome c6 from Monoraphidium braunii[11,12] and cytochrome c7 from Desulfuromonas acetoxidans[13–15] in both oxidation states, and for cytochromes c551 from Ectothiorhodospira halophila[16] and Pseudomonas stutzeri[17], c552 from Hydrogenobacter thermophilus[18] and Nitrosomonas europaea[19], c2 from Rhodobacter capsulatus[20], c3 from Desulfovibrio vulgaris[21] and c6 from Synechococcus elongatus[22] in the reduced state.

The use of heteronuclear NMR is critically dependent on efficient isotope labeling of the protein, often only achieved in heterologous expression hosts such as Escherichia coli, in which the biosynthesis of the heme cofactor, its translocation across the cytoplasmic membrane, and the insertion into the apo-protein pose an additional challenge. Thus, the difficulty in labeling cytochromes is due to the complexity of the protein maturation. Until now, to the best of our knowledge, 13C/15N-labeling has been reported only for R. capsulatus cytochrome c2[23], cytochrome c′ from purple bacteria [24] and Thiobacillus versutus ferrocytochrome c550[25], as well as 15N-labeling of R. sphaeroides cytochrome c2[26], D. vulgaris Hildenborough cytochrome c553[27] and Saccharomyces cerevisiae iso-1-cytochrome c[28].

Greater nutritional adaptability and flexibility to environmental demands often cause bacterial ET chains to display a higher degree of complexity than the mitochondrial counterpart. P. denitrificans is a gram-negative facultative anaerobic bacterium found in soil and sewage. Its energy metabolism depends on ET-linked phosphorylation. A large number of c-type cytochromes has been identified, both soluble and membrane-integrated. The membrane-bound 18-kDa cytochrome c552 was identified as the genuine electron mediator between complexes III and IV; its deletion, or inhibition by a specific antibody, suppresses the ET in membranes [29]. Two soluble fragments of the P. denitrificans cytochrome c552 have recently been over-expressed in E. coli, both lacking the N-terminal membrane anchor domain [30].

In an attempt to learn about the molecular interactions and ET mechanisms of cytochrome c552 with its different redox partners, we determined the solution structure of the P. denitrificans cytochrome c552. The smallest fragment containing the functional domain is a 10.5-kDa soluble protein of 100 residues (cytc hereafter), which interacts with the bacterial oxidase and sustains ET activity. Its expression level in E. coli is sufficient to allow uniform isotope labeling with 15N. Here we report the solution structure of the reduced form of cytc and present a comparison with the recently solved X-ray structure [31].

Materials and methods

Expression and purification

Preparation of nonenriched cytc has been described previously [30]. 15N-enriched cytc was expressed in several 100-mL batches of Bioexpress-1000 medium (Cambridge Isotope Laboratories, Andover, MA, USA) in order to obtain maximal yield.

NMR data collection and processing

We chose pH 5.5 and 298 K for the study of the solution structure of cytc. Under these conditions, the protein was stable for the duration of NMR data collection. NMR samples were prepared at 1.5–4 mm concentration in argon-purged 20 mm phosphate buffer (H2O : D2O = 90 : 10, v/v) at pH 6.0; sodium dithionite in small excess was used for reduction, which caused the pH in the solution to decrease by 0.5 units.

NMR data collection was carried out on a Bruker DMX spectrometer operating at a 1H resonance frequency of 600.13 MHz using a 5-mm triple-resonance (1H/13C/15N) probe with XYZ-gradient capability. Homonuclear and 15N-edited TOCSY and NOESY [32,33] experiments were performed [33a] as reported elsewhere.

The water signal was suppressed using either gradient coherence selection in combination with sensitivity enhancement [34] or the WATERGATE pulse sequence [35]. Quadrature detection in the indirectly detected dimensions was obtained either by the States-TPPI [36] or the echo/antiecho method [37]. Chemical shifts were referenced directly to internal sodium 2,2-dimethyl-2-silapentane-5-sulfonate[38] in order to ensure consistency among all spectra.

A 90° phase-shifted squared-sine-bell function was used for apodization in all dimensions. Forward linear prediction was used to extend the time-domain data in three-dimensional experiments; zero-filling was applied in the indirectly detected dimensions. Polynomial baseline correction was used wherever necessary. The final three-dimensional matrices typically consisted of 1024 × 128 × 256 real data points. The spectral data were processed on a Silicon Graphics Indy workstation using the Bruker xwinnmr 1.3 software package. Peak-picking and data analysis of the transformed spectra were performed using the aurelia 2.5.9 (Bruker) and felix 97 (Molecular Simulations Inc., San Diego, CA) programs.

Restraints generation and structure calculation

The NOE-derived distance restraints were determined from 2D homonuclear NOESY and 3D 15N-edited NOESY-HSQC spectra. Automated assignments of the NOEs, based only on chemical shifts, were obtained with the self-written program nmr2st. An internal calibration, based on the intensities of well-resolved glycine Hα2–Hα3 cross-peaks as well as sequential and medium-range NOEs from residues within the well-defined α-helices near the N-terminus and C-terminus, was used to set the upper distance limits. The cross-peak intensities were thus grouped into different distance categories of 2.5, 3.2, 4.0 and 5.0 Å. Assignment of ambiguous NOE cross-peaks was made by applying a structure-aided filtering strategy in repeated rounds of structure calculations using the redac routine of the diana 2.8 program package [39]. A total of 44 stereospecific assignments of the prochiral methylene (33) and isopropyl methyl (11) groups was obtained using the program glomsa[40]. Pseudo-atom correction for unassigned stereo partners and magnetically equivalent protons was applied as proposed by Wüthrich et al. [41].

In order to introduce the heme moiety into the structure calculations, a new diana residue named CYSH was created, in which the heme CAB carbon atom (the heme nomenclature of the RCSB Protein Data Bank [42] is used throughout) was linked to the sulfur atom of a CYSS residue and inserted into the sequence as CYSH14. The conformation of the heme tetrapyrrole ring from the X-ray structure [31] was used as a rigid entity in the structure calculations; all other single bonds of the prosthetic group were allowed to rotate. In addition, Cys17 Sγ was attached to the CAC carbon atom of the heme residue by the use of several lower and upper limit distance restraints (d[Cys17 SG, Cysh14 CAC] = 1.8 and 1.9 Å, d[Cys17 SG, Cysh14 CBC] = 2.7 and 2.8 Å, d[Cys17 SG, Cysh14 C3C] = 2.8 and 2.9 Å, and d[Cys17 CB, Cysh14 CAC] = 2.8 and 2.9 Å for the lower and upper limits, respectively). Furthermore, the axial ligand atoms His18 Nε2 and Met78 Sδ were fixed to their positions above and below the iron atom in an equivalent fashion using distance restraints to the iron and the four porphyrin ring nitrogen atoms NA, NB, NC and ND. No other restraints, except for NOE restraints derived from NMR spectra, were imposed upon the side-chains of His18 and Met78.

No hydrogen bond restraints were used in the structure calculation. In the analysis, a purely geometrical criterion for the existence of an H-bond was used (dHO < 2.3 Å, ΘNHO > 135°).

The final three-dimensional solution structures of cytc were generated using the program dyana 1.5 [43] which proved to be superior to diana in speed and quality of the obtained structures. Starting ab initio, 300 conformers were calculated by dyana in 8000 annealing steps each. Subsequent energy minimization in the presence of the NMR restraints, carried out with the discover module of the insight 97 software package (Molecular Simulations Inc., San Diego, CA, USA), was performed on the 20 best dyana conformers using the CVFF force field [44] with a dielectric constant equal to r (distance in Å). A force constant of 84 kJ·mol−1·Å2 was used in the NOE restraint term. The tetrapyrrole ring was kept frozen in the X-ray conformation throughout the minimization procedure. The resulting structures were analyzed using the prochecknmr software [45].

Results and discussion

The complete resonance assignment of cytc has been reported elsewhere [33a] and is available at the BioMagResBank ( under accession number BMRB-4471.

The ring-current shifts

Even though the reduced form of cytc is not paramagnetic, a number of rather unusual chemical shift values has been observed [33a]; these values are caused by ring-current effects, mostly from the heme porphyrin ring system. The protons of some residues such as Gly27 (Hα2 at −0.19 p.p.m.), Leu30 (Hδ1 at −0.73, Hδ2 at −1.52 p.p.m.), Val38 (Hγ1 at −0.94 p.p.m.), Phe65 (Hε at 5.54 p.p.m.) and Pro69 (Hγ1 at 0.58 p.p.m., Hγ2 at 0.00 p.p.m.) are situated right above or below the heme plane and thus experience considerable up-field shifts away from the standard random coil values [46]. Notable shifts also occur for the Glu8 Hα (2.38 p.p.m.) and Gly82 HN (4.50 p.p.m.). These effects, however, are attributed to the ring currents of the Tyr95 and Phe80 aromatic side-chains, respectively; their centers are as close as 3.24 ± 0.19 and 3.16 ± 0.17 Å to the above-mentioned atoms.

The down-field shifts of resonances that belong to protons lying within the plane defined as an extension of the heme ring system are less pronounced and therefore less obvious. In the cases of Ala39 HN (9.43 p.p.m.), Ser47 HN (11.04 p.p.m.) and Trp57 Hε1 (11.65 p.p.m.), for example, the exact contribution of ring-current effects remains speculative, because of possible intramolecular H-bonding that would shift the resonances in the same direction.

The largest shifts of side-chain proton resonances, however, are observed for the side-chains of the two axial ligands of the heme iron. In particular, the resonances of His18 Hε1 and Met78 Hε are strongly up-field shifted to 0.53 p.p.m. and −3.36 p.p.m., respectively.

Finally, it may be noted that His18 Nδ1 is protonated and was identified at 165.9 p.p.m., while the corresponding Hδ1 resonance was found at 9.51 p.p.m.

The heme moiety

Assignment of the heme resonances was performed exclusively with NOESY connectivities, except for the two propionate and two thioether groups which also give rise to TOCSY patterns. The four meso-protons have chemical shift values between 8.96 and 9.73 p.p.m. The heme is located in the interior of the protein and thus shows a large number of NOEs (130 in total) to the surrounding amino acid residues.

Secondary structure determination

Figure 1 shows the short-range and medium-range NOE patterns, observed for the backbone protons in the NOESY spectra. Sequential HN–HN connectivities for stretches longer than two amino acids have been observed for residues 5–11, 12–18, 19–22, 24–27, 29–31, 32–37, 38–41, 50–54, 56–58, 60–68, 70–73, 74–76, 79–82 and 85–98; the gap between residues 11 and 12, 22 and 23, 31 and 32, and 55 and 56 may be caused by the near degeneracy of the corresponding HN frequencies, placing the expected HN–HN cross-peak too close to the diagonal for possible detection. Six prolines that interrupt the sequential HN–HN and Hα–HN connectivities are located in positions 4, 28, 49, 59, 69 and 84. All are present in the trans configuration, as indicated by the observation of strong NOE connectivities between their Cδ protons and the Hα of the corresponding preceding residues [46].

Figure 1.

Schematic representation of the sequential connectivities involving HN, Hα and Hβ protons in the reduced form of cytochrome c552 from Paracoccus denitrificans. For sequential connectivities, the thickness of the bars indicates the NOE intensities. Medium-range NOEs are identified by lines connecting the two coupled residues.

Helical structures, characterized by strong sequential HN–HN as well as medium-range Hα–HN(i,i + 3), Hα–Hβ(i,i + 3) and Hα–HN(i,i + 4) NOEs, are present in segments 4–14, 48–53, 61–65, 69–72 and 86–98. The second, third and fourth helix are very short. The presence of a certain number of Hα–HN(i,i + 2) connectivities in all the helices may indicate that they are in part distorted toward a 310-helical conformation.

Solution structure calculations

The experimental NOESY peaks from 2D and 3D spectra were assigned, integrated and transformed into 1893 upper distance restraints. A total of 1486 was found to be meaningful and therefore taken into account by the program dyana[43] in the structure calculations. The number of experimental NOE restraints per residue is shown in Fig. 2; it corresponds to 18.9 experimental and 14.9 meaningful NOEs per residue on the average.

Figure 2.

Histogram representing the number of meaningful NOEs per residue. Bars indicate long-range NOEs (gray), medium-range NOEs (black), and sequential + intraresidual NOEs (white). The heme prosthetic group is treated as residue 101.

A total of 44 stereospecific assignments of diastereotopic groups was obtained with the program glomsa[40]. Consequently, diastereotopic methyl groups with nondegenerate proton resonances were stereospecifically assigned for seven of seven valines and four of five leucines.

Three hundred structures were finally calculated. The structure ensemble obtained using the NOE-derived restraints consisted of the 20 best structures with target functions [40] ranging from 1.01 to 1.15 Å2 and with individual violations not exceeding 0.24 Å. These 20 structures were subsequently subjected to restrained energy minimization. Figure 3 shows the ensemble of the 20 final energy-minimized structures, which were selected to represent the solution structure of cytc in the reduced state. The rmsd of the backbone and all heavy atoms in this region is 0.56 ± 0.10 Å and 1.09 ± 0.09 Å, respectively, for the nonterminal residues. The statistical information for this family of structures is summarized in Table 1. The value of 0.56 Å for the backbone rmsd is the result of some regions with very good definition (global rmsd values per residue < 0.45 Å, e.g. residues 26–30, 37–38, 50–57, 61–80 and 89–93) together with some less-defined regions showing global backbone rmsd values per residue > 0.8 Å (residues 1–6, 20, 23, 42–45 and 96–100). No significant violations of single distance restraints (the average number of violations exceeding 0.2 Å is 4.4 per structure; no single violation exceeds 0.34 Å) and only small violation energies (88.2 ± 10.0 kJ·mol−1 for the sum of the restraint terms) are observed in the ensemble.

Figure 3.

Stereoplot showing the ensemble ofthe 20 final energy-minimized structures of cytochrome c552 from Paracoccus denitrificans in the reduced form. Displayed are the α-carbon traces superposed for residues 2–99. The heme prosthetic group is shown in gray.

Table 1. Structural statistics of the 20 selected structures of cytochrome c552 from Paracoccus denitrificans in the reduced state. The heme prosthetic group is treated as residue 101.
Restraint statistics
 No. of meaningful distance restraints1486
 Intraresidual 297
 Sequential (|i − j= 1) 342
 Medium range (1 < |i − j| < 5) 305
 Long range (|i − j| > 4) 542
 Avg. No. of violations > 0.2 Å per structure   4.4
 Max. violation (Å)   0.34
Structural precision (residues 2–99)Ensemble
 rmsd for the Cα atoms (Å)0.53 ± 0.10
 rmsd for the backbone atoms (Å)0.56 ± 0.10
 rmsd for the heavy atoms (heme included, Å)1.09 ± 0.09

The prochecknmr program [45] was used to further assay the quality of the energy-minimized structure ensemble. Analysis of the backbone dihedral angles φ and ψ shows that > 99% of all the nonglycine/nonproline residues fall within the favorable regions of the Ramachandran plot: 80.4% in the most favored, 18.1% in the additionally allowed and 1.0% in the generously allowed regions. Over all of the 20 structures, 0.5% of the total number of residues (i.e. eight from 1560) fall into disallowed regions.

Description of the global fold

The overall solution structure of cytc(Fig. 4) is well-resolved, with the fold of the protein highly analogous to that of other c-type cytochromes despite substantial differences in the amino acid sequences. The structure is tightly packed and almost globular. It consists of two long helices located at the N-terminus and C-terminus (residues 4–14 and 86–98, respectively) and three shorter helices (residues 48–53, 61–65 and 69–72) in the middle of the sequence.

Figure 4.

Schematic representation of therefined solution structure in the reduced state of cytochrome c552 from Paracoccus denitrificans. (Produced with molscript[47] and Raster3D [48]).

Immediately following the N-terminal helix is a double loop substructure connecting residues 17 and 26 as well as 22 and 30. Two concatenated β-turn-like substructures are observed for residues 30–33 and 33–36, respectively. Another loop connects residues 39 and 46. This loop, together with the second helix (residues 48–53), is part of a larger loop that connects residues 35 and 57. The fourth helix (residues 69–72) is part of the last loop formed by residues 66–83. Despite their various lengths, the loops are generally well-defined in the solution structure with exception of some residues (20, 23 and 42–45) in loops 17–26 and 39–46, as mentioned above. These residues also show increased circular variances in the backbone and side-chain dihedral angles. However, this less-refined structural definition of the two latter loops may in part be due to the fact that some residues in these loops have relatively few interresidual NOEs (Fig. 2), rather than to increased intrinsic mobility. Future 15N-relaxation studies will hopefully help to clarify these questions.

The two terminal helices are involved in tight interactions, as evidenced by a number of NOEs between them (e.g. Ala6 to Asn91, Gly7 to Leu92, Glu8 to Tyr95, Val10 to Leu92, Phe11 to Leu92). The C-terminal helix also interacts with the third helix (residues 61–65) as evidenced by NOEs (e.g. Leu62 to Leu96, Gln63 to Ile93, Glu64 to Arg89). In all cases, the predominant forces are hydrophobic and van der Waal's interactions between the side-chains of the before-mentioned residues.

The heme moiety is almost completely buried inside the protein and is accessible to the solvent only at the HHD, HMD, HBC and HMC groups. The heme ring is slightly nonplanar with a distorted saddle-shaped geometry, as observed in the X-ray structure [31] and many other structures of related cytochromes (see Introduction). The tetrapyrrole ring was kept fixed in the X-ray conformation during minimization to prevent any additional force-field dependent distortion that is not validated by experimental data. The geometry of the axial iron ligands, His18 and Met78, is very well defined. The plane of the His18 imidazole ring is almost perpendicular to that of the porphyrin ring. The concentration of lysines on the surface of the protein surrounding the heme cleft, thought to be of central importance to the formation of the physiologically relevant complexes with cytochrome c reductase and oxidase, is preserved in the present model for the solution structure of cytc. The residues Lys13 and Lys77 are found immediately adjacent to the heme cleft; the side-chain of Lys77 hinders solvent access to the propionate D group. Residues Lys15, Lys19 and Lys70 are found close to the cleft, whereas Lys9, Lys51, Lys74 and Lys85 are more distant, but still on the same ‘hemisphere’(Fig. 5); the other side of the protein, opposite the heme cleft, is completely devoid of lysine residues.

Figure 5.

Solvent accessible surface of cytochrome c552 from Paracoccus denitrificans in the reduced state (A, front view; B, rear view). Lysine residues are depicted in black, the heme in gray.

Surrounding the heme ring are several conserved aromatic residues (Phe11, Phe44, Tyr46, Trp57, Phe65 and Phe80) that could play a role in the redox reaction. The phenyl ring of Phe80 is the one closest to the iron atom (distance from the ring center: 6.51 ± 0.11 Å in the structure ensemble) and lies almost parallel to the heme plane. The next-closest aromatic ring belongs to Phe65 (distance of the ring center to the iron: 7.37 ± 0.09 Å); it is almost perpendicular to the heme plane and completely buried in the interior of the protein structure. Based on the NMR results, the positions of the above-mentioned aromatic rings are comparable with the X-ray structure [31], except for Trp57 and Phe65, as discussed below.

Here, we would like to focus on two residues, Phe44 and Phe80, whose phenyl rings are both solvent-accessible and lie in opposite corners at the entrance to the heme cleft. Phe80 is highly conserved and corresponds to Phe82 of horse heart cytochrome c[5–8], which is believed to play an important role in the biological function of that protein. In the X-ray study [31], small movements of Phe80 are detected in two of the four cytochrome molecules in the asymmetric unit upon change of oxidation state; however, the relevance of this change is not clear. The NMR data show no unusual features for Phe80 in the reduced state. The ring proton resonances of Phe44, however, display extreme line-broadening effects. Similar behavior is observed for the aromatic protons of the adjacent Tyr46 ring, which apparently forms a hydrogen bond with the propionate D of the heme moiety, as discussed below. In mitochondrial c-type cytochromes, the position of Phe44 is occupied either by a tyrosine or phenylalanine residue, while Tyr46 is highly conserved. Therefore, a functionally relevant interaction between these two neighboring residues and the heme system may be possible. Hence, Phe44 and Tyr46 are interesting targets for further mutagenesis studies.

Comparison with the crystal structure of cytc

The solution structure of cytc (pH 5.5) has been compared with the recently determined X-ray structure of the same protein crystallized at pH 4.5 in the reduced form [31]. The rmsd values between the energy-minimized solution structure ensemble of cytc and the crystal structure (deposited under ID code 1QL3 in the RCSB Protein Data Bank) is 0.86 ± 0.05 Å and 1.32 ± 0.06 Å for the backbone and heavy atoms, respectively, excluding the terminal residues. Both structures have a very similar global fold, with all five helices conserved. Minor differences are observed in the positions of the Trp57 and Phe65 aromatic side-chains. The latter are closer together in the solution structure, as evidenced by the NOEs between Trp57 Hε1 and Phe65 Hζ, Phe65 Hζ and heme HAA1/HAA2, as well as Phe65 Hζ and Val38 Hγ1. According to the X-ray structure, these NOEs should either have much smaller intensities or not be observable at all. As a consequence, the center of the Phe65 aromatic ring is shifted away from the iron atom (distance ≈ 7.3 Å) compared with the X-ray structure (distance ≈ 6.4 Å). The origin of these subtle differences between the X-ray and the solution structures is not yet clear. They could be related to differences in pH (the pH of the crystallization medium was 4.5 [31] compared with the solution pH of 5.5 used here), leading to different protonation states, e.g. at the propionate head groups, or to packing effects present in the crystal. It will be interesting to observe whether the shifts in position of the Trp57 and Phe65 rings relative to the X-ray structure also occur in the solution structure of the oxidized form.

The orientation of the physiologically relevant lysine side-chains in the X-ray and the NMR structures is generally preserved in the χ1 dihedral angle; all other side-chain dihedral angles of these lysine residues are rather random in the NMR structures, because of their peripheral position and the relatively small number of NOEs.

Intramolecular hydrogen bonding

In the energy-minimized structures, the residues located in the α-helical regions are found to be involved in the typical CO–HN(i,i + 4) H-bonding. Some other H-bonds in the protein backbone reflect the loop structures. For example, Cys17 O to Gly27 HN and Lys19 O to Leu30 HN are typical for the double-loop structure following the N-terminal helix, Arg36 O to Trp57 HN, Gly54 O to Val38 HN and Trp57 O to Gly35 HN for the large loop connecting residues 35 and 57, and Leu66 O to Leu83 HN for the last loop (residues 66–83).

The propionate head groups of the heme moiety form a number of H-bonds in the X-ray structure [31]. Unfortunately, the carboxyl groups of the propionates are restrained only indirectly in the solution structure due to a lack of proton-derived NOEs. Accordingly, both are relatively disordered. Thus, it remains unclear whether the propionate head groups are under-defined in the structure calculations or actually display increased mobility in solution relative to the X-ray structure. Ala39 N, Tyr46 Oη and Trp57 Nε1 are the proton donors that are consistently close to the propionate A carboxyl group, but show inconsistent H-bonding geometries, while for the propionate D carboxyl group the following possible proton donors have been identified: Ser47 N, Ser47 Oγ, Thr76 Oγ1 and Lys77 N. The downfield shifts of the Ala39 HN (9.43 p.p.m.), Tyr46 Hη (10.27 p.p.m.), Ser47 HN (11.04 p.p.m.) and Trp57 Hε1 (11.65 p.p.m.) resonances could indicate their involvement in H-bonding or may be explained by the ring-current effect of the heme, as mentioned above. However, the detection of the Tyr46 Hη proton resonance is a good indication that the Tyr46 phenol group is involved in stable intramolecular H-bonding.


A biologically fully active 10.5 kDa fragment of the P. denitrificans cytochrome c552 was enriched with the 15N isotope and prepared for high-resolution NMR studies. Its solution structure in the reduced state was obtained through NOE-derived distance information. The average rmsd of the energy-minimized distance geometry ensemble is 0.56 ± 0.10 Å and 1.09 ± 0.09 Å for the backbone and heavy atoms, respectively. The structure is well-resolved in the larger part of the protein, with the exception of two outer loops. The overall fold is similar to that seen throughout the c-type cytochrome family. The X-ray structure differs only slightly in the position of certain aromatic side-chains and the heme propionate head groups. Implications of the NMR-derived solution structure for the biological function of the protein will have to await a comparison of the structural and dynamic differences between the reduced and oxidized forms of cytc.


This work was supported by the Deutsche Forschungsgemeinschaft SFB 472. P. P. acknowledges receipt of a Humboldt fellowship. We thank Ingrid Weber for help with the preparative work, as well as Dr Ulrich Günther and Dr Frank Löhr (all at J. W. Goethe-University of Frankfurt, Germany) for their valuable advice.


  1. *Present address: National Institute of Chemistry, Ljubljana, Slovenia.Note: the coordinates of the functional domain of Paracoccus denitrificans cytochrome c552 in the reduced state have been deposited in the RCSB Protein Data Bank with the PDB ID code 1C7M.