Eur. J. Biochem. 210,931 -936 (1992) 0FEBS 1992

Structural studies of Desulfovibrio vulgaris ferrocytochrome c3 by two-dimensional NMR David L. TURNbK', Carlos A. SALGUEIRO', Jean LeGALL3and Antonio V. XAVIER' Department of Chemistry, University of Southampton,England Centro de Tecnologia Quimica e Biol6gica and Universidade Nova de Lisboa, Portugal Department of Biochemistry, University of Georgia, Athens GA, USA (Received August 18, 1992) - EJB 92 1195

Two-dimensional NMR has been used to make specific assignments for the four haems in Desulfavibrio vulguris (Hildenborough) ferrocytochrome c3 and to determine their haem core architecture. The NMR signals from the haem protons were assigned according to type using two-dimensional NMR experiments which led to four sets of signals, one for each of the haems. Specific assignments were obtained by calculating the ring current shifts which arise from other haems and aromatic residues. Observation of interhaem NOEs confirmed the assignments and established that the relative orientation of the haems is identical to that found in the crystal structure of D. vulgaris (Miyazaki F.) ferricytochrome c3. Assignments were also made for all the aromatic residues cxcept for the haem ligands and F20, which is shifted under the main envelope of signals. The NOEs observed between these aromatic protons and haem protons confirm the similarity between the structures in solution and in the crystal. The assignments reported here are the basis for the cross-assignments of the four microscopic haem redox potentials to specific haems in the protein structure [Salgueiro, C . A,, Turner, D. L., Santos, H., LeGall, J. and Xavier, A. V. (1992) FEBS Lett., in the press].

achieving electronic/protonic energy transduction via a 'redox-Bohr' effect [3 - 51. The structural basis for cooperativity has been studied in several proteins; such studics are considered to be among the most successful achievements of molecular biophysics [6]. Since cytochrome c j is a small, monomeric, and soluble protein, which is diamagnetic in the reduced state, it is amenable to very detailed studies in a variety of solution conditions using high-resolution NMR [2, 7- 111. The use of NMR data together with the information from available X-ray structures [12- 141 should be extremely useful in elucidating the functional mechanism of this interesting protein. Low-spin monohaem cytochromes are among the proteins best studied by NMR [15]. The development of two-dimensional NMR spectroscopy has made it reasonably straightforward to assign the haem protons (Fig. 1, where the haem substituents are numbered according to the IUPAC-IUB nomenclature [16]) in the spectrum of the reduced (diamagnetic) state. Cytochrome c3 poses a much more challenging problem Corresponrlmce to A. V. Xavier, Centro de Tecnologia Quimica e since a detailed NMR study such as that needed to probe the Biolbgica, Apt. 127, P-2780 Oeiras, Portugal structural basis for the intricate mechanisms of cooperativity Abbreviations. c3DvMF, Desulj'ovibrio vulgaris (Miyazaki F . ) cytochrome c3 ; c ~ D v H ,Desulj'ovibrio vulgaris (Hildcnborough) requires not only that the proton signals should be assigned cytochrome c3; TOCSY, total correlation spcctroscopy; Hi, Hi', Mi', to a particular type of haem protons (e.g. haem methyl Mi', and Mi', haem moiety suhstituents, respectively: methine protons at meso Hi, etc.) but that they should also be assigned specifically positions 5, 10, 15 and 20; thioether methine protons, 3' and 8'; to one of the four haems in the three-dimensional structure. methyl groups, 2l, 7', 12' and 18', and thoether methyl groups, 32 A strategy for the specific assignment of haem protons in and 8'. multihaem ferrocytochromes c3 is described in this paper Cytochrome c3 is a small (1 3 kDa) tetrahaem protein only found in Desulfovibrio spp. where it couples the transfer of electrons resulting from the oxidation of organic substrates to the anaerobic respiration of these bacteria [l]. Each haem is covalently bound to two cysteine residues in a -C-X-X-(X)(X)-C-H- sequcnce. This histidine residue is the fifth coordinating ligand to the haem, the sixth one also being a histidine residue. The four c-type haems have different and fairly negative midpoint redox potentials. A comprehensive study of D. gigas cytochromc c j has shown that the four haems have a complex network of interacting redox potentials (homotropic cooperativity) as well as pH-dependent microscopic redox potentials (heterotropic cooperativity) [2, 31. These properties give the protein the necessary ability to function according to a mechanism in which two electrons are transferred in a synchronized manner as well as the transfer of electron(s) being coupled to the transfer of proton(s), thus

932

/

!I3\

(Miyazaki F.) crystal structure [ 131with the classical JohnsonBovey model [23]using eight pairs of current loops to represent each haem, with the parameters obtained by Cross and Wright [24]. Reference shifts of 9.65 ppm for rnesos, 3.59 ppm for methyls, 6.11 ppm for thiocther methines, and 2.22 ppm for thioether methyls were obtained by correcting the observed shifts in the horse cytochrome c spectrum for the effect of aromatic side chains using an average of the ring current shifts calculated from the reduced [25] and oxidised [26] crystal structures for the homologous protein from tuna.

2

RESULTS AND DISCUSSION Assignment of individual haem proton resonances

0-

0-

Fig. 1. Diagram of haem c numbered according to the IUPAC-IUB nomenclature [16). The full lines show the haem protons involved in the short-range connectivities which are easily detected in the 50-ms NOE spectrum, whereas the dashed lines indicate the long-range connectivities also observed in the 400-ms NOE spectrum.

together with its successful application to D. vulgaris (Hildenborough) cytochrome c3 (c,DvH).

MATERIALS AND METHODS Cytochrome cg from Desulfovibrio vulgaris (Hildenborough) bacteria was purified as previously described [ 171. For NMR studies, the sample was lyophilized several times with 99.9% 2 H 2 0 and resuspended in the same solvent to a concentration of approximately 2 mM. The pH was ajusted to 8.9 with either NaO'H or 2HC1 (pH value is the meter reading uncorrected for the isotope effects). Reduction of the protein was achieved under a hydrogen atmosphere in the presence of traces of Desulfovibrio gigas hydrogenase, which was purified as previously described [18]. All 'H two-dimensional spectra were recorded at 298 K on a 500-MHz Bruker AMX-500 spectrometer with the phasesensitive mode by the time proportional phase incrementation method [I 91. The NOE spectra [20] were measured with mixing times of 50, 200 and 400 ms, data size of 512 x 2048 and spectral width of 8065 H z . Total correlation spectroscopy (TOCSY) experiments [21, 221 were performed with the same data size and spectral width but with a mixing time of 60 ms and a spin-lock Geld of 10 kHz. Chemical shifts arc presented in parts per million (ppm) relative to sodium 3-(trimethyl~ilyl)-(2;2,3,3-~H~)propane sulfonate but formate was used as an internal reference. The ring current shifts were calculated using the Brookhaven Protein Data Bank file 2cdv.pdb for the crystal structure of oxidised cytochrome c 3 from D. vulgaris

The general strategy [27] applied to the present case may be outlined as follows. As shown in Fig. 1, meso protons present a characteristic pattern of short-range intrahaem connectivities: H15 protons are not connected to either mcthyl groups or thioether substituents; the H20 protons arc connected to two haem methyls (M2l and M18l); and the only ambiguity arises from H5 and H10 protons which both present connectivities with a thiocther methinc, a thioether methyl and one haem methyl group. To distinguish these last two meso protons the unambigous assignment of the J-coupled thioether methine - methyl connectivities together with a careful analysis of the NOE ones, between thioether methyls and haem methyl groups, must be carried out. Regions of the TOCSY and the 50-ms NOE spectra of the diamagnetic form of cRDvH are shown in Fig. 2. Together, these spectra allow the identification of a11 haem proton resonances according to type (Table 1). As stated above, nzeso protons H I 5 and H20 are immediately idcntified (Fig. 2B), leaving H5 and H10 which need further analysis. The J connectivities between the thioether protons give strong cross peaks in the TOCSY spectrum (Fig. 2A). Then, as the short-range intrahaem connections between the meso protons (H5 and H10) and both the thioether protons as well as the nearest haem methyl group are easily observed in the 50-ms NOE spectrum (Fig. 2B) the identification of meso resonances belonging to H5 or H10 is obtained. The ambiguity is lifted by the observation of the connections between M3'/ M2' and M8'/M7' groups (cf. Fig. l), which are already detectable in the NOE spectrum with short mixing times, and give strong cross peaks at longer mixing times (cf. Table 2) allowing the completion of the haem proton assignments. For the sake of clarity, a full set of intrahaem short-range connectivities for only one haem is shown in Fig. 2. All of the haem proton chemical shifts, except those of the propionates, are listed in Table 1, where the haems are numbered according to the position of the haem-ligated cysteine in the amino acid sequence [28]. NOE experiments carried out with mixing times of 200 ms (not shown) and 400 ms (Fig. 3) confirm the assignment of haem proton resonances to individual haems. Indeed, all the expected meso-to-mcthyl long-range connectivities are observed. In order not to crowd the figure, these connections are indicated in Table 2 together with a systematic comparison of the intrahaem cross peak intensities observcd at 400 ms with the distances taken from the crystal structure [13]. For short distanccs (less than 0.5 nm) all the connectivities are strong. It should be noted that the connectivities involving the pairs H20iH3' and HSiH8' are not always observable since the specific orientation of the thioether groups may lead to a relatively large internuclear distance. The proportional de-

933

F 76

0

*+

r

-7

-8

-9

Fig. 2. Parts of the 500-MHz 'H-TOCSY (A: 60-ms mixing time) and NOE (B: 50-ms mixing time) spectra of reduced c3DvH in 'H20, at 298 K, p2H 8.9. The TOCSY spectrum was transformed with Gaussian apodisation in Fz and cosine multiplication with zero-filling to 1024 points in Fl.The NOESY spectrum was transformed using a 5-Hz Lorentzian line broadening in F2 and a shifted sine-bell multiplication (SSB = 16) with zero-filling to 2048 points in PI. One-dimensional projections show the rcgions including the meso, the aromatic and the thioether methine protons (* - indicates the formate resonance). The boxes drawn in the TOCSY spectrum show the patterns arising from the indicated aromatic residues (Y43, Y65, Y66, H67, and F76); lines arc drawn to connect spectrum B to the cross-peaks which identify the scalar coupling in the thioether bridges of haem 111 in spectrum A (see text). In order not to overcrowd the figure. only the short-rangeconiicctivitics for mesa prolons belonging to haem I11 (identified according to Fig. 1j are shown by lines drawn in spectrum B, with the specific assignments given above the spcctrum. The arrow indicates thc region of thc HI5 nzcso-propionate cross pcaks.

crease of the cross-peak intensities with the distance and/or the mixing time confirm the above assignments. Cross peaks pairing meso protons of the same haem ( z 0.64 nm) were observed in two cases and are also included in Table 2. This information allows the haem resonances to be grouped in four sets, one due to each haem, but it is not sufficient to assign resonances to specific protons in the crystal structure. Cross assignment to the three-dimensional structure

Specific assignments to the haems in the crystal structure were made according to the pattern of shifts predicted for each haem rrom the calculated ring shifts due to the other haems and the aromatic residues (see Materials and Methods). Of the 24 possible permutations for the four sets of haem protons with respect to the crystal structure, one was clearly preferred (cf. Table 1 and Fig. 4) since all four haems simultaneously had their smallest root mean square deviation (rmsd)

from the calculated chemical shifts. The rmsd of the 48 shifts was 0.32 ppm, with deviations of 0.37 (haem I), 0.36 (haem II), 0.25 (haem 111) and 0.29 (haem IV). These deviations may be compared with the rms value of 0.209 ppm found by Cross and Wright after optimising the parameters used in the calculation of the ring current shifts of several protons [24]. The errors inherent in the model are caused by distortions of the haem planes in the crystal structure, but they do not suggest any substantial differences between the structures in solution and in the crystal. Thus, despite the approximate nature of the calculation, the magnitude of the shifts is sufficient to obtain an unambigous assignment. In particular, as depicted in Fig. 4 (cf. Table I), the calculation predicts that the meso proton with the largest upfield shift is an H10 proton belonging to haem I, which agrees with the experimental data in the sense that this meso proton is connected by an NOE to the thioether methyl group (M8') which is also strongly shifted upfield. Analogously, the thioether methine/methyl pair with the largest shift belongs to haem 11. This figure also shows

934 F, ppm

Y43

Y43

I

0 I

d

~

2

0

I-

~

F76

I

F7E

9

Y66

t

d

10

Fig. 3. Part of a 500-MHz 'H-NOESY spectrum of D . vulgaris (Hildenborough) ferrocytochrome cj in 'H20 showing connections to the mcso protons. A mixing time of 400 ms was used, other sample conditions and processing were as defined in Fig. 2. For the sake of clarity, the positions of long-range strong interhaem cross peaks are indicated by boxes labelled a - h (see Table 3). These represent: a, I1101-M2111; b, H10r-M3Zll; c, H101-M32111;d, H5111-M8ZI;e, H5111-H101;f, H20111-M82fV;g, H20111-H101V;h, H101V-M2'III. Cross-peaks arising from NOEs between haem meso protons and aromatic residues are labelled with their specific assignments.

Table 1. Chemical shifts of the haem protons in Desulfovibrio vulgaris (Hildenborough) ferrocytochrome c3, at p2H 8.9 and 27OC.

Assignment

Shift of Haem protons

I

11

111

1v

8.70 9.14 0.25 9.53 3.68 3.13 3.36 3.10 5.05 5.87 0.26 1.99

9.96 9.67 0.17 10.15 4.69 3.95 3.41 3.64 6.83 6.55 2.91 2.86

9.12 9.24 9.98 9.34 3.62 3.01 3.85 3.26 6.01 6.08 1.91 0.60

PPm

H5 1310 H15 H20 M2 M7' M12l M18' H3' HE1 M3' M8'

9.46 8.34 9.47 8.90 2.79 3.22 2.96 3.27 6.31 5.33 2.01 0.23

that the largest downfield shifts of haem methyl and thioether methyl groups belong to haem 111. The remaining haem should have an M82 group strongly shifted upfield which is also in agreement with the experirncntal data (Fig. 4). This assignment was then tested by examining the interhaem NOE connectivities, measured from the 50-ms, 200ms and 400-ms spectra, which were then compared with the

distanccs obtained from the crystal structure of c&MF (Table 3). Fig. 5 shows the strongest interhaem meso-methyl connections which are expected within the haem core; the positions of the corresponding peaks are depicted in Fig. 3 , together with the two expected meso-to-meso interhaem connectivities (H5111-H10,r 0.50 nm and H2OIlr-H1OIV, 0.58 nm). The decrease of the observable interactions at each mixing time with the internuclear distances and the decrease in intensity with the mixing time strongly supports the proposed assignment, despite the fact that the relationship between distance and cross-peak intensity is not quantitative because of slight variations in the degree of spin diffusion [29]. Additional confirmation of the haem assignments is obtained from the NOEs observed with the aromatic side chains. The aromatic protons of all but one of the non-axial ligand residues, i.e. three tyrosines (Y43, Y65, Y66), one histidine (H67) and one phenylalanine (F76) were identified in the TOCSY spectrum (Fig. 2A) and the cross-peaks observed between protons of aromatic residues and the nearest haem meso proton(s), shown in Fig. 3, are in full agreement with the crystal structure. The other aromatic residue is a phenylalanine (F20) which is a conserved one. It could not be assigned at this stage, most probably because its resonances are shifted to the crowded region of the spectrum since it is quite close to two of the haems (I and 111). Furthermore, it should not be surprising that, unlike the other non-coordinated aromatic residues, its signals are broadened by immobilization since in c3DvMF its aromatic ring is inserted

935 Table 2. Comparison of the long-range (400-ms) intrahaem NOE crosspeak intensities with the distances between meso and thioether protons, and carbon atom of the haem and thioether methyl groups, measured from the c3DvMFcrystal structure 1131. For thioether proton groups, a range of distances is defined since they vary according to the torsion angle. The range of intensities is defined as follows:S, strong; M, medium; W, weak; VW, very weak; 0, obscured by other peaks; -, not observable.

Table 3. Comparison of interhaem cross-peak intensities a t 50, 200 and 400 ms with the distances between meso protons and the carbon atom of haem and thioether methyl groups measured from the e3DvMFcrystal structure. NOE intensities were:$ strong; M, medium; W, weak; VW, vcry weak; 0, obscured by other peaks; -, not observable. Conneclivities

Distance

NOE intensities at 400ms

Connectivities

Distances

200ms

50ms

Haem NOE intensities

nm I

I1

111

IV

S S M M S S M M M M W -

S

S

S M

S M

S S

-

-

M S S M M M M

M S S M M M M

nm M2' H3l M2' M3' M2' H5 H5 H8' H5 M8* M7' H8' M7' M8' M7' H10 M12' HI5 HI5 M18' H20 H3' H20 M32 H5 H20 H15 H20

0.31 -0.42 0.34- 0.36 0.54 0.58-0.61 0.55 -0.59 0.38 -0.42 0.30 -0.35 0.54 0.53 0.53 0.54- 0.61 0.57-0.58 0.64 0.64

M 0 W S S M M M M M

-

-

-

vw

w

vw

0.37 0.37 0.41 0.42 0.44 0.53 0.54 0.62 0.63 0.63 0.63 0.65 0.65 0.67 0.67 0.69 0.70 0.71 0.76 0.77 0.77 0.78 0.79 0.79 0.80 0.81 0.84 0.86

M W

vw

W

Observed shift (ppm) Fig. 4. Comparison between the deviations of the measured haem proton shifts from their reference values (see text) and the ring current shifts calculated using the c3DvMF crystal structure 1131 with the classical Johnson-Bovey model 1231. (X) Haem I ; (a)haem IT; ( W ) haem 111; (a)haem IV.

between the planes of haem I and that of a histidine ligand of haem 111 [13]. CONCLIJSIONS

Specific assignment of NMR resonances to individual haem protons of c3DvH allows a preliminary comparison of its structurc in solution (reduced) with that of c3DvMF in the

Fig. 5. flaem core structure of c3DvMF.Thc figure was generated via computer graphics using the X-ray coordinates [13]. The dashed lines indicale the strongest intrahaem meso-methyl NOE connectivities observed for c3DvH (scc also Tablc 3 and Fig. 3).

936 crystal (oxidised). Since all the observed cross-peaks for NOES between haem protons belonging to different haems and between these and the aromatic residue protons are fully consistent with the crystal structure, both the relative position of the four haems and the general folding of the polypeptide chain should be quite similar. It is not possible at this stage to attempt a discussion of structural details. The results presented should be regarded as the cornerstones for subsequent studics which allow the crossassignment of four haems in the three-dimensional structure according to the order of their redox potential [30]. However, it is already possible to show that the architecture of the four haems in cytochromes c3 which are as different in amino-acid sequence and physiological activity as those isolated from D. vulgaris (Hildenborough) and D. haculatus (Norway) [I] is the same. Indeed, the NOES observed between protons belonging to different haems (interhaem) are identical for the two proteins [31]. This was not expected since the X-ray structures published for the c3 D . buculutus (Norway) [12]and D.vulgaris (Miyazaki) [13] proteins differ in the relative orientation of the haems: two of the haems (I and 1V) are rotated by 180" around an axis passing through meso protons H5 and H15. Furthermore, both the X-ray structure [32] and the NMR NOE data (Piqarra-Pereira, M . A., Turner, D. L., LeGall, J. and Xavier, A. V., unpublished work) for a third cytochromc c 3 , isolated from D . gigas, show that the same architecture is maintained. It is quite intriguing that proteins with so little similarity (only 12 amino acid residues plus those which provide the ligation to the haems) can all stabilize that particular architecture. We thank I. Pacheco and F. Matos for technical collaboration

8. Moura, J . J . G.. Saatos, H., Moura, I., LeGall, J., Moore, G. R.,

9. 10.

11.

32. 13.

14.

15.

16. 17. 18.

19.

20. 21. 22. 23.

REFERENCES 1. LeGall, J. & Fauquc, G. (1988) in Biologj ofanaerobic mirroorganisms (Zehnder?A. J. B., ed.) pp. 587-639, John Wiley and Sons, New York. 2. Santos, H., Moura, J. J. G., Moura, I., LcGall. J. & Xavier, A. V. (1984) Eur. J . Riochem. 141, 283-296. 3. Coletta, M., Catarino, T., LeGall, J. & Xavier, A. V. (1991) Eur. J . Bioclrem. 202. 1101 -1106. 4. Xavicr, A. V. (1986) J . Inorg. Biochem. 28, 239-243. 5. Catarino, T.. Coletta, M., LeGall, J. & Xavier. A. V. (1991) Eur. J . Biochem. 202, 1107 - 1113. 6. Perutz, M. (1990) Mechanism c?f cooperativity and allosteric regulnfion in proteins, Cambridge University Press, Cambridge. 7. Dobson, C. M., Hoyle, N. J . , Geraldes, C. F., Rruschi, M., LeGall, J., Wright, P. E. & Williams. R . J. P. (1974) Narure (Lond.) 249,425-429.

24. 25. 26. 27.

28. 29. 30. 31.

Williams, R. J. P. & Xavier, A . V. (1982) Eur. J . Biochem. 127, 151 -155. Fan, K.,Akutsu, H., Kyogoku, Y. &Niki, K. (1990) Biochemistry 29,2257 - 2263. Park, J-S., Kano, K., Niki. K. & Akutsu, H. (1993) FEBS Lett. 285, 149-151. Park, J-S., Kano, K., Morimoto, Y., Higuchi, Y., Yasuoka, N., Ogata, M., Niki, K. & Akutsu, H. (1991) J . Biomol. N M R I , 271 - 282. Pierrot, M., Haser, K., Frey, M., Payan, F. & Astier, J-P. (1982) J . Riol. Chem. 257, 14341 -14348. Higushi, Y . , Kusonoki, M.. Matsuura, Y., Yasuoka, N. & Kakudo, M. (1984) J . Mol. Biol. 172, 109-139. Morimoto, Y.: Tani, T., Okumura. H., IIiguchi, Y. & Yasuoka, N. (1991) J . Biochem. (Tokyo) 110, 532-540. Moore G. R. & Pettigrew, G. W . (1990) Cytocromesc. Evolutionary, structural and physicochemical aspects, Springer-Verlag, Berlin. IUPAC-IUB Joint Commision on Biochemical Nomenclature (1988) Eur. .I. Biochem. 178, 277-328. LcGall, J., Bruschi. H. M. & DerVartanian, D. V. (1971) Biochim. Biuphys. Acta 234, 499-512. LeGall, J., Ljungdahl, P.. Moura, I., Peck, H. D., JR, Xavier, A. V., Moura, J. J. G., Teixeira, M., Huynh, B. H. & DerVartanian, D. V. (1 982) Einchem. Riophju. Res. Commun. 106.610-616. Marion, D. & Wiithrich K . (1983) Biochern. Biophys. Res. Commun. 113. 967 -974. Jeener, J., Meicr, B. H., Bachmann, P. & Ernst, I