Eur. J. Biochem. 77, 53-60 (1977)
Nuclear-Magnetic-Resonance Studies of Pseudomonas aeruginosa Cytochrome c-55 1 Geoffrey R. MOORE, Robert C. PITT, and Robert J. P. WILLIAMS Inorganic Chemistry Laboratory, University of Oxford (Received December 31, 1976)
Nuclear magnetic resonance (NMR) spectroscopy was used to study Pseudomonas aeruginosa cytochrome c-551. Assignments of resonances to specific residues have been made. A low-resolution X-ray structure was used to aid assignments. A structural comparison was made between P. aeruginosa cytochrome c-551 and mammalian cytochrome c, based on comparisons of NMR data.
In order to elucidate the molecular mechanisms of protein function, outline structures of protein molecules are required. This has previously implied a reliance on the methods of X-ray crystallography for determining three dimensional pictures of the structure of proteins. Recent advances in technology have enabled high-resolution nuclear magnetic resonance (NMR) spectroscopy to yield detailed structural information of, for example, hen egg-white lysozyme [I], mammalian cytochrome c [2] and D. vulgaris cytochrome c3 [3]. In addition to static structural information, NMR spectroscopy is capable of yielding information concerning the dynamic aspects of protein structures in solution. The cytochromes c constitute a group of proteins much studied by both X-ray crystallographers and NMR spectroscopists, yet the mechansims of action are still far from fully understood. This is not due to a lack of structural information; sequences of more than a hundred cytochromes c [4] and structures for seven cytochromes c are known [3,5]. All proposed mechansims based on specific amino acid sidechain residues have been shown to be inadequate when protein sequences have been compared, as groups necessary for a particular electron-transfer mechanism are not always conserved. We are left with proposals that electrons tunnel through the heme edge [6] and/or through the protein matrix [7,8]. Much evidence has been accumulated suggesting that electron transfer between proteins takes place within a complex formed between them [8]. The nature of this interaction, and the mechansim for electron transfer are not known, but it is likely that the more Abbreviation. NMR, nuclear magnetic resonance
dynamic structural information, which NMR spectroscopy is capable of providing, is required to understand it. In this paper we shall describe our work on cytochrome c-551, some of which has been reported at Hamburg [9], and we shall compare our results with a preliminary report of more limited scope [lo]. P. aeruginosa cytochrome c-551 has been studied by X-ray crystallographic methods, and a low-resolution structure constructed [ll]. This confirmed the earlier observation, from NMR data [12], that it had the same ironligandsasmammalian cytochromec. Theultimate aim of our studies of P. aeruginosa cytochrome c-551 by NMR is to determine the nature of the complex between cytochrome e-551 and azurin, and the cytochrome c-551 binding site for cytochrome oxidase. Before proceeding to these investigations it is necessary to assign the resonances in the spectra of these proteins to particular protons. It is, incidentally, in the process of this assignment that much structural information is obtained. Investigation of the NMR spectrum of azurin is being carried out in this laboratory by Dr H. A. 0.Hill [13,14]. Here we report on the assignment of the NMR spectrum of cytochrome c-551 prior to an account of the study of the cytochrome-c-551 . azurin complex.
MATERIALS AND METHODS Pseudomonas aeruginosa cytochrome c-55 1 was obtained from Dr J. Melling (Microbiological Research Establishment, Porton Down, Salisbury SP4 OJG, England). It was purified according to the method of Ambler [15], as described (with modification) by Melling [16]. Buffer solutions were 0.05 M
54
NMR Studies of P. arrziginosa Cytochrome c-551
I
I
I
I
I
I
10
9
8
7
6
5
6 (ppm)
Fig. 1 . Aromatic region of the convolution difference spectra phosphate buffer, p H 7.0, 57 " C
of
( a ) ,ferricytochrome c - S j l and ( b ).frrrocytochrome c-5.51, in 0.1 M deuteruted
in acetic acid, the pH being adjusted by addition of ammonia solution. The p H of the cytochrome c551 solution was adjusted to 3.9, and it was adsorbed as a dense band onto a column packed with CM-cellulose cation exchanger (CM23, supplied by Whatman, Ltd, Springfield Mill, Maidstone, Kent, England) which had been equilibrated in buffer at the same pH. It was then eluted using buffer at pH 4.45 to separate it from any azurin present, and deposited on a freshly prepared column at pH 3.9, from which it was eluted as a concentrated band using buffer at pH 6.0. Finally, it was eluted through a Sephadex G-25 (supplied by Pharmacia Fine Chemicals AB, Uppsala, Sweden) column to remove buffer. All columns were 300 mm x 15 mm, also supplied by Pharmacia. The protein was then lyophilized, and dissolved in 0.1 M deuterated phosphate buffer, pH 7.0. Ferrocytochrome c-551 samples were obtained by adding a slight excess of ascorbic acid solution to a solution of ferricytochrome c-551 under argon. Ferrocytochrome c-551 concentrations were determined by inspection of the 551-nm absorption using a Pye Unicam SP8000 ultraviolet spectrophotometer. The 551-nm absorption coefficient was taken to be 28.3 x lo3 M-' cm-' [17]. The NMR spectra were obtained using a Bruker 270-MHz spectrometer operating in the Fourier transform mode, with an Oxford Instruments Co. magnet. Free induction decays were collected in a Nicolet 1085 computer, in which mathematical manipulations were carried out. Convolution difference [ 181, spin-decoupling [19] and spin-echo double resonance [20,21] were carried out as previously described. Acetone and dioxan were used as internal standards but all chemical shifts are quoted in parts per million (ppm) downfield from sodium 2,2-dimethyl-2-silapentane-5-sulphonate. pH values quoted refer to uncorrected meter readings.
RESULTS AND DISCUSSION Assignment of Resonunces to Amino-Acid Type The aromatic region of the convolution difference spectra of ferricytochrome c-551 and ferrocytochrome c-551 in 0.1 M deuterated phosphate buffer, pH 7.0, at 57 "C are shown in Fig. 1. The multiplet nature of many resonances may clearly be seen. P. ueruginosa cytochrome c-553 contains one tyrosine, two phenylalanine, two tryptophan and one histidine residues [15]. The histidine provides the fifth iron ligand and its resonances will be shifted out of the aromatic region of the spectra. Thus the two singlet resonances at 6.80 ppm and 7.13 ppm in the oxidised (ox) spectrum, and at 7.77 ppm and 7.28 ppm in the reduced (red) spectrum may be assigned to the C-2 protons of the two tryptophan residues. Further assignment of the tryptophan resonances comes from identification of the families of coupled resonnces which are unique to tryptophan groups (Table 1). Some of these resonances are readily observable in the convolution difference spectrum (Fig. l), for example the one proton doublet at 8.07ppm (ox) and the one proton triplet at 6.38 ppm (red). Spin-decoupling was employed to discover the resonance positions of other members of these two Families. Some of the coupled tryptophan resonances overlap with other resonances, and in these cases, difference spectra were employed to observe the effect of double irradiation. Difference spectra for triplet to doublet decouplings are often not clear because the difference is only slightly greater than the average noise level (see, for example Fig. 3, of [22]). In order to clarify these results, the recently developed technique of spin-echo double resonance [21] was employed for the assignment of the overlapping tryptophan resonances. This is illustrated in Fig.2. Here, the difference for a
55
G. R. Moore, R. C. Pitt, and R. J. P. Williams
Table .I. Assignmenls of amino acid resonances in spectra of P. aeruginosa cytockrome c-551 Chemical shifts arc measured downfield from sodium 2,2-dimethyl-2-silapentane-S-sulphonate; solutions were 5 mM in 0.1 M deuterated phosphate buffer at pH 7.0, 57 ”C Amino acid
Proton type
Shift in spectrum oxidised
Residue
Primary position
reduced
PPm Tryptophan
Phenylalanine
PPm
C-4 or C-7 C-5 or C-6 C-6 or C-5 C-7 or C-4 (2-2
8.07 7.36 7.03 7.25 7.13
8.02 1.28 7.12 7.61 7.28
C-4 or C-7 C-5 or C-6 C-6 or C-5 C-7 or C-4 C-2
6.81 6.05 6.21 7.18 6.80
736 6.38 5.81 7.08 1.71
ortho meta para
7.08 6.96 7.26
~
Trp-77
’ Trp-56
Phe-7 or Phe-34
7.12
Alanine
Alanine/Threonine
Valine
CH CH3
2.43 6.15
7/CH3 y’CH3 Alanine
7.68 7.20 7.28 7.54 7.32
0
m P
7.35 7.40 7.40
aCH BCH3
3.76 1.36 - 1.46
PCH3
yCH3 y’CH3
2.25 0.97 1.02
1
4.86 1.51 1.51
BCH
C-4 C-5 C-6 c-7 C-2
Val-1 3
BCH3
5.59 1.51
1 1
Ala-14b
aC H PCHi
3.76 1.36- 1.46
Methionine
- SCH3
2.12
1.69
Met-22
- SCHj
2.12
Methionine
SCH3 yC H yCH
16.6
-2.91 -3.53 0.87 -2.73 -0.51
aCH
-
PCH
PCH Histidine
~
C-4
0.15
1
-- SCHj
PCH
2.12 2.64 2.64 2.12 2.12
C-4
7.0 -7.6
yc H
Met-61
His-I6
yCH BCH
Resonance not identified. Assignment of resonances to specific residue is less certain.
triplet decoupling to a doublet is between a positive peak (triplet) decoupling to a negative peak (doublet), leading to a large positive peak in the difference spectrum. The benzenoid resonances of both tryptophan groups in both oxidation states have been identified using these double-resonance techniques, as have those of one of the phenylalanine groups (see Table 1). In a further examination of the aromatic region spectra of ferrocytochrome c-551 were obtained over the range 7 - 77 “C, and spectra of ferricytochrome c-551 over the range 17-67 “C. In neither case were resonances of aromatic groups in slow or intermediate exchange between 180 “C orientations observed [23]. Some temperature-dependent resonances were ob-
served in the aromatic region of the spectra of ferricytochrome c-551 and ferrocytochrome c-551. Many of these resonances may be assigned to NH protons by the fact that they appear only in spectra of freshly prepared solutions. N H proton resonances are much more persistent in spectra of ferrocytochrome c-551 than of ferricytochrome c-551, a phenomenon which has also been noted in studies of mammalian cytochrome c [24]. This implies that N H protons exchange with solvent much more readily in oxidised than in reduced cytochromes of this type. These and other differences in physical properties between ferricytochromes c and ferrocytochromes c have been taken to indicate that there is a conformational difference
56
NMK Studies of P. aeruginosa Cytochrome c-551
h
C
I
I
I
I
I
10
9
8
7
6
6 (wm)
2
1
0
6 (ppm)
Fig. 2. Assignment oftryptophan resonances. (a) Aromatic region convolution difference spectrum of ferrocytochrome c-551; (b) CarrPurcell A (T = 60 ms) spectrum; (c) Carr-Purcell A difference spectrum between (b) and a Carr-Purcell spectrum with irradiation at 8.02 ppm. Conditions as for Fig. 1
Fig. 3. ( a ) Aliphatic region of the convolution difierence spectrum of ferrocytochrome c-5.51 and ( h ) Carr-Purcell spectrum of the Same region, showing a methionine CH, singlet at 1.7ppm. Conditions as for Fig. 1
between the two states, ferrocytochrome c being more compact [2,25]. Such may also be the case for cytochrome c-551. Integration of the resonance intensities was performed on a spectrum of ferrocytochrome c-551 in which the NH resonances had been fully exchanged by maintaining the sample at 90 "C for 2 h. The total intensity of resonances observed in the aromatic region of this spectrum corresponds to 29 f 1 protons. Two one-proton resonances have been assigned to a-CH protons, leaving 27 k 1 protons. The total number of aromatic CH protons whose resonances are expected to fall in this region is 28. We have positively identified the resonances of the meso protons and of two tryptophans and one phenylalanine groups (19 protons), leaving9 unidentified proton resonances. Neither the resonances of the remaining phenylalanine ( 5 protons) nor those of the tyrosine (4 protons) have been identified positively. There are, however, a number of peaks in the spectra of ferricytochrome c-551 and ferrocytochrome c-551 (see Fig. I), corresponding to a total of 8 f 1 protons which have not yet been assigned. One such peak, of 5 f protons intensity, occurs at 7.22 pprn (ox) and 7.65 ppm (red). If the ortho and meta (and para) protons of the tyrosyl (phenylalanyl) residue were magnetically equivalent, the resulting peak would be of 4(5) proton intensity and would show no multiplet structure. A positive assignment will only be possible if magnetic inequivalence can be induced for these protons. Experiments using paramagnetic shift probes are being undertaken with this in mind.
Finally, there is one peak at 5.82 ppm (red), overlapping a tryptophan peak, which appears to be a twoproton doublet. The peak to which it is coupled, either a doublet (tyrosine) or a triplet (phenylalanine) could be shifted under the residual H'HO peak. These assignment have been supported by comparative studies on a series of cytochrome c-551 (Ambler, Moore, Pitt and Williams, unpublished). Resonances of aliphatic amino acids are normally more difficult to assign, because their large numbers, and extensive and similar coupling patterns, lead to strongly overlapping peaks. In favourable cases, however, it is possible to assign the resonances of some aliphatic amino acids. The aliphatic region of the convolution difference spectrum of ferrocytochrome c551 is shown in Fig.3, and compared with the CarrPurcell A (z = 60 ms) spectrum. The three-proton singlet at 1.67 ppm, more clearly observable in the Carr-Purcell A spectrum [20], is assigned to the free methionine residue (Met-22). Coupled methyl resonances are more difficult to assign [26]. However, spin-decoupling irradiation of the one-proton quartet at 5.59 ppm in the spectrum of ferricytochrome c-551 result in a three-proton doublet decoupling to a singlet at 1.51 ppm (Fig. 4). This coupling pattern identifies those resonances as arising from an alanine or threonine residue (see below). Similarly, irradiation of the peak at 4.86 ppm causes a six-proton doublet at 1.51 ppm to decouple to a singlet. Irradiation of the six-proton doublet at 1.51 ppm decouples the peak at 4.86 ppm to a one-proton singlet. Barring coincidental
G. R. Moore, R. C. Pitt, and R. J. P. Williams
d
1.7
1.5 6 (PPm)
1.3
Fig. 4. Decoupling of methyl resonances. (a) A small region of the convolution difference spectrum of ferrocytochrome c-551. (b) The same with irradiation at 4.86 ppm. (c) Difference of a- b. (d) Difference between (a) and a spectrum with irradiation at 5.59 ppm. Conditions as for Fig. 1
overlap of the methyl resonances and their coupled CH resonances, this unusual coupling pattern can only arise from a valine (or leucine) residue when coupling is not observed between the PCH (or yCH) and the aCH (or BCH). Ja-p for valine is 3.7 Hz whereas Jp-y for leucine is 6.5 Hz [27]. We therefore tentatively assign these resonances to a valine residue. Either bond distortion or short spin-lattice relaxation times could account for the lack of observed coupling. Singlet aCH resonances have previously been observed in spectra of lysozyme [28] and mammalian cytochrome c [2]. Similarly, in the spectrum of ferrocytochrome c551, irradiation of the one-proton quartets at 5.96 ppm and 6.15 ppm resulted in three proton doublets at 1.85 pprn and 2.43 pprn respectively decoupling to singlets. These resonance patterns clearly arise from alanine or threonine residues. Assignment to Spec f i e Residues
Assignment of resonances to specific residues supplies direct information concerning the molecular structure ; however, some structural information is required to make unambiguous assignments. This minimal necessary information may be derived either
57
from X-ray crystallographic data or by comparison of the spectrum with spectra of similar proteins of known structure. In the case of P. aeruginosa cytochrome c-551 assignments here will be based on comparison with NMR data from mammalian cytochromes c [2] (and unpublished results of Campbell, Moore and Williams) and on a low-resolution electrondensity map of cytochrome c-551 [ll]. A parallel step in assisting specific assignment is to determine the extent of the perturbations (ringcurrent, pseudocontact and contact shifts) experienced by each resonance. It should prove possible to crossassign all resonances between the two oxidation states, getting independent structural parameters for many of them. This procedure is greatly helped by examining the spectra not only of the separate iron(I1) and iron(II1) oxidation states but also of mixtures of the two. For two species undergoing chemical exchange [29], the rate of exchange compared to the chemical shift differences between resonances determines the character of the NMR spectrum. If the exchange rate is slow, a mixture of the two species produces a spectrum which is simply the two superimposed spectra of each species. In such cases cross-saturation techniques may well be applicable for cross-assignment [30]. At fast rates of exchange, a single sharp resonance appears at an intermediate energy position determined solely by the relative concentrations of each of the two species. For cytochromes in fast exchange titration of the oxidised species with a reducing agent obviously allows resonances to be cross-assigned. For P. aeruginosa cytochrome c-551 a study of the heme. His-16 and Met-61 resonances showed that many were in intermediate exchange between oxidation states [lo]. Addition of as little as 1 of ferricytochrome c-551 to a solution of ferrocytochrome c-551 resulted in appreciable broadening of the heme meso, heme methyl, Met-61 and His-16 resonances. From this broadening the bimolecular rate constant was determined to be 1 . 2 lo7 ~ M-' s-l [lo]. Such a rate is sufficient to place many other less shifted resonances in fast exchange; Fig.5 S ~ O W Sa series of spectra obtained at 27 "C for mixtures of oxidised and reduced cytochrome c-551. Many of the resonances titrate between the oxidised and reduced resonance position without appreciable broadening, for example the oneproton doublet at 8.07 pprn (ox) and 8.02 ppm (red). The resonance positions depend linearly on the ratio of ferricytochrome c-551 to ferrocytochrome c-551 present (Fig. 6). Cross-assignments obtained are presented in Table 1. Confirmation of the predicted [lo] chemical shift value for a contact-shifted heme mesoproton resonance was obtained. The resonance identified at 8.92 ppm (ox, 42 "C) was predicted to be at z 8.7 ppm (ox, 42 "C). However a Met-61 methylene proton resonance predicted to be at z 0 ppm (ox, 42 "C) was identified at -2.65 ppm (ox, 42 "C).
58
NMR Studies of P. ueruginosu Cytochrome (-551
I
I
I
I
I 6
Fig. 5. Titrution between .ferricyrochrome c-551, anclferrocytochrome c-551. Figures on the right refer to the percentage of ferrocytochrome c-551 present. Temperature 27 "C, otherwise conditions as for Fig. 1. Labelled resonances are (a) heme me.so C H ; (b) Trp-77 C-4 or C-7; (c) 4 - 5-proton intensity peak tentatively assigned to the equivalent protons of phenylalanine or tyrosine; (d) Trp-56 C-2; (e) Trp-77 C-2; (9 Trp-56 C-4 or C-7; (g) phenylalanine metu; (h) Trp-56 C-5 or C-6; (i) Trp-56 C-6 or C-5
10.0
c
4
-.-.
10.0
18.0
b
I
5.5
._._
P
.-.-.-
t 5.5
4
._ .-.__.
.__.
.-.--._!_
./.-. '--',!, 1.9
i 0
I
25
50
75
'I,, redu ced
100
Ferrocytochrome c - 551
2 7 2.8
2.9
3 0 3.1
3.2
3.3
3.4 3 5
10'1 T ( K ' )
Fig. 6. Graph showing the chemical shifts of vurious resonances us u function ojpercentu~e,ferrocvtochromrc-551. Conditions and labelling as for Fig. 5 with Q) Met-22 S-CH3
Fig. 7. Temperature-~epen~en~e oj some resonances of' ,ferricytoclzrumr c-551. Labelling as for Fig. 5 and 6
The temperature dependence of the ferricytochrome c-551 resonances (Fig. 7) is in agreement with the cross-assignment data (Fig. 6). With increasing temperature most pseudocontact-shifted resonances
shift back towards their diamagnetic position (pseudocontact shift is proportional to l / T [31]. There are some anomalies however. None of the resonances shown in Fig.7 extrapolate back to their expected
59
G. R. Moore, R. C. Pitt, and R. J. P. Williams
Table 2. Tryptophan resonances of reduced cytochromes c Cytochrome
Chemical shift doublet
triplet
triplet
doublet
6.68 1.67 6.38
5.74 5.75 5.84
7.07 7.10 7.10
PPm Horse Tuna P. aeruginosa c-551
7.60 7.57 7.36
diamagnetic position at infinite temperatures, and some of the resonances do not shift linearly with (temperature)-'. Such behaviour has also been observed for the resonances of horse ferricytochrome c [32]. The temperature dependence of the contactshifted heme resonances of horse ferricytochrome c indicate that there is not a uniquely defined spin state for the iron electrons. A mixture of spin states could also account for the irregular temperature dependence of the pseudocontact-shifted resonances. From the temperature-dependence of the ferricytochrome c-551 resonances (Fig. 7), and the nature of the changes in the NMR spectrum with changing oxidation state (Fig. 5), we can place certain groups with respect to the heme [2]. Thus the - SCH3 group of Met22 is in the shift cone (over the plane of the heme). Similarly, one of the tryptophan residues is nearer to the heme than the other. From the preliminary X-ray structure Trp-56 is close to the heme while Trp-77 is not. We therefore assign the tryptophan resonances accordingly. Note that the Trp-56 resonances exhibit a very similar pattern to those of Trp-59 in mammalian cytochromes c (Table 2) which lies on the edge of the shift cones in both oxidation states [2]. This indicates that the tryptophan-heme orientation in these proteins is similar. It may well be that this is a general feature of the structure of class I cytochromes c [33]. We therefore note with surprise that, in sequence alignments of mammalian cytochromes c and cytochrome c-551, the Trp-56 residue of cytochrome c-551 is placed between residues 75 and 76 of mammalian cytochromes c [ l l ] . This alignment is based on X-ray studies which indicate that the extra 20-25 amino acid residues in mammalian cytochromes c, compared with cytochrome c-551, form a loop between residues 38 and 57 [ l l ] . In view of this the conservation of the tryptophan-heme orientation appears even more remarkable. Assignment of the identified phenylalanine resonances (Table 1) cannot yet be made. However it is clear that this residue is not close to the heme. The resonances are shifted from their primary positions (Table l), but the shifts are not sufficient to allow any
further definite structural conclusions to be drawn at this stage. The identified valine resonances have been assigned to Val-13. As the resonances suffer big shifts which are oxidation-state-dependent the residue is obviously close to the heme. Val-13 is fixed in position by virtue of the Cys-12 and Cys-15 thioether linkages to the heme. The resonances of the parallel Ala-15 in horse ferricytochrome c [2] have been firmly assigned, and these resonances have a similar downfield shift. The quartet resonance at 5.59 ppm (ox), assigned earlier to an alanine or threonine residue may now be assigned to Ala-14 by a similar argument. Experiments involving N M R probes and sequence comparisons, now underway, are required to determine whether it is in general possible to detect amino acid residues, placed between the heme-linked groups in this way. One of the findings from N M R spectra of proteins has been the quantitative determination of aromatic sidechain mobility. The mobility of aromatic amino acids, first observed in lysozyme [34], has been investigated in a number of proteins. Parallel studies allow us to state that the one phenylalanine clearly assigned in the spectrum of cytochrome c-551 is flipping about the Cg-Cy bond. In view of the fact that little temperature-dependence is seen in the rest of the aromatic spectra in the reduced state of the protein and that we see all of the protons at the lowest temperatures, it is exceedingly likely that all phenylalanine and tyrosine residues undergo fast flipping. The mobility of aliphatic amino acid sidechains is generally less easy to observe. Nevertheless, the considerable mobility of various groups, such as Val-109 and Ile-98 of lysozyme, has been studied [35]. In the spectrum of ferricytochrome c-551, the equivalence of the two methyl groups of Val-13 suggests that there is similar rapid motion about the Ca- Cg bond. CONCLUSIONS This study has shown that NMR spectroscopy can yield much detailed structural information for a protein with a minimum of structural input. The structure of P . ueruginosa cytochrome c-551 contains many features common to the structure of mammalian cytochromes c despite having 20 - 25 fewer amino acids. We wish to thank the Medical Research Council and the Science Research Council for their support. This is a contribution from the Oxford Enzyme Group.
REFERENCES 1. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R. SOC. Lond. A Math. Phys. Sci. 345, 23-40. 2. Moore, G. R. & Williams, R. J. P. (1977) Marseille Conference on Electron Transfer Systems, C.N.R.S. Paris, in the press.
60
G. R. Moore, R. C. Pitt, and R. J. P. Williams: NMR Studies of P. aeruginosa Cytochrome c-551
3. Dobson, C. M., Hoyle, N. J., Geraldes, C. F. Wright, P. E., Williams, R. J. P., Bruschi, M. & Le Gall, J. (1974) Nature (Lond.j 249,425 - 429. 4. Dayhoff, M. 0. (1973) in Atlas of Protein Sequence and Structure, vol. 5 suppl., National Biomedical Research Foundation, Washington D.C. 5. Dickerson, R. E. & Timkovich, R. (1975) in The Enzymes (Boyer, P., ed.) pp. 397-547, Academic Press, New York. 6. Salemme, F. R., Kraut, J. & Kamen, M. D. (1973) J. Biol. Chem. 248, 7701 - 7716. 7. DeVault, D., Parkes, J. H. &Chance, B. (1967) Nature (Lond.) 215, 642- 644. 8. Moore, G. R. & Willams, R. J. P. (1976) Coord. Chem. Rev. 18,125- 197. 9. Moore, G. E. &Williams, R. J. P. (1976) Proc. 10th Int. Congr. Biochem., p. 143. 10. Keller, R. M., Wiithrich, K. & Pecht, I. (1976) FEBS Lett. 70, 180-183. 11. Dickerson, R. E., Timkovich, R. & Almassy, R. J. (1976) J. Mol. Biol. 100,473-497. 12. McDonald, C. C., Phillips, W. D., Le Gall, J. & Vinogradov, S. (1970) Abstr. 4th Int. Con5 on Magnetic Resonance in Biological Systems, p. 18. 13. Hill, H. A. O., Leer, J. C., Smith, B. E., Storm, C. B. & Ambler, R. P. (1976) Biochem. Biophys. Res. Commun. 70, 331-338. 14. Hill, H. A. O., Smith, B. E., Storm, C. B. & Ambler, R. P. (1976) Biochem. Biophys. Res. Commun. 70, 783-790. 15. Ambler, R. P. (1963) Biochem. J . 89, 341 -349. 16. Parr, S. R., Barker, D., Greenwood, C., Phillips, B. W. & Melling, J. (1976) Biochem. J. 157, 423-430. 17. Horio, T., Higashi, T., Sasgawa, M., Kusai, K., Nakai, M. & Okuniki, K. (1960) Biochem. J. 77, 194-201. 18. Campbell, I. D., Dobson, C. M., Williams, R. J. P. & Xavier, A. V. (1973) J. Magn. Res. 11, 172-181.
19. Dobson, C. M., Moore, G. R. & Williams, R. J. P. (1975) FEBS Lett. 51, 60-65. 20. Campbell, 1. D., Dobson, C. M., Williams, R. J. P. & Wright, P. E. (1975) FEBS Lett. 57, 96-99. 21. Campbell, I. D. & Dobson, C. M. (1975) .I. Chem. Soc. Chem. Commun. 750-751. 22. Moore, G. R. & Williams, R. J. P. (1975) FEBSLett. 53, 334338. 23. Campbell, I. D., Dobson, C. M., Moore, G. R., Perkins, S. J. & Williams, R. J. P. FEBS Lett. 70, 96-99. 24. Patel, D. J. & Canuel, L. L. (1976) Proc. Natl Acad. Sci. U.S.A. 73, 1398-1402. 25. Margoliash, E. & Schejter, A. (1966) Adu. Protein Chem. 21, 113- 286. 26. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (3975) Proc. R . Soc. Lond. A . Math. Phys. Sci. 345, 41 -51. 27. Roberts, G . C. K. & Jardetzky, 0. (1970) Adu. Protein Chem. 24,447 - 545. 28. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R. SOC.Lond. B. Biol. Sci. 189, 485-502. 29. Pople, J. A,, Schneider, W. G. & Bernstein, H. J. (1959) HighResolution Nuclear Magnetic Resonance, McGraw-Hill, New York. 30. Redfield, A. G. & Gupta, R. K. (1971) Cold Spring Harbour Symp. Quant. Biol. 36,405-411. 31. Wiithrich, K. (1970) Struct. Bonding, 8, 53 - 121. 32. Williams, R. J. P., Moore, G. R . & Wright, P. E. (1977) Vancouver Conference on Bio-Inorganic Chemistry, pp. 369 -401, John Wiley, New York. 33. Ambler, R. P. (1977) Marseille Conjerence on Electron Transjer Systems, C.N.R.S. Paris, in the press. 34. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R . SOC.Lond. B Biol. Sci. 189, 503-509. 35. Dobson, C. M. (1975) D. Phil. Thesis, Oxford.
G. R. Moore, R. C. Pitt, and R . J. P. Williams, Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, Great Britain, OX1 3QU
Nuclear-Magnetic-Resonance Studies of Pseudomonas aeruginosa Cytochrome c-55 1 Geoffrey R. MOORE, Robert C. PITT, and Robert J. P. WILLIAMS Inorganic Chemistry Laboratory, University of Oxford (Received December 31, 1976)
Nuclear magnetic resonance (NMR) spectroscopy was used to study Pseudomonas aeruginosa cytochrome c-551. Assignments of resonances to specific residues have been made. A low-resolution X-ray structure was used to aid assignments. A structural comparison was made between P. aeruginosa cytochrome c-551 and mammalian cytochrome c, based on comparisons of NMR data.
In order to elucidate the molecular mechanisms of protein function, outline structures of protein molecules are required. This has previously implied a reliance on the methods of X-ray crystallography for determining three dimensional pictures of the structure of proteins. Recent advances in technology have enabled high-resolution nuclear magnetic resonance (NMR) spectroscopy to yield detailed structural information of, for example, hen egg-white lysozyme [I], mammalian cytochrome c [2] and D. vulgaris cytochrome c3 [3]. In addition to static structural information, NMR spectroscopy is capable of yielding information concerning the dynamic aspects of protein structures in solution. The cytochromes c constitute a group of proteins much studied by both X-ray crystallographers and NMR spectroscopists, yet the mechansims of action are still far from fully understood. This is not due to a lack of structural information; sequences of more than a hundred cytochromes c [4] and structures for seven cytochromes c are known [3,5]. All proposed mechansims based on specific amino acid sidechain residues have been shown to be inadequate when protein sequences have been compared, as groups necessary for a particular electron-transfer mechanism are not always conserved. We are left with proposals that electrons tunnel through the heme edge [6] and/or through the protein matrix [7,8]. Much evidence has been accumulated suggesting that electron transfer between proteins takes place within a complex formed between them [8]. The nature of this interaction, and the mechansim for electron transfer are not known, but it is likely that the more Abbreviation. NMR, nuclear magnetic resonance
dynamic structural information, which NMR spectroscopy is capable of providing, is required to understand it. In this paper we shall describe our work on cytochrome c-551, some of which has been reported at Hamburg [9], and we shall compare our results with a preliminary report of more limited scope [lo]. P. aeruginosa cytochrome c-551 has been studied by X-ray crystallographic methods, and a low-resolution structure constructed [ll]. This confirmed the earlier observation, from NMR data [12], that it had the same ironligandsasmammalian cytochromec. Theultimate aim of our studies of P. aeruginosa cytochrome c-551 by NMR is to determine the nature of the complex between cytochrome e-551 and azurin, and the cytochrome c-551 binding site for cytochrome oxidase. Before proceeding to these investigations it is necessary to assign the resonances in the spectra of these proteins to particular protons. It is, incidentally, in the process of this assignment that much structural information is obtained. Investigation of the NMR spectrum of azurin is being carried out in this laboratory by Dr H. A. 0.Hill [13,14]. Here we report on the assignment of the NMR spectrum of cytochrome c-551 prior to an account of the study of the cytochrome-c-551 . azurin complex.
MATERIALS AND METHODS Pseudomonas aeruginosa cytochrome c-55 1 was obtained from Dr J. Melling (Microbiological Research Establishment, Porton Down, Salisbury SP4 OJG, England). It was purified according to the method of Ambler [15], as described (with modification) by Melling [16]. Buffer solutions were 0.05 M
54
NMR Studies of P. arrziginosa Cytochrome c-551
I
I
I
I
I
I
10
9
8
7
6
5
6 (ppm)
Fig. 1 . Aromatic region of the convolution difference spectra phosphate buffer, p H 7.0, 57 " C
of
( a ) ,ferricytochrome c - S j l and ( b ).frrrocytochrome c-5.51, in 0.1 M deuteruted
in acetic acid, the pH being adjusted by addition of ammonia solution. The p H of the cytochrome c551 solution was adjusted to 3.9, and it was adsorbed as a dense band onto a column packed with CM-cellulose cation exchanger (CM23, supplied by Whatman, Ltd, Springfield Mill, Maidstone, Kent, England) which had been equilibrated in buffer at the same pH. It was then eluted using buffer at pH 4.45 to separate it from any azurin present, and deposited on a freshly prepared column at pH 3.9, from which it was eluted as a concentrated band using buffer at pH 6.0. Finally, it was eluted through a Sephadex G-25 (supplied by Pharmacia Fine Chemicals AB, Uppsala, Sweden) column to remove buffer. All columns were 300 mm x 15 mm, also supplied by Pharmacia. The protein was then lyophilized, and dissolved in 0.1 M deuterated phosphate buffer, pH 7.0. Ferrocytochrome c-551 samples were obtained by adding a slight excess of ascorbic acid solution to a solution of ferricytochrome c-551 under argon. Ferrocytochrome c-551 concentrations were determined by inspection of the 551-nm absorption using a Pye Unicam SP8000 ultraviolet spectrophotometer. The 551-nm absorption coefficient was taken to be 28.3 x lo3 M-' cm-' [17]. The NMR spectra were obtained using a Bruker 270-MHz spectrometer operating in the Fourier transform mode, with an Oxford Instruments Co. magnet. Free induction decays were collected in a Nicolet 1085 computer, in which mathematical manipulations were carried out. Convolution difference [ 181, spin-decoupling [19] and spin-echo double resonance [20,21] were carried out as previously described. Acetone and dioxan were used as internal standards but all chemical shifts are quoted in parts per million (ppm) downfield from sodium 2,2-dimethyl-2-silapentane-5-sulphonate. pH values quoted refer to uncorrected meter readings.
RESULTS AND DISCUSSION Assignment of Resonunces to Amino-Acid Type The aromatic region of the convolution difference spectra of ferricytochrome c-551 and ferrocytochrome c-551 in 0.1 M deuterated phosphate buffer, pH 7.0, at 57 "C are shown in Fig. 1. The multiplet nature of many resonances may clearly be seen. P. ueruginosa cytochrome c-553 contains one tyrosine, two phenylalanine, two tryptophan and one histidine residues [15]. The histidine provides the fifth iron ligand and its resonances will be shifted out of the aromatic region of the spectra. Thus the two singlet resonances at 6.80 ppm and 7.13 ppm in the oxidised (ox) spectrum, and at 7.77 ppm and 7.28 ppm in the reduced (red) spectrum may be assigned to the C-2 protons of the two tryptophan residues. Further assignment of the tryptophan resonances comes from identification of the families of coupled resonnces which are unique to tryptophan groups (Table 1). Some of these resonances are readily observable in the convolution difference spectrum (Fig. l), for example the one proton doublet at 8.07ppm (ox) and the one proton triplet at 6.38 ppm (red). Spin-decoupling was employed to discover the resonance positions of other members of these two Families. Some of the coupled tryptophan resonances overlap with other resonances, and in these cases, difference spectra were employed to observe the effect of double irradiation. Difference spectra for triplet to doublet decouplings are often not clear because the difference is only slightly greater than the average noise level (see, for example Fig. 3, of [22]). In order to clarify these results, the recently developed technique of spin-echo double resonance [21] was employed for the assignment of the overlapping tryptophan resonances. This is illustrated in Fig.2. Here, the difference for a
55
G. R. Moore, R. C. Pitt, and R. J. P. Williams
Table .I. Assignmenls of amino acid resonances in spectra of P. aeruginosa cytockrome c-551 Chemical shifts arc measured downfield from sodium 2,2-dimethyl-2-silapentane-S-sulphonate; solutions were 5 mM in 0.1 M deuterated phosphate buffer at pH 7.0, 57 ”C Amino acid
Proton type
Shift in spectrum oxidised
Residue
Primary position
reduced
PPm Tryptophan
Phenylalanine
PPm
C-4 or C-7 C-5 or C-6 C-6 or C-5 C-7 or C-4 (2-2
8.07 7.36 7.03 7.25 7.13
8.02 1.28 7.12 7.61 7.28
C-4 or C-7 C-5 or C-6 C-6 or C-5 C-7 or C-4 C-2
6.81 6.05 6.21 7.18 6.80
736 6.38 5.81 7.08 1.71
ortho meta para
7.08 6.96 7.26
~
Trp-77
’ Trp-56
Phe-7 or Phe-34
7.12
Alanine
Alanine/Threonine
Valine
CH CH3
2.43 6.15
7/CH3 y’CH3 Alanine
7.68 7.20 7.28 7.54 7.32
0
m P
7.35 7.40 7.40
aCH BCH3
3.76 1.36 - 1.46
PCH3
yCH3 y’CH3
2.25 0.97 1.02
1
4.86 1.51 1.51
BCH
C-4 C-5 C-6 c-7 C-2
Val-1 3
BCH3
5.59 1.51
1 1
Ala-14b
aC H PCHi
3.76 1.36- 1.46
Methionine
- SCH3
2.12
1.69
Met-22
- SCHj
2.12
Methionine
SCH3 yC H yCH
16.6
-2.91 -3.53 0.87 -2.73 -0.51
aCH
-
PCH
PCH Histidine
~
C-4
0.15
1
-- SCHj
PCH
2.12 2.64 2.64 2.12 2.12
C-4
7.0 -7.6
yc H
Met-61
His-I6
yCH BCH
Resonance not identified. Assignment of resonances to specific residue is less certain.
triplet decoupling to a doublet is between a positive peak (triplet) decoupling to a negative peak (doublet), leading to a large positive peak in the difference spectrum. The benzenoid resonances of both tryptophan groups in both oxidation states have been identified using these double-resonance techniques, as have those of one of the phenylalanine groups (see Table 1). In a further examination of the aromatic region spectra of ferrocytochrome c-551 were obtained over the range 7 - 77 “C, and spectra of ferricytochrome c-551 over the range 17-67 “C. In neither case were resonances of aromatic groups in slow or intermediate exchange between 180 “C orientations observed [23]. Some temperature-dependent resonances were ob-
served in the aromatic region of the spectra of ferricytochrome c-551 and ferrocytochrome c-551. Many of these resonances may be assigned to NH protons by the fact that they appear only in spectra of freshly prepared solutions. N H proton resonances are much more persistent in spectra of ferrocytochrome c-551 than of ferricytochrome c-551, a phenomenon which has also been noted in studies of mammalian cytochrome c [24]. This implies that N H protons exchange with solvent much more readily in oxidised than in reduced cytochromes of this type. These and other differences in physical properties between ferricytochromes c and ferrocytochromes c have been taken to indicate that there is a conformational difference
56
NMK Studies of P. aeruginosa Cytochrome c-551
h
C
I
I
I
I
I
10
9
8
7
6
6 (wm)
2
1
0
6 (ppm)
Fig. 2. Assignment oftryptophan resonances. (a) Aromatic region convolution difference spectrum of ferrocytochrome c-551; (b) CarrPurcell A (T = 60 ms) spectrum; (c) Carr-Purcell A difference spectrum between (b) and a Carr-Purcell spectrum with irradiation at 8.02 ppm. Conditions as for Fig. 1
Fig. 3. ( a ) Aliphatic region of the convolution difierence spectrum of ferrocytochrome c-5.51 and ( h ) Carr-Purcell spectrum of the Same region, showing a methionine CH, singlet at 1.7ppm. Conditions as for Fig. 1
between the two states, ferrocytochrome c being more compact [2,25]. Such may also be the case for cytochrome c-551. Integration of the resonance intensities was performed on a spectrum of ferrocytochrome c-551 in which the NH resonances had been fully exchanged by maintaining the sample at 90 "C for 2 h. The total intensity of resonances observed in the aromatic region of this spectrum corresponds to 29 f 1 protons. Two one-proton resonances have been assigned to a-CH protons, leaving 27 k 1 protons. The total number of aromatic CH protons whose resonances are expected to fall in this region is 28. We have positively identified the resonances of the meso protons and of two tryptophans and one phenylalanine groups (19 protons), leaving9 unidentified proton resonances. Neither the resonances of the remaining phenylalanine ( 5 protons) nor those of the tyrosine (4 protons) have been identified positively. There are, however, a number of peaks in the spectra of ferricytochrome c-551 and ferrocytochrome c-551 (see Fig. I), corresponding to a total of 8 f 1 protons which have not yet been assigned. One such peak, of 5 f protons intensity, occurs at 7.22 pprn (ox) and 7.65 ppm (red). If the ortho and meta (and para) protons of the tyrosyl (phenylalanyl) residue were magnetically equivalent, the resulting peak would be of 4(5) proton intensity and would show no multiplet structure. A positive assignment will only be possible if magnetic inequivalence can be induced for these protons. Experiments using paramagnetic shift probes are being undertaken with this in mind.
Finally, there is one peak at 5.82 ppm (red), overlapping a tryptophan peak, which appears to be a twoproton doublet. The peak to which it is coupled, either a doublet (tyrosine) or a triplet (phenylalanine) could be shifted under the residual H'HO peak. These assignment have been supported by comparative studies on a series of cytochrome c-551 (Ambler, Moore, Pitt and Williams, unpublished). Resonances of aliphatic amino acids are normally more difficult to assign, because their large numbers, and extensive and similar coupling patterns, lead to strongly overlapping peaks. In favourable cases, however, it is possible to assign the resonances of some aliphatic amino acids. The aliphatic region of the convolution difference spectrum of ferrocytochrome c551 is shown in Fig.3, and compared with the CarrPurcell A (z = 60 ms) spectrum. The three-proton singlet at 1.67 ppm, more clearly observable in the Carr-Purcell A spectrum [20], is assigned to the free methionine residue (Met-22). Coupled methyl resonances are more difficult to assign [26]. However, spin-decoupling irradiation of the one-proton quartet at 5.59 ppm in the spectrum of ferricytochrome c-551 result in a three-proton doublet decoupling to a singlet at 1.51 ppm (Fig. 4). This coupling pattern identifies those resonances as arising from an alanine or threonine residue (see below). Similarly, irradiation of the peak at 4.86 ppm causes a six-proton doublet at 1.51 ppm to decouple to a singlet. Irradiation of the six-proton doublet at 1.51 ppm decouples the peak at 4.86 ppm to a one-proton singlet. Barring coincidental
G. R. Moore, R. C. Pitt, and R. J. P. Williams
d
1.7
1.5 6 (PPm)
1.3
Fig. 4. Decoupling of methyl resonances. (a) A small region of the convolution difference spectrum of ferrocytochrome c-551. (b) The same with irradiation at 4.86 ppm. (c) Difference of a- b. (d) Difference between (a) and a spectrum with irradiation at 5.59 ppm. Conditions as for Fig. 1
overlap of the methyl resonances and their coupled CH resonances, this unusual coupling pattern can only arise from a valine (or leucine) residue when coupling is not observed between the PCH (or yCH) and the aCH (or BCH). Ja-p for valine is 3.7 Hz whereas Jp-y for leucine is 6.5 Hz [27]. We therefore tentatively assign these resonances to a valine residue. Either bond distortion or short spin-lattice relaxation times could account for the lack of observed coupling. Singlet aCH resonances have previously been observed in spectra of lysozyme [28] and mammalian cytochrome c [2]. Similarly, in the spectrum of ferrocytochrome c551, irradiation of the one-proton quartets at 5.96 ppm and 6.15 ppm resulted in three proton doublets at 1.85 pprn and 2.43 pprn respectively decoupling to singlets. These resonance patterns clearly arise from alanine or threonine residues. Assignment to Spec f i e Residues
Assignment of resonances to specific residues supplies direct information concerning the molecular structure ; however, some structural information is required to make unambiguous assignments. This minimal necessary information may be derived either
57
from X-ray crystallographic data or by comparison of the spectrum with spectra of similar proteins of known structure. In the case of P. aeruginosa cytochrome c-551 assignments here will be based on comparison with NMR data from mammalian cytochromes c [2] (and unpublished results of Campbell, Moore and Williams) and on a low-resolution electrondensity map of cytochrome c-551 [ll]. A parallel step in assisting specific assignment is to determine the extent of the perturbations (ringcurrent, pseudocontact and contact shifts) experienced by each resonance. It should prove possible to crossassign all resonances between the two oxidation states, getting independent structural parameters for many of them. This procedure is greatly helped by examining the spectra not only of the separate iron(I1) and iron(II1) oxidation states but also of mixtures of the two. For two species undergoing chemical exchange [29], the rate of exchange compared to the chemical shift differences between resonances determines the character of the NMR spectrum. If the exchange rate is slow, a mixture of the two species produces a spectrum which is simply the two superimposed spectra of each species. In such cases cross-saturation techniques may well be applicable for cross-assignment [30]. At fast rates of exchange, a single sharp resonance appears at an intermediate energy position determined solely by the relative concentrations of each of the two species. For cytochromes in fast exchange titration of the oxidised species with a reducing agent obviously allows resonances to be cross-assigned. For P. aeruginosa cytochrome c-551 a study of the heme. His-16 and Met-61 resonances showed that many were in intermediate exchange between oxidation states [lo]. Addition of as little as 1 of ferricytochrome c-551 to a solution of ferrocytochrome c-551 resulted in appreciable broadening of the heme meso, heme methyl, Met-61 and His-16 resonances. From this broadening the bimolecular rate constant was determined to be 1 . 2 lo7 ~ M-' s-l [lo]. Such a rate is sufficient to place many other less shifted resonances in fast exchange; Fig.5 S ~ O W Sa series of spectra obtained at 27 "C for mixtures of oxidised and reduced cytochrome c-551. Many of the resonances titrate between the oxidised and reduced resonance position without appreciable broadening, for example the oneproton doublet at 8.07 pprn (ox) and 8.02 ppm (red). The resonance positions depend linearly on the ratio of ferricytochrome c-551 to ferrocytochrome c-551 present (Fig. 6). Cross-assignments obtained are presented in Table 1. Confirmation of the predicted [lo] chemical shift value for a contact-shifted heme mesoproton resonance was obtained. The resonance identified at 8.92 ppm (ox, 42 "C) was predicted to be at z 8.7 ppm (ox, 42 "C). However a Met-61 methylene proton resonance predicted to be at z 0 ppm (ox, 42 "C) was identified at -2.65 ppm (ox, 42 "C).
58
NMR Studies of P. ueruginosu Cytochrome (-551
I
I
I
I
I 6
Fig. 5. Titrution between .ferricyrochrome c-551, anclferrocytochrome c-551. Figures on the right refer to the percentage of ferrocytochrome c-551 present. Temperature 27 "C, otherwise conditions as for Fig. 1. Labelled resonances are (a) heme me.so C H ; (b) Trp-77 C-4 or C-7; (c) 4 - 5-proton intensity peak tentatively assigned to the equivalent protons of phenylalanine or tyrosine; (d) Trp-56 C-2; (e) Trp-77 C-2; (9 Trp-56 C-4 or C-7; (g) phenylalanine metu; (h) Trp-56 C-5 or C-6; (i) Trp-56 C-6 or C-5
10.0
c
4
-.-.
10.0
18.0
b
I
5.5
._._
P
.-.-.-
t 5.5
4
._ .-.__.
.__.
.-.--._!_
./.-. '--',!, 1.9
i 0
I
25
50
75
'I,, redu ced
100
Ferrocytochrome c - 551
2 7 2.8
2.9
3 0 3.1
3.2
3.3
3.4 3 5
10'1 T ( K ' )
Fig. 6. Graph showing the chemical shifts of vurious resonances us u function ojpercentu~e,ferrocvtochromrc-551. Conditions and labelling as for Fig. 5 with Q) Met-22 S-CH3
Fig. 7. Temperature-~epen~en~e oj some resonances of' ,ferricytoclzrumr c-551. Labelling as for Fig. 5 and 6
The temperature dependence of the ferricytochrome c-551 resonances (Fig. 7) is in agreement with the cross-assignment data (Fig. 6). With increasing temperature most pseudocontact-shifted resonances
shift back towards their diamagnetic position (pseudocontact shift is proportional to l / T [31]. There are some anomalies however. None of the resonances shown in Fig.7 extrapolate back to their expected
59
G. R. Moore, R. C. Pitt, and R. J. P. Williams
Table 2. Tryptophan resonances of reduced cytochromes c Cytochrome
Chemical shift doublet
triplet
triplet
doublet
6.68 1.67 6.38
5.74 5.75 5.84
7.07 7.10 7.10
PPm Horse Tuna P. aeruginosa c-551
7.60 7.57 7.36
diamagnetic position at infinite temperatures, and some of the resonances do not shift linearly with (temperature)-'. Such behaviour has also been observed for the resonances of horse ferricytochrome c [32]. The temperature dependence of the contactshifted heme resonances of horse ferricytochrome c indicate that there is not a uniquely defined spin state for the iron electrons. A mixture of spin states could also account for the irregular temperature dependence of the pseudocontact-shifted resonances. From the temperature-dependence of the ferricytochrome c-551 resonances (Fig. 7), and the nature of the changes in the NMR spectrum with changing oxidation state (Fig. 5), we can place certain groups with respect to the heme [2]. Thus the - SCH3 group of Met22 is in the shift cone (over the plane of the heme). Similarly, one of the tryptophan residues is nearer to the heme than the other. From the preliminary X-ray structure Trp-56 is close to the heme while Trp-77 is not. We therefore assign the tryptophan resonances accordingly. Note that the Trp-56 resonances exhibit a very similar pattern to those of Trp-59 in mammalian cytochromes c (Table 2) which lies on the edge of the shift cones in both oxidation states [2]. This indicates that the tryptophan-heme orientation in these proteins is similar. It may well be that this is a general feature of the structure of class I cytochromes c [33]. We therefore note with surprise that, in sequence alignments of mammalian cytochromes c and cytochrome c-551, the Trp-56 residue of cytochrome c-551 is placed between residues 75 and 76 of mammalian cytochromes c [ l l ] . This alignment is based on X-ray studies which indicate that the extra 20-25 amino acid residues in mammalian cytochromes c, compared with cytochrome c-551, form a loop between residues 38 and 57 [ l l ] . In view of this the conservation of the tryptophan-heme orientation appears even more remarkable. Assignment of the identified phenylalanine resonances (Table 1) cannot yet be made. However it is clear that this residue is not close to the heme. The resonances are shifted from their primary positions (Table l), but the shifts are not sufficient to allow any
further definite structural conclusions to be drawn at this stage. The identified valine resonances have been assigned to Val-13. As the resonances suffer big shifts which are oxidation-state-dependent the residue is obviously close to the heme. Val-13 is fixed in position by virtue of the Cys-12 and Cys-15 thioether linkages to the heme. The resonances of the parallel Ala-15 in horse ferricytochrome c [2] have been firmly assigned, and these resonances have a similar downfield shift. The quartet resonance at 5.59 ppm (ox), assigned earlier to an alanine or threonine residue may now be assigned to Ala-14 by a similar argument. Experiments involving N M R probes and sequence comparisons, now underway, are required to determine whether it is in general possible to detect amino acid residues, placed between the heme-linked groups in this way. One of the findings from N M R spectra of proteins has been the quantitative determination of aromatic sidechain mobility. The mobility of aromatic amino acids, first observed in lysozyme [34], has been investigated in a number of proteins. Parallel studies allow us to state that the one phenylalanine clearly assigned in the spectrum of cytochrome c-551 is flipping about the Cg-Cy bond. In view of the fact that little temperature-dependence is seen in the rest of the aromatic spectra in the reduced state of the protein and that we see all of the protons at the lowest temperatures, it is exceedingly likely that all phenylalanine and tyrosine residues undergo fast flipping. The mobility of aliphatic amino acid sidechains is generally less easy to observe. Nevertheless, the considerable mobility of various groups, such as Val-109 and Ile-98 of lysozyme, has been studied [35]. In the spectrum of ferricytochrome c-551, the equivalence of the two methyl groups of Val-13 suggests that there is similar rapid motion about the Ca- Cg bond. CONCLUSIONS This study has shown that NMR spectroscopy can yield much detailed structural information for a protein with a minimum of structural input. The structure of P . ueruginosa cytochrome c-551 contains many features common to the structure of mammalian cytochromes c despite having 20 - 25 fewer amino acids. We wish to thank the Medical Research Council and the Science Research Council for their support. This is a contribution from the Oxford Enzyme Group.
REFERENCES 1. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R. SOC. Lond. A Math. Phys. Sci. 345, 23-40. 2. Moore, G. R. & Williams, R. J. P. (1977) Marseille Conference on Electron Transfer Systems, C.N.R.S. Paris, in the press.
60
G. R. Moore, R. C. Pitt, and R. J. P. Williams: NMR Studies of P. aeruginosa Cytochrome c-551
3. Dobson, C. M., Hoyle, N. J., Geraldes, C. F. Wright, P. E., Williams, R. J. P., Bruschi, M. & Le Gall, J. (1974) Nature (Lond.j 249,425 - 429. 4. Dayhoff, M. 0. (1973) in Atlas of Protein Sequence and Structure, vol. 5 suppl., National Biomedical Research Foundation, Washington D.C. 5. Dickerson, R. E. & Timkovich, R. (1975) in The Enzymes (Boyer, P., ed.) pp. 397-547, Academic Press, New York. 6. Salemme, F. R., Kraut, J. & Kamen, M. D. (1973) J. Biol. Chem. 248, 7701 - 7716. 7. DeVault, D., Parkes, J. H. &Chance, B. (1967) Nature (Lond.) 215, 642- 644. 8. Moore, G. R. & Willams, R. J. P. (1976) Coord. Chem. Rev. 18,125- 197. 9. Moore, G. E. &Williams, R. J. P. (1976) Proc. 10th Int. Congr. Biochem., p. 143. 10. Keller, R. M., Wiithrich, K. & Pecht, I. (1976) FEBS Lett. 70, 180-183. 11. Dickerson, R. E., Timkovich, R. & Almassy, R. J. (1976) J. Mol. Biol. 100,473-497. 12. McDonald, C. C., Phillips, W. D., Le Gall, J. & Vinogradov, S. (1970) Abstr. 4th Int. Con5 on Magnetic Resonance in Biological Systems, p. 18. 13. Hill, H. A. O., Leer, J. C., Smith, B. E., Storm, C. B. & Ambler, R. P. (1976) Biochem. Biophys. Res. Commun. 70, 331-338. 14. Hill, H. A. O., Smith, B. E., Storm, C. B. & Ambler, R. P. (1976) Biochem. Biophys. Res. Commun. 70, 783-790. 15. Ambler, R. P. (1963) Biochem. J . 89, 341 -349. 16. Parr, S. R., Barker, D., Greenwood, C., Phillips, B. W. & Melling, J. (1976) Biochem. J. 157, 423-430. 17. Horio, T., Higashi, T., Sasgawa, M., Kusai, K., Nakai, M. & Okuniki, K. (1960) Biochem. J. 77, 194-201. 18. Campbell, I. D., Dobson, C. M., Williams, R. J. P. & Xavier, A. V. (1973) J. Magn. Res. 11, 172-181.
19. Dobson, C. M., Moore, G. R. & Williams, R. J. P. (1975) FEBS Lett. 51, 60-65. 20. Campbell, 1. D., Dobson, C. M., Williams, R. J. P. & Wright, P. E. (1975) FEBS Lett. 57, 96-99. 21. Campbell, I. D. & Dobson, C. M. (1975) .I. Chem. Soc. Chem. Commun. 750-751. 22. Moore, G. R. & Williams, R. J. P. (1975) FEBSLett. 53, 334338. 23. Campbell, I. D., Dobson, C. M., Moore, G. R., Perkins, S. J. & Williams, R. J. P. FEBS Lett. 70, 96-99. 24. Patel, D. J. & Canuel, L. L. (1976) Proc. Natl Acad. Sci. U.S.A. 73, 1398-1402. 25. Margoliash, E. & Schejter, A. (1966) Adu. Protein Chem. 21, 113- 286. 26. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (3975) Proc. R . Soc. Lond. A . Math. Phys. Sci. 345, 41 -51. 27. Roberts, G . C. K. & Jardetzky, 0. (1970) Adu. Protein Chem. 24,447 - 545. 28. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R. SOC.Lond. B. Biol. Sci. 189, 485-502. 29. Pople, J. A,, Schneider, W. G. & Bernstein, H. J. (1959) HighResolution Nuclear Magnetic Resonance, McGraw-Hill, New York. 30. Redfield, A. G. & Gupta, R. K. (1971) Cold Spring Harbour Symp. Quant. Biol. 36,405-411. 31. Wiithrich, K. (1970) Struct. Bonding, 8, 53 - 121. 32. Williams, R. J. P., Moore, G. R . & Wright, P. E. (1977) Vancouver Conference on Bio-Inorganic Chemistry, pp. 369 -401, John Wiley, New York. 33. Ambler, R. P. (1977) Marseille Conjerence on Electron Transjer Systems, C.N.R.S. Paris, in the press. 34. Campbell, I. D., Dobson, C. M. & Williams, R. J. P. (1975) Proc. R . SOC.Lond. B Biol. Sci. 189, 503-509. 35. Dobson, C. M. (1975) D. Phil. Thesis, Oxford.
G. R. Moore, R. C. Pitt, and R . J. P. Williams, Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford, Great Britain, OX1 3QU
