Indian Journal of Chemistry Vol. 41A, January 2002, pp.39-45
Dynamics in the electron transfer complexes of plastocyanin with cytochrome f
and cytochrome c
Marcellus Ubbink
Leiden Institute of Chemistry, Leiden University, Gorlaeus Laboratories, P.O. Box 9502, NL-2300 RA Leiden, The Netherlands
e-mail: m.ubbink@chem.leidenuniv.nl Received 25 July 2001
The complex of the photosynthetic redox proteins plastocyanin and cytochrome f is compared with that of plastocyanin and cytochrome c. On the basis of nuclear magnetic resonance chemical shift data, it is concluded that the dynamics of the two complexes differs greatly. The complex of plastocyanin and cytochrome f exists predominantly in a single orientation, stabilised by both electrostatic and hydrophobic interactions. The complex of plastocyanin and cytochrome c consists of an ensemble of orientations of a purely electrostatic nature. The results explain the differences in reaction rates between the complexes.
Introduction
The complex of plastocyanin and cytochrome f
Cytochrome
f
and plastocyanin are two proteins involved in the photosynthetic electron transfer chain that links photosystems 2 and 1 (ref. 1). Cytochromef
is part of the cytochrome bf complex2-4, which is located in the thylakoid membrane in chloroplasts. It has a large soluble N-terminal domain, anchored in the membrane by a single a-helix5
. In cruciferous plants, the soluble domain is cleaved off during isolation of cytochrome
f
6, enabling the purification of a soluble form of cytochromeJ,
which has been widely used for kinetic experiments, crystallisation and NMR. The soluble domain (28 kDa) has an elongated shape and the single, covalently bound haem (c-type) is located in the middle of the protein and is coordinated by a His residue as well as the N- terminal amino group 7. Cytochrome
f
acts as the natural reductant of plastocyanin, which is a soluble type I blue copper proteinB,9. The cytochrome j7plastocyanin reaction is an excellent system for the study of interprotein electron transfer because of its obvious physiological relevance and its high reaction rate (2 x lOB Mls- I at I = 100 mMlO,ll), despite the small driving force and the modest binding constant (7 X 103 MI at I=
100 mM) (ref. 10), It can be shown that the electron transfer rate (ket) in the complex of cytochromef
and plastocyanin at this ionic strength must be 104 S-I or higherlO, From NMR experi- mentsl2, a lower limit ofthe dissociation rate constant(1<off) can be set at 4 x 103 S-I. This means that the interaction is extraordinarily short-lived and that the reaction is probably neither diffusion-controlled nor activation-controlled, since ket ;::: 1<off.
The reaction rate between cytochrome
f
andplastocyanin depends heavily on the ionic strength and demonstrates a bell-shaped curve, with an optimum at 30 mM ionic strength 13 and a strong decrease in rate at higher ionic strength 11.14.15. This is a clear indication that electrostatic attraction is important for fast electron transfer, which is further supported by competition studies with high-valent
1 1 15-17 Th 1 I . .
mo ecu es . e ower rates at very ow 101lIC strength suggest that the initial binding orientation(s) is not active in electron transfer and some kind of rearrangement is required. More evidence for this has been provided by a study that demonstrated that electron transfer is fast in the electrostatic complex but absent in the cross-linked complexlB.
Chemical 1I).0difications 19-22 and mutagenesis 10,23-27 of both proteins have been carried out to determine which residues are responsible for the charge interactions. Most studies indicate that both 'large' (residues 42-45) and 'small' (residues 59-61) acidic patches of plastocyanin contribute to the binding.
Lysine residues 58, 65, 66, 187, 188 and 189 of cytochrome
f
are important in the binding of plastocyanin24,27, These residues form a positive patch, which is responsible for the electrostatic interactions.The complex of plastocyanin and cytochrome c The non-physiological reaction between cyto- chrome c and plastocyanin has been studied for comparison. As in the cytochrome fplastocyanin complex, the partners show strong electrostatic attraction, with the reduction rate of plastocyanin depending on the total charge of cytochrome C28. NMR studies demonstrated that the complex is dynamic29
.3o
and that the electron transfer rate from ferrous cytochrome c to cupric plastocyanin is about 5000 S·I 31. All kinetic and NMR studies as well as modelling32 suggest that cytochrome c binds at the site of the acidic patches of plastocyanin. However, experiments performed by Kostic and co-workers using Zn-substituted cytochrome demonstrate that after the initial complex formation, a rearrangement is required to achieve efficient electron transfer33.36 and this is supported by chemical modification studies3? Kinetic studies on plastocyanin mutants suggest that cytochrome c moves in the direction of the small acidic patch (Glu-59, Glu-60 and Asp-61), towards the hydrophobic patch38
.
The electron transfer site of plastocyanin in complex with cytochrome f
In the absence of a structure of cytochrome
f
(before 1994), it was generally assumed that cytochrome c is a good model for cytochrome
J,
since it is a haem protein and also has a positively charged reaction surface. Implicit in this assumption is that it was expected that the positive charges on cytochromef
are located around the electron transfer site, as in cytochrome c, in which the lysines form a ring around the exposed haem edge. From this, the logical conclusion was that the site of plastocyanin that interacts with the positive charges (the acidic patches) must also be the site where the electron enters plastocyanin to get to the copper atom. This view was supported by the results of He et al 39 that showed a large decrease in reduction rate for YS3L plastocyanin compared to wild type protein. This suggested that Tyr-S3, which is located between the acidic patches and is a neighbour of copper ligand Cys-S4, is a part of the electron transfer route into plastocyanin. However, these results appear not to be reproducible and may have to be attributed to a poor source of cytochrome f40. More recent experiments demonstrate only a small decrease in the reaction rate for YS3L compared to wild type protein26.40.When the structure of cytochrome
f
became available?, it was obvious that the positive patch was not located near the electron transfer site. This opened up the possibility of two interaction sites of plastocyanin, with the hydrophobic patch and the acidic patches being involved in complex formation with the haem area and the positive patch of cytochromeJ,
respectively. Electrons would then enter plastocyanin via copper ligand His-S7 in the hydrophobic patch, in line with other blue copper proteins. This would also account for the effects that mutations of Leu-12 in hydrophobic patch have on the reaction rate23, which is not possible if cytochromef
were to interact only with the acidic patches.
Characterisation by NMR
The complex of plastocyanin and cytochrome f
The interactions between turnip cytochrome
f
and spinach plastocyanin have been studied by NMR41. In a mixture of 15N-labelled plastocyanin and cytochrome f, some amide resonances demonstrate chemical shift changes compared to free plastocyanin and these changes are proportional to the amount of bound plastocyanin. The linewidth of all plastocyanin resonances is increased in the mixture. These phenomena demonstrate that plastocyanin and cytochromef
form a complex in solution in which free and bound proteins are in fast exchange.In these experiments, cytochrome
f
was in the reduced state and the copper in the plastocyanin was replaced by cadmium to prevent electron transfer reactions, which may complicate the NMR spectra.The most perturbed peaks represent the residues most affected by complex formation. These are found at the acidic patches, at the hydrophobic patch around His- S7 and around the copper site (Table 1). In the large acidic patch, residues Glu-43, Asp-44 and Ile-46 are affected and in the small acidic patch residues Ser-56 to Asp-61. The hydrophobic patch is made up of four loops that connect the ~-strands and all of these sense the complex formation. Residues Gly-6 to Ala-13 (except Gly-S) are affected in the '10' loop, Asn-32, Ala-33 and the exposed Phe-35 in the '30' loop, Leu- 62 and Asn-64 in the '60' loop and Cys-S4 and His-S7 to Lys-95 (except Val-93) in the 'SO' loop. Copper ligands Cys-S4, His-S7 and Met-92 are affected, suggesting that the complex formation has effects on the copper site.
Upon oxidation of the cytochrome
J,
additional chemical shift changes are observed for the plastocyanin residues at the hydrophobic patch. TheseUBBINK : DYNAMICS IN ELECTRON TRANSFER COMPLEXES 41
Table I-Residues affecteda by binding of spinach and pea plastocyanin to cytochrome f and cytochrome c respectively, extrapolated to a I: I complex. The shifts in the cytochrome f complex are about 20-fold largerb
Spinach/pea Gly-6 Gly/Ala-7 Asp-9 Gly-lO Ser/Gly-11 Leu-12 Ala-13 Gly-24 Asn-32 Ala-33 Gly-34 Phe-35 His-37 Asn-38 Val-39 Val-40 Phe-41 Glu-43 Asp-44 lIe-46 Ser/Ala-48 Val-50 Asp-51 lie-55
Cytochrome f
+
Cytochrome c
aBased on 30.41
++
+ + ++
++
++
+ + ++
+
++
+++
++
+
+
++
+ +++
+ + ++
+++
+ +++
++
++
+++
+++
bFor a I: I cytochrome j:spinach plastocyanin complex:
(+) 2: 0.1 ppm for proton or 2:0.33 ppm for 15N (++) 2: 0.2 ppm for proton or 2:0.66 ppm for 15N (+++) 2: 0.3 ppm for proton or 2: 1.33 ppm for 15N For a I: I cytochrome c:pea plastocyanin complex:
(+) 2: 0.005 ppm (++) 2: 0.010 ppm (+++) 2: 0.020 ppm eNot observed
additional changes are due to paramagnetic effects caused by the oxidised haem group, suggesting that the hydrophobic patch is located close to the haem in the complex, as paramagnetic effects fall off rapidly with the distance between the observed nucleus and the paramagnetic centre. In the '10' loop residues Leu- 4 and Gly-6 to Ala-13 experience paramagnetic effects, in the '30' loop residues Ans-32, Phe-35, copper ligand His-37 and Asn-38, in the '60' loop, Asn-64 and in the '80' loop Gln-88, Gly-91 and Met- 92. Residue Glu-59 experiences a small negative effect, but otherwise, there are no effects on this side of the protein.
The information from the binding effects and paramagnetic shifts has been used to determine the orientation of plastocyanin ." in the complex with cytochrome
f
The effects on the amide nuclei were translated into distance and angle restraints. Also, some restraints were added to ensure electrostaticSpinach/pea Ser-56 Met-57 SerlPro-58
Glu-59 Glu-60 Asp-61 Leu-62 Asn-64 Ala-65 Tyr-70 Lys-77 Tyr-80 Lys-81 Phe-82 Cys-84 His-87 Gln-88 Gly-89 Ala-90 Gly-91 Met-92 Val-93 Gly-94 Lys/Gln-95
Cytochrome f
+ + + + ++
+ +
+
+ + +++
+++
++
+++
+ +++
++
+
Cytochrome c ++
n.o. c
++
++
++
+ ++
++
+ + +++
+ + +++
+ n.o.C
++
+++
++
Fig. I-The structure of the complex of plant cytochrome f and
plastocyanin (dark)41 .
interaction between the acidic patches and the basic patch of cytochrome
f
10,27. Starting from an arbitrary position, plastocyanin was moved freely around cytochrome J, in order to find an orientation that minimised an error function between input and actual distances. From various starting posItIOns, plastocyanin converged systematically to a single orientation which had the lowest error function and few violations.Figure 1 shows the structure of the complex. Both hydrophobic and acidic patches of plastocyanin make contact with the cytochrome
f
surface. Apparently, both electrostatic and hydrophobic interactions are important to stabilise the complex. The total interface size is 860 (±40) A2. This means that 2 x 860=
1720 A 2 of the total accessible surface area of both proteins is covered upon complex formation, which is 9%. In the acidic patches, residues Asp-42, Glu-43 and Asp- 44 are close to Arg-209, Lys-187 and Lys-185, respectively and residues Glu-59 and Glu-60 are close to Lys-65 and Lys-58, respectively. Plastocyanin residue Asp-51 is close to Lys-187. A tight interaction exists between the hydrophobic patch and the haem area of cytochromef
Plastocyanin residues Gly-lO, Leu-12, Phe-35, Pro-36, Leu-62, Asn-64 and Ser-85 - Ala-90 are close « 4 A) to cytochromef
atoms. A very short electron transfer pathway is present from the haem to the copper atom, since ligand His-87 makes van der Waals contact with Phe-4 and haem ligand Tyr-1. The shortest distance is found for His- 87 NE2 to Tyr-1 C02 (2.9 A). The copper-to-iron dis- tance range is 10.9 (±0.3)A.
This strongly suggests that electron transfer proceeds via His-87.The structure of the complex is based on NMR information from backbone nuclei. Chemical shift changes for side-chain hydrogens have also been observed upon complex formation with reduced cytochrome
f
12. The affected side-chains are in the acidic and hydrophobic patches. The largest shifts are observed for residues Leu-12, His-87, Gln-88 and Ala-90, suggesting that the area around His-87 makes the tightest contact with cytochrome J, perhaps accompanied by water exclusion. The largest shift is observed for L12 H02 with a shift of more than 1 ppm in the 1: 1 complex.The complex of plastocyanin and cytochrome c When reduced horse heart cytochrome c and cadmium-substituted pea plastocyanin are mixed, complex formation can be detected by line broadening and chemical shift changes3o• The chemical shift perturbations are very small (about 20-fold less)
compared to the complex of plastocyanin and cytochrome
f
On plastocyanin, primarily, residues in and between the acidic patches are affected, although small effects on the hydrophobic patch are observed as well (Table 1). Around the acidic patches, residues Val-39 to Phe-41 are affected, as well as Glu-43, Asp- 44, Ile-46, Ala-48, Asp-51, lIe-55, Ser-56, and Pro-58 to Glu-60. At the hydrophobic patch, no effects are observed for the '10' loop, but Gly-34, His-37 and Asn-38 feel complex formation. However, these are buried residues and the effects may well be secondary transmittance effects, of binding occurring at the acidic patches. The clearest direct effects are observed for Asn-64 and Gln-88, located closest to acidic patches.The surprisingly small sizes of the chemical shift changes suggest that the interactions in the interface of the complex are weak and make it unlikely that water is displaced from the protein surface to any significant extent. Yet, the binding constant is not lower than in the case of plastocyanin and cytochrome
f
and is in the order of 105 Ml at low ionic strength 30, so under NMR conditions, most of the protein is in the bound state. The small chemical shift changes are interpreted as evidence for a complex in which both proteins exist in ensemble of different relative orientations, which are in rapid exchange. In such an ensemble, all individual interactions are highly transient and weak. This is further supported by the fact that cytochrome c is too small to interact with the acidic patches and the hydrophobic patch simultaneously, indicating that cytochrome c is not in a single orientation in the complex.In the complex with oxidised cytochrome c, there are no additional changes observed for plastocyanin residues that are caused by paramagnetic effects3o. If cytochrome c were bound to plastocyanin in a single complex, paramagnetic chemical shift changes would be expected that are much larger than the shifts caused by binding. If, however, cytochrome c assumes many different orientations in the complex, the paramagnetic shifts could be averaged to zero.
Therefore, the absence of paramagnetic effects also supports the idea that the complex does not consist of a single orientation but rather of an ensemble of exchanging orientations. The chemical shift changes indicate that cytochrome c interacts primarily with the acidic patches in various orientations, but 'escapes' towJtds the hydrophobic patch every now and then.
Figure 2 gives an impression of what this ensemble of complexes could look like.
UBBINK : DYNAMICS IN ELECTRON TRANSFER COMPLEXES 43
Fig. 2-An impression of the ensemble of orientations of cytochrome c in the complex with plastocyanin. Cytochrome c binds predominantly at the acidic patch, with occasional excursions towards the hydrophobic patch. For clarity, only the haem groups of the various cytochrome c orientations are shown.
A comparison: striking difference in dynamics
Cytochrome c has been used often as a model for cytochrome f, which was related to the absence of a structure of cytochrome
f
and the difficulty to obtain the protein. Both are c-type haem proteins and show a similar ionic strength dependence of the kinetics of electron transfer to plastocyanin. However, it is now clear that cytochrome c differs in many respects from cytochromef
The latter has an elongated shape with a large and a small domain, both of which consists primarily of ~-sheet. The haem is oriented in a different way within the protein and the positive patch does not surround the haem area. This property makes it possible that plastocyanin can bind with both hydrophobic patch and acidic patches simultan- eously (Fig. 1). In cytochrome c, the positive charges surround the haem edge and this is not possible.The NMR results have demonstrated another striking difference between the two proteins.
Cytochrome
f
forms a complex with plastocyanin in which the proteins assume a well-defined, single orientation, at least for a large part of the time. This is supported by the large chemical shift changes, in particular in the hydrophobic patch and by the presence of paramagnetic effects observed for plastocyanin nuclei and caused by the haem group.The structure of the complex indicates that both electrostatic and hydrophobic interactions are important in the complex. The hydrophobic contacts can result in large chemical shift changes, may be
A B
c
Fig. 3--A model for formation of redox protein complexes. A) free proteins, B) ensemble of orientations with electrostatic interactions and C) single orientation, optimal for electron transfer. The equilibrium between states Band C is towards B in the cytochrome c:plastocyanin complex and towards C in the cytochrome J:plastocyanin complex.
related to stripping of water molecules from the protein surfaces upon complex formation. The complex of cytochrome c and plastocyanin, on the other hand, appears to exist as an ensemble of different orientations, with weak, transient interactions only. This is indicated by the small chemical shift changes and the absence of paramagnetic effects from the haem on plastocyanin.
A model for complexformation
The complex formation of redox proteins is believed to exist of at least two steps (Fig. 3). First, the free proteins come together (A) and at relatively long distance the electrostatic forces result in a reorientation of the proteins optimising the charge- charge interactions. The electrostatic forces generally result from a large number of charges, forming a kind of 'plate' 42 and are thus non-specific. In the purely electrostatic complex, both proteins can assume a number of different orientations, which have similar energies and are thus all populated (B). As other interactions are absent, it is also reasonable to assume that the kinetic barriers between them are small and that the various electrostatic conformations are in fast exchange. In the second step a single orientation is stabilised by additional hydrophobic interactions and hydrogen bonds. This results in a specific complex that is favourable for fast electron transfer (C). The Marcus theory43 predicts an exponential decrease of the electron transfer rate with the distance between the metal centres. The decay factor for this decrease is large in particular for through-space electronic coupling. It can thus be expected that electron transfer within the ensemble of electrostatic complexes, with varying donor-acceptor distances and no stable close contacts between the proteins has a low efficiency. A single orientation with a short distance between the metals and close contacts or hydrogen bonds at the interface is more efficient in electron transfer.
NMR results suggest that the conversion from electrostatic ensemble to single orientation complex can be viewed as a reaction equilibrium, which lies towards the ensemble side in one case (cytochrome c:plastocyanin complex) and towards the single orientation side in another (cytochrome fplastocyanin complex). The cytochrome fplastocyanin complex may have evolved towards a specific orientation which is stabilised by specific hydrophobic interactions, in order to obtain very fast electron transfer. The non-physiological complex of cytochrome c and plastocyanin is of purely electrostatic nature like it could occur between any pair of oppositely charged proteins. There is no stabilised specific orientation that favours electron transfer, which explains why electron transfer rates are about lO-fold 10wer23 than between cytochrome
f
and plastocyanin, despite a much more favourable driving force.
As discussed, fast electron transfer within the cytochrome c:plastocyanin complex requires a rearrangement. The NMR results suggest that cytochrome c may bind predominantly in various orientations at the acidic patches, but also interacts weakly with the hydrophobic patch. The rearrangement may represent this movement from acid patches towards the hydrophobic patch (Fig. 2), which is in agreement with the kinetic studies on mutant plastocyanins38. In the case of the complex of cytochrome
f
and plastocyanin, the latter may interact with only the basic patch initially and 'slide' into the orientation shown in Fig. 1, with additional hydrophobic interactions around the haem.Conclusions
The dynamic nature of electron transfer complexes can differ greatly. In the case of a single orientation, the intermolecular paramagnetic effects observed within these complexes contain invaluable information about the relative orientation of the partners and can thus be used to determine the structure of the complex. NMR chemical shift perturbation analysis is an excellent method to map binding sites and distinguish single orientation complexes from orientation ensembles. These methods are not limited to the complexes considered in this work, but, in stead, are broadly applicable to transient protein:protein complexes. Therefore, it can be expected that more structural information about such complexes in solution will become available in the near future.
Acknowledgement
I wish to acknowledge my collaborators Drs Derek S Bendall and B Goran Karlsson.
References
1 Hope A B, Biochim Biophys Acta-Bioenerg, 1456 (2000) 5.
2 Cramer W A, Martinez S E, Furbacher P N, Huang D &
Smith J L, Curr Op Struct Biol. 4 (1994) 536.
3 Cramer W A, Soriano G M, Ponomarev M, Huang D, Zhang H, Martinez S E & Smith J L, Ann Rev Plant Physioi Piant Mol Bioi, 47 (1996) 477.
4 Cramer W A, Soriano G M, Zhang H M, Ponamarev M V &
Smith J L, Physioi Piant, 100 (1997) 852.
5 Gray J C, Photosynthesis Res. 34 (1992) 359.
6 Gray J C, Rochford R J & Packman L C, Eur J Biochem, 223 (1994) 481.
7 Martinez S E, Huang D, Szczepaniak A, Cramer W A &
Smith J L, Structure, 2 (1994) 95.
8 Gross E L, Photosynthesis Res, 37 (1993) 103.
9 Sigfridsson K, Photosynthesis Res, 57 (1998) 1.
10 Kannt A, Young S & Bendall D S, Biochim Biophys Acta- Bioenerg. 1277 (1996) 115.
11 Qin L & Kostic N M, Biochemistry, 31 (1992) 5145.
12 Ejdeback M, Bergkvist A, Karlsson B G & Ubbink M, Biochemistry, 39 (2000) 5022.
13 Meyer T E, Zhao Z G, Cusanovich M A & Tollin G, Biochemistry, 32 (1993) 4552.
14 Niwa S, Ishikawa H, Nikai S & Takahe T, J Biochem (Tokyo), 88 (1980) 1177.
15 Beoku-Betts D, Chapman S K, Knox C V & Sykes A G, Inorg Chem, 24 (1985) 1677.
16 Beoku-Betts D & Sykes A G, inorg Chem, 24 (1985) 1142.
17 Hirota S, Endo M, Hayamizu K, Tsukazaki T, Takabe T, Kohzuma T & Yamauchi 0, JAm chem Soc, 121 (1999) 849.
18 Qin L & Kostic N M, Biochemistry. 32 (1993) 6073.
19 Takenaka K & Takabe T. J Biochem (Tokyo). 96 (1984) 1813.
20 Anderson G P, Sanderson D G, Lee C H. Durell S, Anderson L B & Gross E L, Biochim Biophys Acta, 894 (1987) 386.
21 Takabe T & Ishikawa H, J Biochem. 105 (1989) 98.
22 Gross E L & Curtiss A, Biochim Biophys Acta, 1056 (1991) 166.
23 Modi S, Nordling M. Lundberg L G, Hansson
°
& Bendall D S, Biochim Biophys Acta, 1102 (1992) 85.24 Soriano G M. Ponamarev M V. Piskorowski R A & Cramer W A, Biochemistry, 37 (1998) 15120.
25 Lee B H, Hibino T, Takabe T, Weisbeek P J & Takabe T. J Biochem. 117 (1995) 1209.
26 Illerhaus J, Altschmied L, Reichert J, Zak E, Herrmann R G
& Haehnel W, J bioi Chem. 275 (2000) 17590.
27 Gong X S, Jiang Q W, Fisher N E, Young S, Howe C J, Bendall D S & Gray J C, Eur J Biochem, 267 (2000) 3461. 28 Augustin M A, Chapman S K, Davies D M, Sykes A G,
Speck S H & Margoliash E. J bioi Chem. 258 (1983) 6405.
29 Bagby S, Driscoll P C, Goodall K G, Redfield C & Hill H A 0, Eur J Biochem, 188 (1990) 413.
30 Ubbink M & Bendall D S, Biochemistry, 36 (1997) 6326.
31 King G C, Binstead R A & Wright P E. Biochim Biophys Acta, 806 (1985) 262.
32 Roberts V A, Freeman H C, Olson A J, Tainer J A & Getzoff ED. J bioi Chem. 266 (1991) 13431.
33 Peerey L M & Kostic N M, Biochemistry, 28 (1989) 1861.
UBBINK : DYNAMICS IN ELECTRON TRANSFER COMPLEXES 45
34 Zhou J S & Kostic N M, 1 Am chem Soc, 113 (1991) 7040.
35 Zhou J S & Kostic N M, 1 Am chem Soc, 114 (1992) 3562.
36 Ivkovic-1ensen M M & Kostic N M, Biochemistry, 36 (1997) 8135.
37 Pan L P, Frame M, Durham B, Davis D & Millett F, Biochemistry, 29 (1990) 3231.
38 Ivkovic-Jensen M M, Ullmann G M, Young S, Hansson 0, Crnogorac M M, Ejdeback M & Kostic N M, Biochemistry, 37 (1998) 9557.
39 He S, Modi S, Bendall D S & Gray J C, EMBO 1, 10 (1991) 4011.
40 Bendall D S, Wagner M J, Schlarb B G, Scillick T R, Ubbink
M & Howe C J, in The phototrophic prokaryotes edited by
Peschek et al (Kluwer AcademiclPlenum Publishers, New York), (1999) 315.
41 Ubbink M, Ejdeback M, Karlsson B G & Bendall D S, Structure, 6 (1998) 323.
42 Watkins J A, Cusanovich M A, Meyer T E & Tollin G, Protein Sci, 3 (1994) 2104.
43 Marcus R A & Sutin N, Biochim Biophys Acta, 811 (1985) 265.
