The Interactions of Cyanobacterial Cytochromec 6 and Cytochrome f, Characterized by NMR* 210

During oxygenic photosynthesis, cytochromec 6 shuttles electrons between the membrane-bound complexes cytochrome bf and photosystem I. Complex formation between Phormidium laminosum cytochromef and cytochrome c 6 from bothAnabaena sp. PCC 7119 and Synechococcus elongatus has been investigated by nuclear magnetic resonance spectroscopy. Chemical-shift perturbation analysis reveals a binding site on Anabaena cytochrome c 6, which consists of a predominantly hydrophobic patch surrounding the heme substituent, methyl 5. This region of the protein was implicated previously in the formation of the reactive complex with photosytem I. In contrast to the results obtained for Anabaena cytochromec 6, there is no evidence for specific complex formation with the acidic cytochrome c 6 fromSynechococcus. This remarkable variability between analogous cytochromes c 6 supports the idea that different organisms utilize distinct mechanisms of photosynthetic intermolecular electron transfer.

Electron transport between the membrane-bound complexes cytochrome bf and photosystem I (PSI) 1 is maintained by mobile electron carriers. In plants, this task is fulfilled by plastocyanin (Pc), whereas cytochrome c 6 (cytc 6 ) is the only carrier in certain cyanobacteria. There also exist eukaryotic algae and cyanobacteria, which have the capacity to replace Pc with cytc 6 under copper-depleted conditions (1). Recently, a cytc 6 -like protein has been discovered in the thylakoid lumen of Arabidopsis (2,3). Despite being evolutionarily unrelated, Pc and cytc 6 perform equivalent tasks with common reaction partners. This functional convergence suggests that similar interaction properties exist for both proteins. Not surprisingly, Pc and cytc 6 have comparable redox potentials of around 350 mV. Further-more, both proteins provide similar recognition information, demonstrated by the parallel variation of their isoelectric points, being acidic in green algae while ranging from acidic to basic in cyanobacteria (1,4,5).
To date there has been a considerable amount of kinetic and mutational analysis of the electron transfer reaction between both Pc and cytc 6 and their partner PSI (6 -11). A hierarchy of mechanisms for interprotein electron transport has emerged from this work. Reduction of PSI by Pc or cytc 6 , isolated from different organisms, can follow an oriented collision mechanism (type I), a two-step mechanism requiring complex formation (type II), or a complex formation with rearrangement of the interface before electron transfer occurs (type III). A remarkable homology between the arrangement of charged and hydrophobic recognition patches on the surfaces of Pc and cytc 6 has also been revealed (12,13). In particular, a conserved arginine found in cyanobacterial Pc and cytc 6 was shown to be critical for bimolecular association with PSI (9).
Two structures of the cytf-Pc complex from plant (14) and cyanobacterial (15) sources have been determined. A combination of charged and hydrophobic patches defines the complex interface between turnip cytf and spinach Pc. In contrast, the complex from Phormidium laminosum, a thermophilic cyanobacterium, was found to be predominantly hydrophobic. At present there are no kinetic or structural data for the interaction of cytc 6 and cytf. Although mutagenesis studies have identified key residues for the reaction with PSI, there is no knowledge of the interaction site involved with cytf.
To address the question of molecular recognition in cytc 6 , we have investigated complex formation with cytf using heteronuclear NMR. We have aimed to identify the surface features involved in complex formation and to draw comparisons with cytc 6 -PSI interactions. Two cyanobacterial variants of cytc 6 ( Fig. 1), the basic protein from Anabaena sp. PCC 7119 (pI 9.0) and the acidic protein from Synechococcus elongatus (pI 4.8), have been studied. In this way, the role of electrostatics in protein interactions could be investigated explicitly. Cytf from P. laminosum, with a net charge of Ϫ14, was used as the partner protein. It is important to note that P. laminosum cytf shares 74 and 72% sequence identity with Anabaena cytf and S. elongatus cytf, respectively. Furthermore, the net charge is intermediate of cytf from Anabaena (Ϫ16) and S. elongatus (Ϫ12). This study reports the first structural characterization of the interactions of cytf and cytc 6 .

EXPERIMENTAL PROCEDURES
Protein Preparation-The soluble fragment of cytf was prepared according to previously published methods (15,16). Unlabeled Anacytc 6 was produced in Escherichia coli GM119 (17) transformed with both pEAC-WT (18) and pEC86 (19). The culture was grown in LB * This work was supported by the Research Training Network "TRANSIENT" in the Human Potential Program of the European Commission (Grant HPRN-CT-1999-00095) as well as by the Spanish Ministry of Science and Technology (Grant BMC2000-0444) and the Andalusian Government (Grant PAI, CVI-0198). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ʈ To whom correspondence should be addressed. Tel.: 31-71-527-4628; E-mail: m.ubbink@chem.leidenuniv.nl. 1 The abbreviations used are: PSI, photosystem I; Pc, plastocyanin; cytc, cytochrome c; cytc 6 , cytochrome c 6 ; cytf, cytochrome f; Ana, Anabaena; Syn, Synechococcus; HSQC, heteronuclear single quantum correlation spectroscopy; TOCSY, total correlation spectroscopy; NOESY, nuclear Overhauser enhancement spectroscopy. medium (20) supplemented with 100 g/ml ampicillin, 10 g/ml chloramphenicol, and 1 mM FeCl 3 . Aerobic growth was maintained at 37°C for 18 h before harvesting. A similar expression system in E. coli JM109 was used for the production of uniformly 15 N-labeled Ana-cytc 6 . The culture was grown in M9 minimal media additionally supplemented with 1 g/liter 15 NH 4 Cl and 1 mM thiamine. Aerobic growth was maintained at 37°C for 72 h before harvesting. Isolation and purification of Ana-cytc 6 were achieved as described previously (8,18). The preparation and purification of 15 N, 13 C double-labeled Syn-cytc 6 have been reported previously (21).
NMR Samples-Protein interactions were investigated for the diamagnetic species only, and reducing conditions were maintained in the presence of sodium ascorbate. Protein solutions were concentrated to the required volume using ultrafiltration methods (Amicon; YM3 membrane) and exchanged into 10 mM potassium phosphate, pH 6.0, 10% D 2 O, 1.0 mM sodium ascorbate. Protein concentrations were determined spectrophotometrically using an ⑀ 553 of 26.2 mM Ϫ1 cm Ϫ1 for the ferrous form of both Ana-cytc 6 (22) and Syn-cytc 6 and an ⑀ 556 of 31.5 mM Ϫ1 cm Ϫ1 for the ferrous form of cytf. For the assignment of Ana-cytc 6 , 0.7 mM 15 N-labeled and 1.0 mM unlabeled samples were prepared. To investigate complex formation between Ana-cytc 6 and cytf, microliter aliquots of a 1.9 mM cytf stock solution were titrated into an NMR sample containing 0.3 mM 15 N-Ana-cytc 6 . A reverse titration was also performed in which a sample containing 0.2 mM 15 N-Ana-cytc 6 and 0.5 mM cytf was titrated with a 3.0 mM stock of unlabeled Ana-cytc 6 . To investigate complex formation between Syn-cytc 6 (25). All images were created with a color ramp for positive (blue) or negative (red) surface potentials saturating at 10 kT. Potentials were calculated, for formal charges only, and surfaces were rendered in GRASP (43). The arrows indicate the location of the exposed heme edge. 6 in the presence of two equivalents of cytf (gray).

FIG. 3. Line broadening effects.
A, cross-sections along the F 2 dimension through the 1 H N resonance of Ser-16 in Ana-cytc 6 . The resonance in the free protein (8.07 ppm) has a line width at half-height of 13 Hz. In the presence of two equivalents of cytf, the resonance shifts to 8.18 ppm, and the line width at half-height increases to 34 Hz. B, cross-sections along the F 2 dimension through the 1 H N resonance of Gly-12 in Syn-cytc 6 . Despite the presence of two equivalents of cytf, the resonance is unperturbed (compare Ala-12 of Ana-cytc 6  NMR Spectroscopy-For sequence-specific assignment of the backbone resonances of Ana-cytc 6 , two-dimensional 1 H-15 N HSQC, threedimensional 1 H-15 N NOESY-HSQC (100 ms mixing time), and threedimensional 1 H-15 N TOCSY-HSQC (80 ms mixing time) spectra were recorded on a Bruker DRX 500 NMR spectrometer. For assignment of the side-chain protons, two-dimensional homonuclear NOESY and TOCSY spectra were recorded. XWINNMR was used for spectral processing, and the assignment was performed in XEASY (23). Complete assignments of the 1 H and 15 N resonances were determined for all backbone amides (see Supplementary material, Table S1). The nonexchangeable side-chain protons were also assigned. This assignment is consistent with those reported previously for Monoraphidium braunii cytc 6 (24) and Syn-cytc 6 (25).
Measurements on samples containing cytf and either 15 N-Ana-cytc 6 or 15 N, 13 C-Syn-cytc 6 were performed on a Bruker DMX 600 NMR spectrometer operating at 300 K. Two-dimensional 1 H-15 N HSQC spectra (26) were recorded with spectral widths of 30.0 ppm ( 15 N) and 13.9 ppm ( 1 H). Analysis of the chemical-shift perturbation (⌬␦ Bind ) with respect to the free protein was performed in XEASY.
Binding Curves-Titration curves were obtained by plotting ⌬␦ Bind against the molar ratio (R) of [cytf]:[Ana-cytc 6 ]. For the reverse titration, ⌬␦ Bind was plotted versus [Ana-cytc 6 ]: [cytf]. Non-linear leastsquares fits to a one-site model (27) were performed in Origin (Originlab, Northhampton, MA). This model explicitly treats the concentration of both proteins with R and ⌬␦ Bind as the independent and dependent variables, respectively (27). The binding constant (K a ) and the maximum chemical-shift change (⌬␦ Max ) were the fitted parameters. A global fit was performed in which the curves were fitted simultaneously to a single K a value, whereas ⌬␦ Max was allowed to vary for each resonance.

RESULTS AND DISCUSSION
The Complex of cytf and Ana-cytc 6 -Comparison of 1 H-15 N HSQC spectra of free Ana-cytc 6 and Ana-cytc 6 in the presence of cytf revealed distinct differences arising from complex formation (Fig. 2). When cytf was titrated into Ana-cytc 6 , 33 of the backbone amides experienced chemical-shift perturbation, ⌬␦ Bind Ն 0.03 ppm ( 1 H) and/or Ն 0.10 ppm ( 15 N). A single averaged resonance was observed for each backbone amide, indicating that the free and bound forms of cytc 6 were in fast exchange on the NMR time scale. In addition to chemical-shift perturbation, the presence of cytf caused a general broadening of about 20 Hz of the amide resonances (Fig. 3A), as expected for complex formation (28).
Titration curves of ⌬␦ Bind versus the molar ratio of cytf:Anacytc 6 were plotted for the 15 N nuclei of the four most strongly shifted resonances (Fig. 4A). The curves clearly illustrate that the chemical-shift perturbation increases as a function of the cytf concentration. Despite the addition of two equivalents of cytf, however, saturation of the chemical-shift changes was not observed. The binding curves were fitted to a 1:1 model with a binding constant of ϳ1 ϫ 10 4 M Ϫ1 . The quality of the fit is poor, particularly at the beginning of the curve, which is perhaps due to a small systematic error in the determination of the protein ratio. To investigate this further, a reverse titration was performed in which a sample of cytf was titrated with Ana-cytc 6 . In this case, the binding curves were fitted satisfactorily with a binding constant of 8 (Ϯ2) ϫ 10 3 M Ϫ1 (Fig. 4B). It can be concluded that the binding affinity of cytf for Ana-cytc 6 is ϳ10 4 M Ϫ1 , which is about 2 orders of magnitude greater than the affinity for the physiological partner, P. laminosum Pc (15,29). From the ratio of the observed ⌬␦ Bind to the fitted ⌬␦ Max (Fig.  4B), it was calculated that 63% of the Ana-cytc 6 was bound in the first point of the reverse titration.
Six structures of cytc 6 , from green algal (30 -32), red algal (33), and cyanobacterial (12,25) sources, have been determined previously. All of the known structures exhibit high structural homology. At present, the structure of Ana-cytc 6 is unavailable, and therefore, a model was built in Swiss-MODEL (34) using the NMR structure of Syn-cytc 6 (25) as a template. The chemical-shift map in Fig. 5 illustrates the location of the affected residues in this model with each residue colored according to its observed ⌬␦ Bind . The complex interface consists of a well defined patch, which surrounds the exposed methyl groups (methyl 3, thioether 4 methyl, and methyl 5) of the heme. This patch is composed mainly of three stretches of the primary structure, residues 9 -19, 23-26, and 51-61. The first of these includes the heme-binding motif CXYCH, whereas the sixth ligand, Met-58, occurs in the third stretch. Val-25, which was shown to be important for the interaction between cytc 6 and PSI (8), is found in the second stretch. Cys-17, Met-58, and Ala-60 experience the largest shifts in the complex. Alanine and asparagine are the most abundant residues in the interface, accounting for one-third of all the affected residues. Although 60% of the interface can be classified as hydrophobic, only 4 residues are charged (Lys-55, Lys-66, Glu-68, and Lys-80). Notably the conserved arginine, Arg-64, which has been implicated in the reaction with PSI (9), was not shifted in the complex. Fragata (35) has produced a theoretical model of the complex formed between M. braunii cytc 6 and turnip cytf (Protein Data Bank accession code 1jx8). There is good agreement between the binding site on cytc 6 identified in this model and the experimentally observed interaction site on Ana-cytc 6 .
The complex of cytf and Ana-cytc 6 was also investigated at 50, 100, and 200 mM NaCl. The observed ⌬␦ Bind for the 33 affected amides is plotted as a function of the salt concentration in Fig. 6. As the salt concentration was increased, ⌬␦ Bind decreased. At 200 mM NaCl, the ⌬␦ Bind is zero for most residues. This suggests that the binding constant is considerably re- duced, illustrating the importance of attractive electrostatic interactions in complex formation. The more strongly affected residues such as Cys-17 and Met-58 still experience appreciable shifts at 200 mM NaCl. Furthermore, although the line broadening decreased with increasing salt, there remains ϳ5 Hz broadening at 200 mM NaCl. These observations are indicative of a significant interaction even at high ionic strength.
The Interactions of cytf and Syn-cytc 6 -In contrast to the case of Ana-cytc 6 , titration of cytf into Syn-cytc 6 produced only minor effects in the 1 H-15 N HSQC spectra. Despite the presence of two equivalents of cytf, the amide resonances of Syncytc 6 did not experience chemical-shift perturbation. Line broadening effects on the order of 2 Hz were observed (Fig. 3B), consistent with a weak and highly dynamic interaction (36). Increasing the ionic strength by the addition of 200 mM NaCl had no effect on the protein interactions. In contrast to the clearly defined complex between cytf and Ana-cytc 6 , there is no evidence for specific complex formation with Syn-cytc 6 .
Biological Implications-From our results, it is clear that Ana-cytc 6 and cytf form a well defined complex and that the amount of complex formed is dependent on the ionic strength. It has been shown previously that Ana-cytc 6 forms a complex with Anabaena PSI, the affinity of which decreases with increasing ionic strength (7). Although Ana-cytc 6 has a net charge close to zero, the presence of positive residues in the vicinity of the heme group promotes favorable electrostatic docking to cytf. The role of electrostatics in this complex is therefore analogous to the complex formed between plant cytf and Pc in vitro (14,27,(37)(38)(39). As illustrated in Fig. 5, the interaction site of Ana-cytc 6 is composed mainly of hydrophobic residues, which is necessary to achieve specific complex formation between the two proteins. Such a hydrophobic site is sim-ilar to that proposed previously for the interaction of Ana-cytc 6 with PSI (8).
A type I mechanism has been reported for the reduction of PSI by Syn-cytc 6 (21). In this mechanism, the electron transfer rate is proportional to the number of collisions in which the redox centers of both proteins are aligned. The NMR titration of cytf into Syn-cytc 6 provides no evidence for specific complex formation but suggests a highly dynamic interaction between these partners, in agreement with a type I mechanism. The optimal orientation for productive collisions is facilitated by the prominent acidic patch on the backside of Syn-cytc 6 ( Fig. 1) since the front face will be the energetically more favorable approach with the acidic partners, cytf and PSI. Apparently, cytc 6 from different organisms utilizes different mechanisms for the electron transfer reaction with its partners. A similar conclusion has been reached for the interactions of cytf and Pc from different organisms (15). The source of this different reactivity can be traced to the nature of the protein surfaces and ultimately to variations in the primary structures. The sequences of Ana-cytc 6 and Syn-cytc 6 share 67% identity. Of the 33 residues identified in the Ana-cytc 6 interaction site, 22 are conserved in the Syn-cytc 6 sequence. Only 6 residues, located around the periphery of the complex interface, are significantly different in Syn-cytc 6 . Ala-19 and Gln-26 are replaced by methionines, which have been implicated as endogenous antioxidants (25,40). Ala-49 is replaced by a tyrosine, whereas glutamine and histidine replace Thr-52 and Asn-53, respectively. This cluster of three variations in Syn-cytc 6 results in a bulkier surface, which may hinder interactions with cytf. The replacement of Lys-66 by a threonine in Syn-cytc 6 is representative of the most striking difference between the two sequences. Five additional lysines in Ana-cytc 6 contribute to the significantly higher pI of this protein and enable attractive electrostatic interactions with the acidic cytf.
The different roles of electrostatics and hydrophobics can be further illustrated by comparison with the recently characterized, non-physiological complex of yeast cytochrome c (cytc) and cytf (29). It was found that cytf binds two equivalents of cytc with binding constants of ϳ2 ϫ 10 4 M Ϫ1 and 4 ϫ 10 3 M Ϫ1 . The higher binding affinity, as compared with the cytf-Ana-cytc 6 complex, probably arises from the strong electrostatic attraction between cytf and cytc, which has a pI of 9.7 (41). Fewer residues experienced chemical-shift perturbation in this complex, resulting in a less extensive interaction site (29) (Fig. 7, A  and B). Furthermore, the chemical-shift perturbation was on average 50% smaller than that observed in the complex with Ana-cytc 6 . The prevalence of electrostatic attractions may reduce the amount of time spent in a single orientation, as determined by hydrophobic contacts, and consequently, fewer and smaller chemical-shift changes are observed. Alternatively, the diminished interaction site observed for cytc may be explained in terms of "goodness of fit." It can be assumed that the hydrophobic patch surrounding the heme is better adapted in Ana-cytc 6 than in yeast cytc for binding to cytf. This complimentarity favors closer contact between the protein surfaces and thus a larger interaction site.
Protein docking simulations, using an NMR filter implemented in BiGGER (42), identified the cytc binding sites as the front (Site I) and back (Site II) faces of the heme region of cytf (29) (Fig. 7C). A similar docking simulation using the Ana-cytc 6 model gave slightly different results. Although the front face of the heme remains the favored site of interaction, there is a significant fraction of favorable docking orientations beneath the heme region in the large domain of cytf (Fig. 7D). Notably, there is significant overlap between the top-ranking docking configuration found for Ana-cytc 6 in BiGGER and the docking orientation of M. braunii cytc 6 in the model of Fragata (35). In this model, between the algal cytc 6 and plant cytf, there are favorable electrostatics between complimentary charged patches, a ridge of lysines on the small domain of cytf and a cluster of acidic residues on the side of cytc 6 . This results in a slightly different orientation of cytc 6 in comparison with the cyanobacterial complex, which does not possess such well defined complimentary charged patches.