Conclusions

Within the resolution of the study, the positioning of the transmembrane helices assigned to the major proteins of the core complex (13,15) are consistent with the shape and size of the observed densities in the supercomplex. From the model shown in Fig. 4 we conclude that the 33 kDa protein is located over the D1/D2 heterodimer towards the CP47 side. The two transmembrane helices of CP47, located under the 33 kDa protein, are almost certainly helices 5 and 6 which are joined by a large lumenal loop (14,15). The lumenal ends of transmembrane helices C, D and E of the D2 protein are also covered by the 33 kDa protein. Importantly, the 33 kDa protein is also positioned near to the lumenal ends of the transmembrane helices C, D and E of the D1 protein and located over the lumenal loop joining helices C and D. This CD loop is highlighted in white in Fig. 3c and is modelled using the coordinates for the L subunit of the bacterial reaction centre (17). It is this CD domain of the D1 protein that is likely to bind the Mn cluster involved in water oxidation based on site directed mutagenesis experiments (18,19) and on its close location to the redox active tyrosine Yz (residue 161 on D1) which mediates electron flow from the inorganic cluster to the primary oxidant P680+ (3, 18). Modelled on the coordinates of the M subunit of the bacterial reaction centre (17) the CD loop of the D2 protein is also highlighted in Fig. 4c and, interestingly, like the CD loop of the D1 protein is located under the docking site of the 33 kDa protein. The model presented in Fig. 4 predicts that the 23/17 kDa proteins are located over the lumenal ends of transmembrane helices A and B of the D1 protein and the lumenal loop which joins them (see Fig. 4c). Interestingly, the binding site for the 23/17 kDa extrinsic proteins seems to merge with that of the 33 kDa protein in a region close to the lumenal end of helix C of the D1 protein close to the putative Mn binding site. The model also suggests that the binding site for the 23/17 kDa proteins incorporates two as yet unidentified transmembrane helices possibly those of cytochrome b559 given that this haem protein has been shown to cross link with the 23 kDa OEC protein (20). The docking site for the 23/17 kDa proteins may also incorporate the large lumenal loop joining transmembrane helices 5 and 6 of CP43. This loop corresponds to the similar loop in CP47 and is likely to be located in the same approximate relative position in the PSII map (13). The above assignments do not necessarily imply direct interactions with the lumenal ends of the transmembrane helices of the reaction centre core proteins since the lumenal loops joining these helices are extensive and probably form an interface as suggested by the density distribution shown in Fig. 2d. Not only does the 3D structure of the PSII supercomplex presented here give a framework for the incorporation of the transmembrane helices of the intrinsic subunits, but it also emphasises the tetrameric arrangement of the extrinsic OEC proteins on its lumenal surface. This arrangement is similar to that observed in earlier freeze-etching studies of intact spinach thylakoid membranes (21,22). From this comparison we can conclude that the dimeric organisation of PSII is maintained in vivo and that the supercomplex is indeed a basic structural unit of PSII within the thylakoid membrane of higher plants (23).