Photosystem II, a photosynthetic protein complex, is prone to X-ray damage during crystallography. A high-resolution structure of the undamaged complex now offers a detailed view of its catalytic centre. See Letter p.99
The oxygen we breathe is produced by photosystem II, a large protein complex located in the thylakoid membranes of photosynthetic organisms (Fig. 1). In this issue (page 99), Suga et al.1 describe the first high-resolution crystal structure of undamaged photosystem II, obtained using an X-ray source called a free-electron laser. Their structure provides unprecedented insight into the oxygen-evolving complex of the photosynthetic apparatus.
Photosystem II transforms the energy of sunlight into chemical energy. In this process, the absorption of a photon results in a charge separation in the photosynthetic reaction centre of the protein complex, and the generation of a high-energy electron. This electron is ultimately used to chemically reduce plastoquinone molecules, as part of the electron-transport chain of the light-dependent reactions of photosynthesis.
The electron is replaced by the oxygen-evolving complex (OEC) of photosystem II: the OEC is oxidized with each photon-absorption event until it has been oxidized four times, and then resets itself by splitting water into an oxygen molecule, four protons and four electrons. Understanding the water-splitting reaction and the detailed arrangement of the atoms in the catalytic OEC required for energy-, electron- and proton-transfer reactions is crucial for basic science, and has implications for the design of light-driven batteries.
X-rays are extremely useful for structure determination, but they inflict radiation damage that can destroy the sample being investigated, sometimes before data collection is complete. Previous X-ray crystallographic studies2,3 had provided detailed insight into the molecular architecture of photosystem II and the OEC, using crystals kept at cryogenic temperature during data collection to slow radiation damage. However, spectroscopic studies4 showed that X-ray doses from the synchrotron sources used for structure determination result in serious damage to the OEC, indicating that the manganese atoms in the crystal structures are reduced2 by the ionizing effects of X-rays and by radicals formed during data collection.
This problem now seems to have been solved with the advent of X-ray free-electron lasers (FELs), X-ray sources that provide ultra-bright femtosecond X-ray pulses (1 femtosecond is 10−15 seconds). They allow 'diffraction before destruction' — the ability to detect diffraction patterns before damage occurs. This has led to the development of a technique called serial femtosecond crystallography (SFX)5, in which individual samples are exposed to a single pulse of X-rays; the samples are destroyed by the exposure, and so are replaced after each pulse. Data are thus collected from thousands of micrometre-scale crystals in a serial fashion. Randomly oriented microcrystals are conveniently and efficiently delivered to the X-ray beam in liquid microjets, which allows data to be collected at room temperature using crystals kept in the mother liquor from which they were grown.
Other groups have previously used FELs to collect SFX data from streams of photosystem II microcrystals6,7,8. Spectroscopic measurements9 taken simultaneously with the SFX data showed that the OEC was not damaged, even though the microcrystals absorbed an extremely high dose of X-rays from the femtosecond pulses. However, no detailed information about the OEC's architecture could be derived from the diffraction data because the microcrystals did not diffract to high resolution. Poor diffraction is a common problem in macromolecular crystallography, in particular with membrane proteins. Because of their large size, macromolecules typically form crystals that have a high solvent content and that are held together by relatively few weak interactions. This explains their high fragility and often poor crystalline order.
In their quest to obtain a high-resolution structure of the undamaged OEC, Suga and co-workers used not only the advantages of damage-free data collection afforded by X-ray FELs (in this case, the SACLA instrument in Harima, Japan), but also their expertise in improving the quality of large crystals of photosystem II by dehydration3. Crystal packing, and thus diffraction properties, can be optimized by slow, controlled dehydration of crystals, an approach that has been applied successfully even to large macromolecular complexes.
An additional advantage of using large crystals is that it allows more than one diffraction pattern to be obtained from the same crystal by shifting it to expose a fresh spot. Suga et al. obtained data of excellent quality by rotating crystals for each X-ray exposure, so that each rotation represented a fraction of the crystal's mosaicity (a measure of the spread of crystal-plane orientations). This approach, which can be described as serial femtosecond rotation crystallography (SF-ROX), has previously been used10 to collect data from crystals of the enzyme cytochrome c oxidase, another acutely radiation-sensitive system that evaded analysis using synchrotrons.
The new high-resolution structure depicts undamaged photosystem II in its dark-stable resting state (the S1 state). It reveals the detailed arrangement of the OEC (see Fig. 1a of the paper), which consists of a cluster of four manganese ions and one calcium ion, with five oxygen atoms bridging the metals. Several water molecules are tightly bound close to the OEC. The authors observed slightly shorter distances between the manganese ions, compared to the previously determined high-resolution X-ray-damaged structure3.
Notably, the distances between one of the oxygen atoms (designated O5) and its neighbouring manganese ions are longer than those of the other oxygen atoms — this had previously been observed3, but it was unclear whether this was caused by radiation damage. This observation suggests that O5 is a hydroxide ion (OH−), rather than an oxo ligand (O2−), and so may be one of the OEC's substrate oxygen atoms. The researchers confirmed the accuracy and precision of their structures by obtaining two independent data sets, and by separately analysing data from the two monomers of the dimeric photosystem II complex.
Suga and colleagues' structure provides detailed information that will aid our understanding of the OEC's spectroscopic data, serve as a new starting point for computational studies of the water-splitting reaction, and provide invaluable clues about the mechanism of that reaction. An exciting next step would be to obtain time-resolved measurements using the SF-ROX approach, to provide an equally detailed view of the OEC at work.