pocketz.doi.bio/2025-08-29_143943_nature_nature21400/2025-08-29_143943_nature_nature21400

Light-induced structural changes and the site of O=O bond formation in PSII caught by XFEL | Nature

URL: https://www.nature.com/articles/nature21400

Saved: 2025-08-29T21:39:48.482Z

Letter Published: 20 February 2017 Light-induced structural changes and the site of O=O bond formation in PSII caught by XFEL Michihiro Suga, Fusamichi Akita, Michihiro Sugahara, Minoru Kubo, Yoshiki Nakajima, Takanori Nakane, Keitaro Yamashita, Yasufumi Umena, Makoto Nakabayashi, Takahiro Yamane, Takamitsu Nakano, Mamoru Suzuki, Tetsuya Masuda, Shigeyuki Inoue, Tetsunari Kimura, Takashi Nomura, Shinichiro Yonekura, Long-Jiang Yu, Tomohiro Sakamoto, Taiki Motomura, Jing-Hua Chen, Yuki Kato, Takumi Noguchi, Kensuke Tono, …Jian-Ren Shen Show authors

Nature volume  543,  pages 131–135 (2017)Cite this article

20k Accesses

571 Citations

65 Altmetric

Metrics details

Abstract

Photosystem II (PSII) is a huge membrane-protein complex consisting of 20 different subunits with a total molecular mass of 350 kDa for a monomer. It catalyses light-driven water oxidation at its catalytic centre, the oxygen-evolving complex (OEC)1,2,3. The structure of PSII has been analysed at 1.9 Å resolution by synchrotron radiation X-rays, which revealed that the OEC is a Mn4CaO5 cluster organized in an asymmetric, ‘distorted-chair’ form4. This structure was further analysed with femtosecond X-ray free electron lasers (XFEL), providing the ‘radiation damage-free’5 structure. The mechanism of O=O bond formation, however, remains obscure owing to the lack of intermediate-state structures. Here we describe the structural changes in PSII induced by two-flash illumination at room temperature at a resolution of 2.35 Å using time-resolved serial femtosecond crystallography with an XFEL provided by the SPring-8 ångström compact free-electron laser. An isomorphous difference Fourier map between the two-flash and dark-adapted states revealed two areas of apparent changes: around the QB/non-haem iron and the Mn4CaO5 cluster. The changes around the QB/non-haem iron region reflected the electron and proton transfers induced by the two-flash illumination. In the region around the OEC, a water molecule located 3.5 Å from the Mn4CaO5 cluster disappeared from the map upon two-flash illumination. This reduced the distance between another water molecule and the oxygen atom O4, suggesting that proton transfer also occurred. Importantly, the two-flash-minus-dark isomorphous difference Fourier map showed an apparent positive peak around O5, a unique μ4-oxo-bridge located in the quasi-centre of Mn1 and Mn4 (refs 4,5). This suggests the insertion of a new oxygen atom (O6) close to O5, providing an O=O distance of 1.5 Å between these two oxygen atoms. This provides a mechanism for the O=O bond formation consistent with that proposed previously6,7.

You have full access to this article via your institution.

Download PDF Similar content being viewed by others Oxygen-evolving photosystem II structures during S1–S2–S3 transitions Article Open access 31 January 2024 Structural insights into photosystem II assembly Article 12 April 2021 Structural evidence for intermediates during O2 formation in photosystem II Article Open access 03 May 2023 Main

Figure 1a shows the organization of the electron transfer chain of PSII in pseudo-C2 symmetry with two subunits, D1 and D2. The water-oxidation reaction proceeds via the Si-state cycle8 (with i = 0–4), where dioxygen is produced in the transition S3→(S4)→S0 (Fig. 1b). The high-resolution structures of PSII analysed so far were for the dark-stable S1 state4,5, although a few studies have obtained low-resolution intermediate S-state structures by time-resolved serial femtosecond crystallography (TR-SFX)9,10,11. During the revision of our manuscript, Young et al. reported a two-flash (2F)-illuminated state structure at 2.25 Å resolution in which no apparent changes around O5 were observed12, although estimations of the resolution could yield somewhat different values and therefore small movements of some water molecules may have escaped detection. To achieve a resolution high enough to uncover small structural changes induced by flash illuminations while allowing the Si-state transition to proceed efficiently, we determined the optimal crystal size of PSII with a maximum length of 100 μm, which diffracted up to a resolution of 2.1 Å by a SPring-8 ångström compact free-electron laser (SACLA)–XFEL pulse (Extended Data Fig. 1; Methods). We illuminated the small crystals by 2F (Fig. 1; Methods), which gave rise to a population of 46% S3 estimated by Fourier transform infrared (FTIR) difference spectroscopy (Extended Data Fig. 2; Methods).

Figure 1: TR-SFX of PSII at SACLA–XFEL.

a, Organization of the electron transfer chain of PSII (red arrows). b, Si-state cycle of the water-oxidation reaction of the OEC. hv, photons. c, Experimental setup for the TR-SFX experiments on PSII. d, Schematic representation of the timings of the two flash illuminations and XFEL pulse employed in this experiment. e, 2mFo − DFc maps obtained from the non-pre-flashed dark-state data (grey contoured at 2σ) and mFo − DFc map (green (negative) and red (positive) contoured at ± 5σ) of a region around the ChlD1 site, before locating the two water molecules in the vicinity of ChlD1. f, Stereo image of 2mFo − DFc maps in a region around the OEC, with the grey and cyan contoured at 2σ and 8σ, respectively.

PowerPoint slide

Full size image

The PSII microcrystals were mixed with a grease matrix and delivered to the XFEL beam by a high viscosity micro-extrusion injector13, which greatly reduced the sample consumption and provided a stable flow for uniform excitation. To ensure the turnover of PSII, potassium ferricyanide was added to the samples, and the samples were pre-flashed before the measurements for synchronization of PSII into the S1 state and the oxidation of tyrosine YD. Diffraction data were also collected from samples without the pre-flash (non-pre-flash), and they will be discussed in cases they differed from the pre-flashed samples.

Two data sets were collected for each of the samples (with or without pre-flashes); one is from the dark-state and the other is from 2F-illuminated samples. The two data sets for the pre-flashed samples were analysed to 2.35 Å resolution, and those for the non-pre-flashed samples were analysed to 2.50 Å resolution (Extended Data Fig. 1, Extended Data Tables 1, 2); all of the data sets had very high multiplicities (see Methods). Figure 1e, f shows part of the electron density map from the non-pre-flashed dark-sample (similar maps were obtained with pre-flashed samples, and we focus on the results of pre-flashed samples here unless otherwise stated), where the electron densities for the orientations of amino acid side chains and water molecules are clearly visible. We were able to build 1,626 waters for the dark-state and 1,584 waters for the 2F-state. The manganese atoms and calcium atom in the OEC were distinguished even in the anomalous difference Fourier map with peak intensities well beyond the noise level, owing to the long wavelength of the XFEL used. For example, among the top 30 peaks in the anomalous map of the non-pre-flashed 2F-data (Protein Data Bank (PDB) accession 5GTI), the top 7 peaks with intensities above 15.0σ were identified as the manganese atoms, and the other 23 peaks were identified as the calcium (including the newly identified calcium site at the stromal side of CP43), chloride or sulfur atoms (Extended Data Fig. 3, Extended Data Table 3), indicating that the measurement and integration of the diffraction intensities were accurate. Nevertheless, the interatomic distances within the OEC could not be refined accurately because of the limited resolution. Thus, tight Mn–O distance restraints derived from the S1 state 1.95 Å structure5 were used during the refinement of the dark-state data and relatively loose restraints were applied for the 2F-state data (Methods). Therefore we focus on the structural changes induced by the two flashes based on the isomorphous difference Fourier map between the two data sets.

The two data sets from the two different states were highly isomorphous with a Riso of 0.068 at 2.35 Å resolution, making it possible for us to identify subtle structural changes based on the isomorphous difference Fourier map (Extended Data Table 1). Superposition of the two state structures yielded a root mean square deviation of 0.10 Å. The corresponding Riso values between the pre-flashed and non-pre-flashed samples were more than 0.118 at 2.5 Å resolution; this difference may be caused by the different batches of samples used or differences in the post-crystallization processes employed.

Two areas with major light-induced structural changes were observed in the isomorphous difference Fourier map between the 2F- and dark-states (F(2F) − F(dark) difference map; Fig. 2). These areas are localized around the QB/non-haem iron-binding site and the OEC (dotted red circles in Fig. 2b), and similar changes were observed in both monomers of the dimer (a third area with apparent difference Fourier peaks was found in the vicinity of YD in non-pre-flashed samples; see below and Extended Data Fig. 4). All of these areas are related to electron and proton transfer in PSII, suggesting efficient laser pulse excitation. The peaks were detected at 8.6 for the QB site and 10.7σ for the OEC in monomer A, and 6.9σ for the QB site and 12.7σ for the OEC in monomer B. The largest Fourier difference peak in the other areas of PSII was 4.4σ (Fig. 2d), indicating that the peaks observed in these two areas are well above the noise level.

Figure 2: Isomorphous difference Fourier map between the 2F-state and the dark-state of the pre-flashed samples.

a, b, Isomorphous difference Fourier map between the 2F-state and the dark-state in green (positive) and red (negative) contoured at ±4.5σ with a top view from the luminal side (a) and a side view perpendicular to the membrane normal (b). Red dashed circles in b indicate two regions in which significant difference peaks were observed. c, Stereo image of the region boxed by black dashed lines in b. d, Histogram of the isomorphous difference Fourier map, showing apparent deviations of the difference Fourier peaks from the Gaussian distribution of the noise level beyond around ±5σ. Two arrows indicate the possible largest noise level.

PowerPoint slide

Full size image

QB forms H bonds with His215 and Ser264 of D1; the former is a ligand for the non-haem iron. The isoprenoid tail of QB is located in a hydrophobic environment formed by Phe211, Met214, Phe255 and Phe265 of D1, D2-pheophytin (PhoD2) and lipids (Fig. 3a). In the high-resolution S1 state structures4,5, QB exhibited a weak electron density with a higher B factor than QA, possibly owing to its mixed protonation states. After 2F illumination, positive electron densities were observed in the head group region of QB, whereas a pair of positive and negative densities was observed in the isoprenoid tail in the Fourier difference map, and the average temperature factor of the quinone head group was decreased from 116 Å2 to 91 Å2. This change indicates a rotation of the head group around the axis vertical to the benzene ring by about 10° towards Ser264 and His215 of D1, which shortens the distances between the head group of QB and these two residues by 0.1–0.2 Å and therefore strengthens the H bonds between them. These movements can be explained as follows. As we added potassium ferricyanide to the crystals well before the measurements, the non-haem iron was pre-oxidized to Fe3+, making it able to accept one electron upon flash illumination. Thus, the 2F illumination resulted in a dominant species of QB−, which will make stronger H bonds with Ser264 and His215, resulting in the rotation of the head group and the decrease in the temperature factor. In relation to this process, the hydrophobic residues Phe211, Phe255, Phe265 and Met214 of D1, which surround the isoprenoid tail, also moved slightly to adjust the cavity of the QB site (Fig. 3a, b). Notably, a loop region involving Asn266–Ser268 of D1, which shields the cavity of the QB site in the dark structure, moved by up to 0.8 Å upon 2F illumination, resulting in the partial opening of the QB site, which may suggest a route for the quinone exchange.

Figure 3: Isomorphous difference Fourier map and structural changes around the QB/non-haem iron-binding site and the OEC.

a–d, Isomorphous difference Fourier maps between the 2F-state and the dark-state of the pre-flashed samples in green (positive) and red (negative) contoured at ±5.0σ. D1 and D2 subunits are coloured in yellow and cyan, respectively, in the S1 state and green in the S3 state. a, b, Structural changes around the QB/non-haem iron-binding site. c, d, Structural changes around the OEC. In b–d, movements of residues or groups are indicated by black arrows, and the insertion or displacement of water (oxygen) is indicated by dashed arrows. e, Structure of the Mn4CaO5 cluster after 2F illumination superimposed with the 2mFo − DFc map contoured at 10.1σ. f, Position of the newly inserted oxygen atom O6 relative to its nearby atoms. Light dark balls labeled 1D–4A represent Mn ions; red and cyan balls labeled 1–6 represent oxygen atoms.

PowerPoint slide

Full size image

In the vicinity of QB, a strong negative peak was observed between the bicarbonate (BCT) and the non-haem iron in the difference Fourier map. There was a smaller positive peak on the opposite side of BCT, giving rise to a positional shift of BCT by up to 0.5 Å and an increase in the distances between BCT and the non-haem iron up to 0.2 Å (Fig. 3b). These changes may result from rearrangement of the H bond network around BCT and/or reorientation of BCT upon 2F illumination, which might be caused by the involvement of BCT in the proton transfer for QB protonation or the reduction of the non-haem iron. The possible involvement of BCT in the proton transfer is consistent with the observation that removal of BCT resulted in retardation of the quinone exchange reaction14. In contrast to the QB site, no large changes were found in the QA site.

A number of Fourier difference peaks were found in the region around the OEC (Fig. 3c, d); these peaks allow us to identify the following major structural changes: (i) Mn4 moved slightly towards the outside of the cubane, resulting in a slight elongation of its distance from Mn1 by 0.1–0.2 Å. (ii) Ca2+ also moved away from Mn4 slightly. (iii) A new, strong positive peak was found close to O5, which was modelled as a new oxygen atom O6. (iv) The side chain of Glu189 of D1 moved away from the cubane to accommodate O6. (v) A strong negative peak was found on the water molecule W665, indicating that this water molecule was displaced from this position upon 2F illumination. In relation to this, water molecule W567 (which is H bonded to O4 and W665) moved towards O4, resulting in a smaller distance between W567 and O4. (vi) Accompanying these changes, some other ligand residues also exhibited slight structural changes; these include Asp61, Asp170, His332 and Ala344 of D1.

Among the structural changes described above, the displacement of W665 and the appearance of the new oxygen atom O6 provided important implications for the mechanism of water oxidation. Owing to the displacement of W665, the H bond between this water and W567 was broken. This caused the movement of W567 towards O4, as they are already H bonded (Fig. 3c). The average distance between W567 and O4 upon 2F illumination in the two monomers was determined to be 2.32 Å. This distance is extremely short, which may suggest that W567 is a hydroxide ion rather than a water in the S3 state. This raises the possibility that these two oxygen species represent a state prepared for the upcoming O=O bond formation (Fig. 4). Indeed, studies using W-band 17O electron–electron double resonance-detected NMR have suggested that either O4 or O5 is exchangeable and thus may serve as candidate for the substrate water15. An earlier theoretical study also suggested that either O4 or O5 may be one of the substrates for O=O bond formation16. However, if we consider the possible experimental errors in the interatomic distances determined at the current resolution, these two oxygen atoms may still be connected by an H bond instead of forming a pre-state for O=O bond formation. The structural changes observed therefore strongly suggest the occurrence of proton transfer around this region, in agreement with recent reports that the H bond network starting from O4, W567 and W665 (Extended Data Fig. 5) may serve as a path for protons released from the OEC during the transition from S0 to S1 (refs 17, 18). The movement of W665 observed in the present study may be related to this proton transfer. Based on the proton release pattern of 1:0:1:2 for the S0→S1→S2→S3→S4 transitions, however, the proton transfer from W567 to W665 may occur in the S2→S3 transition.

Figure 4: Possible mechanism of O=O bond formation and proton transfer pathways from the OEC.

The dotted circle indicates displacement of the water molecule (W665) next to the water (W567) H bonded to O4, which results in the approach of W567 to O4. Dashed arrows indicate possible pathways for proton transfer. The two pairs of oxygen atoms enclosed by dashed red lines represent the possible sites of O=O bond formation, with the one between O5 and O6 having a much shorter distance and therefore representing the real site of O=O bond formation.

PowerPoint slide

Full size image

The O6 atom in the 2F illuminated state is very close to O5 (average distance of 1.5 Å for the two monomers; Fig. 3c–f), which enables them to form a peroxide or superoxide, for example19. This strongly suggests that this site is the reaction site for the O=O bond formation (Fig. 4), consistent with the model proposed by Siegbahn et al.6,7,20 as well as the results of electron paramagnetic resonance measurements21,22 and quantum mechanics/molecular mechanics (QM/MM) calculations23. The remaining structural changes can be easily explained by the insertion of this oxygen atom: the elongation of the distance between Mn4 and Mn1 is apparently due to the requirement to accommodate this new oxygen atom. In particular, the movement of Glu189 of D1 away from the cubane is necessary to accommodate O6, as it results in an average distance of 2.70 Å between O6 and Glu189. Without this movement, the distance between O6 and Glu189-OE2 would be 2.30 Å, which would hinder the insertion of the oxygen atom. The movement is made possible by the unique feature of Glu189 as a monodentate ligand to the OEC, in comparison with all of the other ligands that are bidentate. We should also point out that O6 is not necessarily a new water molecule from outside PSII, as a water molecule already present in the S1 state might move to this position4,19,24.

In addition to the two important areas of change in the pre-flashed samples described above, a third area of structural change was found around YD (Tyr160 of D2) in the non-pre-flashed samples. YD is partially H bonded to water molecule W508, which exhibits two conformations, one H bonded (W508I) and the other one not H bonded (W508II) to YD. In the Fourier difference map of the non-pre-flashed samples, a negative peak was found at the position of W508I, suggesting that this water molecule is displaced and the H bond with YD is broken (Extended Data Fig. 4a–c). This finding suggests that proton transfer occurs around this region upon 2F illumination, consistent with the recent QM/MM calculations in which the proximal W508I was found to be unstable upon oxidation of YD25. This in turn suggests that 2F illumination induces partial oxidation of YD in the non-pre-flashed samples, probably owing to inefficient electron donation from the Mn4CaO5 cluster in the crystals. The oxidation of YD will result in proton transfer from YD to the bulk solvent, breaking the H bond between YD and W508I and moving this water molecule to the distal position or bulk solvent25,26. Thus, this structural change is again caused by proton transfer around YD.

Methods

No statistical method was used to predetermine the sample size. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment.

Preparation and crystallization of photosystem II

Highly active PSII was isolated from Thermosynechococcus vulcanus and crystallized as described previously with slight modifications4,27,28. The final PSII core dimers were suspended in 20 mM Mes (pH 6.0), 10 mM NaCl, 3 mM CaCl2, and the final crystallization buffer contained 20 mM Mes, 20 mM NaCl, 40 mM MgSO4, 10 mM CaCl2, 5–7% polyethylene glycol 1,450, 0.85% n-heptyl-β-D-thioglucopyranoside (Dojindo). No re-crystallization procedure was applied. We screened various sizes of PSII crystals and post-crystallization treatment conditions (see below) to prepare suitable samples for TR-SFX by XFEL, and found that too small crystals did not diffract to a high resolution, whereas larger crystals gave rise to a lower efficiency of the Si-state transition induced by the laser excitations. The optimal crystal size was determined to have a maximum length of 100 μm, which diffracted up to a resolution of 2.1 Å by a SACLA–XFEL pulse (Extended Data Fig. 1) and a final population of 46% S3 state upon 2F illumination (see below).

All of the procedures for the preparation, crystallization, pre-flash illumination and diffraction experiments were conducted in the dark or in very dim green light. To prepare a large amount of micro-sized crystals, crystallization was performed in 1.5-ml micro-centrifuge tubes with a sample volume of 50 μl at a concentration of 2.3 mg chlorophyll per ml at 20 °C. The crystals appeared in a few hours; when the crystal size reached a maximum length of 100 μm, 50 μl of the crystallization buffer in which the concentration of PEG1,450 was increased by 1–2% from the crystallization condition was added to stop further growth of the crystals (see Extended Data Fig. 1a for a picture of typical crystals). The crystals were washed several times with this buffer to remove PSII samples that were not crystallized, and stored overnight.

Prior to the XFEL experiments, the crystals were transferred to fresh mother liquid containing 10 mM potassium ferricyanide as an electron acceptor for the pre-flashing samples, and 2 mM potassium ferricyanide for the samples without pre-flash (non-pre-flash). The concentration of ‘cryo-protectant’ was then increased to 22% glycerol, 9% PEG1,450 and 9% PEG5,000 MME by stepwise replacement of the mother solution over 1.5 h. Finally, the crystals were carefully mixed with a grease matrix, Super Lube nuclear grade grease (Synco Chemical Co.), and loaded into a high viscosity micro-extrusion injector as described previously13.

Although SFX can be performed at room temperature and there is no need to freeze the crystals, we found that this post-crystallization procedure to increase the cryo-protectant concentration was important for obtaining good diffracting PSII crystals. When the post-crystallization procedure was not adequate, some crystals gave rise to larger unit cell dimensions of a = 129.1 Å, b = 228.8 Å and c = 305.4 Å (Extended Data Fig. 6a–c). Diffraction spots from the crystals with this larger unit cell were found to be lower than 3.0 Å resolution, probably owing to the loose crystal packing. Indeed, we found that the PSII dimer in this crystal packing harboured two PsbY subunits (one PsbY per monomer PSII; Extended Data Fig. 6e). On the other hand, only one PsbY was found in the PSII dimer (one monomer contained PsbY whereas the other monomer did not) in the structure with unit cell dimensions of a = 126.5 Å, b = 231.2 Å and c = 287.5 Å analysed in this study (Extended Data Fig. 6f). When the two structures were superimposed, one of the two PsbY subunits in the dimer structure with the larger crystal packing interfered with the adjacent monomer in the structure with the smaller crystal packing (Extended Data Fig. 6d–f).

Pre-flash illumination of PSII crystals

To decrease the possible contamination of the S0 state in the dark-adapted crystals, PSII micro-crystals were illuminated with one pre-flash before the post-crystallization treatment. An aliquot of 100 μl solution containing micro-crystals of PSII was transferred into a dialysis button, and the pre-flash was provided by a Nd:YAG laser (Minilite-I, Continuum) at 532 nm with a diameter of 7 mm (large enough to cover the entire sample area) at an energy of ~52 mJ cm−2 at the sample position. After the pre-flash illumination, the sample was stored for 1–3 h in the dark while being transferred into the cryo-protectant conditions. This dark-incubation time was long enough to allow the higher Si-states (S2 and S3) to decay into the S1 state, and therefore to minimize the possible contamination of the S0 state in the dark-adapted crystals, if there are any.

Estimation of the S3 population in the PSII crystals by ATR–FTIR measurements

The population of the S3 state after 2 flashes in the PSII micro-crystals was estimated using light-induced FTIR difference spectroscopy combined with the attenuated total reflection (ATR) method. PSII crystals in the same buffer as the SACLA experiments (including 10 mM potassium ferricyanide) were loaded onto a three-reflection silicon prism and then sealed with a transparent plate and a silicone rubber spacer. For the measurement of standard spectra, the PSII core solution was loaded onto the silicon prism, and the sample temperature was maintained at 20 °C. FTIR spectra were recorded at 4 cm−1 resolution using a Bruker IFS-66/S spectrophotometer equipped with an MCT detector. Flash illumination was provided by a Q-switched Nd:YAG laser (Quanta-Ray INDI-40-10; 532 nm, ~7 ns FWHM, ~21 mJ cm−2). Saturation of the laser flash was confirmed by checking the power dependence of the FTIR signal. After two pre-flashes with subsequent dark adaptation for 30 min, two flashes with an interval of 10 s were applied to the sample and FTIR difference spectra were measured upon each flash. We confirmed that the higher S-states decayed very little during this 10-s interval. This measurement was repeated 10 times to increase the signal-to-noise ratio of the spectra, with each measurement separated by 30 min dark incubation.

The FTIR difference spectra obtained upon the 1st (a) and 2nd (b) flashes of PSII complexes in solution (black lines) and crystals (red lines) are shown in Extended Data Fig. 2. The spectra were normalized to the intensity of the amide II band, which reflects the protein amount. The efficiencies of the S-state transitions in the PSII crystals were estimated following the method previously described29,30. The 1st- and 2nd-flash spectra in the PSII crystal, f1(ν) and f2(ν), respectively, were fitted with linear combinations of the 1st- and 2nd-flash spectra in solution, F1(ν) and F2(ν), respectively, as standard spectra: f1(ν) = c11F1(ν); f2(ν) = c21F1(ν) + c22F2(ν), where c11, c21, and c22 are the coefficients of linear combination. The least-squares fitting was performed in the symmetric COO− region (Extended Data Fig. 2b). The coefficients were estimated to be c11 = 0.73 ± 0.02, c21 = 0.27 ± 0.01, and c22 = 0.64 ± 0.02. The efficiencies of the S1→S2 and S2→S3 transitions are expressed as α0c11 and α0(c22/c11), respectively, and the population of the S3 state after the second flash is calculated as α02c22. Here, α0 is the average efficiency of S-state transitions in solution and it was determined to be 0.85 ± 0.01 from the oscillation pattern of the intensity at 1,400 cm−1 (refs 31, 32) obtained by 12 consecutive flashes. Based on this, the population of the S3 state in the PSII micro-crystals after two flashes is estimated to be 0.46 ± 0.03.

Diffraction experiment at SACLA–XFEL

Single-shot XFEL data collection was performed using femtosecond X-ray pulses from the SACLA at BL3. The pulse parameters of SACLA were as follows: pulse duration, 2–10 fs; X-ray energy, 7 keV; energy bandwidth, 0.5% (FWHM); pulse flux, ~7 × 1010 photons per pulse; beam size 3.0 μm (H) × 3.0 μm (W); repetition rate, 30 Hz.

The PSII crystals mixed with grease were loaded into an injector with a nozzle diameter of 150 μm, and set in a diffraction chamber filled with helium gas in a setup called Diverse Application Platform for Hard X-ray Diffraction in SACLA (DAPHNIS)33. The flow rate was set to 5.6 μl min−1 (5.28 mm s−1) for the 2F data and 2.8 μl m−1 (2.64 mm s−1) for the dark data. The diffraction patterns were recorded using a multiport CCD detector34. Because the excitation laser pulses were provided at 10 Hz and the XFEL pulses had a repetition rate of 30 Hz, each ‘pump-on’ image was followed by two ‘pump-off’ images, which were recorded separately. The pump-on images were used to analyse the 2F-state structure, whereas the diffraction data for the dark-state was collected by a separate run.

To advance the PSII samples to the S3-state, two consecutive excitation laser flashes were provided from two separate Nd:YAG laser sources (Minilite-I, Continuum) to the sample at t = 0 ms and t = 10 ms (that is, the two flashes were separated by 10 ms) for the pre-flashed samples, and t = 0 ms and t = 5 ms (separated by 5 ms) for the non-pre-flashed samples. To ensure sufficient excitation, each of the pump lasers was split into two beams, and each beam from the same pump laser (pump 1 or pump 2) was focused on the same sample point with an angle of 160° with respect to each other (a nearly counter-propagating geometry). The two beams from pump 1 were used for the first flash illumination, whereas the other two beams from pump 2 were used for the second flash illumination. The pump focal diameter of all beams was set to 240 μm (top-hat) and its energy was 42 mJ cm−2 from each direction. The XFEL pulses were provided 93 μm downstream from the pump focal centre at a time of 10 ms for the pre-flashed samples and 15 ms for the non-pre-flashed samples following the second excitation laser (the total time following the first flash is therefore 20 ms in both cases) (Fig. 1d, e). At the flow rate of 5.6 μl min−1, each pump-illuminated crystal for the pump-on XFEL images was separated by 528 μm, which was long enough to avoid influence from the previous excitation lasers, as the illumination point by the last laser extends to only 120 μm from its centre of illumination.

Data processing

The background of the detector was estimated by averaging dark images and subtracted from diffraction patterns. Diffraction images were filtered by the program Cheetah35,36 adapted to SACLA, and processed by the program CrystFEL37. The parameters ‘min-snr’, ‘thresholds’ and ‘min-gradient’ used for peak detection during spot finding were as follows: 6, 500 and 10,000, respectively, for the pre-flashed dark data and pre-flashed 2F data; 5, 500 and 10,000 for the non-pre-flashed dark data; and 5, 500 and 5,000 for the non-pre-flashed 2F data. Indexing was performed using DirAx38 and Mosflm39 with peak integration parameters int-radius = 3, 5, 7, where the unit cell information was provided to avoid integration of the diffraction images from crystals with a longer c axis. Before Monte-Carlo integration, the indexed images with either a diffractionresolutionlimit lower than 4.2 Å or a num_peaks less than 400 were discarded. The numbers of total images collected, the hit images filtered by Cheetah, the indexed images and the number of images used for refinement for pre-flashed samples were as follows: 408,071, 76,047, 64,985 and 27,497, respectively, for the dark data; and 273,550, 60,885, 51,482 and 21,680 for the 2F data. The corresponding numbers for the non-pre-flashed samples are 462,343, 70,083, 54,956 and 22,341 for the dark data; and 876,874, 165,463, 63,711 and 23,903 for the 2F data (Extended Data Table 1). The Lorentz factor for still snapshots was applied manually to the averaged intensities of the pre-flashed dark and 2F data sets40. The statistics for the data collection are given in Extended Data Tables 1, 2, which show that our data for both dark and 2F states had a resolution of 2.35 Å for the pre-flashed samples and 2.5 Å for the non-pre-flashed samples, based on the cut-off value of around 50% for CC1/2. We should point out that the overall multiplicities of all our data are very high, and even at the highest resolution shell, the multiplicity exceeded 500. Together with the facts that: (i) the values of CC1/2 in the highest resolution shells are reasonably high; (ii) CC1/2 decreases gradually without any abnormal disrupt from the low resolution to higher resolution shells; and (iii) the value of CC1/2 in the highest resolution shell is reasonably consistent with the value of I/σ(I) (theoretically 0.5 of CC1/2 corresponds to 2.0 of I/σ(I), and any substantial deviations from these values are indicators of systematic bias and/or problems in the error model41), we consider that the quality of the data in the present study is sufficiently high to allow us to reveal the small structural changes induced by the flash illuminations.

Structural refinement for the pre-flashed dark-adapted S1 state, the non-pre-flashed S1 state and the 2F state

The initial phases up to 4 Å resolution were obtained by molecular replacement with the program Phaser in the CCP4 suite42 using the 1.95 Å resolution XFEL structure of PSII (PDB accession code 4UB6; ref. 5) as the search model, in which the OEC and its direct ligands, QB, waters and glycerol molecules were omitted. After a few cycles of rigid body refinement and subsequent real space density modification using solvent flattening, histogram matching and non-crystallographic symmetry averaging with the program DM in the CCP4 suite42, the electron density map obtained showed features clear enough to allow us to build the model with confidence. Structural refinement was performed with Phenix43 and the model was manually modified with COOT44. After a few cycles of restrained refinement, water molecules were placed in positions corresponding to the water molecules in the higher-resolution structure where positive peaks higher than 3.5σ in the mFo − DFc map were clearly identified. Then, based on the resulting mFo − DFc map, water molecules and glycerol molecules were additionally located when positive peaks higher than 3σ were found. QB, the OEC and its direct ligand residues were modelled in the final step. In our previous XFEL structural analysis at 1.95 Å resolution5, the exact positions of the oxo-bridges were identified in the mFo − DFc map by omitting the oxo-bridged oxygen atoms; however, when the restrained refinement was performed in the same way in this study, the temperature factors of the manganese and/or calcium atoms in OEC were not converged, and gave rise to very high values owing to the limited resolution of 2.35–2.5 Å. Thus, we built the OEC structure using the geometric restraints based on the Mn4CaO5 cluster in the 1.95 Å resolution XFEL structure, and applied tight distance restraints of σ = 0.02 Å to the Mn–O bonds, Ca–O bonds and Mn–ligand residue distances during the refinement. No restraints were given to the Mn–Mn distances and Mn–Ca distances. The Rfactor and Rfree values obtained were 0.133 and 0.171, respectively, for the pre-flashed dark state, 0.139 and 0.186 for the non-pre-flashed dark state, and 0.139 and 0.187 for the non-pre-flashed 2F state (Extended Data Table 1).

Structural refinement for the pre-flashed 2F state

Given the population of the S3 state estimated from the FTIR measurement, the diffraction data obtained from the 2F illuminated sample is expected to contain those partly from the S2 and S1 states of PSII, which means that the resulted electron density would be a mixture of these S-states, including the S3 state. Even with a high-resolution data set, the structural refinement of the mixed states (or mixed structures) would be challenging, especially when the mixed structures are very similar, as in the case of PSII. As we did not have the structure corresponding to the S2 state, we refined the pre-flashed 2F data set as a mixture of the S1 state structure and the 2F state structure45. The occupancies for the S1 structure and the 2F state structure were set to 0.2 and 0.8, respectively, on the basis of the distributions of the temperature factors. The region in which the structure was refined in multiple states was selected on the basis of the large peaks observed in the isomorphous difference Fourier map, and the rest was refined as a single conformation. As a result, thirty-five amino acid residues, five water molecules, one non-haem iron, one QB and one OEC per monomeric PSII were modelled as multiple states. During the restrained refinement, the coordinate for the S1 state structure was fixed. Relatively loose geometric restraints for distance, angle and plane with relatively high sigma values (3–10 times larger than the default values) were applied to the OEC and QB during the refinement. Large positive peaks of 10.3σ for monomer A and 9.5σ for monomer B, found at a position with distances of 1.5 Å and 2.3 Å from O5 and Mn1D, respectively, were modelled as a new oxo-oxygen O6 with an occupancy of 0.4, which gave rise to a temperature factor similar to its nearby atoms. This is also consistent with the population of the S3-state estimated from the FTIR results. The Rfactor and Rfree values thus obtained were 0.129 and 0.176, respectively.

Data availability

The structure factors and atomic coordinates have been deposited in the Protein Data Bank (PDB) with accession numbers 5WS5 and 5WS6 for the pre-flashed dark-stable S1 state and 2F states, and 5GTH and 5GTI for the non-pre-flashed dark-stable S1 state and 2F state, respectively. All other data associated with this manuscript are available from the authors on reasonable request.

Accession codes Primary accessions Protein Data Bank

5GTH

5GTI

5WS5

5WS6

References

Shen, J.-R. The structure of photosystem II and the mechanism of water oxidation in photosynthesis. Annu. Rev. Plant Biol. 66, 23–48 (2015)

Article   CAS   PubMed   Google Scholar

Yano, J. & Yachandra, V. Mn4Ca cluster in photosynthesis: where and how water is oxidized to dioxygen. Chem. Rev. 114, 4175–4205 (2014)

Article   CAS   PubMed   PubMed Central   Google Scholar

Najafpour, M. M. et al. Manganese compounds as water-oxidizing catalysts: From the natural water-oxidizing complex to nanosized manganese oxide structures. Chem. Rev. 116, 2886–2936 (2016)

Article   CAS   PubMed   Google Scholar

Umena, Y., Kawakami, K., Shen, J.-R. & Kamiya, N. Crystal structure of oxygen-evolving photosystem II at a resolution of 1.9 Å. Nature 473, 55–60 (2011)

CAS   ADS   PubMed   Google Scholar

Suga, M. et al. Native structure of photosystem II at 1.95 Å resolution viewed by femtosecond X-ray pulses. Nature 517, 99–103 (2015)

CAS   ADS   PubMed   Google Scholar

Siegbahn, P. E. A structure-consistent mechanism for dioxygen formation in photosystem II. Chemistry 14, 8290–8302 (2008)

Article   CAS   PubMed   Google Scholar

Siegbahn, P. E. Structures and energetics for O2 formation in photosystem II. Acc. Chem. Res. 42, 1871–1880 (2009)

Article   CAS   PubMed   Google Scholar

Kok, B., Forbush, B. & McGloin, M. Cooperation of charges in photosynthetic O2 evolution-I. A linear four step mechanism. Photochem. Photobiol. 11, 457–475 (1970)

Article   CAS   PubMed   Google Scholar

Kern, J. et al. Simultaneous femtosecond X-ray spectroscopy and diffraction of photosystem II at room temperature. Science 340, 491–495 (2013)

Article   CAS   ADS   PubMed   PubMed Central   Google Scholar

Kupitz, C. et al. Serial time-resolved crystallography of photosystem II using a femtosecond X-ray laser. Nature 513, 261–265 (2014)

Article   CAS   ADS   PubMed   PubMed Central   Google Scholar

Kern, J. et al. Taking snapshots of photosynthetic water oxidation using femtosecond X-ray diffraction and spectroscopy. Nat. Commun. 5, 4371 (2014)

Article   CAS   ADS   PubMed   Google Scholar

Young, I. D. et al. Structure of photosystem II and substrate binding at room temperature. Nature 540, 453–457 (2016)

Article   CAS   ADS   PubMed   PubMed Central   Google Scholar

Sugahara, M. et al. Grease matrix as a versatile carrier of proteins for serial crystallography. Nat. Methods 12, 61–63 (2015)

Article   CAS   PubMed   Google Scholar

Shevela, D., Eaton-Rye, J. J., Shen, J. R. & Govindjee . Photosystem II and the unique role of bicarbonate: a historical perspective. Biochim. Biophys. Acta 1817, 1134–1151 (2012)

Article   CAS   PubMed   Google Scholar

Rapatskiy, L. et al. Detection of the water-binding sites of the oxygen-evolving complex of Photosystem II using W-band 17O electron-electron double resonance-detected NMR spectroscopy. J. Am. Chem. Soc. 134, 16619–16634 (2012)

Article   CAS   PubMed   Google Scholar

Yamanaka, S. et al. Possible mechanisms for the O-O bond formation in oxygen evolution reaction at the CaMn4O5(H2O)(4) cluster of PSII refined to 1.9 Å X-ray resolution. Chem. Phys. Lett. 511, 138–145 (2011)

Article   CAS   ADS   Google Scholar

Saito, K., Rutherford, A. W. & Ishikita, H. Energetics of proton release on the first oxidation step in the water-oxidizing enzyme. Nat. Commun. 6, 8488 (2015)

Article   CAS   ADS   PubMed   Google Scholar

Takaoka, T., Sakashita, N., Saito, K. & Ishikita, H. pKa of a proton-conducting water chain in photosystem II. J. Phys. Chem. Lett. 7, 1925–1932 (2016)

Article   CAS   PubMed   Google Scholar

Isobe, H., Shoji, M., Shen, J. R. & Yamaguchi, K. Chemical equilibrium models for the S3 state of the oxygen-evolving complex of photosystem II. Inorg. Chem. 55, 502–511 (2016)

Article   CAS   PubMed   Google Scholar

Siegbahn, P. E. Water oxidation mechanism in photosystem II, including oxidations, proton release pathways, O-O bond formation and O2 release. Biochim. Biophys. Acta 1827, 1003–1019 (2013)

Article   CAS   PubMed   Google Scholar

Cox, N. et al. Photosynthesis. Electronic structure of the oxygen-evolving complex in photosystem II prior to O-O bond formation. Science 345, 804–808 (2014)

Article   CAS   ADS   PubMed   Google Scholar

Pérez-Navarro, M., Neese, F., Lubitz, W., Pantazis, D. A. & Cox, N. Recent developments in biological water oxidation. Curr. Opin. Chem. Biol. 31, 113–119 (2016)

Article   PubMed   Google Scholar

Shoji, M., Isobe, H. & Yamaguchi, K. QM/MM study of the S2 to S3 transition reaction in the oxygen-evolving complex of photosystem II. Chem. Phys. Lett. 636, 172–179 (2015)

Article   CAS   ADS   Google Scholar

Isobe, H. et al. Theoretical illumination of water-inserted structures of the CaMn4O5 cluster in the S2 and S3 states of oxygen-evolving complex of photosystem II: full geometry optimizations by B3LYP hybrid density functional. Dalton Trans. 41, 13727–13740 (2012)

Article   CAS   PubMed   Google Scholar

Saito, K., Rutherford, A. W. & Ishikita, H. Mechanism of tyrosine D oxidation in Photosystem II. Proc. Natl Acad. Sci. USA 110, 7690–7695 (2013)

Article   CAS   ADS   PubMed   PubMed Central   Google Scholar

Nakamura, S. & Noguchi, T. Infrared detection of a proton released from tyrosine YD to the bulk upon its photo-oxidation in photosystem II. Biochemistry 54, 5045–5053 (2015)

Article   CAS   PubMed   Google Scholar

Shen, J.-R. & Inoue, Y. Binding and functional properties of two new extrinsic components, cytochrome c-550 and a 12-kDa protein, in cyanobacterial photosystem II. Biochemistry 32, 1825–1832 (1993)

Article   CAS   PubMed   Google Scholar

Shen, J.-R. & Kamiya, N. Crystallization and the crystal properties of the oxygen-evolving photosystem II from Synechococcus vulcanus. Biochemistry 39, 14739–14744 (2000)

Article   CAS   PubMed   Google Scholar

Noguchi, T. & Sugiura, M. Flash-induced FTIR difference spectra of the water oxidizing complex in moderately hydrated photosystem II core films: effect of hydration extent on S-state transitions. Biochemistry 41, 2322–2330 (2002)

Article   CAS   PubMed   Google Scholar

Suzuki, H., Sugiura, M. & Noguchi, T. pH dependence of the flash-induced S-state transitions in the oxygen-evolving center of photosystem II from Thermosynechoccocus elongatus as revealed by Fourier transform infrared spectroscopy. Biochemistry 44, 1708–1718 (2005)

Article   CAS   PubMed   Google Scholar

Noguchi, T. & Sugiura, M. Flash-induced Fourier transform infrared detection of the structural changes during the S-state cycle of the oxygen-evolving complex in photosystem II. Biochemistry 40, 1497–1502 (2001)

Article   CAS   PubMed   Google Scholar

Suzuki, H., Sugiura, M. & Noguchi, T. Monitoring proton release during photosynthetic water oxidation in photosystem II by means of isotope-edited infrared spectroscopy. J. Am. Chem. Soc. 131, 7849–7857 (2009)

Article   CAS   PubMed   Google Scholar

Tono, K. et al. Diverse application platform for hard X-ray diffraction in SACLA (DAPHNIS): application to serial protein crystallography using an X-ray free-electron laser. J. Synchrotron Radiat. 22, 532–537 (2015)

Article   CAS   PubMed   PubMed Central   Google Scholar

Kameshima, T. et al. Development of an X-ray pixel detector with multi-port charge-coupled device for X-ray free-electron laser experiments. Rev. Sci. Instrum. 85, 033110 (2014)

Article   ADS   PubMed   Google Scholar

Barty, A. et al. Cheetah: software for high-throughput reduction and analysis of serial femtosecond X-ray diffraction data. J. Appl. Crystallogr. 47, 1118–1131 (2014)

Article   CAS   PubMed   PubMed Central   Google Scholar

Nakane, T. et al. Data processing pipeline for serial femtosecond crystallography at SACLA. J. Appl. Crystallogr. 49, 1035–1041 (2016)

Article   CAS   PubMed   PubMed Central   Google Scholar

White, T. A. et al. CrystFEL: a software suite for snapshot serial crystallography. J. Appl. Crystallogr. 45, 335–341 (2012)

Article   CAS   Google Scholar

Duisenberg, A. J. M. Indexing in single-crystal diffractometry with an obstinate list of reflections. J. Appl. Crystallogr. 25, 92–96 (1992)

Article   CAS   Google Scholar

Battye, T. G. G., Kontogiannis, L., Johnson, O., Powell, H. R. & Leslie, A. G. W. iMOSFLM: a new graphical interface for diffraction-image processing with MOSFLM. Acta Crystallogr. D Biol. Crystallogr. 67, 271–281 (2011)

Article   CAS   PubMed   PubMed Central   Google Scholar

Kabsch, W. Processing of X-ray snapshots from crystals in random orientations. Acta Crystallogr. D Biol. Crystallogr. 70, 2204–2216 (2014)

Article   CAS   PubMed   PubMed Central   Google Scholar

Karplus, P. A. & Diederichs, K. Assessing and maximizing data quality in macromolecular crystallography. Curr. Opin. Struct. Biol. 34, 60–68 (2015)

Article   CAS   PubMed   PubMed Central   Google Scholar

Collaborative Computational Project, Number 4. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D Biol. Crystallogr. 50, 760–763 (1994)

Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221 (2010)

Article   CAS   PubMed   PubMed Central   Google Scholar

Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 (2004)

Article   PubMed   Google Scholar

Nango, E. et al. A three-dimensional movie of structural changes in bacteriorhodopsin. Science 354, 1552–1557 (2016)

Article   CAS   ADS   PubMed   Google Scholar

Download references

Acknowledgements

We thank K. Kawakami and N. Kamiya for sharing unpublished information, H. Mino for information on the flash illumination conditions and H. Ago and K. Yamaguchi for discussions. This work was supported by a program for promoting the enhancement of research universities at Okayama University, JSPS KAKENHI Grant Nos JP15H01642, JP16H06162, JP16H06296 (M. Suga), JP16K21181 (F.A.), JP15H05588 (Y.U.), JP15H03841, JP15H01055 (M.K.) and JP24000018 (J.-R.S.), an X-ray Free Electron Laser Priority Strategy Program (J.-R.S., S.Iw.) from MEXT, Japan, an Asahi Glass Foundation (F.A.), a Kato Memorial Bioscience Foundation (F.A.), an Inamori Foundation (M. Suga), the Research Acceleration Program from Japan Science and Technology agency (JST) (S.Iw.), PRESTO from JST (M.K. and F.A.), a grant from Pioneering Project ‘Dynamic Structural Biology’ of RIKEN (M.K.), and the Strategic Priority Research Program of CAS (XDB17030100) (J.-R.S.). The XFEL experiments were performed at beamline 3 of SACLA with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (proposal nos 2013B1259, 2014A1243, 2014A6927, 2014B1281, 2014B6927, 2014B8048, 2015A1108, 2015A6522, 2015B2108, 2015B6522, 2015B8044, 2016A2542, 2016A6621 and 2016A8033), and we thank the staff at SACLA for their help. We also acknowledge computational support from the SACLA HPC system and the Mini-K supercomputer system.

Author information Author notes

Michihiro Suga, Fusamichi Akita, Michihiro Sugahara, Minoru Kubo and Yoshiki Nakajima: These authors contributed equally to this work.

Authors and Affiliations

Research Institute for Interdisciplinary Science and Graduate School of Natural Science and Technology, Okayama University, 3-1-1 Tsushima Naka, Okayama, 700-8530, Japan

Michihiro Suga, Fusamichi Akita, Yoshiki Nakajima, Yasufumi Umena, Makoto Nakabayashi, Takahiro Yamane, Takamitsu Nakano, Shinichiro Yonekura, Long-Jiang Yu, Tomohiro Sakamoto, Taiki Motomura, Jing-Hua Chen & Jian-Ren Shen

Japan Science and Technology Agency, PRESTO, 4-1-8 Honcho, Kawaguchi, 332-0012, Saitama, Japan

Fusamichi Akita & Minoru Kubo

RIKEN SPring-8 Center, 1-1-1 Kouto, Sayo-cho, Sayo-gun, 679-5148, Hyogo, Japan

Michihiro Sugahara, Minoru Kubo, Keitaro Yamashita, Mamoru Suzuki, Tetsuya Masuda, Shigeyuki Inoue, Tetsunari Kimura, Takashi Nomura, Takaki Hatsui, Eriko Nango, Rie Tanaka, Hisashi Naitow, Yoshinori Matsuura, Ayumi Yamashita, Masaki Yamamoto, Makina Yabashi, Tetsuya Ishikawa & So Iwata

Department of Biological Sciences, Graduate School of Science, The University of Tokyo, 2-11-16 Yayoi, Bunkyo-ku, 113-0032, Tokyo, Japan

Takanori Nakane & Osamu Nureki

Institute for Protein Research, Osaka University, Yamadaoka, Suita, 565-0871, Osaka, Japan

Mamoru Suzuki

Division of Food Science and Biotechnology, Graduate School of Agriculture, Kyoto University, Gokasho, 611-0011, Uji, Kyoto, Japan

Tetsuya Masuda

Department of Cell Biology and Anatomy, Graduate School of Medicine, The University of Tokyo, Hongo, Bunkyo-ku, 113-0033, Tokyo, Japan

Shigeyuki Inoue

Department of Chemistry, Graduate School of Science, Kobe University, 1-1 Rokkodai, Nada-ku, 657-8501, Japan, Kobe

Tetsunari Kimura

Department of Picobiology, Graduate School of Life Science, University of Hyogo, 3-2-1 Kouto, Kamigori-cho, Ako-gun, 678-1297, Hyogo, Japan

Taiki Motomura & Jian-Ren Shen

Key Laboratory of Photobiology, Institute of Botany, Chinese Academy of Sciences, No.20 Nanxincun, Xiangshan, 100093, Beijing, China

Jing-Hua Chen & Jian-Ren Shen

Division of Material Science, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, 464-8602, Nagoya, Japan

Yuki Kato & Takumi Noguchi

Japan Synchrotron Radiation Research Institute, 1-1-1 Kouto, Sayo, 679-5198, Hyogo, Japan

Kensuke Tono, Yasumasa Joti, Takashi Kameshima & Makina Yabashi

Department of Cell Biology, Graduate School of Medicine, Kyoto University, Yoshidakonoe-cho, Sakyo-ku, 606-8501, Kyoto, Japan

Eriko Nango & So Iwata

Contributions

J.-R.S. conceived the project; M. Suga, F.A., M.Sugah., M.K., Y.U. and J.-R.S. contributed to the design of the experiment; F.A., Y.N., M.N. and T. Nakano grew the cells and purified the PSII samples; F.A. and Y.N. prepared the PSII crystals; M.Sugah., E.N. and R.T. developed the sample delivery system; F.A., Y.N., M.N., Y.U. and M.Sugah. tested and optimized buffer and crystal suspension conditions for injection; M.Suz., T.Ma., S.In., T.Mo., J.-H.C., H.N., Y.M. and A.Y. operated the injector; K.T., Y.J., T.Ka., T.H., M.Yab., T.I. and S.Iw. developed the diffraction instrumentation; M.K., T.Ki. and T.Nom. designed and optimized the laser excitation scheme and aligned the lasers; M. Suga, F.A., M.Sugah., Y.N., T. Nakane, K.Y., M.N., Y.U., M.Suz., T.Ma., S.In., S.Y., L.-J.Y., T.Mo., J.-H.C., R.T., H.N., Y.M., A.Y. and J.-R.S. participated in collection of the X-ray diffraction data at SACLA; T. Nakane, K.Y., M.Yam., O.N. and S.Iw. developed the data evaluation and/or hit-finding programs; Y.K., F.A. and T.Nog. performed FTIR analysis; M. Suga, T.Y. and T.S. analysed the femtosecond crystallography X-ray diffraction data; M. Suga refined the structure, calculated the electron density maps and made the figures; M. Suga and J.-R.S. wrote the manuscript, and all the authors participated in the discussion of the results and writing of the paper.

Corresponding authors

Correspondence to So Iwata or Jian-Ren Shen.

Ethics declarations Competing interests

The authors declare no competing financial interests.

Additional information

Reviewer Information Nature thanks R. Debus, J. Murray and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Extended data figures and tables Extended Data Figure 1 Micro-sized crystals of PSII, its diffraction image and quality of the diffraction data sets.

a, Micro-sized crystals of PSII used for TR-SFX. b, A diffraction image from a micro-sized PSII crystal obtained with a single pulse from SACLA-XFEL. c–f, Quality of the data sets. CC1/2 (c), Rsplit (d), multiplicity (e) and (f) plotted against resolution for the dark data and the 2F data of the pre-flashed samples.