The structure of plant photosystem I ﬁ supercomplex at 2.6 Å … · philic cyanobacteria, but...

In the format provided by the authors and unedited.

The structure of plant photosystem I supercomplexat 2.6 Å resolutionYuval Mazor*†, Anna Borovikova, Ido Caspy and Nathan Nelson*

Four elaborate membrane complexes carry out the light reaction of oxygenic photosynthesis. Photosystem (PS) I is one oftwo large reaction centres responsible for converting light photons into the chemical energy needed to sustain life. In thethylakoid membranes of plants, PSI is found together with its integral light-harvesting antenna, LHCI, in a membranesupercomplex containing hundreds of light-harvesting pigments. Here we report the crystal structure of plant PSI–LHCI at2.6 Å resolution. The structure reveals the configuration of PsaK, a core subunit important for state transitions in plants, aconserved network of water molecules surrounding the electron transfer centres and an elaborate structure oflipids bridging PSI and its LHCI antenna. We discuss the implications of the structure for energy transfer and theevolution of PSI.

The light reactions of oxygenic photosynthesis determines thecomposition of the Earth’s atmosphere and surface. Twolarge reaction centres are essential parts of the light reactions

and convert absorbed photons into chemical energy. PSII producesthe oxygen in the atmosphere and PSI generates the most negativepotential in nature and thus largely determines the global amountof enthalpy in living systems. Oxygenic photosynthesis evolvedover 3 billion years ago in primordial cyanobacteria-like organ-isms1–3. Later, approximately 1.5 billion years ago, the first photo-synthetic eukaryotes appeared, eventually evolving into landplants roughly 0.5 billion years ago4. The basic building blocks ofphotosynthesis are remarkably conserved. The architectures ofboth photosystems have been determined by numerous tech-niques5–7. The highest resolutions have been obtained from thermo-philic cyanobacteria, but representative structures from eukaryoticchloroplasts, especially plants, are scarce8–11. The first crystal struc-ture of plant PSI was reported in 2003 at 4.4 Å resolution9 andrecently was improved to 2.8 Å resolution12,13.

The plant PSI–LHCI supercomplex consists of 16 subunits,which coordinate 192 light-harvesting pigments. In spite of thisremarkable complexity, the quantum efficiency of this pigmentnetwork is close to unity14. This high quantum efficiency probablyemerges from the orientation of the pigments for optimum energytransfer, and from the interaction of the pigments with the sup-porting protein environment. In nature, the amount and qualityof light available to the organism constantly changes. Plantsadapt by changing the antenna–reaction centre association in aprocess called state transitions, or by dissipating excess energyusing non-photochemical energy quenching pathways15. Thesedynamic aspects of light harvesting are challenging to decipherstructurally. Herein, we report the crystal structure of the plantPSI–LHCI supercomplex at a resolution of 2.6 Å. To achieve thisresult, we used extremely high data multiplicities, which allowedus to recover phase information from the native atoms of thecomplex. The PSI–LHCI structure contains an almost completedescription of PsaK, a subunit present in all known PSI complexesand is important for state transitions. An elaborate network ofwater molecules surround the highly conserved iron-sulfur clusters.

Additionally, more than 35 lipids are present and connect the twolarge membrane complexes. Using Förster energy transfer calcu-lations, we determined the most important pigment pairs respon-sible for energy transfer between LHCI and PSI. Theseconnections may be crucial for the regulation of excitationenergy transfer between the two complexes.

Results and discussionSingle-wavelength anomalous diffraction phasing of PSI–LHCI.To get the most accurate measurement of an anomalous signal, itis vital to obtain highly redundant data. This outcome can beachieved using a low radiation dose, multi axis data collectionstrategy16 but this strategy proved to be problematic in the case ofPSI crystals which typically diffract to medium/low resolution (seeSupplementary Table 1 for a representative data set). Instead, wegenerated a large number (179) of partial datasets collected atmedium beam exposure and assembled them into a large, highlyredundant dataset. Clustering analysis of this large collection ledto a merged dataset composed of 159 individual datasets with anoverall multiplicity of 320 (Supplementary Fig. 1b). The phaseestimates from this dataset were accurate enough to solve thestructure and proved to be important up to the final steps ofrefinement. To obtain a high-resolution dataset, a similar, multi-crystal approach was applied on a smaller collection of crystalsdiffracting to high resolution. The final, high-resolution datasetwas assembled from ten different crystals, which formed a tightcluster with small cell variation and extended to a resolution of2.6 Å (Supplementary Tables 1 and 2).

To summarize, the large number of ligands together with theinherent flexibility of the PSI–LHCI supercomplex resulted in thedeterioration of weak, high-resolution, diffraction data. Carefullymerging data from multiple crystals allowed us to recover thisweak signal and solve the structure of this large membranecomplex to a resolution of 2.6 Å (see Supplementary Fig. 3 forsamples from the electron density map). The final model contains16 protein subunits, 240 ligands (for a comprehensive list seeSupplemental Table 3) and more than 200 water molecules (Fig. 1a).

Department of Biochemistry, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, 69978, Israel. †Present address: School ofMolecular Sciences, The Biodesign Institute, Arizona State University, Tempe, Arizona 85281, USA.*e-mail: [email protected] (Y.M); [email protected] (N.N.)

ARTICLESPUBLISHED: XX XX 2017 | VOLUME: 3 | ARTICLE NUMBER: 17014

NATURE PLANTS 3, 17014 (2017) | DOI: 10.1038/nplants.2017.14 | www.nature.com/natureplants 1

The structure of plant photosystem I supercomplexat 2.6 Å resolutionYuval Mazor*†, Anna Borovikova, Ido Caspy and Nathan Nelson*

Four elaborate membrane complexes carry out the light reaction of oxygenic photosynthesis. Photosystem (PS) I is one oftwo large reaction centres responsible for converting light photons into the chemical energy needed to sustain life. In thethylakoid membranes of plants, PSI is found together with its integral light-harvesting antenna, LHCI, in a membranesupercomplex containing hundreds of light-harvesting pigments. Here we report the crystal structure of plant PSI–LHCI at2.6 Å resolution. The structure reveals the configuration of PsaK, a core subunit important for state transitions in plants, aconserved network of water molecules surrounding the electron transfer centres and an elaborate structure oflipids bridging PSI and its LHCI antenna. We discuss the implications of the structure for energy transfer and theevolution of PSI.

The light reactions of oxygenic photosynthesis determines thecomposition of the Earth’s atmosphere and surface. Twolarge reaction centres are essential parts of the light reactions

and convert absorbed photons into chemical energy. PSII producesthe oxygen in the atmosphere and PSI generates the most negativepotential in nature and thus largely determines the global amountof enthalpy in living systems. Oxygenic photosynthesis evolvedover 3 billion years ago in primordial cyanobacteria-like organ-isms1–3. Later, approximately 1.5 billion years ago, the first photo-synthetic eukaryotes appeared, eventually evolving into landplants roughly 0.5 billion years ago4. The basic building blocks ofphotosynthesis are remarkably conserved. The architectures ofboth photosystems have been determined by numerous tech-niques5–7. The highest resolutions have been obtained from thermo-philic cyanobacteria, but representative structures from eukaryoticchloroplasts, especially plants, are scarce8–11. The first crystal struc-ture of plant PSI was reported in 2003 at 4.4 Å resolution9 andrecently was improved to 2.8 Å resolution12,13.

The plant PSI–LHCI supercomplex consists of 16 subunits,which coordinate 192 light-harvesting pigments. In spite of thisremarkable complexity, the quantum efficiency of this pigmentnetwork is close to unity14. This high quantum efficiency probablyemerges from the orientation of the pigments for optimum energytransfer, and from the interaction of the pigments with the sup-porting protein environment. In nature, the amount and qualityof light available to the organism constantly changes. Plantsadapt by changing the antenna–reaction centre association in aprocess called state transitions, or by dissipating excess energyusing non-photochemical energy quenching pathways15. Thesedynamic aspects of light harvesting are challenging to decipherstructurally. Herein, we report the crystal structure of the plantPSI–LHCI supercomplex at a resolution of 2.6 Å. To achieve thisresult, we used extremely high data multiplicities, which allowedus to recover phase information from the native atoms of thecomplex. The PSI–LHCI structure contains an almost completedescription of PsaK, a subunit present in all known PSI complexesand is important for state transitions. An elaborate network ofwater molecules surround the highly conserved iron-sulfur clusters.

Additionally, more than 35 lipids are present and connect the twolarge membrane complexes. Using Förster energy transfer calcu-lations, we determined the most important pigment pairs respon-sible for energy transfer between LHCI and PSI. Theseconnections may be crucial for the regulation of excitationenergy transfer between the two complexes.

Results and discussionSingle-wavelength anomalous diffraction phasing of PSI–LHCI.To get the most accurate measurement of an anomalous signal, itis vital to obtain highly redundant data. This outcome can beachieved using a low radiation dose, multi axis data collectionstrategy16 but this strategy proved to be problematic in the case ofPSI crystals which typically diffract to medium/low resolution (seeSupplementary Table 1 for a representative data set). Instead, wegenerated a large number (179) of partial datasets collected atmedium beam exposure and assembled them into a large, highlyredundant dataset. Clustering analysis of this large collection ledto a merged dataset composed of 159 individual datasets with anoverall multiplicity of 320 (Supplementary Fig. 1b). The phaseestimates from this dataset were accurate enough to solve thestructure and proved to be important up to the final steps ofrefinement. To obtain a high-resolution dataset, a similar, multi-crystal approach was applied on a smaller collection of crystalsdiffracting to high resolution. The final, high-resolution datasetwas assembled from ten different crystals, which formed a tightcluster with small cell variation and extended to a resolution of2.6 Å (Supplementary Tables 1 and 2).

To summarize, the large number of ligands together with theinherent flexibility of the PSI–LHCI supercomplex resulted in thedeterioration of weak, high-resolution, diffraction data. Carefullymerging data from multiple crystals allowed us to recover thisweak signal and solve the structure of this large membranecomplex to a resolution of 2.6 Å (see Supplementary Fig. 3 forsamples from the electron density map). The final model contains16 protein subunits, 240 ligands (for a comprehensive list seeSupplemental Table 3) and more than 200 water molecules (Fig. 1a).

Department of Biochemistry, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv, 69978, Israel. †Present address: School ofMolecular Sciences, The Biodesign Institute, Arizona State University, Tempe, Arizona 85281, USA.*e-mail: [email protected] (Y.M); [email protected] (N.N.)

ARTICLESPUBLISHED: XX XX 2017 | VOLUME: 3 | ARTICLE NUMBER: 17014

NATURE PLANTS 3, 17014 (2017) | DOI: 10.1038/nplants.2017.14 | www.nature.com/natureplants 1

ThestructureofplantphotosystemIsupercomplexat2.6Åresolution.YuvalMazor,AnnaBorovikova,IdoCaspyandNathanNelson.

Supplementary Figure 1: Cluster analysis of data sets. a. Unit cell based cluster analysis of all collected data sets. b. The final merged datasets composed of 159 sets, could not have been further subdivide without significant signal loss. c. The final high resolution data sets contain a set of virtually isomorphic crystals.

© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

SUPPLEMENTARY INFORMATIONVOLUME: 3 | ARTICLE NUMBER: 17014

NATURE PLANTS | DOI: 10.1038/nplants.2017.14 | www.nature.com/natureplants 1

http://dx.doi.org/10.1038/nplants.2017.14

Supplementary Figure 2: R-factor analysis and correlation statistics of data sets. a. R-factor (calculated using XSCALE 1) plotted for each data set, showing a marked increase after 159 sets. b. Correlation coefficient (CC1/2 calculated using XSCALE) from sets with increasing multiplicities showing an increase in usable data as multiplicity increases.



SUPPLEMENTARY INFORMATION


1X 5X 20X 100X 159X HiRes

Resolution(outershell) 50–2.7(2.75-2.7) 50-2.6(2.64-

2.6)

Beamline SLSPXIII ESRFID30/SLSPXI

Wavelength1Å 1.74 1

spacegroup P212121

a,b,c(Å) 189.7,202.4,213.5 188.8,201.4,213.1

α,β, 90,90,90

Completeness 64.5(8.6)

83.7(19.7)

99.3(87.1)

100.0(100.0)

100(100)

99.9(99.4)

Multiplicity 2.8(1)

10.6(1.1)

41.3(4.1)

190.9(196.6)

332(329)

84.6(83.9)

Totalobservations

392,963(980)

1,976,340(2390)

9,066,979(38,495)

42,240,494(2,135,573)

73,665,482(3,589,408)

21,058,934(1,020,601)

Uniqueobservations

142,337(936)

186,140(2153)

219,708(9458)

221,273(10,863)

221,638(10,897)

248,807(12168)

Rpim 0.125(0.491)

0.075(0.586)

0.051(0.346)

0.031(4.733)

0.032(-12)

0.054(0.792)

I/sigma 3(0)

5.3(0)

9.8(0.2)

16(0.2)

17.4(0)

9.8(1.5)

CC1/2 99.1(19.4) 99.5(0.7) 99.6(33.4) 100(58.2) 100(62.9) 99.7(26.5)

I/sigma(3Å) 0.2 0.6 1.9 2.2 2.3 2.9

CCanom(5.5Å) -10 2 6.5 20.3 20.7 -10

Supplementary Table 1: Data collection and merging statistics. Final scaling and merging was done using AIMLESS 2. The number of partial data sets used to calculate each column is indicated at the top (1X to 159X). Hi Res refers to the collection of datasets shown in Supplementary figure 1C.





HiResResolution(outershell) 50-2.6(2.64-2.6)Rwork/Rfree(%) 20.6/23.7No.ofchains 16No.ofligands 242B-factorligands 107.6No.ofwaters 204B-factorwaters 76.8B-factorprotein 99.7R.m.sdeviationsBondlength(Å) 0.004Bondangles(o) 1.75Ramachandranstatistics:Favored(%) 96Allowed(%) 3.9Outliers(%) 0.1Clashcore 15.6

Supplementary Table 2: Refinement statistics calculated by PHENIX 3.





LigandsinPSI-LHCI

ligand occurrences

4Y28 4XK8 5L8R

Chlorophyll A 146 143 143

Chlorophyll B 9 12 13

Iron sulfur clusters 3 3 3

Phylloquinone 2 2 2

carotenoids

Beta carotene 21 26 26

Lutein 10 5 7

Violaxanthin 0 4 2

Zeaxanthin 1 0 1

lipids

Monogalactosyl

Diglyceride

7 3 20

Phosphatidyl

Diglyceride

5 6 7

Digalactosyl

Diglyceride

1 1 5

Detergents 2 4 10

Ca2+ 1 0 2

240 Supplementary Table 3: Ligands in PSI-LHCI





Additionalfigures

Supplementary Figure 3: Electron density maps. a. 2Fo-Fc maps (contoured at 1.5s) Water molecuels located around Fx. Portions of PsaA and PsaB are shown as well. b. 2Fo-Fc maps (contoured at 1.5s) showing P700 and some of its coordinating residues. c. Omit map showing positive density peak next to the chlorophyll B carbonyl oxygen. For calculation of omit map the carbonyl oxygen was deleted for the model, the coordinates of the remaining atoms were randomized to a mean 0.1Å from their starting positions and all model b factors were set to 90, this model was then used in 10 rounds of phenix.refine. Fo-Fc (in green) contour level is 2s, 2Fo-Fc map (in blue) is contoured at 1.7s. d. electron density map showing lipids head groups in the region separating PSI and LHCI. Four lipids are shown two can only be modeled as head groups and two are shown with some of their acyl tails (2Fo-Fc map countered at 1.5s).





Supplementary Figure 4: Six salt bridges surround FX and coordinate multiple water molecules. Multiple water molecules (red spheres) in the molecular environment of FX are found below the putative membrane surface (indicated by a red line). PsaA in blue, PsaB in pink and PsaC in purple. Positively and negatively charged residues are shown as sticks. Only a subset of the water molecules is shown for clarity.

Supplementary Figure 5: Chlorophyll b molecules are coordinated by a mixture of polar and hydrophobic interactions. a and b. Pigment position 610 in lhca2 and lhca4 is another example where a negatively charged residue coordinates the C7 aldehyde group of chlorophyll-b. c and d. Position 609 is coordinated by hydrophobic interactions between phenylalanine and the p-electrons of the chlorophyll ring. In addition to polar interactions mediated by a water molecule in D and probably in C as well (compared to Figure 5 A, B and C which utilizes the polar amine in the arginine guanidinium group).





1. Kabsch,W.Integration,scaling,space-groupassignmentandpost-

refinement.ActaCrystallogrDBiolCrystallogr66,133-44(2010).2. Evans,P.R.&Murshudov,G.N.Howgoodaremydataandwhatisthe

resolution?ActaCrystallogrDBiolCrystallogr69,1204-14(2013).3. Adams,P.D.etal.PHENIX:acomprehensivePython-basedsystemfor

macromolecularstructuresolution.ActaCrystallogrDBiolCrystallogr66,213-21(2010).





The structure of plant photosystem I ﬁ supercomplex at 2.6 Å … · philic cyanobacteria, but...

Documents

Transcript of The structure of plant photosystem I ﬁ supercomplex at 2.6 Å … · philic cyanobacteria, but...