UC Berkeley

The mechanism of water oxidation in oxygenic photosynthesis, the process generating all oxygen on Earth, has remained elusive despite the maturity of the field of photosynthetic research. Water oxidation takes place at the oxygen-evolving complex (OEC) of photosystem II (PS II), a large dimeric protein located in the thylakoid membrane, and proceeds by the same mechanism in cyanobacteria, algae and all higher plants. Charge separation at the P680 pigment is followed by charge stabilization by reduction of a plastoquinone and oxidation of the OEC. Over the five steps of the Kok cycle and controlled by absorption of four photons at P680, the OEC stores four oxidizing equivalents prior to oxidizing two substrate water molecules to dioxygen. These steps take place on a microsecond to millisecond scale and may be activated in sequence by illuminating dark-adapted PS II with short flashes of visible light.

Studying the mechanism of water splitting at the oxygen-evolving complex (OEC) in photosystem II presents several challenges: it must be studied by a time-resolved method in order to track transient states in the cycle, at room temperature in order for the OEC to advance to these states, and with a method capable of resolving the atomic-level structure of a large transmembrane protein. A pump-probe "diffraction-before-destruction" experiment at an X-ray free electron laser (XFEL) addresses all these challenges. Diffraction is collected from each in a series of microcrystals, which may be advanced to a transient state and delivered to the XFEL beam under ambient conditions.

This dissertation describes a series of XFEL experiments that revealed the structure of PS II at high resolution in multiple illuminated states. PS II microcrystals are delivered to the XFEL beam with either a liquid jet or a drop-on-demand system in which droplets containing microcrystals are deposited by acoustic droplet ejection onto a kapton conveyor belt. Using visible lasers positioned along the path of the jet or droplets, crystals are illuminated to uniformly advance OEC centers, and the diffraction patterns from hundreds of thousands of individual crystals are combined to generate the diffraction dataset. X-ray emission spectra from the same crystals are collected simultaneously for evaluation of the redox state of the cluster to confirm turnover.

This work focuses on the XFEL data processing methods developments that enabled these experiments and analyzed the diffraction datasets they produced. An overhaul of real-time data processing at the beamline included development of the cctbx.xfel graphical user interface, which was used to filter crystallization batches and sample delivery conditions and to provide feedback on quality and completeness of datasets. Optimization of the crystal models allowed filtering of multiple crystal forms of PS II and resolved some apparent nonisomorphism in the remaining distribution of unit cells due to uncertainties during indexing. A position-dependent correction was applied to integrated intensities to account for a highly asymmetric shadow on the detector, and several improvements to the merging program cxi.merge were critical to successfully merging these data. Finally, structure solution and analysis of a series of datasets were streamlined with various custom tools for automation, parallelization and calculations on the atomic positions.

PS II structures are reported in four metastable and two transient states of the Kok cycle, of which four have never been reported to high resolution and two are reported at the highest resolution at room temperature to date. Analysis of these structures reveals water insertion between 150 and 400 µs after illumination of the S2 state, and a detailed analysis of the series of structures reveals possible channels for substrate water approach. Another structure in the S3 state with ammonia bound reveals the probable position of a water coordinating the OEC that does not participate in the water oxidation mechanism. The aggregated evidence from these structures excludes at least one proposed mechanism and produces three favored mechanisms for water oxidation, involving some subset of the following in O-O bond formation: water W3 coordinated to Ca, water W2 coordinated to Mn4, bridging oxo O5 and inserted water Ox. Investigation of additional transient states near O-O bond formation may distinguish between these mechanisms and resolve the water oxidation mechanism.