Discovering oxygen channel topology in photosystem II using implicit ligand sampling and wavefront propagation

Highlights

• We examine oxygen migration pathways in the large multi polypeptide photosystem II protein complex.

• We introduce a new strategy for finding small molecule pathways in large multiprotein complexes.

• Our approach combines molecular dynamics, implicit ligand sampling and wavefront propagation within a three dimensional free energy map.

• Our results show the presence of two putative oxygen escape channels in photosystem II.

Abstract

Photosystem II (PSII) of photosynthesis uses water as substrate for its photochemical reaction. Molecular oxygen is one of the products of this complex reaction which uses the energy of light to oxidize water and reduce plastoquinone. The active site of PSII is buried deep within the protein, which raises the question of whether there are specific access channels guiding substrate water to the site of catalysis and product oxygen away from it. Substrate/product channels have been proposed to exist and serve various functions in PSII, however the preferred paths have not been unambiguously identified. We investigated oxygen transport between the active site of PSII and the solvent. For this purpose, we have applied molecular dynamics simulations followed by implicit ligand sampling. We then found minimal cost pathways for the oxygen through the protein and obtained topology maps of the oxygen-accessible part of a chemical labyrinth. Application of this new strategy led to identification of two oxygen channels in the protein. Both channels connect the protein surface with region of high oxygen affinity near the active site.

Introduction

Oxygen plays an important role in many biological reactions. It is produced through the light-driven oxidation and splitting of water in plants and cyanobacteria. Catalytic sites of many different proteins interact with oxygen. Formerly, the idea of unconstrained oxygen diffusion through proteins was generally accepted [1]. However, this point of view is changing. While the role of channels in relation to oxygen entry in myoglobin is still a matter of intense debate [2], a number of studies (hydrogenase [3], truncated hemoglobin [4], [5], aquaporin [6], copper amine oxidazes [7], flavoenzymes [8], cytochrome oxidase [9], fluorescent proteins [10], [11] and lipoxygenases [12], [13]) have described specific oxygen diffusion channels. The experimental localization of molecular oxygen within proteins is difficult, as it is highly mobile and usually is not resolved in X-ray crystal structures. In some cases mimicking oxygen with xenon or krypton, which have higher electron densities, has allowed the detection of potential oxygen binding sites [14]. Other techniques, like tryptophan fluorescence quenching, suffer from relatively low spatial resolution. The application of computational methods is an alternative and promising approach to determine the localization of oxygen in a protein. In this article we describe a new computational strategy for investigation of oxygen pathways in large protein complexes and apply it to photosystem II (PSII) of photosynthesis.

PSII is the first protein complex in the chain of oxygenic photosynthesis. It uses the energy of sunlight to extract electrons from water molecules. Products of this reaction: molecular oxygen, protons and electrons are used to power a variety of cellular metabolic processes [15], [16]. PSII uses an abundant energy supply and ultimately powers most forms of life on our planet. However, structural and mechanistic details of this process are not well understood. PSII is a ∼350 kDa protein complex of 20 polypeptide subunits embedded in the thylakoid membranes of green plant chloroplasts, or internal membranes of cyanobacteria [17]. The catalytic site of PSII is the oxygen-evolving complex (OEC), a Mn₄CaO₅ cluster embedded in the D1 polypeptide and separated from the bulk water by extrinsic protein subunits. In cyanobacteria these extrinsic subunits are O, U and V (Fig. 1).

Fast oxygen diffusion away from the catalytic site has been suggested to be important for PSII function as it would minimize potential oxidative damage to the active site [18]. Water channels have been suggested to be important for optimal binding and orientation of the substrate water as well as for restricting the access of unwanted solutes and reactions associated with excessive amounts of water [19]. The elucidation of the structure of PSII [20], [21] gave the first insights into details of substrate and product conduction. Several channels were identified in a number of computational studies examining internal solvent accessible surfaces found within the crystallographic structures of PSII [22], [23], [24]. These channels were tentatively assigned to water/oxygen/proton channels based solely on their width and number of hydrophilic amino acids. A significant limitation of these studies is that only one particular static conformation of the protein complex was considered. This is a physically unrealistic picture as proteins and the channels within them are dynamic at physiological temperatures [3], [25]. To overcome these limitations molecular dynamics (MD) simulations of a PSII core complex in the presence of explicit solvent water molecules were performed to identify water channels from a dynamic point of view [26], and to determine the energetic barriers for water permeation though PSII extrinsic proteins [27]. These studies found a system of branching pathways of water diffusion in PSII leading to the OEC and connecting to a number of distinct entrance points on the lumenal surface. Gibbs free energy profiles for water permeation indicated that water access is restricted in all of the observed channels. The main constriction sites were identified in water pathways [27]. These sites were suggested to serve as selectivity filters that restrict both the access of solutes detrimental to the water oxidation reaction and loss of Ca²⁺ and Cl⁻ from the active site. Molecular dynamics studies supported the original idea that regulatory mechanisms could exist to control substrate access to the Mn₄Ca05 cluster [27] and revealed mechanistic details of this process.

Thus, computational approaches have been proven to be useful in studies of water permeation. Oxygen permeation in PSII, however, has not yet been studied computationally. The only experimental result related to molecular oxygen diffusion within PSII revealed two krypton binding sites [24]. While water is abundant in the interior of PSII protein and it constitutes its natural environment, oxygen is not. Therefore computational approaches based on equilibrium molecular dynamics and diffusion tensor imaging [26] or accelerated water injection [27] cannot be used to identify oxygen pathways. In this paper, we discuss current strategies of finding small molecule pathways through molecular labyrinths and present a new approach for the solution of this problem. We then apply this method to investigate the topology and energetics of the pathways of molecular oxygen within PSII.