Proteorhodopsin Activation Is Modulated by Dynamic Changes in

Nov 12, 2015 - A complex series of dynamic hydrogen-bonding networks involving water molecules exists, facilitated by water channels and hydration sit...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/biochemistry

Proteorhodopsin Activation Is Modulated by Dynamic Changes in Internal Hydration Jun Feng and Blake Mertz* The C. Eugene Bennett Department of Chemistry, West Virginia University, 217 Clark Hall, Morgantown, West Virginia 26506, United States S Supporting Information *

ABSTRACT: Proteorhodopsin, a member of the microbial rhodopsin family, is a seven-transmembrane α-helical protein that functions as a light-driven proton pump. Understanding the proton-pumping mechanism of proteorhodopsin requires intimate knowledge of the proton transfer pathway via complex hydrogen-bonding networks formed by amino acid residues and internal water molecules. Here we conducted a series of microsecond time scale molecular dynamics simulations on both the dark state and the initial photoactivated state of blue proteorhodopsin to reveal the structural basis for proton transfer with respect to protein internal hydration. A complex series of dynamic hydrogen-bonding networks involving water molecules exists, facilitated by water channels and hydration sites within proteorhodopsin. High levels of hydration were discovered at each proton transfer sitethe retinal binding pocket and proton uptake and release sitesunderscoring the critical participation of water molecules in the proton-pumping mechanism. Water-bridged interactions and local water channels were also observed and can potentially mediate long-distance proton transfer between each site. The most significant phenomenon is after isomerization of retinal, an increase in water flux occurs that connects the proton release group, a conserved arginine residue, and the retinal binding pocket. Our results provide a detailed description of the internal hydration of the early photointermediates in the proteorhodopsin photocycle under alkaline pH conditions. These results lay the fundamental groundwork for understanding the intimate role that hydration plays in the structure−function relationship underlying the proteorhodopsin proton-pumping mechanism, as well as providing context for the relationship of hydration in proteorhodopsin to other microbial retinal proteins.

I

occur in nature and have the potential for use in applications such as optogenetics.15,16 PRs are categorized into two major classes according to their absorption maxima and marine dwelling zone: green proteorhodopsin (GPR), which is commonly found in surfacedwelling bacteria, and blue proteorhodopsin (BPR), which is typically found at lower ocean depths.9 A single amino acid residue at location 105 (Leu in GPR and Gln in BPR) is responsible for the color-tuning between the two classes.17 In addition, the photocycle of GPR is 10 times faster than that of BPR.18 Nevertheless, GPR and BPR both share a high degree of sequence identity and are structurally homologous.19,20 PR acts as a vectorial proton pump.21 Under native conditions (alkaline pH), protons are transported outward, i.e., from the cytoplasm to the extracellular medium, but at acidic pH, proton transport is inverted, leading to the uptake of protons by the cell. This pH-dependent behavior makes PR unique from prototypical archaeal proton pumps such as bR

nternal water molecules and their role in membrane protein function have been intensively studied,1−6 leading to the consensus that water is an indispensable component in facilitating protein function. Proteins such as bacteriorhodopsin (bR), an archaeal proton pump, utilize protein-bound water molecules to facilitate proton transport across the cell membrane.7 Proteorhodopsin (PR) represents one of the newest members of the rhodopsin superfamily, a collection of seven-transmembrane helical retinylidene proteins. Since their initial discovery in marine bacteria,8 PRs have been found to be widely distributed among all three domains, including oceanic plankton,9 freshwater bacteria,10 archaea,11 soil-bound bacteria,12 and eukaryotes.13 PR functions as a light-driven proton pump, utilizing the chromophore retinal, which is covalently bound to a conserved lysine residue via a protonated Schiff base, as the light-harvesting engine.14 Isomerization of retinal from an all-trans to a 13-cis configuration following photon absorption triggers a photocycle of conformational changes through a series of photointermediates, culminating in the translocation of a proton across the bacterial cell membrane (Figure 1). Understanding the structure−function relationship of PR activation will provide an invaluable model to apply to the ever-growing number of retinal proteins that are found to © XXXX American Chemical Society

Received: August 19, 2015 Revised: November 3, 2015

A

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

microbial retinal protein photocycle.28 However, the similarities with and differences between the role of hydration in PR and other retinal proteins such as bR remain poorly understood. Obtaining the structural details of the role of hydration in the early photointermediates of the PR photocycle is essential in contributing to our overall understanding of the structure− function relationship in PR. Molecular dynamics (MD) simulations are a tool that can effectively probe the hydration dynamics and hydrogenbonding networks within PR at the atomic level of detail necessary for understanding its structure−function relationship.29−32 In this study, we conducted a series of microsecond time scale MD simulations to investigate the formation and dynamics of interior water channels and water molecules surrounding the amino acid residues implicated in the ground state and the initial photointermediate (K state) of PR. Our results show that hydration plays a critical role in the structure−function relationship of the early photointermediates in the proteorhodopsin photocycle. The ability of water to aid in shuttling protons is directly affected by the isomerization of the retinal cofactor, leading to a rearrangement of the hydrated regions in the vicinity of the retinal binding pocket and the proton release group.

Figure 1. Proteorhodopsin, a retinal protein proton pump. (a) Tertiary structure of blue proteorhodopsin (PDB entry 4JQ6). The covalently bound chromophore, retinal, initiates the proton-pumping mechanism, interacting with several amino acid residues that facilitate proton transfer from the cytoplasmic (CY) side to the extracellular (EC) side of the protein. Black arrows indicate specific proton transfer steps during proton pumping. (b) The retinal binding pocket is the central hub for proton pumping. Upon photoisomerization of retinal from an all-trans to a 13-cis conformation, the Schiff base of retinal is deprotonated, with proton transfer taking place via a water-mediated interaction with the proton acceptor, D97. D227 acts as a counterion to stabilize the complex between the Schiff base, water, and D97 during the proton transfer. Proton release to the outside of the cell potentially occurs through the putative proton release group, E142. The Schiff base is reprotonated from the cytoplasmic side of the protein through the proton donor, E108. H75 acts to stabilize D97 in the dark state, and R94 interacts with both D227 and E142. (c) The photocycle of proteorhodopsin is initiated upon absorption of a photon, leading to retinal isomerization and progression from the dark state (PR) to the initial photointermediate (K).

■

COMPUTATIONAL METHODS Simulation Setup. We chose the X-ray crystal structure of BPR in the dark state20 (PDB entry 4JQ6) as the starting structure in our simulations, which is structurally similar to other microbial rhodopsins, with root-mean-square deviations (RMSDs) of 1.4 and 2.4 Å from bR25 (PDB entry 1C3W) and sensory rhodopsin33 (PDB entry 1H68), respectively. Missing loops were added using the backbone structural model of GPR.19 All amino acid numbering represents the corresponding residue numbers in GPR (SwissProt entry Q9F7P4.1); there is an 18-residue numbering shift in the BPR structure, unless otherwise specified. MD simulations were conducted at alkaline pH to model PR in its native environment. However, conventional all-atom MD simulations do not allow for a change in protonation of amino acid side chains once they are initially set. In addition, little information regarding the protonation states of internal ionizable residues in BPR (H75, D97, E108, E142, and D227) is available. On the basis of the distance (2.86 ± 0.14 Å) between the Nδ atom of H75 and the carboxylic oxygen atom of D97 in the BPR X-ray crystal structure (2.86 ± 0.14 Å),20 H75 is likely to be protonated on the Nδ atom in the imidazole ring, stabilizing the hydrogen bond with the deprotonated side chain of D97. However, the protonation state of H75 in GPR remains under debate.34,35 D97, the proton acceptor, has a pKa of ∼7.2−7.6 in GPR21,34,36 but is unknown in BPR. E108, the proton donor, has an unusually high pKa (>8.5) in GPR21 but is unknown in BPR. In our simulations, we have modeled E108 as being deprotonated. It has been shown that the protonation state of the proton donor in other microbial retinal proteins can directly affect the hydration of the cytoplasmic channel, regardless of the photointermediate.37 For this reason, the majority of our analysis focuses on the retinal binding pocket and the extracellular channel. E142 has a pKa of 9 in GPR,38 and D227 is deprotonated at pH 9 in GPR39 but is unknown in BPR. Therefore, all glutamic and aspartic acid residues were deprotonated. H75 was protonated on the δ-

and is reflected in the amino acid residues directly implicated in PR function (Figure 1). PR possesses a proton acceptor (D97), a proton donor (E108), and a complex counterion (D227) much like bR, but lacks a well-defined proton release group (PRG) (putatively E142) and also contains a highly conserved histidine that stabilizes the proton acceptor (H75) but is not present in bR. Internal water molecules also play a crucial role in retinal protein function, forming complex hydrogen-bonding networks with protein residues7,22 as well as creating water channels23 that extend through the interior of the protein. These channels contribute to the pumping mechanism by shuttling protons between protein side chains that lie beyond acceptable distances for direct proton transfer. Although few internal waters were resolved in the BPR X-ray crystal structure,20 crystallographic structures of bR possess water clusters near the Schiff base and in the proton release and uptake sites.24−26 In addition, Fourier transform infrared (FTIR) spectroscopy studies have shown how proton “switches” (protonation/ deprotonation of key amino acids) in retinal proteins are facilitated via interactions with water molecules.22,27 This complex interplay between variable protonation states of titratable amino acid residues and dynamical rearrangement of water molecules forms the structural basis for the progression of distinct spectroscopic intermediates in the B

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

Figure 2. Internal hydration of BPR reaches equilibrium in the dark state and undergoes noticeable fluctuations in the K state. (a) Time evolution of the number of internal waters over the course of both dark and K state simulations: (left) dark state (simulations I−III) and (right) K state (simulations 1−3), (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3). (b) Average 3D water density from dark state simulation I. (c) Average 3D water density from K state simulation 1. 3D water density plots were rendered using an isosurface with an isovalue of 0.75: (top) cytoplasmic side of BPR and (bottom) extracellular side of BPR. See the Supporting Information for the 3D water density of other simulations. (d) Average water density with a bin size of 0.5 Å through the interior of BPR along the z axis from the extracellular (negative z) to the cytoplasmic (positive z) side of the protein. Color schemes correspond to plots in panel a: () dark state and (---) K state.

the starting structure of K state simulations 1−3, respectively. Isomerization of retinal from all-trans to 13-cis, 15-anti was simulated by applying two modified torsional potentials:48 (1) a single minimum at the cis position and a barrier height of 60 kcal mol−1 to the C12−C13C14−C15 dihedral of retinal and (2) a single minimum at the anti position and a barrier height of 60 kcal mol−1 to the C14−C15Nζ−Cε dihedral of the PSB. The second torsional potential was applied to maintain the Schiff base in the anti position during retinal isomerization, in agreement with the suggested PSB configuration of bR in the K state.49,50 Simulations were run with modified potentials for 2 ps, and retinal isomerization was completed within 0.5 ps. Thereafter, each simulation was continued for 3 μs without the modified potentials. Retinal remained in the 13-cis, 15-anti configuration throughout the simulation. The aggregate simulation time was 18 μs. Analysis. Hydrogen bond interactions were identified using a distance cutoff of 3.0 Å between donor and acceptor atoms. The number of internal water molecules was counted as the number of waters (heavy atom) residing in the interior of the protein with a dlipid−water of ≥4 Å and a |zwater| of ≤14 Å (using heavy atoms only), with the lipid bilayer centered at the origin on the x−y plane and a hydrophobic core approximately 28 Å in length. The hydrophobic core thickness was estimated as the average distance between the corresponding C2 atoms of the sn-1 acyl chains in the upper and lower leaflets of the bilayer. Three-dimensional (3D) water density was calculated with a grid size of 1 Å using MDAnalysis.51 Calculations were applied to dark state trajectories after 0.75 μs and K state trajectories after 0.2 μs based on the time evolution of internal water molecules (Figure 2a). The dark state simulations started from the crystal structure with relatively no internal waters. Therefore, a longer equilibration time was required when the 3D water density was calculated. The water occupancy of each

nitrogen. All-trans retinal was covalently bound to K231 via a protonated Schiff base (PSB). The replacement method of the Membrane Builder40,41 module from the CHARMM-GUI server (http://www. charmm-gui.org) was used to place BPR in the center of an explicitly hydrated lipid bilayer of 129 lipids (65 in the upper leaflet and 64 in the lower leaflet because of the asymmetry of the protein), composed of a 3:1 1-palmitoyl-2-oleoyl-snglycero-3-phosphoethanolamine (POPE):1-palmitoyl-2-oleoylsn-glycero-3-phosphoglycerol (POPG) molar ratio to mimic a bacterial membrane;42 38 sodium and 2 chloride ions were added to neutralize the system along with approximately 7000 TIP3P waters, resulting in a final system size of 74 Å × 74 Å × 73 Å. Minimization and equilibration protocols were taken from CHARMM-GUI and conducted using the molecular simulation package CHARMM43 c37b1 with the CHARMM c36 force field. The force field for retinal was obtained from Feller and co-workers.44,45 For production simulations, CHAMBER46 was used to convert CHARMM format topology and coordinate files into AMBER format input files. AMBER 12 with GPU acceleration (pmemd.cuda)47 was used to run all simulations. Three dark BPR systems (simulations I−III) were run in the NPT ensemble at a constant temperature of 300 K in Langevin dynamics with a friction coefficient of 5 ps−1 and a constant pressure of 1 bar using the Berendsen pressure scaling algorithm with a relaxation time of 8 ps. The particle mesh Ewald method was used to calculate electrostatics with a nonbonded cutoff of 8 Å, as suggested in the AMBER manual. All hydrogen atoms were constrained with the SHAKE algorithm. Each simulation was integrated at a time step of 2 fs for a running time of 3 μs. To simulate BPR activation, three snapshots were randomly taken from dark state simulations I−III after 500 ns and used as C

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry residue was calculated as the percentage of time that heavy atoms of water were within 3 Å of heavy atoms of the protein. A water-bridged interaction was identified when a water molecule simultaneously formed a hydrogen bond with two protein residues. A water channel was said to be present when two protein residues were connected by two or more water molecules via hydrogen bonding. Analysis was performed using CHARMM, MDAnalysis, and in-house scripts. Molecular figures were made using VMD52 and PyMOL,53 and plots were made using matplotlib.54

■

RESULTS AND DISCUSSION We analyzed trajectories of both the dark state (simulations I− III) and the K state (simulations 1−3) of BPR. No large scale conformational changes of PR took place. The average RMSDs from the Cα atoms of the transmembrane helices in the BPR Xray crystal structure in the last 2 μs of the dark state simulations were 1.9 ± 0.1, 1.8 ± 0.1, and 2.0 ± 0.2 Å, respectively, while the rmsds in the last 2 μs of the K state simulations were 2.1 ± 0.2, 1.9 ± 0.2, and 2.4 ± 0.2 Å, respectively (see the Supporting Information for rmsd plots). The main focus of our results is on protein internal hydration and the potential effect it has on PR function. Statistics of Protein Internal Hydration. Though very few water molecules are present in the dark state X-ray crystal structure of BPR, we observed a significant influx of bulk water into the interior of the protein, reaching equilibrium after approximately 0.75 μs in the dark state simulations (Figure 2a). This influx leads to the formation of transient water channels in distinct regions of PR (Figure 2b,c), each of which contains amino acid residues directly involved in the proton-pumping mechanism (Table 1) and is consistent with other proton-

Figure 3. Several functionally relevant interior residues are wellhydrated in PR. Water occupancy, defined as the percentage of time that heavy atoms of water were within 3 Å of heavy atoms of the protein, is given for each amino acid. Statistics were collected using trajectories after 0.75 μs in dark state simulations (top) and 0.2 μs in K state simulations (bottom). Loop residues are colored blue: (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3).

switch (Q105) as well as the acidic amino acids that undergo protonation changes during the photocycle: D97 (proton acceptor), E108 (proton donor), E142 (putative proton release group), and D227 (PSB counterion). In addition, other residues proximal to these essential amino acids undergo hydration: S61 (hydrogen-bonded to the proton donor E108), R94 (hydrogen-bonded to the counterion D227), T101, N220, Y223 (hydrogen-bonded to the proton release group E142), N224, and N230 (hydrogen-bonded to Q105). Altogether, these residues combine to form a putative proton shuttling pathway with the retinal cofactor located at the nexus in the core of the protein. The Retinal Binding Pocket Possesses a Highly Flexible Hydrogen-Bonding Network That Stabilizes Chromophore Movements. The retinal binding pocket is well-hydrated in both the dark and K state simulations. In particular, high water density exists in the vicinity of D97 and D227, which act as a complex counterion to the PSB.55,56 H75, which is highly conserved among PRs, stabilizes D97 through the formation of a hydrogen bond.20,34 In the dark and K state simulations, this hydrogen-bonding pair remains stable through either direct hydrogen bonding or indirect water-bridged interactions (Figure 4a,d). Likewise, the interaction between the proton acceptor D97 and the PSB is frequently stabilized in the dark state by utilizing a water molecule. However, upon isomerization of the retinal to the 13-cis conformation, the D97−PSB interaction is broken, occasionally connected through water-mediated interactions via a single water bridge (Figure 4c) or a water wire formed by two or more water molecules (Table 2). Although occurrences of water-mediated interactions and ordered water arrangement vary significantly among the simulations, indicating statistical nonconvergence on the microsecond time scale, internal water molecules connecting key residues in the retinal binding pocket are captured in our simulations and are a common functional theme among other microbial rhodopsins such as bR.7 According to the dark state crystal structure of BPR, the carboxylic side chain of the counterion D227 potentially forms a weak hydrogen bond (with an average distance of 3.35 ± 0.1 Å among the three crystal subunits) with the PSB. In the dark state simulations, this interaction remained stable, fluctuating between direct contact of the residue side chains and indirect hydrogen bonding (Figure 4b). However, the D227−PSB

Table 1. Transient Water Channel Formation between Critical Protein Residues as well as the Schiff Base in BPR dark state (%)a

a

K state (%)

interaction

I

II

III

1

2

3

E108−PSB PSB−R94 R94−E142

0.0 4.5 42.7

0.0 2.1 23.3

0.0 0.5 2.8

3.6 0.8 35.9

38.7 1.1 35.2

2.3 0.1 28.0

Percentage of time over the entire simulation.

transporting proteins.7,30 In general, internal water density is largely similar between the dark and K states. The water channel on the cytoplasmic side is readily distinguishable, as it connects bulk water, the proton uptake site, and the retinal binding pocket. However, on the extracellular side of the protein, no formation of a water channel between the bulk solvent and the retinal binding pocket occurs. This channel is prevented from fully forming by the presence of the side chain of R94, which lies halfway between the retinal binding pocket and the putative proton release group. After isomerization of retinal from the all-trans to 13-cis configuration, a marked increase in the level of hydration of the extracellular region occurs in all of our simulations (Figure 2d). The majority of internal hydration sites in PR are located on transmembrane helices B, C, and G (Figure 3). Many of the well-hydrated amino acids either actively participate in or directly interact with residues that actively participate in the proton-pumping mechanism. Several key amino acid residues extensively interact with water, including the color-tuning D

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

is absent in bR (A215). The presence of this polar residue may be directly linked to the color-tuning switch between the green and blue variants in PR; in the BPR crystal structure, N230 interacts with Q105 via a hydrogen bond with the amide side chain of the glutamine. Our simulations show that N230 can form a hydrogen bond with the amide nitrogen of Q105 through either its backbone or its side chain oxygen. However, the hydrogen bond interaction was transient, and distances fluctuated greatly (Figure 5a,d). In addition, spectroscopic studies of a N230S mutant showed that mutation of N230 does not alter the environment or structure of the Schiff base, despite its proximity to the Schiff base linkage.57 Thus, the specific functional role of N230 remains elusive. The same spectroscopic study also demonstrated that Q105 of BPR interacts with the Schiff base in both the dark and K states.57 The mechanism by which this interaction leads to the blue shift observed in BPR is controlled through both spatial and electrostatic interactions. Recent UV−visible studies showed that the volume occupied by the side chain at residue 105 is directly correlated to PR absorption wavelength, where smaller side chains did not disrupt the hydrogen-bonded network within the retinal binding pocket that stabilizes the PSB with its complex counterions.58 However, solid state NMR and time-resolved flash photolysis studies revealed that the Q105−PSB interaction directly affects the hyperconjugation of electrons along the polyene chain, localizing the positive charge of the PSB that leads to the blue shift of BPR.59 The dark state X-ray crystal structure provides a partial picture of this essential interaction: Q105 is in close contact with the C13-methyl group of retinal, and it is also reproduced in our dark state simulations (Figure 5e). Surprisingly, in the K state simulations, the carbonyl oxygen of Q105 shifts to form a direct hydrogen bond with the Schiff base proton on the microsecond time scale in two of three trajectories (Figure 5c), in accordance with the previously mentioned FTIR and solid state NMR studies.57,59 Furthermore, water-bridged interactions between Q105 and the PSB occurred, often when direct interactions were broken, in all three K state simulations, in agreement with the UV−visible studies.58 This indicates a twofold cause for the color-tuning mechanism between BPR and GPR: (1) The photoisomerization of retinal to the 13-cis conformation allows Q105 to directly interact with the PSB throughout much of the PR photocycle, and (2) water molecules play a significant role in modulating the color-tuning interaction, aided by the electrostatic attractions that exist between Q105, water, and the PSB. Altogether, the side chain arrangements of several key residues within the retinal binding pocket are clearly maintained through a network of water molecules that is dynamic enough to adjust to the conformational changes that take place during the initial stages of photoactivation in BPR. The Proton Donor Can Be Stabilized by Versatile Hydrogen-Bonding Partners. At alkaline pH, PR takes up a proton from the bulk cytoplasm and subsequently donates the proton to the photoactive site in the retinal binding pocket. E108 serves as the proton donor to the Schiff base during proton pumping and has an unusually high pKa (>8.5) in GPR21 but is unknown in BPR. In addition, the hydrogenbonding network localized around E108 in GPR is stronger than in the corresponding residue of bR (D96),21 indicating an environment slightly different from typical microbial proton pumps. The channel between E108 and the PSB in the BPR structure is similar to bR (∼12 Å and hydrophobic), and it was shown through a combination of FTIR experiments and MD

Figure 4. Water-bridged interactions help stabilize counterion residues in both the dark and K states during BPR activation. (a) Representative snapshot of the retinal binding pocket in the dark state illustrating the water-mediated interactions that stabilize key residues in the initial protonation event in BPR activation: orange sticks for retinal, green sticks for amino acid residues (GPR numbering), and spheres for water. (b) Time evolution of the interaction between the PSB and the side chain oxygen of D227 in dark (top) and K state (bottom) simulations and formation of waterbridged interactions between the PSB and D227. (c) Interaction between the PSB and the side chain oxygen of D97 in dark and K state simulations and formation of water-bridged interactions between the two. (d) Interaction between the side chain oxygen of D97 and Nδ of H75 in dark and K state simulations and formation of water-bridged interactions between the two: (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3) and water-bridged interactions shown as horizontal bars at the bottom of each respective frame, color-coded by simulation.

Table 2. Frequency of Water-Mediated Interactions between the Proton Acceptor D97 and the PSB dark state (%) no water single water bridge water wire

K state (%)

I

II

III

1

2

3

52.4 46.2 1.4

76.9 21.7 1.5

100.0 9.3 1.9

77.6 2.0 20.4

94.7 2.5 2.8

93.5 0.4 6.1

interaction was broken after the all-trans → 13-cis isomerization of retinal leading to rotation of the Schiff base toward the cytoplasmic side of PR, which is also known to take place in bR.49,50 Both FTIR and pump probe flash photolysis studies made a compelling case that the D227−PSB association is maintained in the K state through a water-mediated interaction and is supported by the loss of decay of the M photointermediate with a D227N mutant.55,56 Although we did not observe water-mediated interactions between D227 and the PSB in our K state simulations, the distance between the two (∼5 Å) is shorter than the distance between D97 and the PSB and is much more stable, indicating that this interaction is important for the K → M transition and subsequent proton transfer from the PSB to the proton acceptor (D97) (Figure 4). In addition to the complex counterion residues within the retinal binding pocket, both GPR and BPR possess a polar residue (N230) adjacent to the Schiff base linkage (K231) that E

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

Figure 5. Color-tuning switch in PR possesses dynamic interactions with the PSB in the K state. (a) Time evolution of the interaction between the side chains of N230 and Q105 in dark (top) and K state (bottom) simulations and formation of water-bridged interactions between N230 and Q105. (b) Representative snapshot (top) of the retinal binding pocket in the K state illustrating the direct interactions that stabilize the color-tuning switch with the backbone of N230 and the PSB during BPR activation. Representative snapshot (bottom) in the K state showing how water can facilitate interactions between the side chains of N230 and Q105 in the binding pocket: orange sticks for retinal, green sticks for amino acid residues (GPR numbering), and spheres for water. (c) Interaction between the PSB and the side chain oxygen of Q105 in dark and K state simulations and formation of water-bridged interactions between the two. (d) Interaction between the backbone oxygen of N230 and the side chain oxygen of Q105 in dark and K state simulations and formation of water-bridged interactions between the two. (e) Interaction between the C13-methyl group of retinal and the side chain oxygen of Q105 in dark and K state simulations: (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3) and water-bridged interactions shown as horizontal bars at the bottom of each respective frame, color-coded by simulation.

simulations on bR that proton transfer occurs only through transient formation of a Grotthuss-type proton transfer chain.30,60 In the X-ray crystal structure of BPR (obtained under acidic conditions), the E108 side chain forms a hydrogen bond with the backbone oxygen of an adjacent serine (S61), which is one helical turn above the side chain−side chain interaction observed between D96 and T46 in bR.7 However, in our deprotonated E108 simulations (I−III and 1−3), a hydrogenbonding interaction can form between the side chains of S61 and E108, often mediated through the participation of a water molecule that is part of a channel extending from the bulk cytoplasm to the retinal binding pocket (Figure 6a,b). The increase in the interior volume of BPR along with the rearrangement of the side chain of E108 allows for formation of the water wire from the PSB to E108 and bulk solvent in the cytoplasm. This behavior closely reflects the proposed model that facilitates regeneration of the Schiff base in bR, in which D96 and T46 are hydrogen-bonded to the same water, forming one end of the water wire that connects the proton donor to the Schiff base.30 The relatively short time scales in which hydrogen-bonding rearrangement (backbone and side chain of S61 and water) of E108 was observed in each set of simulations provide context for the potential mechanism by which the protonation switch of the proton donor at the cytoplasmic proton uptake site takes place. Furthermore, the short time scale is in agreement with previous studies of bR that demonstrated how the proton donor (D96) protonation state acts as a “latch” for opening (deprotonated) and closing (protonated) of the cytoplasmic channel connecting the bulk solution to the retinal binding pocket, irrespective of the photointermediate state.37

Figure 6. Proton donor of BPR undergoes conformational fluctuations influenced by changes in hydration and protonation state. (a) Time evolution of the interaction between the side chains of S61 and E108 in dark (top) and K state (bottom) simulations and formation of water-bridged interactions between S61 and E108: (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3) and waterbridged interactions shown as horizontal bars at the bottom of each respective frame, color-coded by simulation. (b) Representative snapshot of the proton donor (E108) in the dark state forming a hydrogen-bonding network with S61 through the presence of a water channel from a cytoplasmic bulk solution: green sticks for amino acid residues (GPR numbering) and spheres for water.

Retinal Photoisomerization Propagates Changes in Hydration in the Vicinity of the Putative Proton Release Group. During the PR photocycle, a proton is released to the extracellular side of the membrane. However, the exact residues involved in the proton release are still unclear. In bR, a pair of glutamates (E194 and E204), an arginine residue (R82), and a group of hydrogen-bonded water molecules function as the proton release group.61 The proton is stored between multiple binding sites consisting of several water molecules and the F

DOI: 10.1021/acs.biochem.5b00932 Biochemistry XXXX, XXX, XXX−XXX

Article

Biochemistry

Figure 7. Putative proton release group of PR undergoes dynamic changes in hydration from the dark to K states. (a) Time evolution of the interaction between the side chains of E142 and Y95 in dark (top) and K state (bottom) simulations and formation of water-bridged interactions between E142 and Y95: (black) simulation I(1), (red) simulation II(2), and (gray) simulation III(3) and water-bridged interactions shown as horizontal bars at the bottom of each respective frame, color-coded by simulation. (b) Representative snapshot of the putative proton release group, E142, in the dark state forming a complex hydrogen-bonded network with Y95, Y208, Y223, R94, and D227 through the presence of a water channel from the extracellular bulk solution: green sticks for amino acid residues (GPR numbering) and spheres for water. RET denotes retinal. (c) Interaction between the side chains of R94 and D227 in dark and K state simulations and formation of water-bridged interactions between the two. (d) Interaction between the side chains of E142 and Y223 in dark and K state simulations and formation of water-bridged interactions between the two. (e) Interaction between the side chains of E142 and Y208 in dark and K state simulations and formation of water-bridged interactions between the two.

play a role in the process of proton release in later photointermediates of the PR photocycle. In addition, a conserved arginine residue (R94) that is homologous to R82 in bR forms a water-mediated hydrogenbonded network (via Wat403 and Wat405 from the BPR X-ray crystal structure) with E142 in the BPR X-ray crystal structure, similar to the water cluster reported in the bR proton release site. Mutation of R94 abolishes proton-pumping function,20 indicating that this arginine residue may be directly implicated in the proton release group. As mentioned earlier, R94 forms a critical barrier separating areas of high water density between the putative proton release group and the retinal binding pocket. The area of water density located proximal to the retinal binding pocket is consistently observed in all simulations (Figure 2). However, the region of water density located between the bulk extracellular solvent and R94 does not exist in dark state simulation III. This is due to the rotation of the E142 side chain toward the extracellular side of BPR, exposing it to bulk solvent (Supporting Information). Interestingly in the K state simulations, the water density increased between E142 and R94 (Figure 2, the water density along z), leading to formation of a semicontinuous water channel extending from the retinal binding pocket to the extracellular bulk solvent. Besides the change in hydration from the dark to the K state, it appears that the interaction between R94 and D227 is important in the early photointermediates of PR; this residue pair either directly forms a hydrogen bond or possesses a watermediated hydrogen bond for the vast majority of 18 μs of overall simulation time (Figure 7c). Although this is a clear indication that R94 acts as a hub for extracellular hydration dynamics in PR, elucidating the specific role of R94 in the proton release mechanism will require more extensive studies.

aforementioned residues, upon which the proton is localized to either E194 or E204 after movement of R82 away from the proton acceptor.61 This motif is not uniformly conserved throughout microbial retinal proteins,7 as PR lacks a pair of glutamates homologous to the primary sequence of bR. However, E142 lies in a similar region, and it is stabilized via a hydrogen-bonding network with three tyrosine residues (Y95, Y208, and Y223) as observed in the X-ray crystal structure.20 The protonation state of E142 is variable at environmental pH; NMR spectroscopy identified a pKa (>9) in GPR that would favor protonation,62 whereas FTIR difference spectroscopy inferred a lower pKa (