Crystal Structures of Archaerhodopsin-1 and -2: Common Structural Motif in Archaeal Light-driven Proton Pumps

doi:10.1016/j.jmb.2006.02.032

J. Mol. Biol. (2006) 358, 675–685

Crystal Structures of Archaerhodopsin-1 and -2: Common Structural Motif in Archaeal Light-driven Proton Pumps Nobuo Enami1, Keiko Yoshimura1, Midori Murakami1, Hideo Okumura1 Kunio Ihara2 and Tsutomu Kouyama1,3* 1

Department of Physics Graduate School of Science Nagoya University, Nagoya 464-8602, Japan 2 Center for Gene Research Nagoya University, Nagoya 464-8602, Japan 3

RIKEN Harima Institute/ SPring-8, 1-1-1, Kouto Mikazuki, Sayo, Hyogo 679-5148, Japan

Archaerhodopsin-1 and -2 (aR-1 and aR-2) are light-driven proton pumps found in Halorubrum sp. aus-1 and -2, which share 55–58% sequence identity with bacteriorhodopsin (bR), a proton pump found in Halobacterium salinarum. In this study, aR-1 and aR-2 were crystallized into 3D ˚ , cZ117.6 A ˚ ) and C2221 (aZ crystals belonging to P43212 (aZbZ128.1 A ˚ ˚ ˚ 122.9 A, bZ139.5 A, cZ108.1 A), respectively. In both the crystals, the asymmetric unit contains two protein molecules with slightly different conformations. Each subunit is composed of seven helical segments as seen in bR but, unlike bR, aR-1 as well as aR-2 has a unique omega loop near the N terminus. It is found that the proton pathway in the extracellular half (i.e. the proton release channel) is more opened in aR-2 than in aR-1 or bR. This structural difference accounts for a large variation in the pKa of the acid purple-to-blue transition among the three proton pumps. All the aromatic residues surrounding the retinal polyene chain are conserved among the three proton pumps, confirming a previous argument that these residues are required for the stereo-specificity of the retinal isomerization. In the cytoplasmic half, the region surrounded by helices B, C and G is highly conserved, while the structural conservation is very low for residues extruded from helices E and F. Structural conservation of the hydrophobic residues located on the proton uptake pathway suggests that their precise arrangement is necessary to prevent a backward flow of proton in the presence of a large pH gradient and membrane potential. An empty cavity is commonly seen in the vicinity of Leu93 contacting the retinal C13 methyl. Existence of such a cavity is required to allow a large rotation of the sidechain of Leu93 at the early stage of the photocycle, which has been shown to accompany water translocation across the Schiff base. q 2006 Elsevier Ltd. All rights reserved.

*Corresponding author

Keywords: proton pump; retinal; water translocation; X-ray crystallography; protein crystallization

Introduction Since bacteriorhodopsin (bR) was discovered in the cell membrane of an extremely halophilic archaeon Halobacterium salinarum,1 the number of archaeal retinal proteins for which amino acid sequences are available have increased to O25.2,3

Abbreviations used: heptane triol, 1,2,3-trihydroxylheptane; bR, bacteriorhodopsin; aR-1 and -2, archaeal rhodopsins 1 and 2. E-mail address of the corresponding author: [email protected]

They are classified into three types of rhodopsin according to their physiological functions; i.e. lightdriven proton pumps (e.g. bacteriorhodopsin), chloride ion pumps (e.g. halorhodopsin in H. salinarum) and phototaxis receptors (e.g. sensory rhodopsin I and II in H. salinarum). 4–7 Their biochemical and spectroscopic studies have shown that the light-induced isomerization of retinal around the C13–C14 double bond is a crucial event that occurs commonly in all the archaeal rhodopsins.8,9 Recent crystallographic studies of bacteriorhodopsin, halorhodopsin and sensory rhodopsin II have revealed that the stereospecificity of retinal isomerization is attributable

0022-2836/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.

676 largely to a spatial arrangement of bulky residues within the retinal-binding pocket, which are highly conserved through the evolution of archaeal rhodopsins.10–16 More recently, retinal proteins with the same photochemical property as archaeal rhodopsins have been shown to exist in eubacteria and uni-cell eukaryotes. 17–19 Structural and functional comparisons of these retinal proteins have widened our understanding as to how a prototype rhodopsin had evolved to acquire different functions. Such analyses would also provide information as to how an individual function had been optimized under different environments. For quantitative discussion about a common structural motif utilized for an individual function, we need to accumulate structural data of retinal proteins from various organisms. We have recently investigated the crystal structures of archaerhodopsin-1 and -2 (aR-1 and aR-2). These archaeal rhodopsins have been found in the claret membranes of Halorubrum sp. aus -1 and -2, respectively, that thrive in saturated brines in western Australia.20,21 aR-1 and aR-2 share 85% sequence identity to each other, while their sequence identities with bR are much lower (i.e. 55–58%).2,3 aR-1 as well as aR-2 functions as a light-driven proton pump; i.e. their photoreaction cycles are described as reported for bR.22,23 But there are significant differences in thermodynamic properties for aR-1 and aR-2. In some aspects, aR-1 is rather similar to bR. For example, the thermal equilibrium between the two isomers in the dark adapted state is shifted largely toward the trans isomer in aR-2, while the 13-cis isomer is relatively more stable in aR-1 and bR. This difference is believed to come from substitution of a residue (Met145 in bR) contacting the retinal ionone

Crystal Structures of Archaerhodopsin-1 and -2

ring, since the same thermodynamic property as observed for aR-2 is reproduced when Met145 in bR is substituted to phenylalanine.24 It was reported, on the other hand, that the pKa of the acid purple-to-blue transition is much higher in aR-2 than in aR-1 or bR.22 It has not been clarified whether this difference correlates with a difference in the amino acid sequence. Here, the structural data of aR-1 and aR-2 are compared with each other to find out a structural reason for variation in the thermodynamic properties of these proteins. They are also compared with the previously reported structures of bR to reveal the common structural motif that is required for an efficient proton pumping. It was then found that hydrophobic residues existing in the postulated proton uptake pathway are highly conserved during the evolution. The present result also provided an important insight into the functional role of micro-cavities located on the proton uptake pathway, which are probably necessary for relocation of internal water molecules that has been recognized as a key event for the unidirectional movement of protons.25

Results Sequence alignments of aR-1, aR-2 and bR Here, the amino acid sequences of aR-1 and aR-2 are numbered after bR (Figure 1). As the N-terminal polypeptide of aR-1 is six residues longer than that of bR, the seventh residue from the N terminus is designated as Asp1. Residues preceding Asp1 are numbered in the conventional rule but distinguished

Figure 1. Amino acid sequences of bR, aR-1 and aR-2. The alignment of amino acid residues is optimized by introducing gaps (denoted as K). Residues conserved among the three proteins are shown in black; otherwise they are shown in color. Residues that are clearly identified in the crystal structures are shown in bold. The rectangles and the orange arrows below the sequences indicate the regions forming a-helices and a b-sheet, respectively. The region forming an omega loop in aR-1 and aR-2 is marked by the blue arrow. The amino acid sequences of aR-1 and -2 are numbered after bR.

677

Crystal Structures of Archaerhodopsin-1 and -2

by adding an asterisk; e.g. the N-terminal residue is expressed by Thr1*. In the case of aR-2, Asp1 is the first residue that is visible in the electron density map and, thus, no special care is paid for residues preceding Asp1. But, as the BC loop of aR-2 is one residue longer than that of bR or aR-1, a special numbering rule is adopted for residues in this loop; namely, the residue following Ser73 is designated as Gly73* and the next is called Thr74. (It will be later shown that insertion of Gly73* at the turn in the antiparallel beta-sheet has a profound effect on the structure of the proton release channel.) Crystals of aR-1 and aR-2 aR-1 molecules aggregate to form claret patches (the so-called claret membrane) in the cell membrane of Halorubrum sp. aus-1. Atomic force micrographs showed that claret membrane is isolated as a sheet with a thickness of 5 nm and a diameter of 200–600 nm. X-ray diffraction data from stacked claret membranes indicated that aR-1 molecules are arranged on a hexagonal lattice ˚ , in which aR-1 with a cell dimension of 63.6 A forms a trimeric complex as observed for purple membrane (unpublished data). The structural similarity between claret membrane and purple membrane may suggest that the membrane fusion method developed for crystallization of bR is applicable to aR-1. But we have not yet succeeded in converting claret membrane into uniformly sized vesicles, which are requisite for growth of the P622 crystal.26,27 Instead, a bi-pyramidal crystal was found to grow at a high detergent:protein weight ratio (w1:1). This crystal exhibits claret and its absorption spectrum is characterized by fine peaks at 453, 478, 509 and 546 nm; i.e. bacterioruberin, the second chromophore of claret membrane, is incorporated into the crystal. The absorbance of this chromophore becomes negligibly small when the polarization plane of the measuring light is perpendicular to the 4-fold axis of the crystal. On the contrary, the absorbance of the retinal chromophore becomes highest in the same direction of the polarization plane. In the light-adapted crystal, the retinal chromophore has an absorption peak at 565 nm (at 100 K); it is blue-shifted by 10 nm as compared to that observed for bR. The claret membrane of Halorubrum sp. aus-2 is also isolated as a sheet but, in this membrane, aR-2 does not form a detectable ordered lattice. For crystallization of aR-2, the membrane was partially delipidated and then mixed with nonylglucoside at a detergent:protein weight ratio of w1:1. When ammonium sulfate was used as a precipitant, three crystal forms were obtained at different pH values. Among them, a rhombic crystal grown at pH 5.2 ˚ resolution (Table 1). Although diffracted up to 2.5 A a hexagonal crystal was obtained, it was too small to provide useful structural data. The rhombic crystal of aR-2 contains no trace of bacterioruberin, which is abundantly present in the native claret

Table 1. Data collection and final refinement statistics Data set Data collection ˚) Resolution (A Space group ˚) Unit cell (A

Data completion (%) (outer shell) Number of unique reflections Multiplicity Rsyma (%) (outer shell) I/s (outer shell) Refinement Resolution limit ˚) (A Protein residues Number of water Number of sulfate Number of detergent Rcrystb (%) Rfree (%) rms deviation of ˚) Bond length (A rms deviation of Bond angle (deg.)

Archaerhodopsin-1

Archaerhodopsin-2

40.0–3.4 P43212 aZ128.1 bZ128.1 cZ117.6 aZbZgZ908 90.4 (99.9)

120.0–2.5 C2221 aZ122.9 bZ139.5 cZ108.1 aZbZgZ908 95.8 (97.3)

12,632

30,865

9.9 6.0 (10.3)

3.3 4.4 (49.3)

6.5 (1.3)

9.5 (1.9)

40.0–3.4

15.0–2.5

474 0 0 0

464 29 3 12

22.9 28.2 0.008

24.1 26.8 0.007

1.16

1.11

Here, the statistics as to the structural models deposited in PDB (lauz and lvgo) P are P shown.P P a Rsym Z hkl i jIi KhIij hkl i Ii , where Ii is the intensity of an individual reflection and hIi is the mean intensity obtained from multiple reflections. P observations of symmetry-related P b Rcryst Z hkl ðjFobs jKjFcalc jÞ= hkl jFobs j (5% randomly omitted reflections were used for Rfree).

membrane. In this crystal form, the retinal chromophore has an absorption peak at 555 nm (at 100 K). Crystal structure of aR-1 Our preliminary crystallographic study has shown that the bi-pyramidal crystal of aR-1 belongs to the space group P43212.28 In this study, the crystal structure was further refined using the diffraction ˚ resolution (Table 1.). The result of data at 3.4 A refinement shows that the asymmetric unit contains two subunits with nearly identical conformations (Figure 2(c)). Their main-chains are superimposed ˚ . In each subunit, the with an rms deviation of 0.44 A polypeptide chain is folded into seven a-helices (A through G) and the loop between helices B and C forms an antiparallel b-sheet. All-trans retinal is bound to a lysine residue (Lys216) located in the middle of helix G and the retinal polyene chain is surrounded by the same aromatic residues as observed for bR. Unlike bR, aR-1 has an ordered structure near the N terminus; i.e. the polypeptide from Asp1 to Gly6 forms an omega loop, in which the carboxyl group of Asp1 hydrogen bonds to the backbone amides of residues 4, 5 and 6 (Figure 2(d)). This loop anchors a stretched N-terminal polypeptide from Thr1* to

678

Crystal Structures of Archaerhodopsin-1 and -2

Figure 2. Crystal packing of aR-1 in the P43212 crystal. (a) and (b) Views along the c axis and in perpendicular to the caxis. (c) Wire model of aR-1 dimer with retinal (red). (d)–(g) 2FoKFc maps of protein–protein contact regions, contoured at 1.1s and overlaid on the structural model. Carbon atoms in subunit-I and -II are drawn in gold and light blue, respectively; nitrogen, oxygen and sulfur atoms are in blue, red and green, respectively. This Figure and Figures 3–7 were drawn with XtalView.46

Ala6*, which plays an important role in the head-tohead association of two subunits; i.e. the backbone amides and carbonyls of Ala3* and Gly5* of one subunit (subunit-I) interacts with the BC loop of the adjacent monomer (subunit-II), forming a b sheet comprising of three strands (Figure 2(g)). However, this head-to-head association is not very strong. In fact, the N-terminal polypeptide of subunit-II is far from the BC loop of subunit-I; instead, it is involved in association of two dimers that are related by a crystallographic 2-fold axis (Figure 2(b)). The inter-dimer association is strengthened by a hydrogen-bonding interaction between Thr183 OH of subunit-I and Tyr135 OH of subunit-II and by hydrophobic interactions between residues in helix F of subunit-I and those in helices F and G of subunit-II (Figure 2(f)). Interactions between neighboring tetramers are mediated by hydrophobic residues in helix A of subunit-I and those in helices A and B of subunit-I (Figure 2(e)). Bacterioruberin has no clear electron density and its structural role in the crystal packing has remained to be clarified. (Although a long string of electron density is seen

along the hydrophobic surface of protein, it is not clear enough to provide structural information about bacterioruberin.) A detailed comparison of the two subunits in the asymmetric unit suggests that the AB and the BC loops are flexible so that their conformations are altered noticeably by the protein–protein interactions. Except for these loops, there is no essential structural difference between the two subunits. Crystal structure of aR-2 The C2221 crystal of aR-2 is composed of membranous layers, in each of which the proteins are arranged on an orthorhombic lattice (Figure 3(b)). The overall structure of aR-2 is similar to that of aR-1. Two subunits in the asymmetric unit associate tightly with each other via a hydrogen-bonding network that is composed of ionic residues (Asp36 in the AB loop, Arg163 and Glu166 in the EF loop and Arg227 at the C-terminal end of helix G) and two sulfate ions (Figure 3(d)). In this contact region, two aspartic residues (Asp36) from the two subunits contact each

Crystal Structures of Archaerhodopsin-1 and -2

679

Figure 3. Crystal packing of aR-2 in the C2221 crystal. (a) and (b) Views along the b axis and the c-axis. (c) aR-2 dimer with retinal (red) and detergent molecules (yellow). (d)–(g) 2FoKFc maps of protein–protein contact regions, contoured at 1.1s and overlaid on the structural model. Carbon atoms in subunit-I and -II are drawn in gold and light blue, respectively; nitrogen, oxygen and sulfur atoms are in blue, red and green, respectively.

other, forming a paired structure. Besides the head-tohead association, there are three different types of protein–protein contact (Figure 3(b)). In one of them, Thr183 OH in one subunit (subunit-I) hydrogenbonds to Tyr135 OH in the DE loop of the adjacent subunit (Figure 3(e)). In a protein–protein contact around a crystallographic 2-fold axis, the indole nitrogen of Trp32 is hydrogen-bonded to the backbone carbonyl of Arg40, the side-chain of which is hydrogen-bonded to the backbone carbonyl of

residue 32 (Figure 3(f)). In another inter-dimer contact, Thr136 OH of the B chain is hydrogenbonded to Tyr151 OH in the middle of helix E of a different subunit (Figure 3(g)). This protein–protein contact is further stabilized by pi–pi interactions between aromatic residues (Phe141 and Phe147) from the two subunits. It is noteworthy that Tyr151 and Phe147 in aR-2 are replaced by other residues in aR-1 and bR. This implies that the protein packing found in the C2221 crystal is unique to aR-2.

680 The two subunits in the asymmetric unit are ˚ . But superimposed with an rms deviation of 0.44 A a significant structural difference is seen in the proton release complex existing near the extracellular membrane surface, which is composed of two glutamate residues (Glu194 and Glu204) and the nearby water molecules (Figure 4). These glutamate residues form a paired structure in subunit-I, while this paired structure is broken in subunit-II. In the latter subunit, Glu194 directs towards Arg134, the side-chain of which mediates interactions between helices D and E. This structural difference is accounted for by a tilting of the BC loop toward the DE loop in subunit-I. In subunit-I, the backbone amide of Ala72 in the BC loop is hydrogen-bonded to the polypeptide carbonyl of Lys129 in the DE loop. Formation of this hydrogen-bonding accompanies a small distortion in the DE loop. It is possible that the latter distortion is caused by the protein–protein contact and that it affects the conformation of the BC loop as well as the orientation of Arg134 and, thereby, altering the structure of the proton release complex. The N-terminal polypeptide (Asp1 to Arg7) of aR-2 forms an omega loop as found in aR-1. But no residue before Asp1 is visible in the electron density map. A previous biochemical assay indicated that the N-terminal polypeptide of aR-2 resists digestion.29 This resistance may be explained by supposing that the omega loop is so rigid as to prevent full access of a peptidase to Asp1. Thus the possibility

Figure 4. Structural comparison of the two subunits in the dimeric structure of aR-2. 2FoKFc map of the proton release channel in subunit-I, contoured at 1.1s and overlaid on its structural model. Carbon, nitrogen and oxygen atoms in subunit-I are drawn in gold, blue and red, respectively. Carbon atoms in subunit-II that is superimposed on subunit-I are drawn in light blue. Red and cyan spheres represent water molecules in subunit-I and -II, respectively.

Crystal Structures of Archaerhodopsin-1 and -2

cannot be excluded that Asp1 is actually the first residue in the matured form of aR-2.

Discussion A detailed structural comparison of aR-2 with aR1 indicates that the proton release channel of aR-2 is opened as compared with that of aR-1 (Figure 5). The structural difference in this region is more significant when aR-2 is compared with bR. It is noteworthy that, irrespective of a higher sequence homology between aR-1 and aR-2, the structural correlation of the polypeptide backbone around the BC loop is lower between aR-1 and aR-2 than between aR-1 and bR. The low structural correlation between aR-2 and aR-1 is attributable to difference in the length of the BC loop, which is one residue longer in aR-2 than in aR-1 (or bR). As the hydrogen bonds within the b-sheet are conserved between aR-2 and aR-1, insertion of one residue (Gly73*) at the turning point in the antiparallel b-sheet is accompanied inevitably by a large distortion in the b-sheet structure, which propagates to its neighbors. In fact, the structural difference between aR-2 and aR-1 is enhanced at residues Gly63–Gly65 (i.e. at the junction between the b-sheet and helix B) and at residues Tyr80–Arg82 (i.e. at the junction between the b-sheet and helix C); note that helix C is two residues longer in aR-2 than in aR-1 or bR. The latter difference accompanies alteration in the backbone conformation of Arg82 and re-orientation of its side-chain; i.e. the guanidinium ion of Arg82 directs towards Glu204 in aR-2, while it directs towards Tyr57 in aR-1 and bR (Figure 5(b)). As a consequence of these alterations, the outlet of the proton release channel is enlarged in aR-2, in which the carboxyl group of Glu194 is surrounded by a larger number of water molecules. Such distribution of water molecules may weaken the paired structure of Glu194 and Glu204. Indeed it was observed that the paired structure is broken in one of the subunits contained in the asymmetric unit (Figure 5(b)). Since the paired structure of Glu194 and Glu204 has been shown to break in the acid blue form of bR,30 the observed structure of aR-2 is correlated well with the previous report that the pKa value of the purple-to-blue transition of aR-2 is much higher (pKaw5) than observed for aR-1 or bR (pKaw3.5–4).22 As long as the central part of the protein is discussed, there is no essential structural difference among bR, aR-1 and aR-2 (Figure 5(a)). All the aromatic residues (Trp86, Trp139, Trp182, Tyr185, Trp189) surrounding the retinal polyene chain are conserved among the three proton pumps. This observation is in accordance with the previous argument that these residues are absolutely required for fixation of the C5–C12 portion of the polyene chain and thus for donating the stereospecificity of the retinal isomerization.10 However, a residue (Met145 in aR-1 and bR) contacting the C5 methyl of retinal is substituted to phenylalanine in

Crystal Structures of Archaerhodopsin-1 and -2

681 result is in line with the previous observation of a higher content (77%) of the trans isomer in the dark adapted state of aR-2.24 In Figure 6, the structures of the cytoplasmic halves of the three proton pumps (bR, aR-1 and aR-2) are compared. It shows that the structural conservation is very low for residues extruded from helices E and F. This would be expected from the low sequence homology for residues in these helices (Figure 1). It is also clear that most residues extruding from the protein surface into the lipid bilayer have varied during the evolution of archaeal rhodopsins. This implies that the topology of the protein surface contacting lipid molecules can be altered without loss of the proton pumping activity. In contrast to a large variation in the topology of the protein surface, the region surrounded by helices B, C and G is highly conserved among the three proton pumps (enclosed by a red broken line in Figure 7). Especially the structure between the retinal Schiff base and Asp96 is conserved perfectly. This structural conservation is understood form the functional point of view, as Asp96 acts as a proton donor to the Schiff base in the M-to-N transition.31,32 It is noteworthy that most residues

Figure 5. Structural comparison of aR-1, aR-2 and bR. 2FoKFc density maps of the retinal-binding pocket (a) and the proton release channel (b) in the subunit-I of aR-2 are contoured at 1.1s and overlaid on its structural model; carbon, oxygen and nitrogen atoms in aR-2 are drawn in gold, red and blue, respectively. The structural models of aR-1 (subunit-I) and bR are superimposed on aR-2; carbon atoms in aR-1 and bR are drawn in yellow and grey, respectively.

aR-2.3 Since the side-chain of Met145 pushes the ionone ring downwards, the retinal polyene chain of the all-trans retinal is bent largely in aR-1 or bR. In this architecture of the retinal binding pocket, the 13-cis isomer seems to be more stabilized as compared with the trans isomer. In aR-2, on the other hand, the benzene ring of Phe145 is far from the C5 methyl. Since the confliction between the ionone ring and its surrounding is much smaller, the all-trans retinal can be accommodated with a less significant bending of the polyene chain. This

Figure 6. Structural comparison of the cytoplasmic halves of aR-1, aR-2 and bR. Carbon atoms in aR-2 (subunit-I), aR-1(subunit-I) and bR are drawn in gold, yellow and cyan, respectively, and nitrogen and oxygen atoms in all the proteins are in blue and red, respectively.

682

Crystal Structures of Archaerhodopsin-1 and -2

Figure 7. Micro-cavities in the light-adapted states of aR-2 (a), aR-1 (b) and bR (c) and in the L intermediate of bR (d). In (a) and (b), two subunits contained in the asymmetric unit are superimposed and micro-cavities in the subunits I and II are drawn in yellow and cyan, respectively; the micro-cavities observed commonly in both the subunits are in green. In (c) and (d), micro-cavities are drawn in cyan. Some important residues (Arg82, Leu93, Asp96, Lys216 and retinal) are represented with a ball-and-stick model. Red spheres in (a), (c) and (d) represent water molecules. Since the proton release channel in aR-2 is so opened that cavity VI is actually connected to the solvent region, this cavity is visualized by placing an artificial polypeptide near the outlet of the proton release channel. This Figure was drawn with SwissPdbViewer.48

in the postulated proton uptake pathway are hydrophobic (e.g. Phe27, Leu93, Leu97, Leu99, Leu100, Phe171, Leu174, Phe219, Leu223 and Leu224). These residues are conserved in all the archaeal proton pumps whose amino acid

sequences are available, while most of them are substituted to different amino acids in archaeal sensory rhodopsins. 2–4 A simple and likely explanation for their conservation among the archaeal proton pumps is that a precise arrange-

683

Crystal Structures of Archaerhodopsin-1 and -2

ment of these hydrophobic residues is necessary for blocking a backward flow of proton in the presence of a large pH gradient and membrane potential. Indeed water molecules are excluded from the proton uptake pathway in the unexcited initial state of aR-2 and bR, in which the pKa of Asp96 is kept very high (pKaO11).33 But, as a transient decrease in the pKa of Asp96 takes place during the proton pumping cycle, the hydrophobic residues in the proton uptake pathway must be arranged with such a plasticity that allows water molecules to access to Asp96 in the photoreaction N intermediate.34 It has been hypothesized that water access to Asp96 increases when the cytoplasmic half of helix F tilts outward in the M-to-N transition.35–37 But, this hypothesis was recently questioned by a crystallographic study of the N intermediate.38 The low conservation of residues in helix F among the archaeal proton pumps also suggests the possibility that the outward tilting of helix F is not essential for active transport of protons; i.e. it follows another essential structural change induced within the highly conserved region. A previous structural analysis of the L intermediate of bacteriorhodopsin has shown that the sidechain of Leu93 rotates in the K-to-L transition.25 This rotation creates a micro-cavity above the retinal Schiff base, which accommodates a water molecule (Wat602) that is dragged up across the Schiff base. Figure 7 shows the distribution of micro-cavities as well as internal water molecules in aR-1, aR-2 and bR. Some cavities are seen in only one of the two subunits in the asymmetric unit (e.g. cavity II in aR-1 and aR-2), suggesting that the volume of each cavity is influenced by the lattice force. Nonetheless, a micro-cavity existing in the vicinity of Leu93 (cavity I) is commonly seen in the unexcited initial state. Existence of this cavity is required for rotation of the side-chain of Leu93 in the K-to-L transition. Indeed cavity I vanishes transiently in the L intermediate, in which the volumes of cavities II and III increase so as to accommodate water molecules. From structure comparison with the M intermediate generated at room temperature,39 it is suggested that a series of volume changes in the micro-cavities control water relocation within the proton pathway. Since the protonation states of the Schiff base and Asp96 are dependent on how they interact with water molecules, it is predicted that a more comprehensive analysis of micro-cavities of archaeal proton pumps will offer a simple molecular mechanism that accounts for the directional movement of protons.

Materials and Methods Protein purification and crystallization Claret membrane-1 and -2 were isolated from Halorubrum sp. aus-2 and -1, respectively, according to the

procedure utilized for preparation of purple membrane of Halobacterium salinarum.40 For crystallization of aR-1, a mixture (40–80 ml) of claret membrane-1 (5 mg/ml), 5 mg/ml b-octythioglucoside (Dojin, Kumamoto, Japan), 1.0 M ammonium sulfate, 0.3% heptane-triol, 0.04% (w/v) NaN3 and 40 mM sodium citrate (pH 5.2) was concentrated at 10 8C by the sitting-drop vapor diffusion method using a commercial crystallization kit (Vapor Diffusion Crystalplate, ICN Biomedicals, Inc., Ohio), each reservoir well of which was filled with 0.5 ml of 2.8–2.9 M ammonium sulfate and 100 mM citrate buffer. Incubation for three to six weeks yielded bi-pyramidal crystals with a size of 0.1 mm! 0.1 mm!0.05 mm. For crystallization of aR-2, claret membrane-2 in 0.1 M NaCl at pH 8.0 was partially delipidated with 0.5% Tween20 at 20 8C for 30 min41 and, subsequently, a mixture (40–80 ml) of delipidated claret membrane-2 (5 mg/ml), 4 mg/ml nonylglucoside, 1.0 M ammonium sulfate, 0.04% NaN3 and 40 mM sodium citrate (pH 5.2) was concentrated at 10 8C by the sitting-drop vapor diffusion method, using 0.5 ml of 3.2–3.3 M ammonium sulfate and w100 mM citrate buffer as the reservoir solution. Incubation for three to six weeks yielded rhombic crystals with a size of 0.1 mm!0.1 mm! 0.02 mm. Measurements of absorption spectra Absorption spectra of the crystals of aR-1 and aR-2 were measured using a micro-spectrophotometer, in which monochromatic light from a double monochromator (Shimadzu UV350A) was focused on a single crystal by a combination of quartz lenses and two diaphragms in a home-made microscope and the measuring light passing through the crystal was focused again on a pinhole attached before a photomultiplier tube.42 For measurements at cryogenic temperature, a frozen crystal was mounted on a goniometer head attached to the micro-spectrophotometer, by which the crystal orientation was adjusted. The temperature of the crystal was controlled in flow of cold nitrogen gas from a cryostreamer (Oxford CC-12). Data collection and scaling X-ray diffraction measurements were performed at the beamlines SPring8-BL38B1 and -BL41XU, where a frozen crystal kept at 100 K was exposed to a monochromatic ˚ with an X-ray flux rate X-ray beam at wavelength of 1.0 A of w2!1012 photons/mm2 per s. Diffraction data were collected using a CCD detector (ADSC Quantum 4R), with an oscillation range of 18 and an X-ray flux of 4!1013 photons/mm2 per image. Indexing and integration of diffraction spots were carried out with Mosflm 6.1.43 The scaling of data was done using SCALA in the CCP4 program suites.44 Structural refinement Structural analyses of aR-1 and aR-2 were performed ˚ with CNS45 and Xtalview.46 Diffraction data at 3.4 A resolution from the P43212 crystal of aR-1 were analyzed by molecular replacement using the structure of bR (pdb id: 1iw6) as an initial model.47 After a rigid body refinement, the conformations of two aR-1 molecules in the asymmetric unit were modified manually on the ground of the 2FoKFc map. Subsequent refinement

684 including simulated annealing and individual B-factor refinement resulted in an Rcryst of 22.9% and an Rfree of 28.2% (Table 1). The final model is composed of two polypeptide chains, each of which is composed of 237 residues (shown by bold letters in Figure 1). ˚ resolution from the C2221 Diffraction data at 2.5 A crystal of aR-2 were analyzed in a similar way as mentioned above but using the structural model of aR-1 (pdb id: 1auz) as an initial model. After a number of cycles of manual refinement, simulated annealing and individual B-factor refinement, clear electron densities became visible along two polypeptide chains in the asymmetric unit, each of which is composed of 234 residues (shown by bold letters in Figure 1). Then water molecules, sulfate ions and detergent molecules were added on the basis of the 2FoKFc map. Subsequent refinement resulted in an Rcryst of 24.3% and an Rfree of 26.9% (Table 1). The final model contains 29 water molecules, 12 nonylglucoside and three sulfate ions in the asymmetric unit. ˚ 3) in archaeal proton pumps Micro-cavities (O15 A were visualized with SwissPdbViewer.48 Protein Data Bank accession codes Crystallographic coordinates of archaerhodopsin-1 and -2 have been deposited in the RCSB Protein Data Bank (accession codes: 1auz and 1vgo, respectively).

Acknowledgements We express our gratitude to Drs N. Shimidzu and K. Hasegawa for helping data collection at the beamlines BL38B1 and BL40B2 of SPring-8. This work was supported by grant-in-aids from the Ministry of Education, Science, and Culture of Japan and partly by National Project on Protein Structural and Functional Analyses.

References 1. Oesterhelt, D. & Stoeckenius, W. (1971). Rhodopsinlike protein from the purple membrane of Halobacterium halobium. Nature, 233, 149–152. 2. Mukohata, Y. (1994). Comparative studies on ion pumps of the bacterial rhodopsin family. Biophys. Chem. 50, 191–201. 3. Ihara, K., Umemura, T., Katagiri, I., Kitajima-Ihara, T., Sugiyama, Y., Kimura, Y. & Mukohata, Y. (1999). Evolution of the archaeal rhodopsins: evolution rate changes by gene duplication and functional differentiation. J. Mol. Biol. 285, 163–174. 4. Blanck, A. & Oesterhelt, D. (1987). The haloopsin gene II. Sequence primary structure of halorhodopsin and comparison with bacteriorhodopsin. EMBO J. 6, 265–273. 5. Blanck, A., Oesterhelt, D., Ferrando, E., Schegk, E. S. & Lottspeich, F. (1989). Primary structure of sensory rhodopsin I, a prokaryotic photoreceptor. EMBO J. 8, 3963–3971. 6. Oesterhelt, D. (1998). The structure and mechanism of the family of retinal proteins from halophilic archaea. Curr. Opin. Struct. Biol. 8, 489–500.

Crystal Structures of Archaerhodopsin-1 and -2

7. Landau, E. M., Pebay-Peyroula, E. & Neutze, R. (2003). Structural and mechanistic insight from high resolution structures of archaeal rhodopsins. FEBS Letters, 555, 51–56. 8. Fodor, S. P. A., Pollard, W. T., Gebhard, R., van den Berg, E. M. M., Lugtenburg, J. & Mathies, R. A. (1988). Bacteriorhodopsin’s L550 intermediate contains a C14C15 s-trans-retinal chromophore. Proc. Natl Acad. Sci. USA, 85, 2156–2160. 9. Yan, B., Takahashi, T., Johnson, R., Derguini, F., Nakanishi, K. & Spudich, J. L. (1990). All-trans/13cis isomerization of retinal is required for phototaxis signaling by sensory rhodopsins in Halobacterium halobium. Biophys. J. 57, 807–814. 10. Nishikawa, T., Okumura, T. & Kouyama, T. (2005). Crystal structure of the 13-cis isomer of bacteriorhodopsin in the dark-adapted state. J. Mol. Biol. 352, 319–328. 11. Belrhali, H., Nollert, P., Royant, A., Menzel, C., Rosenbusch, J. P., Landau, E. M. & Pebay-Peyroula, E. (1999). Protein, lipid and water organization in bacteriorhodopsin crystals: a molecular view of the purple ˚ resolution. Proteins: Struct. Funct. membrane at 1.9 A Genet. 7, 909–917. 12. Luecke, H., Schobert, B., Richter, H.-T., Cartailler, J.-P. & Lanyi, J. K. (1999). Structure of bacteriorhodopsin at ˚ resolution. J. Mol. Biol. 291, 899–911. 1.55 A 13. Kolbe, M., Besir, H., Essen, L. O. & Oesterhelt, D. (2000). Structure of the light-driven chloride pump ˚ resolution. Science, 288, halorhodopsin at 1.8 A 1390–1396. 14. Royant, A., Nollert, P., Edmani, K., Neutze, R., Landau, E. M., Pebay-Peyroula, E. & Navarro, J. (2001). X-ray ˚ resolution. structure of sensory rhodopsin II at 2.1-A Proc. Natl Acad. Sci. USA, 98, 10131–10136. 15. Luecke, H., Schobert, B., Lanyi, J. K., Spudich, E. N. & Spudich, J. L. (2001). Crystal structure of sensory rhodopsin II at 2.4 angstroms: insights into color tuning and transducer interaction. Science, 293, 1499–1503. 16. Edman, K., Royant, A., Nollert, P., Maxwell, C. A., Pebay-Peyroula, E., Navarro, J. et al. (2002). Early structural rearrangements in the photocycle of an integral membrane sensory receptor. Structure, 10, 473–482. 17. Vogeley, L., Sineshchekov, O. A., Trivedi, V. D., Sasaki, J., Spudich, J. L. & Luecke, H. (2005). Anabaena sensory ˚. rhodopsin: a photochromic color sensor at 2.0 A Science, 306, 1390–1393. 18. Be´ja`, O., Spudich, E. N., Spudich, J. L., Leclerc, M. & DeLong, E. (2001). Proteorhodopsin phototrophy in the ocean. Nature, 411, 786–789. 19. Ga¨rtner, W. & Losi, A. (2003). Crossing the borders: archaeal rhodopsins go bacterial. Trends Microbiol. 11, 405–407. 20. Mukohata, Y., Sugiyama, Y., Ihara, K. & Yoshida, M. (1988). An Australian Halobacterium contains a novel protein pump retinal protein: archaerhodopsin. Biochem. Biophys. Res. Commun. 151, 1339–13445. 21. Sugiyama, Y., Maeda, M., Futai, M. & Mukohata, Y. (1989). Isolation of a gene that encodes a new retinal protein, archaerhodopsin, from Halobacterium sp. aus-1. J. Biol. Chem. 264, 20859–20862. 22. Mukohata, Y., Ihara, K., Uegaki, K., Miyashita, Y. & Sugiyama, Y. (1991). Australian halobacteria and their retinal protein ion pumps. Photochem. Photobiol. 54, 1039–1045. 23. Lukashev, E. P., Govindjee, R., Kono, M., Ebrey, T. G., Sugiyama, Y. & Mukohata, Y. (1994). pH dependence

Crystal Structures of Archaerhodopsin-1 and -2

24.

25.

26.

27.

28. 29.

30.

31.

32.

33.

34.

of the absorption spectra and photochemical transformations of the archaerhodopsins. Photochem., Photobiol. 60, 69–75. Ihara, K., Amemiya, T., Miyashita, Y. & Mukohata, Y. (1994). Met-145 is a key residue in the dark adaptation of bacteriorhodopsin homologs. Biophys. J. 67, 1187–1191. Kouyama, T., Nishikawa, T., Tokuhisa, T. & Okumura, H. (2003). Crystal structure of the L intermediate of bacteriorhodopsin: evidence for vertical translocation of a water molecule during the proton pumping cycle. J. Mol. Biol. 335, 531–546. Kouyama, T., Yamamoto, M., Kamiya, N., Iwasaki, H., Ueki, T. & Sakurai, I. (1994). Polyhedral assembly of a membrane protein in its three-dimensional crystal. J. Mol. Biol. 236, 990–994. Takeda, K., Sato, H., Hino, T., Kono, M., Fukuda, K., Sakurai, I. et al. (1998). A novel three-dimensional crystal of bacteriorhodopsin obtained by successive fusion of the vesicular assemblies. J. Mol. Biol. 283, 463–474. Enami, N., Okumura, H. & Kouyama, T. (2002). X-ray crystallographic study of archaerhodopsin. J. Photosci. 9, 320–322. Uegaki, K., Sugiyama, Y. & Mukohata, Y. (1991). Archaerhodopsin-2, from Halobacterium sp. aus-2 further reveals essential amino acid residues for light-driven proton pumps. Arch. Biochem. Biophys. 286, 107–110. Okumura, H., Murkami, M. & Kouyama, T. (2005). Crystal structures of acid blue and alkaline purple forms of bacteriorhodopsin. J. Mol. Biol. 351, 481–495. Butt, H. J., Fendler, K., Bamberg, E., Tittor, J. & Oesterhelt, D. (1989). Aspartic acids 96 and 85 play a central role in the function of bacteriorhodopsin as a proton pump. EMBO J. 8, 1657–1663. Gerwert, K., Souvignier, G. & Hess, B. (1990). Simultaneous monitoring of light-induced changes in protein side-group protonation, chromophore isomerization, and backbone motion of bacteriorhodopsin by time-resolved Fourier-transform infrared spectroscopy. Proc. Natl Acad. Sci. USA, 87, 9774–9778. Szaraz, S., Oesterhelt, D. & Ormos, P. (1994). pHinduced structural changes in bacteriorhodopsin studied by Fourier transform infrared spectroscopy. Biophys. J. 67, 1706–1712. Zscherp, C., Schlensinger, R., Tittor, J., Oesterhelt, D. & Heberle, J. (1999). In situ determination of transient pKa changes of internal amino acids of bacteriorhodopsin by using time-resolved attenuated total reflection Fourier-transform infrared spectroscopy. Proc. Natl Acad. Sci. USA, 96, 5498–5503.

685 35. Subramaniam, S., Gerstein, M., Oesterhelt, D. & Henderson, R. (1993). Electron diffraction analysis of structural changes in the photocycle of bacteriorhodopsin. EMBO J. 12, 1–8. 36. Han, B.-G., Vonck, J. & Glaeser, R. M. (1994). The bacteriorhodopsin photocycle: direct structural study of two substrates of the M-intermediate. Biophys. J. 67, 1179–1186. 37. Vonck, J. (1996). A three-dimensional difference map of the N intermediate in the bacteriorhodopsin photocycle: part of the F helix tilts in the M to N transition. Biochemistry, 35, 5870–5878. 38. Schobert, B., Brown, L. S. & Lanyi, J. K. (2003). Crystallographic structures of the M and N intermediates of bacteriorhodopsin: assembly of a hydrogen-bonded chain of water molecules between Asp-96 and the retinal Schiff base. J. Mol. Biol. 330, 553–570. 39. Takeda, K., Matsui, Y., Kamiya, N., Adachi, S., Okumura, H. & Kouyama, T. (2004). Crystal structure of the M intermediate of bacteriorhodopsin: allosteric structural changes mediated by sliding movement of a transmembrane helix. J. Mol. Biol. 341, 1023–1037. 40. Oesterhelt, D. & Stoeckenius, W. (1974). Isolation of the cell membrane of Halobacterium halobium and its fragmentation into red and purple membrane. Methods Enzymol. 31, 667–678. 41. Fukuda, K., Ikegami, A., Nasuda-Kouyama, A. & Kouyama, T. (1990). Effect of partial delipidation of purple membrane on the photodynamics of bacteriorhodopsin. Biochemistry, 29, 1997–2002. 42. Sakai, K., Matsui, Y., Kouyama, T., Shiro, Y. & Adachi, S. (2002). Optical monitoring of freeze-trapped reaction intermediates in protein crystals: a microspectrophotometer for cryogenic protein crystallography. J. Appl. Crystallog. sect D, 35, 270–273. 43. Steller, I., Bolotovsky, B. & Rossmann, M. G. (1997). An algorithm for automatic indexing of oscillation images using Fourier analysis. J. Appl. Crystallog. sect. D, 30, 1036–1040. 44. CCP4. (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760–763. 45. Bru¨nger, A. T., Adams, P. D., Glore, G. M., Delano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography and NMR system (CNS): a new software system for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. 46. McRee, D. E. (1993). Practical Protein Crystallography, Academic Press, San Diego. 47. Matsui, Y., Sakai, K., Murakami, M., Shiro, Y., Adachi, S., Okumura, H. & Kouyama, T. (2002). Specific damages induced by X-ray radiation and structural changes in the primary photoreaction of bacteriorhodopsin. J. Mol. Biol. 324, 469–481. 48. Guex, N. & Peitsch, M. C. (1997). SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling. Electrophoresis, 18, 14–23.

Edited by R. Huber (Received 18 October 2005; received in revised form 9 February 2006; accepted 14 February 2006) Available online 3 March 2006