Bothnia dystrophy is caused by domino-like rearrangements in cellular retinaldehyde-binding protein mutant R234W

Abstract

Cellular retinaldehyde-binding protein (CRALBP) is essential for mammalian vision by routing 11-cis-retinoids for the conversion of photobleached opsin molecules into photosensitive visual pigments. The arginine-to-tryptophan missense mutation in position 234 (R234W) in the human gene RLBP1 encoding CRALBP compromises visual pigment regeneration and is associated with Bothnia dystrophy. Here we report the crystal structures of both wild-type human CRALBP and of its mutant R234W as binary complexes complemented with the endogenous ligand 11-cis-retinal, at 3.0 and 1.7 Å resolution, respectively. Our structural model of wild-type CRALBP locates R234 to a positively charged cleft at a distance of 15 Å from the hydrophobic core sequestering 11-cis-retinal. The R234W structural model reveals burial of W234 and loss of dianion-binding interactions within the cleft with physiological implications for membrane docking. The burial of W234 is accompanied by a cascade of side-chain flips that effect the intrusion of the side-chain of I238 into the ligand-binding cavity. As consequence of the intrusion, R234W displays 5-fold increased resistance to light-induced photoisomerization relative to wild-type CRALBP, indicating tighter binding to 11-cis-retinal. Overall, our results reveal an unanticipated domino-like structural transition causing Bothnia-type retinal dystrophy by the impaired release of 11-cis-retinal from R234W.

Bothnia dystrophy is an autosomal recessive disease of the retina with high prevalence in northern Sweden (1). The disease, which affects 1 in 4500 births, is characterized by early-onset night blindness followed by macular degeneration and eventual loss of vision in adulthood (2). Bothnia dystrophy is caused by the arginine-to-tryptophan missense mutation in position 234 (R234W) in the RLBP1 gene, which has been mapped to chromosome 15q26 (1). Several other clinical phenotypes, including autosomal recessive retinitis pigmentosa (3), retinitis punctata albescens (4), fundus albipunctatus (5), and Newfoundland rod-cone dystrophy (6), could be associated with up to 8 different amino acid substitution mutations in the same gene. However, the molecular basis of the retinal disorders that result from these RLBP1 mutations is not well understood because of a lack of structural data.

The RLBP1 gene encodes the 36-kDa cellular retinaldehyde-binding protein (CRALBP), a soluble retinoid carrier that first was isolated from the cytoplasm of bovine retina in complex with endogenous 11-cis-retinal and 11-cis-retinol (3, 7). CRALBP possesses high stereoselectivity toward 11-cis-retinoids, and it has been proposed that CRALBP protects these retinoids from premature photoisomerization and from enzymatic reverse isomerization (8, 9). In vivo and in vitro studies suggest that CRALBP is essential for the proper function of both rod and cone photoreceptors (10, 11). Interestingly, CRALBP is not present in photoreceptor cells but is expressed abundantly in the adjacent retinal pigment epithelium (RPE) and in the Müller glial cells of the retina (12, 13). In both cell types, CRALBP participates in the regeneration of active 11-cis-retinoids from the inactive 11-trans products of the rhodopsin photocycle and in the de novo synthesis of these retinoids from 11-trans metabolic precursors. The cycling of retinoids between photoreceptor and adjacent pigment epithelium cells is known as the “visual cycle” (14, 15). CRALBP of the RPE cells binds 11-cis-retinol, which is the direct product of the isomerohydrolysis of all-trans-retinylesters in the visual cycle and which also is formed upon activation of stored 11-cis-retinylesters (16). CRALPB functions as a retinoid carrier, assisting the oxidation of 11-cis-retinol to 11-cis-retinal in the next step of the visual cycle and preventing uncontrolled storage of 11-cis-retinol by esterification (14). In Müller glial cells CRALBP is reported to have the opposite function—namely, assisting the esterification of 11-cis-retinol (17). Stable supramolecular complexes between components with retinoid-processing activity and CRALBP have been isolated from bovine retina by co-purification and co-precipitation (18, 19). Functional and structural interactions of CRALBP with the membrane-bound cis-retinoid-specific dehydrogenase and with the multivalent plasma membrane linker in the RPE have been demonstrated both in vitro and in vivo (20). Ligand-binding studies with CRALBP bearing mutations in the lipid-binding pocket implicate at least 10 hydrophobic protein-to-ligand interactions in retinoid binding (21, 22). The R234W mutant of CRALBP has a 2-fold increased affinity for 11-cis-retinal. The M226K and R151Q mutants have lost all retinoid-binding affinity (23). CRALBP contains 2 conserved domains termed “CRAL-TRIO” (pfam03765: residues 46–127; pfam00650: residues 133–314) that are characteristic for the SEC14p-like family of lipid binders and include a flexible helical gate giving access to a lipid-binding cavity (24, 25). Crystal structures of 4 homologous SEC14p-like lipid binders and a Ras-specific GTPase-activating protein (RasGap) suggest structural and functional conservation of the CRAL-TRIO fold (26–30).

Here we present the crystal structures of wild-type human CRALBP and of its disease-causing R234W mutant, both in complexes with their endogenous ligand, 11-cis-retinal. Extensive van der Waals interactions between the ligand-binding pocket and 11-cis-retinal explain the high stereoselectivity of CRALBP and how it protects the ligand against premature photoisomerization. In the R234W 11-cis-retinal complex, ligand binding is affected by domino-like conformation changes. Comparison of the binding-pocket conformation of wild-type and R234W CRALBP reveals enhanced packing density in the mutant resulting in a tighter binding of the ligand. A consequence of tighter binding by the R234W mutant is the reduced rate of ligand photoisomerization. Tighter binding also slows down the release of 11-cis-retinal from the protein, explaining how and why the R234W mutation inhibits retinal metabolism in the human eye. The structures of CRALBP and of the R234W mutant provide an explanation of the molecular mechanism underlying Bothnia dystrophy.

Results and Discussion

Crystallization and Structure Determination.

Apo-CRALBP did not afford diffracting crystals, probably because of aggregation-induced heterogeneity of the apoprotein. The addition of different non-ionic detergents to apo-CRALBP did not improve crystal quality. Addition of a 1.5 molar excess of 11-cis-retinal reduced aggregation and improved crystal quality to <4.0 Å diffraction, but the crystals were twinned. Replacement of high-entropy surface residues (lysine, glutamate) by alanine (31) and truncation of the protein from both ends did not improve crystal quality. Gel filtration chromatography to remove unbound 11-cis-retinal and strict selection of the fractions with the lowest polydispersity index (<15%) by dynamic light scattering (DLS) yielded CRALBP that produced hexagonal crystals of space group P6₅22 diffracting to 3 Å. New surface mutants were constructed with the goal of improving resolution further; 1 of these mutants was the disease-causing mutation R234W (23). Crystals of the selenomethionine-substituted R234W 11-cis-retinal complex (R234W) diffracted to 1.7 Å. Phases were determined using R234W and single-wavelength anomalous dispersion (SAD) (32). The structure of wild-type CRALBP was solved subsequently by molecular replacement using the atomic model of R234W. The data collection, refinement, and validation statistics are reported in supporting information (SI) Table S1.

Overall Fold of Wild-Type CRALBP.

The structure model of wild-type CRALBP comprises residues 23–306of 317 residues (Fig. 1). The protein folds into the prototypical SEC14-like domain structure and is most similar to α-tocopherol transfer protein (PDB code 1OIP; overall sequence identity of 29%; ref. 26). CRALBP is composed of 2 domains, the N-terminal all-helical α domain and the C-terminal 3-layer αβα sandwich domain. (The topology of the secondary structure of CRALBP is depicted in Fig. S1.) The N-terminal α domain comprises helices α1–α5. The core of the C-terminal αβα domain comprises a β-sheet with 1 antiparallel and 4 parallel strands, β1–β5, and 6 helices. Helices α6 and α7 are packed against the concave face, and the 4 helices α9–α12are packed against the convex face of the sheet. A third structural motif is formed between the aperiodic segment of the N terminus (residues 23–42) and 2 helices (α13, α14) of the C terminus of CRALBP. The retinal-binding cavity is delimited by the convex side of the β-sheet and the 4 adjacent helices. The interdomain contacts are governed by hydrophobic interactions between helix α5 of the N-terminal α domain and helices α7 and α8 of the C-terminal αβα domain. Side-chain hydrogen bonds between Y117 of helix α5 and E185 of helix α7 and between Y124 of helix α5 and D225 of helix α8 control the orientation of the interdomain contact.

Fig. 1.

Monomeric structure of CRALBP. (A) Ribbon diagram of wild-type CRALBP bound to 11-cis-retinal. The helices of the N-terminal domain are indicated in green; C-terminal helices are indicated in blue; the helical gate (helices α11 and α12) is indicated in gray; β-strands are indicated in red. The position of R234 is indicated as a yellow sphere, the 11-cis-retinal ligand is shown as orange sticks, and the cavity surface is shown in orange. The cavity surface was calculated with VOIDOO (33). (B) View after rotation of A by 90° on the vertical axis. Images were generated with PYMOL (54).

Structure of the R234W Mutant of CRALBP.

The structure model of the R234W mutant comprises residues 57–306 of 317 residues. The R234W mutant crystallized in the space group C2, unlike wild-type CRALBP, which crystallized in space group P6₅22. Because of the different crystal contacts, the N-terminal residues 1–56 are disordered in the mutant. The functionally important R234W mutation is located in the loop between helix α10 and β-strand β4 (Fig. 1). In wild-type CRALBP, R234 is part of a cluster of 3 pairs of positively charged residues (R103-K104, K236-R234, and K261-R265) at the protein surface. (Electrostatic surface potentials of wild-type CRALBP are shown in Fig. S2.) This cluster is not in direct contact with the retinoid ligand, but it binds a dianionic tartrate molecule that was present in the crystallization buffer (Fig. 2A). In the R234W mutant, the apolar indole ring is buried by van der Waals contacts between helix α4 of the N-terminal α domain and helix α7 of the C-terminal αβα domain, with contact to residues F99, A102, and L188, respectively. The conformation change at residue position 234 dislocates nearby residues R234, K104, and K261 over distances of 7.7, 5.4, and 3.4 Å, respectively, and disrupts the coordination of tartrate. The burial of W234 also induces a conformation change of the loop between α10 and β4 (residues 227–238) by an average 2.3-Å rms deviation. This conformation change triggers a series of side-chain flips including F198, F235, and I238, which are part of the retinal-binding pocket (Fig. 2B). As a consequence the side-chain Cδ methyl group of I238 is rotated into the retinal-binding cavity. This rotation is the only significant structural alteration in the retinal-binding pocket of the R234W mutant. As described later, it probably is the decisive factor for the functional disorder of this mutated CRALBP.

Fig. 2.

Comparison of wild-type and R234W CRALBP. (A) Stereoview of the critically charged cleft of CRALBP superimposed on R234W. Tartrate and interacting residues of the CRALBP basic cleft are depicted as gray sticks. Dashed black lines indicate possible hydrogen bonds and short contacts; distances are given in Å. Corresponding residues of the R234W basic cleft are depicted as salmon sticks. (B) Stereoview of the domino-like structural transition in CRALBP by the R234W mutation. Side chains of wild-type CRALBP (gray) that participate in a cascade of side-chain flips in R234W (salmon) are depicted as sticks, the protein backbones are shown as a worm model. The cavity surface of the wild-type CRALBP and the 11-cis-retinal chromophore are shown in orange.

The Retinoid-Binding Pocket.

11-cis-retinal is sequestered completely from bulk solvent in the binding pockets of both wild-type and R234W CRALBP (Fig. 3A). The pocket volumes were calculated using the program VOIDOO (33) and a rolling probe with a radius of 1.0 Å. The R234W volume is 5.989 × 10²Å³, which is 7.2% less than that of wild-ype CRALBP. The volume reduction can be attributed to the I238 conformation change that is triggered by the R234W mutation, as described earlier. The side-chain conformations of the other residues lining the pocket remain unaltered between mutant and wild-type CRALBP. The I238 Cδ methyl group improves the shape complementarity between the ligand and the binding pocket by protruding into a pit at the junction between the β-ionone ring and the polyene tail (Fig. 3A). The better fit between I238 and the ligand increases the local packing density and tightens the binding of 11-cis-retinal. Structural evidence for tighter packing is provided by the greater electron density of 11-cis-retinal. In the mutant CRALBP, the electron density map of all carbons from the β-ionone ring to the terminal aldehyde (C1–C15) is well defined (Fig. 3B). In wild-type CRALBP, the electron density map of the C11–C15 terminal tail is poorly defined, suggesting that this part of the ligand is slightly disordered in the less tightly packed ligand-binding pocket. The excellent resolution of the R234W structure and the more restricted conformation of the 11-cis-retinal ligand provide a detailed picture of the retinoid conformation and of the retinoid-to-protein interactions (Fig. 3B). The absolute conformation of the 11-cis-retinal is 6-s-trans, 11-cis, twisted 12-s-cis. The polyene chain is almost planar up to C12. The β-ionone ring is twisted by −7° around the C6–C7 bond relative to the plane of the polyene chain. The tail of the polyene chain from C12 to the α,β-unsaturated aldehyde and including the methyl group at C13 also is planar. The 2 planes defined by C7–C12 and C13–C15 of the polyene are twisted by 45° around the C12–C13 bond. 11-cis-retinal interacts with the side chains of the binding pocket as follows: The α,β unsaturated aldehyde is stacked between the phenyl ring of F161 and the sulfur of M223. The carbonyl oxygen of the aldehyde serves as the hydrogen bond acceptor for the phenol group of Y180 and the acidic oxygen of E202 (2.8 Å and 2.6 Å hydrogen bonding distances). The β-ionone ring and the polyene chain are fixed by van der Waals interactions with the apolar side chains V224, L227, I238, V254, and L258 and I163, W166, F173, L215, L220, F240, F247, and Y251, respectively. In the R234W mutant, the I238 side chain improves the shape complementarity by a particularly close van der Waals contact with the methyl group C16 of the β-ionone ring. Recent fluorescence titration of CRALBP with different ligands adds evidence for stronger ligand binding by the R234W mutant (23). R234W has a 2-fold stronger affinity (K_d ∼ 10 nM) for 11-cis-retinal than does wild-type CRALBP. Furthermore, the same study shows that wild-type and mutant CRALBP have a 2.5-fold higher affinity for 11-cis-retinoids than for 9-cis-retinoids (23). The structural basis for this 11-cis versus 9-cis stereoselectivity is the shape complementarity between the binding pocket and the 11-cis-retinal with a 45° out-of-plane twist between C12 and C13, which cannot be realized with the 9-cis isomer. For the same reasons, apo-CRALBP cannot combine with either 13-cis-retinal or all-trans-retinal. The 3-fold stronger affinity of CRALBP for 11-cis-retinal than for 11-cis-retinol (23, 34) may be caused by the stronger stacking interaction of F161 with pi orbitals of a carbonyl group than with sigma orbitals of a primary alcohol and by better hydrogen-bonding geometry between the E202 and Y180 donors and the planar sigma orbitals of the aldehyde than with the tetrahedral sigma orbitals of a primary alcohol. The specificity of the CRALBP retinoid-binding pocket reflects the need for stringent control of retinoids in metabolic processes, particularly the 11-cis-retinoids in the eye that are sensitive to light. Similarly, the closest structural homologue of CRALBP, the α-tocopherol transfer protein (α-TTP), selectively binds RRR-α-tocopherol out of 8 vitamers of the vitamin E group (26). A structural comparison of the 2 proteins reveals the similarity of the overall backbone architecture of the 2 binding pockets, with an average 1.9-Å rms deviation over 205 Cα atoms (Fig. 4). Furthermore, the side chains of both ligands adopt similarly bent shapes through van der Waals interactions with apolar side chains. However, the positions and orientations of the respective ligands within their binding pockets are grossly different. Our findings suggest that both proteins accomplish shape complementarity between cognate ligand and binding pocket by individual fine tuning of shape and size of a set of amino acid side chains.

Fig. 3.

Ligand binding pocket of CRALBP. (A) Cavity-overlay of wild-type CRALBP (gray) onto R234W (orange). The corresponding 11-cis-retinal ligands are shown as gray and orange sticks, respectively. (B) Close-up view of the ligand-binding pocket of R234W. Side chains (gray) that form van der Waals contacts with 11-cis-retinal (orange) or participate in hydrogen bonding in the ligand binding pocket are shown. The protein backbone is shown as a worm model. The electron density of the ligand is shown as a mesh at 1 σ in the 2F_o–F_c map. Dashed black lines indicate hydrogen bonds; distances are given in Å.

Fig. 4.

3-D overlay of the binding pockets of wild-type CRALBP and α-TTP. Ribbon diagrams of CRALBP (salmon) and of α-TTP (gray) are shown with a numbering scheme according to the CRALBP secondary structure. 11-cis-retinal (salmon) and RRR-α-tocopherol (gray) are depicted as sticks; oxygen atoms are shown in red.

Protective Role of CRALBP Against Premature Ligand Photoisomerization.

The photoisomerization rate of 11-cis-retinal in complex with CRALBP is only 4% of the rate observed in complex with rhodopsin, suggesting that a function of CRALBP is to protect the chromophore from premature photoisomerization (8). A difference between rhodopsin and CRALBP is that 11-cis-retinal is covalently bound as Schiff-base in rhodopsin but is bound non-covalently in CRALBP. Another difference is the unequal conformations of 11-cis-retinal in the binding pockets of CRALBP and rhodopsin (35, 36). (Chromophore conformations are depicted in Fig. S3.) The molecular basis of retinoid chemoprotection by CRALBP was not well understood because of the lack of structural data. The geometry of 11-cis-retinal, as seen in the x-ray structures of wild-type and mutant CRALBP, may offer the following “molecular” explanation of this difference. In ground-state bovine rhodopsin the 11-cis-retinal Schiff-base adopts a 6-s-cis, 11-cis, 12-s-trans conformation (37), and the polyene chain is helically distorted between C10 and C13. This distortion facilitates light induced π–π* transitions of the chromophore (38). Moreover, the cis–trans isomerization of the electronically excited state is favored in rhodopsin by the torsional moment provided by the intramolecular interactions between groups of the polyene chain—namely, steric repulsions between the hydrogen (C10) and the methyl group (C20) at C13 and dipole–dipole repulsion between the methyl groups (C18 and C20) at C5 and C13. This strain is released in the distorted all-trans conformation of bathorhodopsin that can be trapped within a few hundred femtoseconds following light absorption. In CRALBP, in contrast, the central region of 11-cis-retinal adopts a planar conformation that is less favorable for light-induced π–π* transitions (38). In addition, because of the 6-s-trans conformation (as opposed to 6-s-cis in rhodopsin) the dipole–dipole repulsion between methyl groups (C18 and C20) at C5 and C13 is relieved. In conclusion, the hydrophobic pocket of CRALBP imposes conformational strain upon the 11-cis-retinal that disfavors photoisomerization.

Nevertheless, illumination of 11-cis-retinal complexed to CRALBP or R234W results in a loss of absorbance at 425 nm and an increase at 380 nm, consistent with the production of the all-trans geometrical isomer of retinal that is released for lack of affinity for CRALBP (39). We have determined first-order rate constants for the photoisomerization of uncomplexed 11-cis-retinal in solution or bound to CRALBP and R234W using the method previously described (8) (see SI Methods and Materials for details). The measurements revealed a 5-fold reduction of R234W photosensitivity (191.5 M⁻¹ cm − ¹) relative to wild-type CRALBP (Fig. 5). Interestingly, the binding of 11-cis-retinal to wild-type CRALBP or R234W results in an equal decrease of the extinction coefficient from 25,000 to 15,400 M⁻¹ cm⁻¹. This finding indicates that the observed 5-fold reduction in photosensitivity of 11-cis-retinal bound to R234W must be attributed to the local increase in packing density through the I238 Cδ methyl group and not to a change in the extinction coefficient. Overall, the estimated free energy gain for the burial of a single tryptophan (−1.6 to −2.6 kcal/mol) (40) in R234W is relayed over a distance of 15 Å toward the central core where it results in increased packing through the I238 Cδ methyl group (−0.6 to −1.0 kcal/mol) (41).

Fig. 5.

Time courses of the photoisomerization of 11-cis-retinal in the presence of ethanol, wild-type CRALBP, and R234W. The amount present at t = 0 is a_0; the amount present after illumination is a. For a first-order decay process, a = a₀e^−kt. The first-order rate constants calculated from this experiment are k_ethanol = 0.859 × 10⁻³ s⁻¹; k_CRALBP = 0.197 × 10⁻³ s⁻¹; k_R234W = 0.035 × 10⁻³ s⁻¹.

Other mutations of CRALBP Associated with Retinitis Pigmentosa.

Maw et al. (3) first reported a case of autosomal recessive retinitis pigmentosa associated with a pathological RLBP1 gene defect, the 4763G-A nucleotide substitution. Known allelic variants of the gene encoding CRALBP now encompass missense mutations, frameshift mutations that result in truncations, and mutations affecting a canonical splice donor site that prevent translation of the protein.

Of the allelic variants identified thus far, 3 missense mutations (R151Q, R234W, and M226K) have been characterized in vitro with the recombinant proteins (23). Recombinant CRALBP R151Q was shown to be largely insoluble and to lack retinoid-binding capability; M226K was moderately soluble but also lacked retinoid-binding. As shown in Fig. 6, residue R151 is located at the end of beta-strand β1 and is connected by a network of hydrogen bonds (2.9 Å, 3.0 Å, and 3.0 Å) to N190 and T193 of helix α8 and to the carbonyl oxygen of residue G155, respectively. Helix α7 and the adjacent helix α10 represent a central building block of the CRAL-TRIO fold defining 1 wall of the 11-cis-retinal-binding cavity. Disruption of the connecting hydrogen bonds toward helix α8 in R151Q may destabilize the overall fold and most probably is the cause of the association of R151Q with retinitis punctata albescens (3, 4). A similar clinical phenotype has been reported for a patient who had the related R151W mutation of RLBP1 (42). M226 is located in the hydrophobic interface between helices α7 and α10 and thus affects the same area of the 11-cis-retinal binding cavity as R151Q. Substitution of an uncharged methionine residue by a positively charged lysine at position 226 is associated with the loss of retinoid-binding capability and retinitis punctata albescens (4). The M226K mutation may weaken van der Waals interactions by introducing a positive charge between helices α7 and α10. A phenotype of retinitis punctata albescens similar to M226K is found in association with R103W (4, 43). The R103W mutation maps to the critically charged cleft of wild-type CRALBP that harbors R234 as well (Fig. 2A). The CRALBP structural model indicates that both R103 and R234 are involved in the coordination of dianionic tartrate deriving from the crystallization buffer. The comparison with the R234W model has revealed burial of R234 and disruption of the tartrate binding site. It is likely that the R103W mutation may exert its pathologic effect through a similar mechanism. Saari et al. (44) have proposed that the CRALBP basic cleft is physiologically relevant, stimulating retinoid transfer through specific binding of acidic phospholipids, particularly the dianionic glycerophospholipid phosphatidic acid. Accordingly, the interaction of acidic phospholipids with the basic cleft has been suggested to serve CRALBP for membrane docking (26, 44, 45). Structural superimposition of CRALBP on human α-TTP and on phosphatidylinositol transfer protein (SEC14p) from yeast reveal that the basic cleft motif is highly conserved (26–30). The R103W mutation of CRALBP represents the equivalent of the cleft mutation R59W in α-TTP that causes ataxia with vitamin E deficiency (26). Furthermore, K236 and R265 of the CRALBP cleft map to the pathologic cleft mutations R192H and R221W in α-TTP (46). In addition, R265 is equivalent to K239A in yeast SEC14p that is reported to abolish phosphatidylinositol transfer activity completely in vitro (47). These findings suggest that the cleft motif in CRALBP, α-TTP, and SEC14p has an essential role in stimulating the transfer of cognate ligands by acidic phospholipids. It is tempting to speculate that the CRALBP basic cleft may be involved in routing retinoids for the cytoplasmic leaflet of the plasma membrane because of the relatively high content of acidic phospholipid headgroups of the leaflet (48).

Fig. 6.

Missense mutations in CRALBP associated with retinitis pigmentosa. The side chains of residues mutated in retinitis pigmentosa are shown as green sticks on a worm model of the protein backbone; the interacting amino acid residues are shown in gray. The 11-cis-retinal ligand is shown as an orange stick, and the cavity surface is shown in orange. Dashed black lines indicate hydrogen bonds; distances are given in Å.

Concluding Remarks.

Mutations in CRALBP are a possible cause of defective retinal metabolism that leads to the clinical phenotype of retinitis pigmentosa. The crystal structure of CRALBP provides insight into stereospecific binding and into the chemical protection of 11-cis-retinal in the human eye. The crystal structure of the pathologic R234W mutation of CRALBP reveals impaired 11-cis-retinal release through stabilization of the ligand complex and disruption of a conserved basic surface patch with putative loss of lipid stimulation of retinoid transfer. The results of our study suggest impaired 11-cis-retinal release may be a major cause of night blindness and retinal pathology in patients carrying the R234W missense mutation in the RLBP1 gene (1, 2). This study may help elucidate the issue of visual cycle regulation in the human eye.

Materials and Methods

Cloning.

Human RLBP1 cDNA was obtained from Deutsches Ressourcenzentrum für Genomforschung GmbH and was cloned into the NdeI and XhoI sites of the pET-28a vector (Stratagene). Point mutations were introduced by using the QuikChange kit (Stratagene). A detailed description of methods used is given in SI Materials and Methods.

Expression and Purification.

Escherichia coli BL21 (DE3) cells were transformed and cultured overnight with agitation at 37 °C in LB medium containing 30 μg/mL kanamycin. The cultures were grown at 20 °C to an OD₆₀₀ of 0.7 and then were induced with isopropyl-thiogalactopyranoside at a final concentration of 1 mM for 16 h. Cells were harvested by centrifugation at 5000 × g for 45 min and were resuspended in 250 mL of ice-cold lysis buffer (20 mM imidazole; 100 mM NaCl; 20 mM Tris-HCl, pH 7.4; 1% wt/vol Triton X-100). The cells were disrupted by ultrasonication for 20 min, and the lysate was centrifuged at 20,000 × g for 35 min to remove debris. Proteins were purified by nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography and subsequent size-exclusion chromatography. The CRALBP mutant R234W was expressed and purified in the same way as the wild-type protein. Selenomethionine-labeled R234W was prepared by the method of methionine biosynthesis inhibition (49) and purified in the same way as the wild type.

Illumination of 11-cis-Retinal.

11-cis-retinal and its complexes with CRALBP and R234W were illuminated according to the methods of Saari and Bredberg (8). The time-dependent photoisomerization of 11-cis-retinal in ethanol, in CRALBP, and in the mutant R234W was recorded by measuring the absorbance at 380 nm, which was used to calculate the amounts of 11-cis-retinal (ε = 24,935 M⁻¹ cm⁻¹ at 380 nm) and all-trans-retinal (ε = 42,880 M⁻¹ cm⁻¹ at 383 nm). The photosensitivities of 11-cis-retinal in presence of ethanol, CRALBP, and R234W were determined with the photosensitivity being defined as the product of the quantum yield and the molar extinction coefficient of the substrate.

Dynamic Light Scattering.

DLS analysis has been carried out on a DynaPro Plate reader (Wyatt Technology). The protein concentration was set at 1 mg/mL, and analyses were performed using Dynamics 5.25 software provided by the manufacturer.

Crystallization.

Hexagonal crystals of wild-type CRALBP in complex with 11-cis-retinal were obtained at 20 °C by the sitting-drop, vapor-diffusion method, in 1.0 M K/Na tartrate, 0.1 M Tris-HCl (pH 7.0), 0.2 M Li₂SO₄. The crystals belong to space group P6₅22 with a = 71.92 Å, b = 71.92 Å, c = 303.20 Å, α = 90°, β = 90°, γ = 120°. Monoclinic crystals of R234W in complex with 11-cis-retinal were obtained using the same method in a condition of 20% (wt/vol) PEG 3000, 0.1 M Hepes (pH 7.5), 0.2 M NaCl. The crystals belong to the space group C2 with a = 87.93 Å, b = 57.88 Å, c = 75.15 Å, α = 90°, β = 122.846°, γ = 90°.

Data Collection and Structure Solution.

Crystals were flash-cooled in a nitrogen stream at 110 K after 15% glycerol was added. Data for wild-type CRALBP were collected at 100 K at the beamline ID29 (European Synchrotron Radiation Facility, France). Data for the R234W mutant were collected at beamline X06SA (Swiss Light Source, Paul Scherrer Institute, Switzerland) at 100 K. The structure was solved by SAD using a single crystal of selenomethionine-labeled R234W diffracting to 1.7 Å (32). Data were integrated and scaled with the x-ray detector software for processing single-crystal monochromatic diffraction data (XDS) (50). Selenium positions were determined by heavy-atom substructure search (32, 51). The SAD data set was submitted to a software suite for improvement and objective interpretation of crystallographic electron density maps and automatic construction and refinement of macromolecular models (ARP/wARP) (52), and the primary model was refined using the phenix.refine program from the PHENIX program package (51). The final R234W model contained residues 57–306, 11-cis-retinal, and 374 water molecules. The structure of wild-type CRALBP subsequently was solved by molecular replacement using the atomic model of R234W and the program Phaser (53) and was refined to 3.0-Å resolution using the program phenix.refine. Coordinates and structure factors of both structures have been deposited in the RCSB Protein Data Bank with ID codes 3HX3 and 3HY5.