Disease-causing Mutations in the Cellular Retinaldehyde Binding Protein Tighten and Abolish Ligand Interactions* 210

Mutations in the human cellular retinaldehyde binding protein (CRALBP) gene cause retinal pathology. To understand the molecular basis of impaired CRALBP function, we have characterized human recombinant CRALBP containing the disease causing mutations R233W or M225K. Protein structures were verified by amino acid analysis and mass spectrometry, retinoid binding properties were evaluated by UV-visible and fluorescence spectroscopy and substrate carrier functions were assayed for recombinant 11-cis-retinol dehydrogenase (rRDH5). The M225K mutant was less soluble than the R233W mutant and lacked retinoid binding capability and substrate carrier function. In contrast, the R233W mutant exhibited solubility comparable to wild type rCRALBP and bound stoichiometric amounts of 11-cis- and 9-cis-retinal with at least 2-fold higher affinity than wild type rCRALBP. Holo-R233W significantly decreased the apparent affinity of rRDH5 for 11-cis-retinoid relative to wild type rCRALBP. Analyses by heteronuclear single quantum correlation NMR demonstrated that the R233W protein exhibits a different conformation than wild type rCRALBP, including a different retinoid-binding pocket conformation. The R233W mutant also undergoes less extensive structural changes upon photoisomerization of bound ligand, suggesting a more constrained structure than that of the wild type protein. Overall, the results show that the M225K mutation abolishes and the R233W mutation tightens retinoid binding and both impair CRALBP function in the visual cycle as an 11-cis-retinol acceptor and as a substrate carrier.

Mutations in the human gene RLBP1 encoding the cellular retinaldehyde-binding protein (CRALBP) 1 cause retinal pathology and have been associated with autosomal recessive retinitis pigmentosa (1), Bothnia dystrophy (2,3), retinitis punctata albescens (4), fundus albipunctatus (5), and Newfoundland rod-cone dystrophy (6). These diseased phenotypes are all characterized by photoreceptor degeneration and night blindness (delayed dark adaptation) but differ in age of onset, rate of progression, and severity. The molecular basis for the clinical differences in these related retinal dystrophies is not well understood and no effective therapies exist for the pathology resulting from impaired CRALBP function.
CRALBP is an abundant, 36-kDa protein in the cytosol of the retinal pigment epithelium (RPE) and Mü ller cells of the retina where it carries endogenous 11-cis-retinol and 11-cis-retinal (7). The CRALBP ligand binding cavity is mapped in the accompanying report (8). In vivo studies (9) show that CRALBP serves as a major 11-cis-retinol acceptor in the isomerization step of the visual cycle (7,10,11), stimulating the enzymatic isomerization of all-trans-to 11-cis-retinol in the rod visual cycle. However, CRALBP appears to function within an RPE protein complex (12) and to serve multiple functions. In vitro, CRALBP facilitates the oxidation of 11-cis-retinol to 11-cisretinal by 11-cis-retinol dehydrogenase (12,13), retards 11-cisretinol esterification in the RPE by lecithin:retinol acyltransferase (13), and is required for hydrolysis of endogenous RPE 11-cis-retinyl ester (14).
Six CRALBP mutations have been linked with retinal pathology, including three missense mutations (R150Q, M225K, and R233W), a frameshift mutation and two predicted splice junction alterations (1,2,4,6). Recombinant CRALBP (rCRALBP) containing the R150Q mutation lacks the ability to bind 11-cis-retinal and exhibits low solubility (1). Toward a better understanding of the molecular basis of retinal pathology associated with RLBP1 gene defects, we report here characterization of rCRALBP containing the M225K or R233W disease-causing mutations (Fig. 1).
Mutagenesis and Production of rCRALBP Mutants M225K and R233W-Mutant rCRALBP cDNA carrying either the R233W or M225K substitutions were created using The QuikChange site-directed mutagenesis method (Stratagene). Briefly, WT human CRALBP cDNA in the pET19b vector (16) was cleaved with XbaI and HindIII and the coding region subcloned into pBlueSK. The following complimentary oligonucleotides were used to substitute a Lys for a Met at residue 225 (underlined) in mutant M225K: sense, 5Ј-GAAGATGGTGGACAAGCT-CCAGGATTCCTT-3Ј; antisense, 5Ј-AAGGAATCCTGGAGCTTGTCCA-CCATCTTC-3Ј. To substitute a Trp for an Arg at residue 233 (underlined) in the R233W mutant, complimentary oligonucleotides were also used: sense, 5Ј-ATTCCTTCCCAGCCTGGTTCAAAGCCATCC-3Ј; antisense, 5Ј-GGATGGCTTTGAACCAGGCTGGGAAGGAAT-3Ј. Each mutagenesis mix was transformed into Escherichia coli strain XL1-Blue (Stratagene), mutant clones identified by restriction mapping with NspI for M225K and MspI for R233W, were amplified, cleaved with XbaI and HindIII, and ligated back into expression vector pET19b (Novagen). Each insert was sequenced in both directions using the ABI PRISM Dye Terminator Cycle Sequencing kit and the model 377 DNA sequencer (PerkinElmer Life Sciences, Applied Biosystems). WT and rCRALBP mutants M225K and R233W were expressed in E. coli strain BL21(DE3)LysS with a N-terminal His tag and purified using nickelnitrilotriacetic acid-agarose columns (Qiagen) (16). Recombinant protein was quantified according to Bradford (17) using WT rCRALBP previously quantified by amino acid analysis for the standard reference protein.
Mass Spectrometry, Amino Acid Analysis, and Electrophoresis-The masses of the intact mutant proteins were determined by LC ESMS using a PerkinElmer Life Sciences Sciex API 3000 triple quadrupole electrospray mass spectrometer, a Vydac C4 column (1.0 ϫ 150 mm), an Applied Biosystems model 140D high-performance liquid chromatography system and aqueous acetonitrile/trifluoroacetic acid solvents at a flow rate of 50 l/min (18,19). Phenylthiocarbamyl amino acid analysis was performed with an Applied Biosystems model 420H/130/920 automated system and vapor phase HCl hydrolysis (20). Purified rCRALBP mutants M225K and R233W (ϳ100 pmol each) were digested overnight with trypsin, and the peptide digests were analyzed with a PE Biosystems Voyager DE Pro MALDI-TOF mass spectrometer using ␣-cyano-4-hydroxycinnamic acid as matrix (21,22). SDS-PAGE was performed  1. Location of the M225K and R233W mutations in the CRALBP retinoid binding pocket. Amino acid sequence surrounding residues Met-225 and Arg-233 in the CRALBP retinoid binding cavity are shown for human (h), bovine (b), and mouse (m) CRALBP (8,19,34). Species differences are shaded, and residues influencing CRALBP retinoid binding are shown in boldface.

FIG. 2. SDS-PAGE analysis of rCRALBP mutants M225K and R233W.
Crude bacterial lysates (ϳ15 g of total protein) and purified recombinant proteins (1-2 g of each) from cells expressing wild type rCRALBP, mutant M225K, or mutant R233W were analyzed by SDS-PAGE on a 12% gel and stained with Coomassie Blue. S, crude soluble lysate; RP, re-suspended pellet (inclusion body) fraction; P, purified protein; and MW, molecular weight markers. on 10% or 12% acrylamide gels using the Bio-Rad Mini-Protein II system (8).
Analysis of Retinoid Binding Function-Retinoid labeling of purified apo-rCRALBP with 11-cis-retinal or 9-cis-retinal, removal of excess retinoid, bleaching, and analysis by UV visible spectroscopy and fluorescence spectroscopy were performed in the dark, under dim red illumination as previously described (8,19).
Analysis of 11-cis-Retinol Dehydrogenase Activity-Human recombinant 11-cis-retinol dehydrogenase (rRDH5) was expressed in Hi-5 insect cells using a baculovirus vector kindly provided by Dr. K. Palczewski (12,23) and purified to apparent homogeneity by nickel affinity chromatography. rRDH5 oxidation activity was measured at pH 7.5 (8,24) and reduction activity was measured at pH 5.5 (25) using purified mutant or WT rCRALBP or equimolar amounts of free 11-cisretinol or 11-cis-retinal as substrate (8). Control assays with free retinoid as substrate were done in the absence of any carrier protein.
Solution State Heteronuclear Single Quantum Correlation NMR- 15 N uniformly labeled WT and mutant R233W rCRALBP were prepared by biosynthetic incorporation in E. coli strain BL21(DE3)LysS grown in defined minimal media (8,26). Purified mutant and WT rCRALBP with bound 11-cis-retinal (ϳ0.3 mM) were adjusted to 8% D 2 O (v/v) and transferred to 250 l of microcell NMR tubes (Shigemi Inc., Allison Park, PA) (8). All NMR experiments were performed at 25°C with a Varian INOVA 500-MHz spectrometer equipped with a triple resonance probe. Sensitivity enhanced two-dimensional 1 H-15 N heteronuclear single quantum correlation experiments were recorded using water-flipback for water suppression. Data was processed on a Sun UltraSPARC workstation using NMRPipe and Pipp software (8,27,28). Holo-protein preparations were maintained in the dark or under dim red illumination to prevent retinoid isomerization.

Expression and Structural Integrity of rCRALBP Mutants
M225K and R233W-WT and mutant rCRALBPs were produced in bacteria and SDS-PAGE of the crude soluble bacterial lysates and re-suspended pellet fractions showed that the M225K mutant was less soluble than the R233W rCRALBP mutant (Fig. 2). The R233W mutant was present in the soluble lysate fraction in amounts comparable to that of the WT protein (Fig. 2). The purified mutant proteins were characterized FIG. 3. Retinoid binding analysis of CRALBP mutants M225K and R233W. UV-visible absorption spectra are shown before and after exposure to bleaching illumination following retinoid labeling with either 11-cis-or 9-cis-retinal and removal of excess retinoid. With bound 11-cis-retinal, the ligand absorbance maxima for mutant R233W ( max ϭ 425.3 Ϯ 0.6 nm, n ϭ 3) and the WT protein are indistinguishable. With bound 9-cis-retinal, the ligand absorbance maxima for R233W ( max ϭ 408.3 Ϯ 1.2 nm, n ϭ 3) is slightly red-shifted compared with the human WT protein ( max ϭ 400 nm) (15). The absorption spectra from mutant M225K shows no chromophore absorbance near 425 or 400 nm indicating no bound retinoid; the ϳ380-nm absorbance is from free retinoid and/or nonspecific interaction with retinoid. by amino acid analysis (Table I) Fig. S1).
Retinoid Binding Properties of Mutants M225K and R233W-UV-visible spectral analysis of purified mutant R233W and WT rCRALBP with bound 11-cis-retinal were essentially identical, however, with bound 9-cis-retinal the ligand absorbance is slightly red-shifted for the R233W mutant (Fig.  3). Upon exposure to bleaching illumination, the chromophore absorbance for both holo-proteins shifts to ϳ380 nm due to the formation of unbound all-trans-retinal. In contrast, UV-visible spectra of purified mutant M225K exhibit no evidence for binding of either 11-cis or 9-cis-retinal (Fig. 3). Retinoid labeling performed in bacterial lysates prior to protein purification yielded the same UV-visible spectral results (data not shown).
Apparent equilibrium dissociation constants (K d ) for mutant R233W rCRALBP complexed with 11-cis-or 9-cis-retinal were determined by fluorescence titration of the apo-protein, monitoring the decrease in the intrinsic fluorescence of the protein upon ligand binding (Table II). The determined K d values reveal significantly greater affinity of the R233W mutant for both 11-cis-retinal (K d ϳ 10 nM) and 9-cis-retinal (K d ϳ 24 nM) relative to WT rCRALBP (K d ϳ 21 nM for 11-cis-retinal and ϳ53 nM for 9-cis-retinal). The average number of binding sites extracted from the mutant R233W titration data was about 0.6 for either retinoid.
Mutant Substrate Carrier Function for 11-cis-Retinol Dehydrogenase-When rRDH5 was assayed in the presence of the M225K rCRALBP mutant plus free retinoid, about 3-fold greater K m values were obtained relative to WT rCRALBP for both oxidation and reduction (Table III). When rRDH5 was assayed using R233W-bound retinoid as substrate, determined K m values were 4-to 7-fold higher than WT rCRALBP (Table  III). Relative to the free retinoid controls, neither mutation appeared to significantly effect V max (Table III). In vivo CRALBP is thought to facilitate the RDH5-catalyzed oxidation of 11-cis-retinol to 11-cis-retinal (13) and in vitro the V max for WT rCRALBP was ϳ20% greater in the oxidation reaction relative to free 11-cis-retinol and ϳ10 -16% greater than the mutants. No significant difference in V max was observed using WT rCRALBP or free 11-cis-retinal in the reduction reaction. The results support impaired substrate carrier function in the RDH5 oxidation reaction for the M225K and R233W rCRALBP mutants.
Conformational Differences between WT rCRALBP and Mutant R233W-The 1 H-15 N heteronuclear single quantum correlation (HSQC) NMR spectra for 15 N uniformly labeled holo-R233W rCRALBP and WT holo-rCRALBP were recorded in the dark and overlaid for comparison (Fig. 4). Resonances for Trp-165, Trp-244, Met-208, Met-222, and Met-225 in HSQC NMR spectra of WT rCRALBP were assigned in the accompanying report (8). Tentative assignments have also been made for the other four Met in WT rCRALBP (8) and randomly designated M a , M b , M c , and M d (Fig. 4). Mutant R233W has a total of three Trp, and Trp-233 was assigned in Fig. 4 based on the one additional resonance in the downfield chemical shift region characteristic of Trp side-chain NH. Conformational differences in holo-protein structures between WT rCRALBP and mutant R233W are demonstrated (Fig. 4) by the WT (blue) and R233W (red) resonances that do not superimpose, such as M d . Structural differences within the retinoid binding cavity are exemplified by Met-222, an apparent retinoid binding pocket component (8), which clearly does not align with any other signal in the R233W HSQC spectrum. In separate experiments, HSQC NMR spectra were recorded before and after exposure of 15 N uniformly labeled holo-R233W rCRALBP to bleaching illumination. The results (Fig. 5A) show that the majority of resonances remain unaffected by ligand isomerization. A few chemical shift changes are apparent in the R233W HSQC spectra upon light-induced retinoid isomerization, but the changes are much less extensive than those observed upon bleaching the WT protein (Fig. 5B).

DISCUSSION
Functionally impaired CRALBP was first associated with autosomal recessive retinitis pigmentosa (arRP) in 1997 when the missense mutation R150Q was found in the RLBP1 gene from a family in India (1). Retinitis pigmentosa is a family of inherited diseases with many forms and causative genes and classification of the disease types continue to evolve (29). Since 1997, five other recessive defects in the RLBP1 gene have been found to cause retinal dystrophies, including the two missense mutations M225K and R233W associated with retinitis punctata albescens and Bothnia dystrophy (2,5). Pathological mutations in the CRALBP gene have now been associated with other phenotypes and detected in pedigrees from Europe, the Middle East, Newfoundland, and India (1)(2)(3)(4)(5)(6). RLBP1 gene defects are thought to be a rare cause of retinal disease (4); however, in northern Sweden the high prevalence of Bothnia dystrophy caused by the R233W mutation constitutes a significant medical problem for which therapies are sought (2,3). To better understand the molecular basis of retinal pathology associated with impaired CRALBP and possibilities for therapeutic intervention, we have pursued structure-function studies of the mutant rCRALBPs containing the disease causing substitutions M225K and R233W.
The primary structural integrity of the purified M225K and R233W mutant recombinant proteins was confirmed by amino acid analysis and mass spectrometry. In contrast to the largely insoluble R150Q rCRALBP associated with arRP (1), the  R233W mutant exhibits solubility comparable to that of WT rCRALBP, whereas the M225K mutant is less soluble than WT rCRALBP but significantly more soluble than the R150Q mutant. With regard to retinoid binding properties, UV-visible spectral analysis revealed that the M225K mutant resembled the R150Q mutant and completely lacked the ability to bind cis-retinoids (1). In contrast, mutant R233W bound stoichiometric amounts of 11-cis-or 9-cis-retinal based on absorbance spectral ratios (19). Fluorescence titrations yielded apparent equilibrium disso-ciation constants for the R233W mutant that demonstrated nanomolar affinities for 9-cis-and 11-cis-retinal that were about 2-fold tighter than determined for WT rCRALBP. The variability of the retinoid affinity data (relative standard error of the mean was ϳ15-23%) was within the limits of experimental error of the titration methodology and due, in part, to the low aqueous solubility of retinoids and variable apo-protein stability. Furthermore, the protein concentration (0.5 M) used in the fluorescence titrations was significantly higher than the apparent K d values, and, under these conditions, K d values FIG. 4. Heteronuclear single quantum correlation NMR spectra of uniform 15 N-labeled WT and mutant R233W rCRALBP. HSQC NMR spectra for WT holo-rCRALBP (blue) and holo-R233W rCRALBP (red) were recorded separately in the dark then overlaid. These experiments correlate directly bonded 1 H-15 N pairs within the protein structures and show that substantial conformational differences exist between the WT and mutant R233W holo-proteins. Met-222 in the WT spectra does not overlay with any resonance in the R233W spectra, demonstrating that the proteins differ in retinoid binding pocket conformation. Trp-233 in the R233W spectra was assigned by an additional characteristic Trp resonance in the downfield chemical shift region relative to WT rCRALBP. Other residue assignments were determined elsewhere (8).
FIG. 5. Heteronuclear single quantum correlation NMR spectra of uniform 15 N-labeled mutant R233W and WT rCRALBP before and after bleaching. The 1 H-15 N correlation spectrum for the proteins with bound 11-cis-retinal was recorded in the dark (red), and the sample was then exposed to bleaching illumination and re-analyzed (blue) by the same 1 H-15 N correlation experiment. A, mutant R233W HSQC spectra. The vast majority of the resonances overlay in both experiments, indicating very little conformational change occurs upon ligand isomerization in the R233W ligand binding pocket. B, WT rCRALBP HSQC spectra. Although most residues remain unchanged upon bleaching (8), more chemical shift changes are apparent than in the R233W spectra.
should be considered to be upper limits for the actual values (30). The difference between the binding affinities of WT rCRALBP and its R233W mutant may therefore be larger than observed here.
Studies with crude extracts from bovine RPE microsomes (13), and purified proteins (12) strongly support a substrate carrier interaction between CRALBP and RDH5. The kinetic analyses performed here with purified recombinant proteins demonstrate that rCRALBP harboring either the M225K or the R233W mutations decrease the apparent affinity of rRDH5 for 11-cis-retinol and 11-cis-retinal relative to the WT protein. The K m of rRDH5 for 11-cis-retinoid in the presence of the M225K mutant approximated that for the enzyme reaction with free retinoid (12). An even higher K m was obtained for rRDH5 using the holo-R233W mutant as substrate, reflecting lower affinity between enzyme and retinoid due to the tighter binding of retinoid by R233W. The R233W mutant likely hinders movement of the hydrophobic substrate to the active site of enzyme. These data are consistent with the notion that CRALBP affects the activity of RDH5 by "channeling" of retinoids to the enzyme, much like cellular retinoic acid binding protein II facilitates delivery of retinoic acid to the retinoic acid receptor (31,32).
Clues to the structural basis for the tighter retinoid binding properties of the R233W mutant were obtained by two-dimensional NMR. HSQC NMR spectra showed that the three-dimensional structure of the R233W mutant with bound ligand was significantly different than that of WT holo-rCRALBP, including different retinoid-binding cavity conformations. Furthermore, conformational changes observed by NMR after photoisomerization of 11-cis-retinal in the ligand binding pocket were much more pronounced for the WT protein (8) than for the R233W mutant (Fig. 5). Residues Arg-233 and Met-225 are conserved within human, bovine, and mouse CRALBP (Fig. 1), and the present results are consistent with ligand interactions with both residues. Replacing positively charged Arg-233 with apolar tryptophan likely strengthens nonionic interactions within the hydrophobic ligand binding pocket, resulting in a more constrained, less flexible R233W protein structure. Likewise, inserting charged lysine in place of apolar Met-225 appears to disrupt critical interactions within the retinoid binding cavity, perhaps by opening the hydrophobic region to greater solvent accessibility, which in turn precludes specific ligand interaction and lowers M225K protein solubility.
The function of CRALBP in the visual cycle depends upon the rapid association and dissociation of retinoid from the ligand binding pocket. The results of this study implicate impairment of both retinoid binding and release as causes of the night blindness and retinal pathology reported for human patients with CRALBP mutations. CRALBP serves as the major 11-cis-retinol acceptor in the isomerization reaction of the rod visual cycle, therefore the lack of retinoid binding by M225K rCRALBP significantly slows the enzymatic conversion of alltrans to 11-cis-retinol, as observed for the CRALBP knockout mouse (9). Binding of 11-cis-retinol by apoCRALBP, coupled with its oxidation to 11-cis-retinal by RDH5, provides a strong driving force for isomerization (13,33). Interactions with other proteins likely facilitate the release of ligand from the CRALBP binding pocket, promoting catalytic rather than stoichiometric retinoid binding (9,12). The R233W mutation results in tighter rCRALBP retinoid binding and lower rRDH5 affinity for rCRALBP bound retinoid. The overly tight retinoid binding caused by the R233W mutation appears to hinder the rapid release of ligand, resulting in a "full house" effect that impairs 11-cis-retinol acceptor function and slows the isomerization of all-trans to 11-cis-retinol.