Duplication and Divergence of Zebrafish CRALBP GenesUncovers Novel Role for RPE- and Muller-CRALBP inCone Vision

Ross Collery,1 Sarah McLoughlin,1 Victor Vendrell,1 Jennifer Finnegan,1 John W. Crabb,2

John C. Saari,3 and Breandan N. Kennedy1

PURPOSE. During vertebrate phototransduction 11-cis-retinal isisomerized to all-trans-retinal. Light sensitivity is restored byrecombination of apo-opsin with 11-cis-retinal to regeneratevisual pigments. The conversion of all-trans retinal back to11-cis-retinal is known as the visual cycle. Within the retina,cellular retinal-binding protein (CRALBP) is abundantly ex-pressed in the retinal pigment epithelium (RPE) and Mullerglia. CRALBP expressed in the RPE is known to facilitate therate of the rod visual cycle. Recent evidence suggests a role forMuller glia in an alternate cone visual cycle. In this study, therole of RPE- and Muller-CRALBP in cone vision was character-ized.

METHODS. The CRALBP orthologues rlbp1a and rlbp1b wereidentified in zebrafish by bioinformatic methods. The spatialand developmental expression of rlbp1a and rlbp1b wasdetermined by in situ hybridization and immunohistochem-istry. Depletion of the expression of the correspondingCralbp a and Cralbp b proteins was achieved by microinjec-tion of antisense morpholinos. Visual function was analyzedin 5-day post fertilization (dpf) larvae using the optokineticresponse assay.

RESULTS. The zebrafish genome contains two CRALBP ohno-logues, rlbp1a and rlbp1b. These genes have functionally di-verged, exhibiting differential expression at 5 dpf in RPE andMuller glia, respectively. Depletion of CRALBP in the RPE orMuller glia results in abnormal cone visual behavior.

CONCLUSIONS. The results suggest that cone photoreceptorsincorporate 11-cis-retinoids derived from the rod and conevisual cycles into their visual pigments and that Muller-CRALBPparticipates in the cone visual cycle. (Invest Ophthalmol VisSci. 2008;49:3812–3820) DOI:10.1167/iovs.08-1957

The visual system transforms light impulses detected byphotoreceptors in the retina into images of the external

environment. Rod photoreceptors function during dim night-light conditions, whereas vision in daylight is mediated by conephotoreceptors. Cones can adapt to �10 log units of illumina-tion, and subtypes of cones have distinct spectral sensitivitiesenabling daylight color vision.1,2

In vertebrate photoreceptors, light is sensed by photopig-ments that consist of an opsin G-protein-coupled receptor andthe light-sensitive chromophore 11-cis-retinal (for review, seeRef. 3). Different opsin proteins are present in each rod andcone, resulting in different spectral sensitivities. In rods, lightphotoisomerizes 11-cis-retinal in the photopigment to all-trans-retinal. This process induces a conformational change in rho-dopsin, activating the phototransduction cascade. This cascadesends signals via the optic nerve to the visual cortex of thebrain, where impulses are interpreted as a visual image. Theprocess of regeneration of the rod photopigment is known asthe rod visual cycle and requires regeneration of 11-cis-retinalfrom all-trans-retinal (for reviews see Refs. 4, 5). The processinvolves chemical modification and shuttling of retinoid inter-mediates within rods and the surrounding retinal pigmentepithelium.

Cellular retinaldehyde-binding protein (CRALBP) is a cyto-plasmic protein, abundantly expressed in the retinal pigmentepithelium (RPE) and Muller glia of the retina and in the pinealgland.6–8 Structurally, human CRALBP is a �36-kDa mono-meric protein, proposed to adopt an “open” or “closed” con-formation, depending on whether it is carrying an endogenousligand.9 CRALBP interacts structurally and functionally with11-cis-retinol dehydrogenase (RDH5), an enzyme of the visualcycle in RPE.10 CRALBP is a member of the CRAL_TRIO familyof proteins that share a lipid-binding domain derived from theyeast Sec14 protein.11 Residues 120-313 comprise the ligand-binding pocket within which retinoid-interacting residues, in-cluding W165, Y179, F197, C198, M208, Q210, M222, V223,M225, and W244, have been identified.9,12–15 The C terminusof CRALBP binds to the PDZ-domains of ezrin-radixin-moesin(ERM)–binding phosphoprotein50/sodium hydrogen ex-changer regulatory factor-1 (EBP50/NHERF-1), which in turnbinds to ezrin and actin, proteins localized to the apical pro-cesses of the retinal pigment epithelium and Muller glialcells.16,17 The (N/D)TA(L/F) minimum binding motif at theCRALBP C terminus is found in multiple CRALBP orthologuesand is proposed to bind CRALBP to apical processes of RPEcells and apical microvilli of Muller cells.

CRALBP contains a high-affinity binding site for 11-cis-reti-nal or 11-cis-retinol, vitamin A metabolites uniquely associatedwith sensing light.18 RPE-CRALBP facilitates 11-cis-retinal re-generation during the rod visual cycle. The functional require-ment of CRALBP in retinal Muller cells and in the pineal glandis unknown. Mutations in the single human CRALBP gene,RLBP1, can lead to forms of blindness that reflect retinitispunctata albescens, a photoreceptor degeneration accompa-nied by subretinal, white-to-yellow punctate deposits and de-

From the 1UCD Conway Institute and UCD School of Biomolecularand Biomedical Sciences, University College Dublin, Dublin, Ireland;the 2Cole Eye Institute, Cleveland Clinic Foundation, Cleveland, Ohio;and the 3Department of Ophthalmology, University of Washington,Seattle, Washington.

Supported by research project Grant RP/2006/277 from the IrishHealth Research Board (BNK). Zpr-1 was provided by the ZebrafishInternational Resource Center (ZIRC) which is supported by NationalInstitutes of Health-National Center for Research Resources (NIH-NCRR) Grant P40 RR12546.

Submitted for publication March 1, 2008; revised April 18, 2008;accepted July 24, 2008.

Disclosure: R. Collery, None; S. McLoughlin, None; V. Ven-drell, None; J. Finnegan, None; J.W. Crabb, None; J.C. Saari, None;B.N. Kennedy, None

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be marked “advertise-ment” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Corresponding author: Breandan N. Kennedy, UCD Conway Insti-tute, UCD School of Biomolecular and Biomedical Sciences, UniversityCollege Dublin, Dublin 4, Ireland; brendan.kennedy@ucd.ie.

Investigative Ophthalmology & Visual Science, September 2008, Vol. 49, No. 93812 Copyright © Association for Research in Vision and Ophthalmology

layed rod and cone resensitization.19–22 These mutations cantighten or abolish CRALBP ligand binding.10,21 Rlbp1�/�

(knockout) mice have delayed rhodopsin regeneration anddark adaptation after illumination coupled with diminished11-cis-retinal production.23 However, photoreceptor degener-ation characteristic of missense mutations in human CRALBPare not recapitulated in the Rlbp1�/� mouse.

RPE-CRALBP is an established facilitator of 11-cis-retinalregeneration during the rod visual cycle, accelerating theisomerization of all-trans- to 11-cis-retinol.23–25 Recently, evi-dence of a novel pathway that regenerates cone photopig-ments, the cone visual cycle, has accumulated.26–30 This cycleappears to involve a novel biochemical pathway for regener-ating 11-cis-retinal involving enzymatic modification and shut-tling of intermediates within cones and surrounding Mullerglial cells.28 The expression of CRALBP in Muller cells and thebinding of CRALBP to 11-cis-retinoids suggest that Muller-CRALBP plays a key role in the cone visual cycle.

The function of Muller-CRALBP has not been selectivelyassessed. Although patients with missense CRALBP mutationsand CRALBP knockout mice display abnormal cone responses,the phenotype could arise from defective RPE- and/or Muller-CRALBP, as there is a single mammalian gene.19,23 Approachesto dissecting the role(s) of retinal CRALBP include generatingmodels with selective loss of RPE- or Muller-CRALBP. This losscan be achieved in mice by using conditional knockout ap-proaches, but is time-consuming and expensive and requiresthe maintenance of several transgenic lines. An alternative isthe zebrafish, which has abundant cones and is amenable togenetic manipulation.

In the current study, we demonstrated that zebrafish con-tain two CRALBP ohnologues and that these duplicated geneshave diverged such that zebrafish rlbp1a/Cralbp a is predom-inantly expressed in the RPE and zebrafish rlbp1b/Cralbp b ispredominantly expressed in Muller glia. Using antisense mor-pholino technology for selective knockdown of RPE-CRALBPor Muller-CRALBP, we demonstrated that depletion of eitherpool results in abnormal cone-mediated vision. This findingsuggests that RPE- and Muller-CRALBP function in the conevisual cycle.

MATERIALS AND METHODS

Animal Breeding and Maintenance

Zebrafish were treated in accordance with the ARVO Statement for theUse of Animals in Ophthalmic and Vision Research. They were main-tained and raised under standard conditions at 28.5°C on a 14-hourlight:10-hour dark cycle. Wild-type Tubingen zebrafish embryos wereobtained by natural spawning and raised in embryo medium.31 Larvaeused for in situ hybridization were raised in embryo medium supple-mented with 0.003% phenylthiourea (Sigma-Aldrich) to inhibit pigmen-tation. Larvae were staged according to their age in days post fertiliza-tion (dpf).

Sequence Analysis

Human CRALBP protein sequence (P12271) was used as a probe tosearch Ensembl, build version 7 (http://www.ensembl.org/32) for or-thologous sequences in multiple species using the BLAST algorithm.Two zebrafish ohnologues, rlbp1a and rlbp1b (NP_991253;NP_956999), were identified, along with orthologues from chimpan-zee (Pan troglodytes, XP_510580), cow (Bos taurus, NP_776876),mouse (Mus musculus, NP_065624), rat (Rattus norvegicus,NP_001099744), chicken (Gallus gallus, NP_001019865), tropicalfrog (Xenopus tropicalis, NP_001005455), African clawed toad (Xe-nopus laevis, AAH54209), pufferfish (two ohnologues; Fugu rubripes;SINFRUP00000129793, SINFRUP00000164753), and cavefish (twoohnologues; Tetraodon nigriviridis; CAG01050, CAF99866). Protein

alignments and phylogenetic tree-building were performed inCLUSTAL W (http://www.ebi.ac.uk/Tools/clustalw2/index.html/ pro-vided in the public domain by European Bioinformatics Institute,European Molecular Biology Laboratory, Heidelberg, Germany) withdefault settings. The results were annotated in image-analysis software(Illustrator, ver. 11; Adobe Systems, San Jose, CA).

Wholemount In Situ Hybridization

PCR primers amplified full-length cDNAs of rlbp1a and rlbp1b andintroduced 5� clamps and restriction sites to facilitate cloning (in italic)(rlbp1a-F1, cgcggatcc CTCTCACAGAACGTTGCATTG; rlbp1a-R1, cgc-gaattc GACAAAGAACAGGAATGCCTGG; and rlbp1b-F1, cgcggatcc TG-GAGAATCTGAGCACTATGGC; rlbp1b-R1, cgcgaattc CTGGAAAAC-CAATATGGGTTAAAACGACG). Total RNA was extracted from adultzebrafish eyes (TRIzol; Invitrogen, Carlsbad, CA), genomic DNA wasdigested (RQ1 DNase; Promega, Madison, WI), and cDNA was synthe-sized with an RT-PCR system (Thermoscript; Invitrogen), with randomhexamers used to initiate synthesis. rlbp1a and rlbp1b cDNAs werecloned into the expression vector pCS2P� and used as templates forsynthesis of antisense and sense digoxigenin-labeled riboprobes. Larvalzebrafish were fixed in 4% formaldehyde/PBS. In situ hybridization wasperformed essentially as previously described.33 Larvae were photo-graphed under 100% glycerol (StereoLumar V12 microscope; Carl ZeissMeditec, Inc., Dublin, CA). Photographs were oriented on computer(Photoshop ver. 5 software; Adobe Systems).

Western Blot Analysisand Immunohistochemistry

Adult zebrafish were dark-adapted before dissection of whole eyes,RPE, and neuroretina. Tissues were homogenized and boiled in 1�SDS-PAGE electrophoresis buffer (16 mM Tris [pH 6.8]), 8% glycerol,0.6% SDS, 270 mM �-mercaptoethanol, and 0.003% bromophenolblue). Samples were separated on a 15% SDS-polyacrylamide gel for 2hours before overnight transfer to a nitrocellulose membrane. CRALBPisoforms were immunodetected by using a polyclonal antibody tobovine CRALBP (UW55; kind gift of one of the authors; JCS) followedby HRP-conjugated goat anti-rabbit IgG (Sigma-Aldrich) and chemilu-minescence reagents (ECL; GE Healthcare, Piscataway, NJ).

Adult and larval zebrafish eyes were fixed with 4% formaldehydeand cryoprotected by a sucrose series before embedding in OCT(Sakura Fintek, Torrance, CA). Twelve-micrometer sections were cutand thaw-mounted onto charged slides (Superfrost Plus; Fisher Scien-tific, Pittsburgh, PA). The sections were rehydrated, blocked in 2%(vol/vol) normal goat serum, 1% bovine serum albumin, and 1% TritonX-100 in PBS before incubating overnight at 4°C with polyclonalanti-CRALBP antibody. Polyclonal blue opsin (kind gift of David Hyde,Department of Biological Sciences, University of Notre Dame, NotreDame, IN) and monoclonal zpr-134 antibodies probed for cone-specificmarkers. Slides were rinsed in PBS before incubating with Cy2 orCy3-conjugated goat anti-rabbit/mouse secondary antibody as appropri-ate (Jackson ImmunoResearch Europe Ltd., Newmarket, UK). Afterrinsing in PBS and counterstaining with 300 nM DAPI, slides weremounted (Vectashield; Vector Laboratories, Burlingame, CA). Sectionswere examined with a laser scanning confocal microscope (LSM 510Meta; Carl Zeiss Meditec, Inc.). Images were taken using with anoil-immersion lens (Plan-Apochromat 63�/1.4 Oil DIC objective lens;Carl Zeiss Meditec, Inc.) with a 12-bit pixel depth and a resolution of512 � 512 pixels.

Morpholino Knockdown

Morpholino oligonucleotides targeting ATG translational start sites ofrlbp1a and rlbp1b were designed by Gene Tools (Gene Tools LLC,Philomath, OR) from cDNA sequences (rlbp1a start blocking MO,TTCCACTAACAACCGCCATTGCTTC; rlbp1b start blocking MO,CGAAAACTTCCAGTAGCAGCCATAG). Morpholino oligonucleotideswere resuspended in nuclease-free water and injected into wild-type,

IOVS, September 2008, Vol. 49, No. 9 Duplication and Divergence of Zebrafish CRALBP Ohnologues 3813

one- to two-cell zebrafish embryos along with 0.01% phenol red tracerdye.

Optokinetic Response Assay

The optokinetic response assay was performed essentially as previ-ously described.35 This assay uses eye movement to calculate visualresponse to a moving stimulus. Briefly, 5 dpf larval zebrafish wereplaced in a Petri dish containing 9% methylcellulose to immobilize thelarvae, while allowing respiration to continue. A drum containingalternating black-and-white stripes (18° per stripe) was rotated at aspeed of 20 rpm. Larvae were recorded during 5 seconds of norotation, 25 seconds of clockwise rotation, 5 seconds of no rotation,and 25 seconds of anticlockwise rotation. The number of saccades perminute was quantified with a microscope equipped with a digital

camera (model SZX16 microscope with a model DP71 digital cameraand Cell F software; Olympus, Tokyo, Japan).

RT-PCR of the 5�-UTR

Total RNA was extracted from adult zebrafish eyes (TRIzol; Invitrogen)and resuspended in nuclease-free water. The concentration and purityof RNA were measured with a spectrophotometer (NanoDrop Tech-nologies, Wilmington, DE), and contaminating genomic DNA was re-moved using RQ1 DNase (Promega). RNA was stored at �80°C untilused. Reverse transcription was performed on 1 �g total RNA with anRT-PCR system (Thermoscript; Invitrogen) at 50°C, after priming withrandom hexamers. Synthesized cDNA was stored at �20°C until used.cDNA was used in standard PCR reactions with 1 �L cDNA per 25 �LPCR reaction, in standard PCR conditions, with extension times ad-

FIGURE 1. Zebrafish CRALBP orthologues. (A) A phylogenetic alignment of CRALBP protein orthologues was built (DS Gene ver. 1.5; Accelrys,Inc., San Diego, CA). The branch length is proportional to the amount of inferred evolutionary change. Duplication of the fish genome resultedin two zebrafish and tetraodon CRALBP ohnologues. (B) The exon–intron structures of rlbp1a (Bi) and rlbp1b (Bii) are shown in a genomiccontext. Exons are numbered 1 to 7 based on current Ensembl notation, and extra rlbp1a exons discovered during this study are numbered �3to �1. Blue: coding regions. (C) Alignment of human, mouse, and zebrafish CRALBP protein sequences. Numbers represent amino acid positions.Red: missense mutations associated with inherited blindness. Blue horizontal bar: ligand-binding pocket of human CRALBP; green: residuesinvolved in ligand binding. Bold red: residue M225, also a disease mutation. PDZ-domain target motifs at the C termini of orthologues are boxed.

3814 Collery et al. IOVS, September 2008, Vol. 49, No. 9

justed to 1 minute per kilobase of target amplicon. Primers weredesigned complementary to expressed sequence tags (ESTs) identifiedafter BLAST analysis of EST databases with full-length rlbp1a mRNA

sequence as the probe (EST: AGENCOURT_21412034, GenBank Acc:CN177714) (rlbp1a �ex(�4)-F1, GAGCTCTGTCATTCTGCGGGTC;rlbp1a ex(�3)-F1, GGCATGTTCCCAGAGCTCTGTCA; rlbp1a�ex(�2)-F1, GGAAACACCTCAACAGCAATG; rlbp1a �ex(�1)-F1,GAGGTCGCAGTACAAATGAGTGG; rlbp1a �ex(�1)-R1, GCT-TCAAATACTCAATGCAACG).

RESULTS

Two Cralbp Ohnologues in Zebrafish

Using the human CRALBP protein sequence as a probe, weidentified two CRALBP ohnologues in the zebrafish genome.The teleost genome has duplicated since its radiation fromother vertebrates, accounting for the presence of two CRALBPgenes in zebrafish (Danio rerio), cavefish (Tetraodon nigri-viridis), and pufferfish (Fugu rubripes)36 (Fig. 1A). rlbp1a andrlbp1b are located on zebrafish chromosomes 25 and 7, re-spectively (Fig. 1B). The encoded proteins, zebrafish Cralbp aand Cralbp b, have �64% and �56% protein identity, respec-tively, to human CRALBP, �81% protein identity to each other,and predicted sizes of �35.5 and 35.7 kDa, respectively (Fig.1C). Of the 10 residues specifically implicated in the ligand-binding pocket of human CRALBP, 8 are evolutionarily con-served in zebrafish. In addition, residues R150 and R233, asso-

FIGURE 2. Expression of rlbp1a and rlbp1b in the pineal during larval development. In situ hybridizationshows rlbp1a and rlbp1b expression in the pineal at 1 dpf (A, E) and increasing at 3 to 5 dpf (B, C, F, G).Expression of rlbp1b remains strong in the pineal at 7 dpf (H). (F, Inset) The pineal with the parapinealshowed expression to the left (blue bracket). Rlbp1a and rlbp1b did not show diurnal or circadianexpression (I–N). Dorsal images of 3 dpf larvae harvested in light (I, J, K) and dark (L, M, N) phases andprobed for rlbp1a and rlbp1b and aanat2 transcripts. Arrows: the pineal.

TABLE 1. Lengths of rlbp1a and rlbp1b Coding Exons and Introns

rlbp1a rlbp1b

Exon1 44 bp (12 bp) 43 bp (9 bp)2 129 bp 129 bp3 205 bp 205 bp4 179 bp 179 bp5 159 bp 159 bp6 111 bp 111 bp7 533 bp (129 bp) 624 bp (147 bp)

Intron1–2 356 bp 268 bp2–3 1256 bp 2267 bp3–4 2219 bp 6093 bp4–5 76 bp 768 bp5–6 886 bp 3157 bp6–7 4697 bp 1529 bp

The length of each exon is shown, and the length of the codingregions in partially coding exons is shown in parentheses.

IOVS, September 2008, Vol. 49, No. 9 Duplication and Divergence of Zebrafish CRALBP Ohnologues 3815

ciated with inherited forms of blindness, are conserved acrossall orthologues.19–21 Whereas Cralbp b contains the C-terminalDTAL sequence associated with binding to the PDZ-domains ofEBP50/NHERF-1,16,37 Cralbp a does not, suggesting alteredfunction or subcellular localization (Supplementary Fig. S1,online at http://www.iovs.org/cgi/content/full/49/9/3812/DC1).We note that human proteins named CRALBP-like 138 andCRALBP-like 2 (NP_001010852) have been reported; however,as these proteins have only �41% and �43% protein identity tohuman CRALBP, respectively, they were not considered trueCRALBP orthologues.

Exon–Intron Structure of the ZebrafishCRALBP Genes

The coding regions of rlbp1a and rlbp1b comprise sevenexons and six introns (Fig. 1 and Supplementary Figs. S2,S3). The size of the coding exons is highly conserved,although the intron sizes have changed considerably (Table1). Searches of ESTs reveal three putative noncoding exons(exon �3, �2, and �1) in the 5�UTR of rlbp1a but not ofrlbp1b. Alignments with genomic sequence demonstratethese exons to extend �12 kb upstream of the rlbp1atranslational start site with introns of 1375, 9116, and 1301

nt, respectively (Supplementary Fig. S4). In agreement withthe bioinformatic analyses, PCR using primers extendingfrom these noncoding exons to exon 1 demonstrate they areincorporated into transcripts in adult eye mRNA. It is un-clear whether these noncoding exons represent alternativepromoter start sites or alternative splice forms.

Expression of rlbp1a and rlbp1b in Larval Pinealand Eye

The spatial and temporal expression of rlbp1a/Cralbp a andrlbp1b/Cralbp b transcripts was determined by wholemount insitu hybridization. Both rlbp1a and rlbp1b were expressed in thedeveloping eye and pineal gland from 1 dpf and were stronglyexpressed in those organs from 3 to 5 dpf (Figs. 2, 3). At 7 dpf,rlbp1a was expressed in the eye, but was not detected in thepineal gland, whereas rlbp1b was expressed in both the eye andpineal. Approximately 50% of larvae examined showed parapinealexpression of rlbp1b. The parapineal gland is located ventrolat-eral to the pineal and is thought to regulate asymmetric braindevelopment.39 To determine whether expression of either ze-brafish CRALBP gene exhibited circadian or diurnal regulation,we performed wholemount in situ hybridization on embryoscollected during mid-light and mid-dark time points (Fig. 2). No

FIGURE 3. Differential expression of CRALBP isoforms in adult zebrafish retina. (A) Immunohistochem-istry with a pan-anti-CRALBP antibody labeled adult RPE and Muller glia. (B–D) Muller-CRALBP colocalizedwith GFP expressed under control of a Muller-specific GFAP promoter. (E) Western blot showing twoCRALBP isoforms present in the whole zebrafish eye, one exclusively expressed in the neural retina, andthe other in the RPE. Bovine and mouse eye samples showed the expected single band of CRALBP.

3816 Collery et al. IOVS, September 2008, Vol. 49, No. 9

obvious change in the pineal expression level of rlbp1a or rlbp1bwas observed during light and dark phases. Larvae probed foraanat2 showed the expected cyclic dark-phase upregulation and

light-phase downregulation.40 Thus, neither Cralbp a nor Cralbp bdisplayed evidence of circadian or diurnal expression profiles inthe pineal.

FIGURE 4. Differential expression of rlbp1a and rlbp1b in RPE and Muller glia. Wholemount in situ hybrid-ization showed rlbp1a and rlbp1b to be expressed in the eye at 5 dpf (A, D), although rlbp1a was predominantin the RPE and rlbp1b in the inner nuclear layer (B, C, E, F). rlbp1a and rlbp1b were expressed at the interfacebetween the retina and the olfactory placodes and rlbp1b in the ciliary epithelium (B, E). Cryosections probedwith pan-anti-CRALBP antibody showed extensive staining in the RPE and Muller glial cells in wild-type larvae(G). Morpholino knockdown of Cralbp a depleted CRALBP expression in the RPE (H), whereas knockdown ofCralbp b depleted CRALBP expression in Muller glia (I). Knockdown of Cralbp a (K, N) or Cralbp b (L, O) doesnot affect cone morphology or cone marker expression (blue opsin [short-single cones] or zpr-1 [doublecones]) compared with wild-type (J, M). RPE, retinal pigment epithelium; ONL, outer nuclear layer; INL, innernuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; OP, olfactory placode; CMZ, ciliary marginalzone; CE, ciliary epithelium. Scale bars: (A–I) 50 �m; (J–O) 20 �m.

IOVS, September 2008, Vol. 49, No. 9 Duplication and Divergence of Zebrafish CRALBP Ohnologues 3817

Expression of Zebrafish CRALBP Isoforms inMuller Glia and RPE

Analysis of CRALBP protein expression in zebrafish was per-formed with a polyclonal antibody raised against recombinantbovine CRALBP8 that does not distinguish between zebrafishCralbp a and Cralbp b isoforms (Fig. 3). Consistent with otherspecies, immunohistochemical staining of retinal sections con-firmed extensive expression of zebrafish Cralbp a/Cralbp b inthe RPE and Muller glia (Fig. 3A). Expression was strongesttoward the ganglion cell layer where the Muller end feet arelocated (Fig. 3A). Muller glial expression was verified by colo-calization with GFP expressed under the control of a promoterfor glial fibrillary acidic protein (GFAP), an established Mullermarker (Fig. 3B).41 Western blot analyses provided the firstindication that the RPE and Muller glia express unique CRALBPisoforms. The analysis revealed distinct bands of �33 to 35 kDain protein extracted from the entire retina. The slower mobilityisoform was specific to the RPE, and the faster mobility isoformspecific to the neuroretina (Fig. 3E).

As duplicated genes often diverge in function, we hypoth-esized that the zebrafish CRALBP ohnologues evolve distinctexpression profiles in the retina. In situ hybridizations withgene-specific probes indicated that at 5 dpf rlbp1a was pre-dominantly expressed in the RPE and rlbp1b in Muller glia(Figs. 4A–F). Retinal cryosections demonstrated robust expres-sion of rlbp1b in Muller cells (Fig. 4E), which had projectionsspanning the retina and nuclei in the inner nuclear layer.42

rlbp1a and rlbp1b were also expressed at the interface be-tween the retina and the olfactory placodes (Figs. 4B, 4E) andrlbp1b was expressed in the ciliary epithelium, as shown inother species.43,44

The results conflicted with the Western blot data as Cralbpb is predicted to have a slower mobility than Cralbp a. Manyproteins, including proteins found in the retina, do not segre-gate at the predicted size on Western blots, which may be dueto posttranslational modification. To confirm the in situ find-ings at the protein level, we microinjected antisense morpho-linos, for selective translational block of each zebrafish CRALBPohnologue, and analyzed Cralbp a/Cralbp b expression in cor-responding retinal sections using the pan-CRALBP antibody.Cralbp a knockdown resulted in a specific depletion of

CRALBP labeling in the RPE, but not in Muller glia, whereasCralbp b knockdown resulted in a specific depletion ofCRALBP labeling in Muller glia but not in the RPE (Figs. 4G–I).No change in gross cone morphology or expression of cone- orrod-specific markers was observed after specific knockdown ofeither Cralbp a or Cralbp b (Figs. 4J–O and data not shown).Thus, the zebrafish genome contains duplicated CRALBP genesthat, at 5 dpf, exhibit exclusive patterns of expression withinthe RPE (rlbp1a) or Muller glia (rlbp1b).

Muller- and RPE-CRALBP in Cone Vision

At 5 dpf, zebrafish vision is mediated by cone photoreceptors,and the rod photoreceptors, though present, do not contributeto visual function.45 The ability to deplete RPE-CRALBP orMuller-CRALBP selectively at 5 dpf enabled us to assess theircontribution to cone vision by using the optokinetic responseassay. In this assay, larval fish are immobilized in a viscousmedium that allows free rotation of the eyes, and a drum linedwith alternating black and white stripes rotates about thelarvae.35 A saccade, or change in eye angle greater than 20°C inresponse to this moving stimulus is easily observed, and thenumber of saccades as a function of time is used as a quanti-tative test for visual response. Wild-type and control morpho-lino-injected larvae responded with �25 to 30 saccades perminute of drum rotation (Fig. 5). Knockdown of RPE-CRALBPor Muller-CRALBP resulted in a statistically significant reduc-tion in saccade response, showing that zebrafish CRALBP ex-pressed in RPE and Muller glia are independently essential fornormal cone vision.

DISCUSSION

Although the complete functional profile of CRALBP is un-known, its physiological importance is underlined by geneticassociation with heritable forms of blindness.19–21 CRALBP is a

FIGURE 5. Knockdown of Cralbp a or Cralbp b caused impaired op-tokinetic response. Optokinetic response was measured in wild-typelarvae and larvae injected with control morpholino or cralbp a andcralbp b translation blocking morpholinos at 5 dpf. A Student’s t-testcalculated significance of differences in number of saccades. ***P�0.001 and **P � 0.01; n � 10 animals per group; error bars, SD. MO,morpholino.

FIGURE 6. Proposed model of CRALBP function in the rod and thecone visual cycles. RPE-CRALBP and Muller-CRALBP contribute reti-nol/al to the photoreceptors after retinoid regeneration. Pathwaysdepicted are as follows: (1) 11-cis-retinal is exported to rods from theRPE16,24; (2) 11-cis-retinol may be exported to cones from the Mullerglia; (3) 11-cis-retinol may be exported to cones from the RPE; (4)11-cis-retinal may be exported to cones from the RPE. Although 11-cis-retinal bound to CRALBP has been purified from the neuroretina, itsexact location is unknown and is not identified in this proposed model.CRALBP is shown bound to 11-cis-retinal and 11-cis-retinol in the RPE,though only 11-cis-retinal bound to CRALBP has been isolated from thistissue.22,54

3818 Collery et al. IOVS, September 2008, Vol. 49, No. 9

known component of the rod visual cycle regeneration of11-cis-retinal, where it functions as an acceptor of 11-cis-retinoland a substrate carrier for 11-cis-retinol dehydrogenase(RDH5).12,46 Insights into CRALBP function have been gath-ered from rod-dominant models.12,16,23 In the present study,using the cone-dominant zebrafish, we identified two CRALBPohnologues and characterize their expression and function.

Both zebrafish CRALBP genes were expressed in the sen-sory pineal at early developmental stages, but by 7 dpf, onlyrlbp1b continued to exhibit pineal expression. CRALBP ex-pression in the pineal was observed earlier than in the retina,consistent with the rapid organogenesis of the zebrafish pi-neal.40 Many phototransduction components, though expressedin mammalian and nonmammalian pineals, are probably evo-lutionary relics, as entrainment of circadian output from themammalian pineal is controlled by light-sensitive ganglion cellsin the retina.47 However, in zebrafish, signals from the eye arenot necessary for circadian entrainment, and the pineal con-tains conelike photoreceptors that are directly sensitive tolight.48,49 We found no evidence that either zebrafish CRALBPohnologue is under diurnal or circadian regulation in the pi-neal. However, it is plausible that pineal CRALBP(s) regulateentrainment of circadian rhythms to new light–dark cycles.Further studies are needed to resolve the potential role forCRALBP in olfaction. Our data indicate expression of zebrafishCRALBP genes in olfactory placodes, consistent with a recentreport implicating Pax6 as a regulator of CRALBP expression inthe developing mouse brain50 and the established role of Pax6in eye and olfactory system development.51,52

In the retina, recent work in cone-dominant models hashypothesized a cone visual cycle involving Muller-CRALBP.26,28,29 Characterization of the cone visual cycle iswarranted given that loss of cone vision results in debilitatingblindness.53 CRALBP knockout mice have abnormal cone phys-iology, but this effect cannot be attributed solely to either RPE-or Muller-CRALBP, as expression is eliminated in both loca-tions.23 The duplication and divergence of expression of ze-brafish CRALBP ohnologues has enabled us to analyze the roleof each ohnologue selectively. In our study, rlbp1b was pre-dominantly expressed in Muller glial cells and rlbp1a in theRPE. Knockdown of either RPE-CRALBP or Muller-CRALBPresulted in abnormal cone vision. Based on the known bio-chemistry of CRALBP, abnormal cone vision resulting fromzebrafish Cralbp a/Cralbp b depletion probably resulted fromimpaired retinoid metabolism, culminating in delayed regener-ation of 11-cis-retinal and diminished visual pigment assem-bly.23 In summary, our data (1) provide primary evidenceconfirming that Muller-CRALBP contributes to cone vision and(2) demonstrates a novel role for RPE-CRALBP in contributingto cone vision.

We propose a revised model of CRALBP function in the rodand cone visual cycles (Fig. 6). RPE-CRALBP facilitates provi-sion of the 11-cis-retinal required by rods for visual pigmentregeneration as previously described.16,24 Muller-CRALBP, andperhaps RPE-CRALBP, facilitate the provision of 11-cis-retinolto cones. Unlike rods, cones have an 11-cis-retinol dehydroge-nase activity that oxidizes 11-cis-retinol to 11-cis-retinal.26,55

Furthermore, cones, but not rods, can regenerate visual pig-ments from 11-cis-retinol or 11-cis-retinal.55 Thus, we speculatethat RPE-CRALBP facilitates the provision of 11-cis-retinal tocones. This suggests two cell sources and three pathways thatprovide 11-cis-retinoids to cones compared with one for rods.However, CRALBP purified from RPE has been found boundonly to 11-cis-retinal, whereas CRALBP purified from the neu-roretina is bound to 11-cis-retinal and 11-cis-retinol.54 Thus,further investigation and refinement of the model is warranted,as the model does not pinpoint a location in the neuroretinawhere CRALBP is bound to 11-cis-retinal.

A recent report provides evidence of the existence of apreviously unannotated, noncoding exon in the human RLBP1gene.56 The data suggest the presence of alternative transcrip-tion start sites in the human CRALBP gene, although it remainsto be determined whether in vivo these are uniquely or pref-erentially used during development or in specific cells. We alsoidentify previously unidentified, noncoding exons in the ze-brafish rlbp1a gene and confirm their presence in transcriptscontaining the rlbp1a coding sequence. Zebrafish represent anexcellent model system with which to characterize the regu-lation of CRALBP genes in vivo because of their amenability totransient transgenesis, rapid development, and transparency.In addition, the duplication and divergence of expression ofzebrafish CRALBP genes provides a serendipitous model todistinguish the transcriptional regulators of RPE- and Muller-specific expression.

Acknowledgments

The authors thank Stephan Neuhauss and Helia Schonthaler for helpfuldiscussions, Ann Cullen for assistance with confocal microscopy, andthe Conway Institute Biotechnical Services for animal husbandry.

References

1. Normann RA, Werblin FS. Control of retinal sensitivity. I. Light anddark adaptation of vertebrate rods and cones. J Gen Physiol.1974;63(1):37–61.

2. Ebrey T, Koutalos Y. Vertebrate photoreceptors. Prog Retin EyeRes. 2001;20(1):49–94.

3. Blomhoff R, Blomhoff HK. Overview of retinoid metabolism andfunction. Neurobiology. 2006;66(7):606–630.

4. Saari JC, Crabb JW. Genetic and proteomic analyses of the mousevisual cycle. In: Chalupa LM, Williams R, eds. Eye, Retina, andVisual System of the Mouse. Cambridge, MA: The MIT Press. Inpress.

5. Lamb TD, Pugh EN Jr. Dark adaptation and the retinoid cycle ofvision. Prog Retin Eye Res. 2004;23(3):307–380.

6. Saari JC, Futterman S, Bredberg L. Cellular retinol- and retinoicacid-binding proteins of bovine retina: purification and properties.J Biol Chem. 1978;253(18):6432–6436.

7. Futterman S, Saari JC. Occurrence of 11-cis-retinal-binding proteinrestricted to the retina. Invest Ophthalmol Vis Sci. 1977;16(8):768–771.

8. Bunt-Milam AH, Saari JC. Immunocytochemical localization of tworetinoid-binding proteins in vertebrate retina. J Cell Biol. 1983;97(3):703–712.

9. Wu Z, Hasan A, Liu T, Teller DC, Crabb JW. Identification ofCRALBP ligand interactions by photoaffinity labeling, hydrogen/deuterium exchange, and structural modeling. J Biol Chem. 2004;279(26):27357–27364.

10. Golovleva I, Bhattacharya S, Wu Z, et al. Disease-causing mutationsin the cellular retinaldehyde binding protein tighten and abolishligand interactions. J Biol Chem. 2003;278(14):12397–12402.

11. Panagabko C, Morley S, Hernandez M, et al. Ligand specificity inthe CRAL-TRIO protein family. Biochemistry. 2003;42(21):6467–6474.

12. Crabb JW, Nie Z, Chen Y, et al. Cellular retinaldehyde-bindingprotein ligand interactions: Gln-210 and Lys-221 are in the retinoidbinding pocket. J Biol Chem. 1998;273(33):20712–20720.

13. Liu T, Jenwitheesuk E, Teller DC, Samudrala R. Structural insightsinto the cellular retinaldehyde-binding protein (CRALBP). Pro-teins. 2005;61(2):412–422.

14. Wu Z, Yang Y, Shaw N, et al. Mapping the ligand binding pocketin the cellular retinaldehyde binding protein. J Biol Chem. 2003;278(14):12390–12396.

15. Wu Z, Bhattacharya SK, Jin Z, et al. CRALBP ligand and proteininteractions. Adv Exp Med Biol. 2006;572:477–483.

16. Nawrot M, West K, Huang J, et al. Cellular retinaldehyde-bindingprotein interacts with ERM-binding phosphoprotein 50 in retinalpigment epithelium. Invest Ophthalmol Vis Sci. 2004;45(2):393–401.

IOVS, September 2008, Vol. 49, No. 9 Duplication and Divergence of Zebrafish CRALBP Ohnologues 3819

17. Nawrot M, Liu T, Garwin GG, Crabb JW, Saari JC. Scaffold proteinsand the regeneration of visual pigments. Photochem Photobiol.2006;82(6):1482–1488.

18. Saari JC, Bredberg DL, Noy N. Control of substrate flow at a branchin the visual cycle. Biochemistry. 1994;33(10):3106–3112.

19. Burstedt MS, Sandgren O, Holmgren G, Forsman-Semb K. Bothniadystrophy caused by mutations in the cellular retinaldehyde-bind-ing protein gene (RLBP1) on chromosome 15q26. Invest Ophthal-mol Vis Sci. 1999;40(5):995–1000.

20. Morimura H, Berson EL, Dryja TP. Recessive mutations in theRLBP1 gene encoding cellular retinaldehyde-binding protein in aform of retinitis punctata albescens. Invest Ophthalmol Vis Sci.1999;40(5):1000–1004.

21. Maw MA, Kennedy B, Knight A, et al. Mutation of the geneencoding cellular retinaldehyde-binding protein in autosomal re-cessive retinitis pigmentosa. Nat Genet. 1997;17(2):198–200.

22. Saari JC, Crabb JW. Focus on molecules: cellular retinaldehyde-binding protein (CRALBP). Exp Eye Res. 2005;81(3):245–246.

23. Saari JC, Nawrot M, Kennedy BN, et al. Visual cycle impairment incellular retinaldehyde binding protein (CRALBP) knockout miceresults in delayed dark adaptation. Neuron. 2001;29(3):739–748.

24. Stecher H, Gelb MH, Saari JC, Palczewski K. Preferential release of11-cis-retinol from retinal pigment epithelial cells in the presenceof cellular retinaldehyde-binding protein. J Biol Chem. 1999;274(13):8577–8585.

25. Winston A, Rando RR. Regulation of isomerohydrolase activity inthe visual cycle. Biochemistry. 1998;37(7):2044–2050.

26. Mata NL, Radu RA, Clemmons RC, Travis GH. Isomerization andoxidation of vitamin a in cone-dominant retinas: a novel pathwayfor visual-pigment regeneration in daylight. Neuron. 2002;36(1):69–80.

27. Mata NL, Ruiz A, Radu RA, Bui TV, Travis GH. Chicken retinascontain a retinoid isomerase activity that catalyzes the direct con-version of all-trans-retinol to 11-cis-retinol. Biochemistry. 2005;44(35):11715–11721.

28. Trevino SG, Villazana-Espinoza ET, Muniz A, Tsin AT. Retinoidcycles in the cone-dominated chicken retina. J Exp Biol. 2005;208:4151–4157.

29. Muniz A, Villazana-Espinoza ET, Thackeray B, Tsin AT. 11-cis-Acyl-CoA:retinol O-acyltransferase activity in the primary culture ofchicken Muller cells. Biochemistry. 2006;45(40):12265–12273.

30. Villazana-Espinoza ET, Hatch AL, Tsin AT. Effect of light exposureon the accumulation and depletion of retinyl ester in the chickenretina. Exp Eye Res. 2006;83(4):871–876.

31. Westerfield M. Zebrafish Book. A Guide for the Laboratory Use ofZebrafish (Danio rerio). 5th ed. Eugene, OR: University of OregonPress; 2007.

32. Hubbard T, Barker D, Birney E, et al. The Ensembl genomedatabase project. Nucleic Acids Res. 2002;30:38–41.

33. Barthel LK, Raymond PA. Subcellular localization of alpha-tubulinand opsin mRNA in the goldfish retina using digoxigenin-labeledcRNA probes detected by alkaline phosphatase and HRP histo-chemistry. J Neurosci Methods. 1993;50(2):145–152.

34. Larison KD,. Bremiller R. Early onset of phenotype and cell pat-terning in the embryonic zebrafish retina. Development. 1990;109(3):567–576.

35. Brockerhoff SE, Hurley JB, Janssen-Bienhold U, Neuhauss SC,Driever W, Dowling JE. A behavioral screen for isolating zebrafishmutants with visual system defects. Proc Natl Acad Sci U S A.1995;92(23):10545–10549.

36. Amores A, Force A, Yan YL, et al. Postlethwait. Zebrafish hoxclusters and vertebrate genome evolution. Science. 1998;282(5394):1711–1714.

37. Nawrot M, Liu T, Garwin GG, Crabb JW, Saari JC. Scaffold proteinsand visual pigment regeneration. Photochem Photobiol. 2006;82(6):1482–1488.

38. Kong YH, Ye GM, Qu K, et a l. Cloning and characterization of anovel, human cellular retinaldehyde-binding protein CRALBP-like(CRALBPL) gene. Biotechnol Lett. 2006;28(17):1327–1333.

39. Gamse JT, Thisse C, Thisse B, Halpern ME. The parapineal medi-ates left-right asymmetry in the zebrafish diencephalon. Develop-ment. 2003;130(6):1059–1068.

40. Gothilf Y, Coon SL, Toyama R, Chitnis A, Namboodiri MA, KleinDC. Zebrafish serotonin N-acetyltransferase-2: marker for develop-ment of pineal photoreceptors and circadian clock function. En-docrinology. 1999;140(10):4895–4903.

41. Bernardos RL, Raymond PA. GFAP transgenic zebrafish. Gene ExprPatterns. 2006;6:1007–1013.

42. Raymond PA, Barthel LK, Bernardos RL, Perkowski JJ. Molecularcharacterization of retinal stem cells and their niches in adultzebrafish. BMC Dev Biol. 2006;6:36.

43. Eisenfeld AJ, Bunt-Milam AH, Saari JC. Immunocytochemical local-ization of retinoid-binding proteins in developing normal and RCSrats. Prog Clin Biol Res. 1985;190:231–240.

44. Martin-Alonso JM, Ghosh S, Hernando N, Crabb JW, Coca-PradosM. Differential expression of the cellular retinaldehyde-bindingprotein in bovine ciliary epithelium. Exp Eye Res. 1993;56(6):659–669.

45. Branchek T, Bremiller R. The development of photoreceptors inthe zebrafish, Brachydanio rerio. I. Structure. J Comp Neurol.1984;224(1):107–115.

46. Crabb JW, Carlson A, Chen Y, et al. Structural and functionalcharacterization of recombinant human cellular retinaldehyde-binding protein. Protein Sci. 1998;7(3):746–757.

47. Hattar S, Liao HW, Takao M, Berson DM, Yau KW. Melanopsin-containing retinal ganglion cells: architecture, projections, andintrinsic photosensitivity. Science. 2002;295(5557):1065–1070.

48. Kennedy BN, Li C, Ortego J, Coca-Prados M, Sarthy VP, Crabb JW.CRALBP transcriptional regulation in ciliary epithelial, retinal Mul-ler and retinal pigment epithelial cells. Exp Eye Res. 2003;76(2):257–260.

49. Allwardt BA, Dowling JE. The pineal gland in wild-type and twozebrafish mutants with retinal defects. J Neurocytol. 2001;30(6):493–501.

50. Holm PC, Mader MT, Haubst N, Wizenmann A, Sigvardsson M,Gotz M. Loss- and gain-of-function analyses reveal targets of Pax6 inthe developing mouse telencephalon. Mol Cell Neurosci. 2007;34:99–119.

51. Nomura T, Haba H, Osumi N. Role of a transcription factor Pax6 inthe developing vertebrate olfactory system. Dev Growth Differ.2007;49(9):683–690.

52. Collinson JM, Quinn JC, Hill RE, West JD. The roles of Pax6 in thecornea, retina, and olfactory epithelium of the developing mouseembryo. Dev Biol. 2003;255(2):303–312.

53. Hicks D, Sahel J. The implications of rod-dependent cone survivalfor basic and clinical research. Invest Ophthalmol Vis Sci. 1999;40(13):3071–3074.

54. Saari JC, Bredberg L, Garwin GG. Identification of the endogenousretinoids associated with three cellular retinoid-binding proteinsfrom bovine retina and retinal pigment epithelium. J Biol Chem.1982;257(22):13329–13333.

55. Jones GJ, Crouch RK, Wiggert B, Cornwall MC, Chader GJ. Reti-noid requirements for recovery of sensitivity after visual-pigmentbleaching in isolated photoreceptors. Proc Natl Acad Sci U S A.1989;86(23):9606–9610.

56. Vogel JS, Bullen EC, Teygong CL, Howard EW. Identification of theRLBP1 gene promoter. Invest Ophthalmol Vis Sci. 2007;48(8):3872–3977.

3820 Collery et al. IOVS, September 2008, Vol. 49, No. 9