Isolation of a Zebrafish Rod Opsin Promoter to Generate a Transgenic Zebrafish Line Expressing Enhanced Green Fluorescent Protein in Rod Photoreceptors*

To exploit zebrafish as a transgenic model, tissue-specific promoters must be identified. We isolated a 20-kilobase (kbp) zebrafish rod opsin genomic clone, which consists of 18 kbp of 5′-flanking region, the entire coding region, and 0.5 kbp of 3′-flanking sequence. Polymerase chain reaction, Southern blotting, and DNA sequencing revealed the rod opsin gene lacks introns. The transcription start site was localized 94 nucleotides upstream of the translation initiation site. Sequence alignment with orthologous promoters revealed conserved cis-elements includingglass, NRE,OTX/Bat-1,Ret-1/PCE-1,Ret-4, and TATA box. A 1.2-kbp promoter fragment was cloned upstream of the enhanced green fluorescent protein (EGFP) cDNA and microinjected into 1- to 2-cell stage zebrafish embryos. EGFP expression was detected in the ventral-nasal eye at 3 days postfertilization and spread throughout the eye. Progeny of the positive founder fish, which were identified by polymerase chain reaction amplification of fin genomic DNA, exhibited EGFP expression in the retina, confirming the germline transmission of the transgene. Frozen eye sections demonstrated the EGFP expression was rod-specific and exhibited a similar developmental expression profile as the rod opsin protein. This stable transgenic line provides a novel tool for identification of genes regulating development and maintenance of rod photoreceptors.

The zebrafish, Danio rerio, has emerged as a novel vertebrate model system that is amenable to mutagenesis and transgenesis. High fecundity, rapid oviparous development, and a translucent embryo make zebrafish a prolific experimental model (1). Furthermore, the zebrafish eye possesses distinct advantages for studying the development, function, and inherited diseases of the retina. Eye ontogenesis proceeds rapidly, completing the laminae of the adult retina by 3 days postfertilization (dpf 1 ; Ref. 2). The zebrafish eye is relatively large and accessible, and the position and morphology of the rod and cone classes are readily distinguishable (3). Finally, the integrity of visual system structure and function can be evaluated by morphological, behavioral, and electrophysiological methods (4 -6). Chemical mutagenesis screens previously identified over 65 zebrafish visual system mutants exhibiting either behavioral (4,6,7) or morphological abnormalities (5). Insertional mutagenesis screens generated additional eye mutants (8,9). Overall, Ϸ35 mutants possess visual system-specific defects (4 -7). Transposition, random insertion, and retroviral integration provide the means for introducing transgenes into the zebrafish genome (10 -12). Accurate tissue-specific and developmental expression has been recapitulated in nonretinal tissue using endogenous zebrafish promoters (13,14).
Rhodopsin, comprising the rod opsin protein bound to 11-cisretinaldehyde, constitutes the photosensitive visual pigment of vertebrate rod photoreceptors (15). Rhodopsin is concentrated in the outer segments of the rod photoreceptor cells, which mediate vision in dim light. Mutations in the human rod opsin gene (RHO) can result in retinitis pigmentosa, a slow retinal degeneration resulting in blindness (16). Drosophila melanogaster, Mus musculus, and Xenopus laevis have served as genetic models for studying rhodopsin function and generating models of inherited human visual disease (17)(18)(19). However, none of these models combines the vertebrate eye with features of amenability to mutagenesis screens and transgenesis, as found in zebrafish. The developmental expression of zebrafish rod opsin initiates in two phases. It is expressed initially in a ventral-nasal patch of cells, and subsequently rod opsin-expressing cells arise in a scattered pattern throughout the remainder of the retina (20,21). Furthermore, rod photoreceptors are added to the fish retina throughout life from two retinal sources: stem cells in the circumferential germinal zone and rod precursor cells, which are found in the outer nuclear layer (22)(23)(24).
To develop tools for transgene expression in the zebrafish retina, we cloned the rod opsin gene. Characterization of the gene revealed that it lacks introns and contains conserved cis-regulatory sequences. We identified a 1.2-kbp promoter fragment that directs enhanced green fluorescent protein (EGFP) expression specifically to rods. Furthermore, this transgene was integrated stably into the germline and transmitted to the subsequent generation. The expression of transgenes in zebrafish photoreceptors will enhance our ability to investigate visual gene function, gene regulation, and gene therapy.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes, T4 polynucleotide kinase, and Taq DNA polymerase were purchased from Fisher. Hybond-Nϩ nylon mem-branes were purchased from Amersham Pharmacia Biotech. The Ultraspec RNA II isolation system and Hybrisol II were obtained from Cinna Biotecx and Intergen, respectively. The zebrafish genomic library was purchased from Stratagene. The Thermoscript reverse transcriptase system and ThermoSequenase Radiolabeled Terminator cycle sequencing kit were obtained from Life Technologies, Inc. and U. S. Biochemical Corp., respectively. The pEGFP-1 plasmid and microcapillaries were purchased from CLONTECH and World Precision Instruments, Inc., respectively.
Northern Blotting-Total RNA was isolated using the Ultraspec RNA II isolation system. RNA aliquots (Ϸ18 g) were electrophoresed in 1% agarose gels containing 6.6% formaldehyde, blotted to Hybond-N nylon membranes in 20ϫ SSC, and UV cross-linked. Membranes were hybridized with a probe in Hybrisol II containing 35% (v/v) formamide at 55°C. Final high stringency washing conditions were 0.08ϫ SSC, 0.1% SDS at 65°C for 30 min twice.
Genomic Library Screens-Approximately 10 7 recombinant phage in a Fix II zebrafish genomic library were screened with a radiolabeled zebrafish rod opsin cDNA (26). Escherichia coli strain XL1-MRAP2 was infected with titrated phage and plated on 150-mm NZY Petri dishes. Plaques were lifted and UV-crosslinked onto Hybond-Nϩ nylon membranes (27). Hybridization conditions were as described under "Southern Blotting." Polymerase Chain Reaction (PCR) Mapping-Genomic, phage, and plasmid DNA templates were PCR-amplified using primers spanning the zebrafish rod opsin cDNA. PCR parameters were 30 cycles of denaturation at 94°C for 30 s, anneal primers at 65-67°C for 30 s, and extension at 72°C for 5 min using Taq DNA polymerase.
Primer Extension-A primer complementary to the 5Ј end of the zebrafish rod opsin cDNA (RR1) was end-labeled with 32 P using T4 polynucleotide kinase. 12 g of zebrafish eye or body total RNA were hybridized with the labeled primer for 1 h at either 60 or 42°C (in 50% formamide). The annealed primer was extended using the Thermoscript reverse transcriptase system supplemented with 50 ng/l actinomycin D. Products were extracted with phenol/chloroform/isoamyl alcohol, ethanol-precipitated, and electrophoresed on a 6% polyacrylamide-urea sequencing gel. The ThermoSequenase Radiolabeled Terminator cycle sequencing kit was used for DNA sequencing.
Embryo Microinjections-A 1.2-kbp (EcoRI/XbaI) promoter fragment of the zebrafish rod opsin gene was subcloned upstream of the EGFPcoding sequence in pEGFP-1, creating pZOP-EGFP. This vector was linearized and resuspended in water plus India ink at 50 ng/l. Zebrafish embryos were microinjected with linearized vector at the 1-to 2-cell developmental stage (28). Injection needles were pulled from 1-mm (inner diameter) microcapillaries on a Sutter Instrument Co. model P-97 Flaming/Brown micropipette puller. Embryos positioned on an agarose injection chamber (28) were injected using air pressure on a Narishige U. S. A., Inc. micromanipulator. Some fish were reared in water treated with 0.003% 1-phenyl-2-thiourea to inhibit melanin formation (28).
Screening for Transgenic Fish-Fish were anesthetized and placed on a depression slide for fluorescence microscopy using a Carl Zeiss, Inc. Axiovert 100 microscope and cooled charge-couple deviced camera under a ϫ10 objective. Fish were viewed using a Carl Zeiss, Inc. narrow bandpass fluorescein isothiocyanate filter set and compared with a rhodamine filter set to differentiate between a true EGFP signal and autofluorescence. Fish exhibiting EGFP expression were isolated and pooled for long term characterization. Caudal and anal fins were clipped from adult founders, and genomic DNA was isolated (28). These samples were screened for the presence of the EGFP cassette using the EGFP-F (5Ј-GCCACAAGTTCAGCGTGTCC) and EGFP-R (5Ј-GATGC-CCTTCAGCTCGATGC) primers and the PCR parameters described above.
Whole-mount and Retina Section Antibody Labeling-Larval and juvenile transgenic G 1 fish were fixed in 4% paraformaldehyde, 5% sucrose, PBS pH 7.4. Antibody labeling of whole-mount and frozen sections (10 m) was performed as described using polyclonal antirhodopsin diluted 1:5,000 (29).

The Zebrafish Genome Contains a Single Rod Opsin
Gene-We examined the potential for duplicate rod opsin genes in the zebrafish genome by Southern blots. The zebrafish rod opsin cDNA typically detects single genomic fragments, although digestion with either ApaI or EcoRI yielded two hybridizing fragments (Fig. 1). The presence of two ApaI fragments is consistent with an internal ApaI site in the cDNA probe. However, because no internal EcoRI sites are known, the two EcoRI fragments are likely due to a sequence polymorphism. Overall, the pattern indicates the zebrafish rod opsin gene is likely single copy.
Isolation of a Zebrafish Rod Opsin Genomic Clone-A 20-kbp rod opsin genomic clone was isolated from a zebrafish library. The phage clone was mapped by Southern blot analysis using regions of the rod opsin cDNA as probes (data not shown). The data suggested that the zebrafish rod opsin gene contained minimal, if any, intron sequence. A 3-kbp (EcoRI-NotI) fragment detected by the cDNA probe was subcloned into the pCR2.1 plasmid. We examined this genomic subclone for the presence of introns by comparing the PCR amplification products from genomic DNA, the genomic clone, and the cDNA clone using primers spanning the zebrafish rod opsin cDNA (Fig. 2). The genomic and cDNA templates consistently produced PCR products of the same size (Fig. 2), which further demonstrates that the zebrafish rod opsin gene lacks introns.
Localization of the Transcription Start Site for Zebrafish Rod Opsin-We examined the size of the rod opsin mRNA to approximate the length of the 5Ј-untranslated sequence. The rod opsin cDNA clone that we previously characterized was 1,576 nucleotides with 92 nucleotides of 5Ј-untranslated sequence (26). We probed zebrafish eye and body total RNA with the rod opsin cDNA clone on Northern blots. A 1.9-kb transcript was detected in the eye RNA but not in the body RNA (Fig. 3A). The size and tissue specificity are consistent with other rod opsin orthologues (15). The transcription start site of the zebrafish rod opsin gene was determined by primer extension. Consistent with the Northern results, primer extension products were observed with eye RNA templates but not with body RNA (Fig.  3B). DNA sequencing of the 3-kbp genomic subclone with the identical primer used for primer extension localized the transcriptional start sites to the cytosine and guanine nucleotides 94 and 91 base pairs (bp) upstream of the translation start site, respectively. The presence of more than one primer extension product may be because of differential methylation at the 5Ј end of the transcript (30).
A Zebrafish Rod Opsin Promoter Directs EGFP Expression to Photoreceptors-A 1.2-kbp fragment containing the putative zebrafish rod opsin promoter, the transcriptional start site, and 52 bp of 5Ј-untranslated sequence were cloned upstream of the EGFP cassette to generate pZOP-EGFP. Zebrafish embryos at the 1-to 2-cell stage were microinjected with linearized pZOP-EGFP. Approximately 50% of injected embryos continued developing, with 90% of them hatching and 65% surviving beyond 15 dpf. An average of 6% of the injected embryos expressed the EGFP transgene at variable levels, although expression was always restricted to the eye (Fig. 6, A and B). EGFP expression was detected first at 3 dpf in the ventral-nasal eye and retinal sections localized the EGFP expression to the photoreceptor layer (data not shown). The localization and morphology of the EGFP-expressing cells was consistent with rod photoreceptorspecific expression.
Germline Transmission of the ZOP-EGFP Transgene-24 fish that were EGFP-positive by fluorescence microscopy survived to adulthood. The EGFP cassette was PCR-amplified from fin genomic DNA in seven of these fish (data not shown). Based on fluorescence microscopy, one of these fish exhibited germline transmission of the transgene to Ϸ6% of its progeny. These EGFP-positive G 1 progeny were isolated, and their EGFP expression patterns were examined. At 4 dpf, the EGFP expression was limited to the ventral-nasal portion of the ret-ina, although rhodopsin protein expression was detected throughout the retina (Fig. 6C). Careful inspection of the EGFP-expressing cells in the whole-mount-labeled retinas confirmed rhodopsin expression in the outer segments of the EGFP-positive cells (Fig. 6D). At 17 and 54 dpf, EGFP expression in G 1 fish spanned the photoreceptor layer of the retina, but with an apparent dorsal-ventral gradient (Fig. 6, E and F,  respectively). Close examination of the EGFP-positive cell mor- phology in the 54-dpf retina revealed the characteristic rod photoreceptor structure (Fig. 6G). Furthermore, labeling frozen sections with rhodopsin antiserum demonstrated that the retinal distribution of EGFP was restricted to rod outer segments and rod nuclei (Fig. 6H). DISCUSSION The advent of zebrafish as a prolific model system for vertebrate genetics directed us to isolate and characterize the zebrafish rod opsin gene. As previously reported for other teleosts, the zebrafish gene lacks introns, contrasting with the typical gene structure of five exons and four introns found in nearly all other vertebrate opsin genes (42). Our localization of the transcription start site predicts a 1,660-bp nonpolyadenylated mRNA, which is consistent with the 1.9-kilobase polyadenylated message identified by Northern blot. In zebrafish, genome duplication has resulted in multiple alleles of some zebrafish genes (43). Occasionally, this creates distinct regulatory profiles for the alleles that additively account for the expression pattern of the gene (43). Also, gene duplication can mask recessive and dominant mutations generated in mutagenesis screens. For zebrafish rod opsin, our Southern data are consistent with a single-copy gene or copies that are highly conserved.
We identified evolutionary conserved sequences in the 5Јflanking region of the rod opsin gene that may control rod opsin expression in zebrafish. The conserved cis-elements (glass, NRE, Ret-1/PCE-1, and Ret-4) include sites for transcription factors (Crx, Erx, Glass, Mash-1, Nrl, and Rx) that regulate retina/photoreceptor/rod-specific expression and that segregate with visual defects (32, 34, 37-41, 44 -47). Rod opsin expression is regulated in a combinatorial fashion that depends on the synergistic activities of several transcription factors (34,39). The presence of these regulatory elements in the ZOP-EGFP transgene likely helped to ensure its robust level of rod-specific expression. Although rod opsin gene expression may be regulated in a complex fashion, its promoter is excellent for expressing transgenes in the rod photoreceptors. Rod opsin promoters have successfully directed gene expression to rod photoreceptors in both mouse and Xenopus (48 -51). Also, the promoter in its native state drives a high level of expression during the terminal differentiation of rod cells and maintains that expression level through the life of the cell.
In this study, we identified a 1.2-kbp zebrafish promoter that directs EGFP expression to rod photoreceptors. The expression pattern of the transgene appears to recapitulate the tissuespecific and developmental progression of the zebrafish rod opsin gene (20,21,52). At 3 dpf, EGFP-positive cells are clustered in the ventral-nasal retina and begin to appear singly in other retinal regions at 6 -7 dpf (data not shown). However, a more critical analysis at the cellular level reveals two subtle divergences from endogenous rod opsin expression. First, the temporal expression of ZOP-EGFP is delayed slightly relative to rod opsin. Rhodopsin expression is observed first in the ventral retina, nasal to the choroid fissure. Although differentiating rods accumulate rapidly and to a high density in this ventral patch, rods are added at a slower rate and in a random, isolated fashion in other retinal regions (20). In contrast, the EGFP expression pattern at 4 dpf appears like the pattern of rhodopsin expression seen in developing fish at 55-60 h postfertilization (20). Thus, rods appear to differentiate and express rhodopsin prior to expressing EGFP. This result may suggest that an additional factor necessary for the initiation of rod opsin expression during early differentiation acts beyond our 1.2-kbp cloned promoter fragment. Second, in adults, not all rod photoreceptors express EGFP, and the expression may occur in a dorsal-ventral gradient. A gradient of transgene expression was reported previously with a bovine rod opsin promoter directing transgene expression in mice (51). These temporal and spatial divergences from rhodopsin expression may be because of restricted expression of the transgene to a subpopulation of rods such as the rod precursor cells, position effects related to the integration site of the transgene, or perhaps cytotoxicity by the EGFP (53). Further examination of the transgenic retinas may shed light on native regulation of rod opsin gene expression. Although the expression of EGFP is delayed, the persistent rod-specific expression makes this a useful promoter for expressing transgenes in rod cells during both development and adulthood.
EGFP was transiently expressed in Ϸ6% of injected embryos, consistent with percentages previously described (13,14). The frequency of injected embryos exhibiting germline transmission was 0.25-1.75%. However, the frequency of transmission may have been underestimated because of gene silencing and technical difficulties in detection of EGFP in the eye. In addition to screening for transgene expression by fluorescence microscopy, adult fish were screened secondarily for stable transgene integration by PCR amplification of fin genomic DNA. Adults that typed positive were outcrossed, and their progeny were screened by fluorescence microscopy to verify transmission of the transgene in the germline. As the fish age, microscopic observations of the retina become obscured by the envelope of melanophores and iridiphores around the eyes. Treatment of fish with 1-phenyl-2-thiourea inhibits the synthesis of melanin, thus permitting observation of EGFP through the translucent eye at 3 dpf (the onset of rhodopsin expression). However, the 1-phenyl-2-thiourea does not inhibit the production of the gold-or silver-reflecting components of the iridiphores so that at 6 -8 dpf, EGFP fluorescence is often observed only as it exits through the lens. Thus, proper orientation of the fish, as previously reported for Xenopus (54), is critical to ensure the accurate detection of EGFP expression.
The identification of a functional zebrafish rod opsin promoter will provide several significant applications. The promoter will be used to drive the expression of putative dominantly active transgenes to generate models of human retinal diseases. Also, as candidate genes are identified for zebrafish retinal mutants, the expression of wild-type genes in the mutant will be a powerful method to confirm gene identification. This type of "gene rescue" has been widely exploited in Drosophila. In addition, transgenic lines that express markers such as EGFP will be useful tools in mutagenic screens, in which subtle phenotypes may be identified more easily by the loss of a readily observable marker like the EGFP. Because the cis-elements in rod opsin promoters are conserved between zebrafish and mammals, transgenic zebrafish may prove useful in characterizing the developmental regulation of rod opsin and other photoreceptor genes. Transgenic zebrafish also may prove useful in elucidating the current controversy over the nature of circadian photoreception. The controversy stems from whether the circadian photoresponse is initiated in the eye, pineal, or other tissues and whether the photoreceptor is opsinbased, a cryptochrome, or an unknown photoreceptor (55,56). Apparent system redundancy and species differences complicate the issue (57,58). In zebrafish, six opsin genes expressed in four retinal photoreceptor classes and six cryptochrome genes expressed in the body, brain, and eye have been characterized (26,57). Mutagenesis screens to identify circadian mutants and the ablation of specific retinal cell populations by targeted expression of toxic genes may prove useful to reveal genes that function in circadian photoreception.