Tbx2b is required for ultraviolet photoreceptor cell specification during zebrafish retinal development

Abstract

The vertebrate rod and cone photoreceptors are highly specialized sensory neurons that transduce light into the chemical and electrical signals of the nervous system. Although the physiological properties of cones and rods are well known, only a handful of genes have been identified that regulate the specification of photoreceptor subtypes. Taking advantage of the mosaic organization of photoreceptors in zebrafish, we report the isolation of a mutation resulting in a unique change in photoreceptor cell fate. Mutation of the lots-of-rods (lor) locus results in a near one-for-one transformation of UV-cone precursors into rods. The transformed cells exhibit morphological characteristics and a gene-expression pattern typical of rods, but differentiate in a temporal and spatial pattern consistent with UV-cone development. In mutant larvae and adults, the highly ordered photoreceptor mosaic is maintained and degeneration is not observed, suggesting that lor functions after the specification of the other photoreceptor subtypes. In genetic chimeras, lor functions cell-autonomously in the specification of photoreceptor cell fate. Linkage analysis and genetic-complementation testing indicate that lor is an allele of tbx2b/fby (from beyond). fby was identified by a pineal complex phenotype, and carries a nonsense mutation in the T-box domain of the tbx2b transcription factor. Homozygous fby mutant larvae and lor/fby transheterozygotes also display the lots-of-rods phenotype. Based upon these data, we propose a previously undescribed function for tbx2b in photoreceptor cell precursors, to promote the UV cone fate by repressing the rod differentiation pathway.

Vertebrates have evolved 2 major classes of retinal photoreceptors: rods, which mediate dim light vision, and cones, which detect light of greater intensity, have a faster temporal resolution, and mediate color vision. Largely from the analysis of mutations in mice and humans, a transcriptional network regulating photoreceptor cell development has been proposed. The presumptive photoreceptor progenitors sequentially express the homeobox transcription factors (TFs) Otx2 and Crx (1–3), and in their absence, photoreceptor precursors are not specified or fail to differentiate. Rod specification requires the additional expression of the Maf-family TF Nrl and its target Nr2e3 (4, 5). Nrl acts as a molecular switch; in its absence, precursors adopt the short-wavelength (S) opsin cone fate (5, 6), and mis- and over-expression of Nrl transforms most if not all cone precursors into functional rods (7, 8). NR2E3 expression, which is disrupted in enhanced S-cone syndrome in humans and the rd7 mouse, is required for the repression of cone-specific genes in rod precursors (4, 9, 10). However, it remains to be determined if a reciprocal system exists in cone precursors for repressing rod-specific genes.

The zebrafish retina, in addition to rods, possesses 4 cone subtypes, each with a distinct morphology and expressing a unique opsin (11–17). The spatial and temporal differentiation of the photoreceptors leads to the formation of a highly ordered, precisely defined arrangement (16, 18). The photoreceptor mosaic provides an opportunity to systematically uncover genetic mechanisms regulating vertebrate photoreceptor subtype specification, similar to the studies of Drosophila ommatidial assembly initiated several decades ago.

We identified a mutation called lots-of-rods (lor^p25bbtl) that results in an increase in the number of rods and a decrease in the number of UV cones in the larval and adult zebrafish retina. The lor^p25bbtl phenotype demonstrates many features opposite to those observed in enhanced S-cone syndrome and mutations of Nr2e3 or Nrl in mice. Genetic analysis revealed that lor^p25bbtl is an allele of tbx2b, a T-box TF with essential roles in development (19–22). Our data suggest that during photoreceptor cell differentiation, tbx2b acts cell-autonomously to promote the UV-cone fate by repressing the rod fate in zebrafish photoreceptor cell progenitors.

Results

To identify genes essential for vertebrate photoreceptor development, we screened 5- to 6-day postfertilization (dpf) zebrafish larvae for ethyl nitrosourea-induced mutations, leading to alterations in rod patterning (23). Rods first appear in the ventral retina coincident with the expression of the first cone opsin (Fig. 1B) but differentiate in the dorsal and central retina in a sporadic pattern subsequent to the differentiation of the cones (Fig. 1C) (15, 17). In contrast, lor^p25bbtl mutants displayed a higher number of rods across the entire retina, with few gaps in the central or dorsal regions (Fig. 1E). Between 3 and 5 dpf, rod immunolabeling in mutant larvae spread in a continuous front from the ventral to central and finally to the dorsal retina, reminiscent of the wave-like fashion of cone differentiation (14) (Fig. 1F). The increased number of rods was confirmed by labeling with either monoclonal or polyclonal antibodies against rhodopsin, and was mirrored by GFP expression when lor^p25bbtl was placed on the XOPS-GFP transgenic background (18). In teleosts, rods are continuously added to the postembryonic retina from a population of mitotically active cells called the rod progenitors (24); however, BrdU labeling detected no increased mitotic activity in lor^p25bbtl mutants (data not shown). lor^p25bbtl mutants were also morphologically normal (Fig. 1 A and D), demonstrated a robust optokinetic response (OKR), and could be routinely reared to fertile adults.

Fig. 1.

lor^p25bbtl mutants display increased labeling for rods. Ventral views of bright field images (A and D) and confocal immunofluorescent images of rod-specific labeling of eyes from WT (B and C) and a lor^p25bbtl mutant larvae (E and F) (dorsal is up). At 3 dpf, WT larvae display rod labeling first in the ventral retina (B) followed at 5 dpf, by sporadic labeling of individual cells in the central and dorsal retina (C). At 3 dpf, lor^p25bbtl mutant larvae display increased rod immunolabeling across the ventral and central retina (E) that is evenly distributed at 5 dpf (F). (G) Linkage analysis placed lor^p25bbtl between SSLP markers Z31583 and Z13481 on the MGH panel. The number of recombinants in 87 mutant larvae is shown (red). The interval containing the lor mutation and the tbx2b locus is shown.

The lor^p25bbtl mutation was mapped to an interval of chromosome 15, near simple sequence length polymorphism (SSLP) markers z22430/z25911 (Fig. 1G), where tbx4 and tbx2b/fby were localized (21). tbx2b is a TF mainly associated with transcriptional repression during cell cycle control, limb, heart and endoderm development (19, 20, 22), and cancer (25). A single mutant allele of tbx2b has been reported in zebrafish. The fby mutation results from a T-to-A transversion generating a premature stop codon within the T-box sequence, and was isolated based upon a pineal gland phenotype (21). Complementation testing indicated that lor^p25bbtl is an allele of tbx2b. Crosses between carriers of the lor^p25bbtl mutation and the fby mutation revealed that fby failed to complement lor^p25bbtl; ≈25% of the progeny of the intercross displayed the increased rod phenotype (data not shown). In addition, homozygous fby mutant larvae, confirmed by sequencing the tbx2b gene, demonstrated the “lots-of-rods” phenotype, whereas phenotypically WT siblings were either homozygous for the WT allele or heterozygous.

tbx2b expression was examined by in situ hybridization in WT and lor^p25bbtl mutant embryos (Fig. 2 A–D). tbx2b expression was greater in the dorsal retina and absent in the ventral retina adjacent to the choroid fissure (see Fig. 2 A and B) (26, 27), the region of precocious rod differentiation in embryos and of highest rod density in larvae (see Fig. 1 B and C). In lor^p25bbtl mutants, the dorsal/ventral gradient of tbx2b retinal expression was still detectable, but labeling was much fainter than in the WT embryos (see Fig. 2 B and D). At 44 (Fig. 2D) and 60 hpf (data not shown), tbx2b expression persisted in the inner retina, in the regions of continued neurogenesis at the retinal margins and cells adjacent to the forming ONL, but was diminished in the central retina. In mutant larvae, labeling was greatly reduced across the retina. RT-PCR of RNA extracted from WT and lor mutant embryos confirmed the lower expression of tbx2b at 20 and 28 hpf in the mutant (Fig. 2E). Sequencing of tbx2b cDNA from 4-dpf WT and lor^p25bbtl larvae revealed no changes in the tbx2b coding region (data not shown), suggesting that lor^p25bbtl represents a mutation in a regulatory sequence.

Fig. 2.

tbx2b expression is reduced in lor mutant embryos. Sagittal sections of WT (A) and lor^p25bbtl mutant (C) embryos at 28 hours postfertilization (hpf) following in situ hybridization for tbx2b show labeling throughout the eye, but the lack of expression in the ventral retina near the choroid fissure (CF) and lens (L). Transverse sections of labeled WT (B) and lor^p25bbtl embryos (D) at 44 hpf. tbx2b expression is most intense at the dorsal retinal margin and cells adjacent to the developing outer nuclear layer (ONL) as well as the inner nuclear layer (INL) (optic nerve; ON). Note the dramatically reduced expression in the lor^p25bbtl embryos. (E) Amplification of tbx2b by RT-PCR from RNA extracted from WT and lor^p25bbtl embryos at 20 and 28 hpf demonstrates a reduction in the amount of tbx2b mRNA in the mutant.

To test if the increased number of rods resulted from a change in fate of 1 of the 4 cone subtypes, serial sections of larvae from intercrosses of lor^p25bbtl heterozygotes were co-immunolabeled with 1 of 4 polyclonal antisera to the cone opsins and a rod-specific monoclonal antibody to distinguish mutant from WT larvae [supporting information (SI) Fig. S1]. In the lor^p25bbtl retinas, labeling for the UV opsin was nearly absent (see Fig. S1D′). Immunolabeling for the 3 other cone opsins (see Fig. S1) and markers of other retinal cell types did not reveal any other alteration, nor was increased apoptosis or retinal degeneration found.

Confocal images of eyes enucleated from whole-mount immunolabeled WT and mutant larvae confirmed the significant changes in the number of rods and UV cones (Fig. 3 A, B, D). In WT larvae, the UV cones are regularly spaced across the entire retina with the few rods distributed across the central and dorsal regions (see Fig. 3A). In lor^p25bbtlmutant larvae, rods were evenly distributed across the entire retina and the number of UV cones was markedly reduced (see Fig. 3B). The number and distribution of UV cones varied from mutant to mutant and between the 2 eyes from a single animal, and no cells were observed that labeled for both UV opsin and markers of rods simultaneously. Cell counts and spatial pattern analysis of rod labeling in lor^p25bbtlmutants mirrored those typically observed for the UV cones in WT larvae. Immunolabeling of fby mutant larvae revealed a similar yet stronger phenotype (Fig. 3C), and transheterozygous larvae (lor^p25bbtl/fby) revealed an intermediate phenotype (Table S1). Expression analysis by RT-PCR of opsins and photoreceptor arrestins in 5-dpf larvae confirmed the increase in expression of rod opsin and rod arrestin and the decrease in expression of the UV opsin in lor^p25bbtl mutant larvae relative to controls (Fig. 3 E and F). Based upon the genetic data, we propose that lor^p25bbtl is a hypomorphic allele of tbx2b and together, fby and lor^p25bbtl form an allelic series. Therefore, we shall refer to the lor^p25bbtl allele as tbx2b^p25bbtl.

Fig. 3.

Increased rod number and fewer UV cones in lor^p25bbtl and fby mutants. (A–C) Confocal images of eyes from 5-dpf WT (A), lor^p25bbtl (B), and fby (C) larvae, immunolabeled for UV opsin (green) and rods (red). Eyes from lor^p25bbtl and fby mutant larvae demonstrate a dramatic deficit in UV cones, with fby displaying a more severe phenotype; in this particular sample, no UV cones were detected. (D) Graph showing the average number of rods (red) and UV cones (green) per unit area quantified from confocal images of lor^p25bbtl mutant and WT retinas (mean + SD). (E and F) Expression analysis of opsins (E) and arrestins (F) by RT-PCR from RNA extracted from 5-dpf WT and lor^p25bbtl larvae. The RT-PCR product for UV opsin was reduced and products for rod opsin and rod arrestin were increased in lor^p25bbtl (red boxes).

Histological analysis of the photoreceptor cell mosaic was performed in WT and tbx2b^p25bbtl adults. In transverse sections (Fig. 4 A and C), other than showing fewer UV cones, tbx2b^p25bbtl homozygous adults demonstrated no significant difference in laminar organization of the photoreceptors (Fig. 4 A and C). Rods appeared equally abundant in WT and lor, most likely because of the persistent mitotic activity of the rod progenitors. Across the entire retina, the morphology of the rods and remaining UV cones was normal (see Fig. 4C). There were no intermediate or hybrid cell types, such as the so-called “cods” or “rones” that occur in the Nr2e3 mutant mice (4, 28). In tangential sections taken near the outer limiting membrane (OLM), the photoreceptor mosaic was distorted in tbx2b^p25bbtl adults; the double cone nuclei were not in orderly rows, and the numerous rod inner segments were aggregated together as they passed through the OLM (data not shown). However, more distal sections through the level of the blue-cone inner segments and remaining UV-cone outer segments revealed that the alternating rows of single and double cones were maintained, and the occasional UV cone alternated in rows with the blue cones (Fig. 4D). Therefore, the most plausible explanation for the disruption of the mosaic near the OLM is that the physical absence of the UV cones led to a distortion of the orderly packing of the remaining photoreceptors.

Fig. 4.

Maintenance of the photoreceptor mosaic in tbx2b^p25bbtl mutant adults. Transverse (A and C) and tangential (B and D) histological sections through the ONL from WT and tbx2b^p25bbtl adult animals (rods are GFP⁺) (green), immunolabeled for UV opsin (red), and counterstained with DAPI (blue). (A and C) Tiering of the rod and cone photoreceptor cells is maintained although the UV cones (asterisks) in the tbx2b^p25bbtl retinas are diminished in number and appear to have a greater width than in WT retinas. The approximate plane of the sections in (B) and (D) are indicated [arrows in (A) and (C)]. (B) Immunolabeling of tangential sections for the UV opsin reveals the photoreceptor cell mosaic composed of rows of UV cones and blue cones (arrowheads). (D) In tbx2b^p25bbtl the regular arrangement of the rows of blue cones (gaps in labeling; arrowheads) and occasional UV cones are maintained although the width of the cones is greater and there is some distortion of the row mosaic.

Numerous soluble factors have been implicated in the differentiation of rods and cones (29, 30), yet as a TF, we would predict that tbx2b functions cell-autonomously in photoreceptor cell specification. To test this hypothesis, we generated genetic chimeras between WT and tbx2b^p25bbtl mutant embryos (31). Cells from WT blastula stage embryos (donors) injected with rodamine-dextran were transplanted into age-matched tbx2b^p25bbtl hosts carrying the XOPS-GFP transgene (18). The resulting chimeras displayed retinas with a mixture of donor WT cells among regions of mutant cells with GFP⁺ rods. Immunolabeling for UV opsin and rhodopsin demonstated that the donor-derived WT cells (red) often colabeled for UV opsin, but few colabeled for rod opsin (Fig. 5A). In contrast, among the neighboring mutant cells, UV-opsin expression was absent and GFP⁺ rods were abundant (see Fig. 5A). In 4 mutant retinas, cell counts showed of the 113 WT donor cells located in the ONL, 44 cells (39%) colabeled for UV opsin, and only 9 (8%) for rod opsin. In reciprocal transplants, out of 148 tbx2b^p25bbtl mutant donor cells, 5 cells (3.4%) colabeled for UV opsin and 65 cells (44%) colabeled for rod opsin (Fig. 5B), consistent with a cell-autonomous role in regulating photoreceptor cell fate.

Fig. 5.

tbx2b^p25bbtl acts cell-autonomously. Genetic chimeras were generated by blastula transplantation and allowed to develop to 80 hpf. Donor cells were labeled by rhodamine-dextran (red); tbx2b^p25bbtl deficient rods expressed the XOPS-GFP transgene. Chimeras were whole-mount immunolabeled for UV opsin (A and B) (blue) and for rods (A) (green). (A and B) Stacks of confocal images taken tangential to the photoreceptor cell layer and in the orthogonal plane (insets). (A) WT donor cells (red) located in the ONL of tbx2b^p25bbtl hosts frequently colabel for the UV opsin (inset). (B) tbx2b^p25bbtl mutant donor cells (red) in the ONL of WT hosts frequently differentiate as GFP⁺ rods and rarely label for the UV opsin. Inset shows dextran/GFP positive rods neighboring UV opsin-positive/dextran negative cone outer segments.

Nr2e3, an early marker and key regulator of rod differentiation in most vertebrates, is transiently expressed in all presumptive photoreceptors in the zebrafish; then, before the onset of opsin expression, Nr2e3 becomes restricted to cells of the rod lineage (4, 32). As such, we would anticipate that in tbx2b^p25bbtl mutant larvae, Nr2e3 expression would persist in the population of progenitors transfated to become rods. Immunofluorescent images of WT retinas showed that at 2 dpf, before genesis of the ONL, cells in the central retina and near the future ONL expressed Nr2e3 (Fig. 6A). By 3 dpf, almost all of the Nr2e3 labeling was restricted to the ONL, and as the first rods started developing in the ventral retina, colocalization of nuclear labeling for Nr2e3 and rod labeling was detected (Fig. 6B Inset). As previously reported (4), by 4 dpf the expression of Nr2e3 was down-regulated in the presumptive cones and became restricted to cells sporadically distributed across the ONL that colabeled for rod opsin (Fig. 6C). Near the retinal margin, individual cells expressing Nr2e3 and not colabeled for the rod marker can be observed. In tbx2b^p25bbtl mutants, Nr2e3 expression persisted in many cells across the ONL, most of which colabeled for rod markers (Fig. 6D). These data suggest that the reduction or absence of tbx2b leads to the persistent expression of Nr2e3 in a subset of cells and their subsequent differentiation as rods.

Fig. 6.

Persistent Nr2e3 expression in lor. (A–D) Transverse cryosections from 2- (A), 3- (B), and 4-dpf (C and D) WT and tbx2b^p25bbtl embryos immunolabeled for rods (4C12, green) and Nr2e3 (red), and counterstained with DAPI (blue); dorsal is up. (A) At 2 dpf, labeling for Nr2e3 is observed in the inner and outer retina before differentiation of photoreceptors. (B) At 3 dpf, Nr2e3 labeling is restricted to developing photoreceptors in the ONL. In the ventral retina, Nr2e3 nuclear labeling colocalizes with immunolabeling for rods (inset). (C and D) At 4 dpf, expression of Nr2e3 persists in differentiating rods in WT and tbx2b^p25bbtl retinas.

Discussion

Taking advantage of the precisely defined photoreceptor mosaic in zebrafish, the data show that lor^p25bbtl mutants exhibit an increase in the number of rods and a dramatic decrease in the number of UV cones because of a change in cell fate. We present genetic evidence that tbx2b is a regulator of photoreceptor cell fate in zebrafish and is essential for the proper specification of the UV cone. Most striking is that the function of tbx2b during photoreceptor development is opposite to that of Nrl (5).

Our data significantly add to the most widely held model of neuronal specification during retinal development. The current model proposes that multipotent retinal progenitor cells pass through a series of competence states, such that at a specific time, cells can adopt only one or a few particular cell fates in response to extrinsic signals or internal cues (33). Consistent with this model, early- and late-born photoreceptor cell progenitors express the TFs Otx, Crx, and NeuroD (1–3, 34). Yet, in early-born photoreceptor cell precursors, thyroid hormone receptor beta regulates the expression of middle wavelength-sensitive opsin versus S-opsin (29), and in later-born precursors, Nrl acts as a molecular switch to drive expression of rod genes and repress the S-cone fate (5, 6). Thus, the S-cone was positioned as the default photoreceptor. Our data identify a previously unknown pathway essential for specification of the UV cone fate, the likely homologue of the S-cone in mammals. Previous studies have described Tbx2 as a transcriptional repressor (19, 20, 25). Based upon these observations, we propose a mechanism by which photoreceptor cell diversity is maintained by the expression of discrete genes that act to suppress alternative cell fates within precursors that share a common molecular signature.

What can be concluded about this strikingly conserved relationship between SWS1-cone precursors and rod progenitors in teleosts and mammals? As noted, the phenotype observed by mutation of tbx2b in zebrafish is directly opposite to the phenotype elicited by loss of Nrl function in mammals, suggesting a conserved ontology. The evidence suggests that tbx2 orthologues may also play a role in photoreceptor identity in mammals. Consistent with the genetic data, the dorsal-ventral pattern of expression of Tbx2 in the retina is remarkably conserved across vertebrates, suggesting a conservation of function (26, 27, 35–37). Additionally, a search for genetic modifiers of the rd7 mouse uncovered several loci that suppress the retinal degeneration and restore the normal photoreceptor number (38). One of these mapped to chromosome 11, in close proximity to the Tbx2 locus. Moreover, as part of an in silico study of human promoter sequences, TBX2 was identified as a potential target for the photoreceptor-specific TFs NRL, NR2E3, and CRX (39). Unfortunately, targeted mutagenesis of Tbx2 in mice causes severe cardiac defects (20), and mutant embryos die between embryonic day 10.5 and 14.5, preventing an analysis of photoreceptor cell fate.

The dorsal-ventral gradient of tbx2b expression and its conspicuous absence from the ventral retina, an area of precocious rod genesis in zebrafish, suggests that the function of tbx2b is to repress the rod cell fate. Consistent with this hypothesis, the reduction or absence of tbx2b expression in genetic mutants underlies the precocious differentiation of rods across the entire retina. The expression data suggest that the timing of tbx2b action precedes photoreceptor differentiation. tbx2b mRNA is expressed in the neuroepithelium and at the retinal margin before formation of the ONL, yet is down-regulated in photoreceptors. Similarly, we report that Nr2e3 expression is observed in cells of the central retina before formation of the ONL, at the retinal margin, and is expressed by mitotic photoreceptor progenitors (32). In addition, others have shown that Crx and NeuroD are also expressed in a comparable pattern (34, 40). Thus, in a rapidly developing vertebrate such as the zebrafish, photoreceptor specification may occur before or be coincident with cell cycle exit and laminar positioning. Interestingly, prior work suggested that tbx2b is essential for neuronal differentiation in the dorsal retina (26), but we did not find evidence for a similar defect in homozygous tbx2b^p25bbtl or fby mutant larvae. However, a second zebrafish orthologue of mammalian tbx2, tbx2a (also called tbx2b-like), shares 78% identity with tbx2b (41) and shows a similar pattern of expression. The previously reported morpholino experiments likely revealed a conserved role for tbx2 orthologues during the earlier stages of vertebrate retinogenesis.

In summary, our unique finding for tbx2b in zebrafish is one of only a small collection of studies in vertebrates, to show a specific role for a TF in photoreceptor-cell subtype specification. The phenotype also suggests a highly conserved relationship between rod progenitors and the SWS1-cone precursors in teleost and mammals. It is anticipated that as genetic screens for alterations in the teleost retinal mosaic continue, additional genes that regulate vertebrate photoreceptor subtype specification will be isolated and allow us to test specific hypotheses of the mechanisms of cell fate determination in the nervous system.

Methods

Zebrafish Maintenance.

Rearing, breeding, and staging of zebrafish (Danio rerio) were performed according to standard methods (42). The tbx2b^p25bbtl mutant was isolated from a screen of ethyl nitrosourea-mutagenized zebrafish immunolabeled for developing rods at 5 to 6 dpf (23). Mutagenesis was performed at the University of Pennsylvania as previously described (43). The transgenic zebrafish line expressing GFP in rods (18) and the fby^c144 line were previously described (21). All procedures were approved by the Florida State University Animal Care and Use Committee.

PCR, Cloning, and Sequencing.

Primers for cloning the ORF of tbx2b were provided by Dr. Jeffrey Gross. RNA was isolated with TRIzol (Life Technologies), and cDNA was synthesized using oligo(dT) primer. The sequences of the gene specific primers can be provided upon request. The RT-PCR products from 4-dpf WT and tbx2b^p25bbtl embryos were cloned into the TOPO vector (Invitrogen) and sequenced.

The amplification and sequencing of the T-box in the tbx2b gene from the fby line was performed as follows: The progeny of fby heterozygous parents were fixed at 5 dpf, immunolabeled with 4C12 antibody (specific for rods), and screened for WT and “lots-of-rods” phenotype. DNA was extracted from 3 individual larva of each condition and subjected to PCR. The 1,603-bp PCR product was sequenced and the resulting data for WT and tbx2b^p25bbtl were compared.

Immunohistochemistry.

Immunolabeling and fluorescence microscopy of whole-mount larvae and frozen sections (10 μm) were performed as described previously (18, 44). BrdU incorporation was performed as described (32). Sections and enucleated eyes from whole-mounted immunolabeled larvae were imaged with a Zeiss 510 Scanning Laser Confocal microscope equipped with either a 20× (NA 0.75) objective or 40× water-immersion objective (NA 1.2) as described (18).

Whole-Mount in Situ Hybridization.

Whole-mount in situ hybridization was performed essentially as described (45).

Cell Transplantation.

Genetic chimeras were generated as described (31). Donors were labeled at the 1- and 2-cell stage by injection with lysine-fixable dextran-conjugated Alexa Fluor 594 (Invitrogen). Transplant was performed at the 1,000-cell stage. The chimeras were fixed at 80 hpf and immunolabeled as described above.

Mapping.

Linkage analysis was performed at the Zebrafish Mapping Facility at the University of Louisville from DNA isolated from 100 WT siblings and 100 tbx2b^p25bbtl embryos using SSLP markers (46).

Quantitative Analysis.

Confocal images from whole eyes immunolabeled for UV cones and rods (4C12 antibody) were analyzed with the Scion Image Software (Scion) (18). An area corresponding to 3,500 μm² in the central retina was used to count UV cones and rods in 4- and 5-dpf WT and lor retinas. The following number (n) of 4- or 5-dpf retinas and cells types were analyzed: WT, n = 4 (UV cones) and n = 6 (rods); tbx2b^p25bbtl, n = 6 (UV cones) and n = 6 (rods); lor/fby, n = 7 (UV cones and rods). The average number of UV cones or rods and the standard deviation were reported.