In light of the distinct expression of barhl1 and barhl2 in the tetrapod retina, we aimed at clarifying the retinal expression domains spanned by each of the three paralogs to get further insight into a possible non-redundant function of barhl paralogs in the zebrafish retina. We performed in situ hybridization on embryos at different developmental stages, starting from the beginning of retinal differentiation around 30 hours-post-fertilization (hpf), until 70 hpf. Transcripts of barhl2 are detected in the retina as early as 35 hpf (Figure 1B). At this stage, most RGCs and the first ACs are born 15. Barhl2 transcripts are initially localized in a few cells in the central retina and subsequently expressed in the whole retina (Figure 1A-C), in a pattern matching the wave of AC genesis. At 40 hpf, expression appears mostly restricted to the inner part of the INL (Figure 1J) and later on it extends to the GCL (Figure 1C, K). By 70 hpf expression is maintained in both the INL and the RGC layer (data not shown). Interestingly, transcripts of barhl1.2 were detected in the retina already at 28-30 hpf (Figure 1D). This stage corresponds to the peak of RGC genesis. Similarly to barhl2, its expression is initially restricted to the central retina and subsequently spreads temporally and dorsally (Figure 1D, E). However, this expression appears largely restricted to the RGC layer (Figure 1E, L and 1M) and it later becomes restricted to the ciliary marginal zone (CMZ) (50 hpf, Figure 1F). In contrast, no evident expression of barhl1.1 was found in the retina at all stages analyzed, although transcripts could be clearly detected in the brain (Figure 1G-I). This observation confirms that zebrafish barhl1.1 is the ortholog of other vertebrate barhl1; which were never found expressed in the retina of tetrapod 78.

Given the observed non-redundant expression of barhl2 and barhl1.2 transcripts in the retina, we then investigated in greater detail the relative spatio-temporal distribution of these two transcripts by double fluorescent in situ hybridization (FISH). At 35 hpf, a few barhl2-FITC (in green) positive cells could be detected in the central retina, located within a broader domain of barhl1.2-Cy3 (in red) positive cells (Figure 2A, D and 2G). Within this domain, expression of barhl1.2 and barhl2 appears mostly non-overlapping (white arrows in Figure 2A, D and 2G). By 40 hpf, transcripts of barhl1.2 are mostly restricted to the GCL while the ones of barhl2 are mostly found in the inner part of the INL where ACs are present (Figure 2B, E and 2H). Interestingly, overlapping and distinct patterns of expression of barhl1.2 and barhl2 can be observed also in the brain (see example in the diencephalon in Figure 2C, F and 2I).

To map out the temporal and spatial relationship between barhl1.2 and barhl2 expression and RGC differentiation, we compared their expression to the one of atoh7 16. The proneural bHLH (basic helix-loop-helix) transcription factor Atoh7 (previously named Ath5), homolog of Drosophila Atonal, has been shown to be essential for RGC differentiation in mouse, zebrafish and human 17181920 and is transiently expressed in retinal precursors fated to become mainly RGCs, but also horizontal, photoreceptor and specific subpopulations of ACs 152122. FISH performed with barhl1.2-Cy5 (in red) and atoh7-FITC (in green) shows that at 30 hpf transcripts of both genes co-localize in the central retina (Figure 3A-C). At 40hpf, atoh7 transcripts start to be downregulated in the mature RGCs of the central retina (Figure 3E, F), while barhl1.2 expression is maintained in this area (Figure 3D-F). At this stage, we begin to observe a co-localization of barhl1.2 with atoh7 within the CMZ (Figure 3F) clearly visible at 50hpf, (Figure 3G-I). Thus, barhl1.2 expression overlaps and follows the expression of atoh7 with a slight delay, suggesting that atoh7 might directly or indirectly influence expression of barhl1.2 in progenitors adopting the RGC fate. We then analyzed the expression of barhl2 with respect to the one of atoh7. Interestingly, in this case barhl2 expression is mostly complementary to the one of atoh7 (Figure 4). At 35 hpf, barhl2 starts to be expressed in a central retinal domain where atoh7 transcripts are already downregulated (Figure 4A-E). Later on, barhl2 expression expands towards the peripheral retina in a fan-like wave while atoh7 gets restricted to the CMZ (see 45 hpf and 50 hpf in Figure 4 F-J and 4 K-O respectively). At all stages analyzed, barhl2 and atoh7 expressions appear mostly mutually exclusive (examples are highlighted with asterisks in Figure 4I-J, N-O), although few cells co-expressing both genes are always visible at the expression boundaries (as indicated by arrows in Figure 4D-E, and 4N-O). Therefore, in contrast with barhl1.2, the complementary expression of barhl2 and atoh7 suggests a reciprocal negative regulation between these two genes. To further investigate this aspect in vivo, we took the advantage of the atoh7-/- mutant embryos (lakritz). In the atoh7-/- retina RGCs fail to exit the cell cycle and to differentiate as a result of a loss-of-function mutation within the atoh7 gene 19. We tested how atoh7 loss-of-function and therefore lack of RGC differentiation affect the expression of barhl1.2 and barhl2 (Figure 5). In situ hybridization on embryos at 40 hpf shows that barhl1.2 transcripts are completely missing in the GCL of the atoh7-/- retina (Figure 5E-F), while barhl2 expression is retained in this mutant retina (Figure 5A-B). The expression of both barhl paralogs in other brain areas, such as the tectum and the rhombic lips, is not affected by Atoh7 loss (Figure 5C-D and 5 G-H). Thus, barhl1.2 appears to be sustained by atoh7 while barhl2 onset of expression appears independent on atoh7. It remains to be demonstrated whether a negative feedback interaction exists between barhl2 and atoh7.

As it is commonly accepted that all teleosts underwent one further round of WGD, we wanted to test whether other teleosts have also retained more than two barhl and how their expression in the retina evolved. The medaka fish is a well-established model system and is therefore very suitable for a comparison with zebrafish 23. So far, one barhl has been described in the medaka fish (olbarhl, 14); based on phylogenetic analysis olbarhl has been clustered within the Barhl1 group of paralogs (olbarhl 6). Interestingly, olbarhl1 is also expressed in the retina and in particular in the developing GCL 14. In order to identify other putative barhl paralogs, we performed a BLAST search against the medaka genome (EnsEMBL Release 58, see methods) using zebrafish Barhl proteins as queries. We could identify a second member of the medaka barhl family on chromosome 4. By performing a multiple sequence alignment of Barhl proteins, the newly identified medaka Barhl was assigned to the Barhl2 group (Figure 6). The newly identified Barhl medaka protein has an asparagine residue instead of an alanine at position 15, an aspartic acidic residue instead of a glutamic acid at position 34 as well as an alanine instead of serine at position 38 (highlighted by asterisks in Figure 6C). The homeobox sequences of vertebrate Barhl differ at these positions and clearly divide the genes in the two mentioned groups. The FIL domains represent another set of highly conserved motifs 78121424. Only one of the FIL domains, FIL2, can be found in both medaka Barhl1 and Barhl2 (Figure 6A, B). To further characterize the evolutionary origin of the new medaka ortholog, a phylogenetic tree was constructed using Drosophila BarH1 and BarH2 protostome outgroup (Figure 7). Additionally, we rooted the tree with Barhl protein sequences of the ancestral deuterostomes Ciona savignyi and Branchiostoma floridae, which we identified by BLAST search. Ciona savignyi Barhl and Branchiostoma floridae Barhl also cluster with the outgroup, as those two deuterostomes have not undergone the vertebrate-specific WDG and have only one ancestral Barhl protein. The Barhl proteins are clearly clustered in two groups, with the newly identified medaka sequence belonging to the Barhl2 group. Thus, medaka has one Barhl1 and one Barhl2 that relate to the homologs in other teleosts, as we would expect from the relationship among these teleost species. Our tree clearly illustrates the relationship between zebrafish Barhl1 paralogs and other teleosts Barhl1 proteins and suggests that zebrafish Barhl1.2 has split basally from the teleost Barhl1 group.

We then sought to compare the expression patterns of both medaka barhl1 and medaka barhl2 in the retina by in situ hybridization, using antisense probes against transcripts amplified from mixed stages-brain and eye cDNA (see Methods). Medaka barhl1 starts to be expressed in the central retina at stage 25 1425. This stage more or less corresponds to 28-30 hpf in zebrafish, when atoh7 expression and RGC differentiation begin 2326. Later on, expression extends to the whole RGC layer 14. We found that at stage 30, the medaka barhl1 is still expressed in the GCL but is already being downregulated in the central retina (Figure 8A). Interestingly, this is the stage at which atoh7 expression also starts to be downregulated in the central retina in medaka 26. On the other hand, the medaka barhl2 is still strongly expressed in the central retina (Figure 8C). By stage 35, medaka barhl1 transcripts can be detected only in the CMZ (Figure 8B) while medaka barhl2 is still strongly expressed in both the GCL and the inner part of the INL (Figure 8D). Thus, the expression domain of the medaka barhl1 appears very similar to that of the zebrafish barhl1.2 as both transcripts can be found very early in the GCL and are very soon after restricted to the CMZ. Conversely, the medaka expression of barhl2 resembles the one observed for zebrafish barhl2, as both can be found strongly expressed in the INL and GCL until late stages of development.

Has the retinal expression domain been lost by one of the barhl duplicates, or has it been acquired after divergence of the two duplicated genes? Since the teleost branch underwent a WGD that did not affect other vertebrates, we expect to find two loci for each barhl gene in fish, which would explain the presence of two barhl1 paralogs in zebrafish, but only one paralog would have been retained in medaka. To further investigate this hypothesis, we compared the loci of barhl genes and looked for genes that are conserved in synteny. It has been proposed that genes that play important roles in development are surrounded by highly conserved elements and also show highly conserved gene synteny 27. Kikuta et al. elaborated on the highly conserved synteny between the human and zebrafish barhl1 loci that might extend to the regulatory level 27. For a more in-depth analysis we used additional teleost species to take into account syntenic conservation within these rapidly evolving fish species 28. We searched for genes that are in synteny between zebrafish, stickleback, medaka, Tetraodon and human, and found that for both barhl2 and barhl1, the conservation of the locus is very high, both for non-coding sequence as well as syntenic genes. Genomic locations of all genes used in the analysis can be found in the Additional file 1: Table1. Figure 9A shows a model of the synteny relations between medaka, zebrafish and human. Human barhl1 is located on chromosome 9. Orthologs of the genes surrounding barhl1 in human can be found on zebrafish chromosomes 5 (where barhl1.2 is present) and 21 (where barhl1.1 is present) and medaka chromosomes 9 (where no barhl can be found) and 12 (where barhl1 is present). This scenario is most likely the result of WGD in teleosts and highlights that the second medaka barhl1 paralog, expected to be conserved in synteny on chromosome 9 has been lost during evolution. For barhl2 we can observe a similar pattern: The genes surrounding barhl2 on human chromosome 1 can be found distributed between two chromosomes each in zebrafish and medaka. Notably, in both teleost species only one duplicate of barhl2 has been retained. The fact that we found three or more genes that are conserved in their position as neighbours of barhl2 in loci that did not contain barhl2 (zebrafish chromosome 2, medaka chromosome 17) suggests that these chromosomal regions represent duplicated regions that did not retain a second barhl2 paralog. The loss of one barhl2 copy has been consistently observed in other teleosts species such as Tetraodon and fugu whereas the barhl1 locus seems to have undergone different genomic rearrangements. Indeed, in Tetraodon we found that syntenic genes around barhl1 are distributed to three distinct loci, whereas in stickleback, they can be found in a single linkage group (see Figure 9B for an overview on the distribution of the syntenic genes surrounding the barhl locus in all species considered in this analysis). Altogether, the evolution of the barhl1 locus in teleosts, coupled to WGD and species-dependent regional duplication, seems to be to be under less evolutionary constraint than in the case of barhl2, where a stereotypic pattern of gene distribution can be observed.

Breeding and rising of zebrafish followed standard protocols 34. Zebrafish embryos were treated with 0.0045% 1-Phenyl-2-Thiourea (PTU) in medium after gastrulation to prevent pigment formation. Medaka embryos were kept in ERM medium containing 1 g/l NaCl, 30 mg/l KCl, 40 mg/l CaCl2 × 2 H2O and 163 mg/l MgSO4 × 7 H2O in deionized water.

All fish are housed in the fish facility of our laboratory, which was built according to the local animal welfare standards (Tierschutzgesetz 111, Abs. 1, Nr. 1) and in accordance with European Union animal welfare guidelines. The facility is under the supervision of the local representative of the animal welfare agency. No animal experiments were performed. Embryos of medaka (Oryzias latipes) and zebrafish (Danio rerio) were used exclusively at stages prior to hatching (not considered as animals according to German law and European union regulations). Zebrafish and medaka were raised and maintained as described previously 26. The following strains were used for zebrafish embryos: wild type WIK/AB and for medaka embryos: the Cab wild type strain."

An 800 bp fragment was polymerase chain reaction (PCR)-amplified from mixed stage medaka cDNA eye and brain using the forward primer (GAGATAGACACCGTGGGAACTGG) and reverse primer (CTGATGGAGTCCGGTACATGCTG) designed to bind in exons 1 and 4 of olbarhl2 (ENSORLT00000002844, EnsEMBL v58). Cycling conditions: five cycles 95°C, 10 sec, 65°C, 20 sec, 72°C, 4 min; followed by 28 cycles with annealing at 60°C. A Taq DNA-Polymerase was used for A-Tailing, incubation was for 30 min at 72°C. The PCR product was cloned into pCRII TOPO TA vector (Invitrogen) and sequenced using T7 and SP6 promoters. The sequence has been submitted to the EMBL [GenBank: JQ008931].

Single whole-mount in situ hybridization of barhl genes was carried out as previously described in 30, for the zebrafish embryos and in 31, for the medaka embryos. Riboprobes where labelled with digoxigenin-UTP (Roche Applied Science). Hybridization with the probe was carried on over night at 65°C/68°C. Anti-DIG primary antibody coupled to alkaline phosphatase (Roche Molecular Biochemicals) and NBT-BCIP (Roche Molecular Biochemicals) was used for signal detection. For the FISH on zebrafish embryos, modifications were applied to the method described in 30, as suggested by Stephan Kirchmaier (unpublished). Embryos were washed with 100 μl Tyramide Signal Amplification (TSA) solution and incubated with FITC in TSA. Incubation with the barhl2 probe was for 30 minutes, incubation with the atoh7 probe was for 40 minutes. Embryos were then kept in the dark for all following steps. For detection and staining of the antisense probes, embryos were washed 5 × 10 min with TNT (0.1M Tris pH7.5, 0.15M NaCl, 0.1% Tween20), incubated with 1% H2O2 in TNT for 20 min and washed again 5 × 10 min. Embryos were blocked in TNB (2% DIG Block in TNT) for 1 h at RT and afterwards incubated with Anti-Digoxigenin-POD Fab fragments diluted 1:100 in TNB. For signal detection, Fluorescein (FITC), Cyanine 3 (Cy3) or Cyanine 5 (Cy5) Fluorophore Tyramide by PerkinElmer was used. Embryos were then incubated in 1 × 4',6-Diamidin-2-phenylindol (DAPI) in TNT over night at 4°C and washed several times in TNT the next day. Embryos stained with NBT/BCIP were mounted in 87% Glycerol on microscope slides and imaged with a Leica DM5000B, 10x or 20x air objectives, Leica CD500 microscope camera and Leica FireCam 1.7.1 software. Double fluorescent embryos were mounted in 100 × 15 mm glass bottom dishes in 1.5% low melting agarose. Confocal stacks were taken using the Leica SP5 confocal microscope, 20x water immersion objective and Leica Application Suite (LAS) software. FITC was excited at 488 nm by Argon laser, Cy3 by the 568 nm Helium-Neon laser, Cy5 at 633 nm by Helium-Neon and DAPI by an UV laser. Emission was sensed at 500-550 nm for FITC, 650-700 nm for Cy3, 650-800 nm for Cy5 and 400-500 nm for DAPI. Emission channels were imaged sequentially to avoid bleed-through of the two fluorescent signals. Pictures were processed using the open source software ImageJ version 1.43 and Adobe Photoshop CS3.

Protein sequences were obtained from EnsEMBL Genome Browser (v58) after using TBLASTN to perform a search for the zebrafish protein sequence of Barhl2 and Barhl1.1, respectively, against the genomic DNA of the used species to check the integrity of the annotated protein sequences. The following protein sequences were used: Danio rerio Barhl1.1 (Accession number ENSDARP00000016114) Danio rerio Barhl1.2 (Acc. No. ENSDARP00000051473) Danio rerio Barhl2 (Acc. No. ENSDARP00000093436) Homo sapiens BARHL1 (Acc. No. ENSP00000263610) Homo sapiens BARHL2 (Acc. No. ENSP00000359474) Mus musculus BARHL1 (Acc. No. ENSMUSP00000053147) Mus musculus BARHL2 (Acc. No. ENSMUSP00000084005) Xenopus tropicalis barhl1 (Acc. No. ENSXETP00000013720) Xenopus tropicalis barhl2 (Acc. No. ENSXETP00000051744) Gasterosteus aculeatus Barhl1 (ENSGACP00000023974) Gasterosteus aculeatus Barhl2 (ENSGACP00000005730) Tetraodon nigroviridis Barhl2 (ENSTNIP00000016761) Tetraodon nigroviridis Barhl1.1 (ENSTNIP00000008103) Tetraodon nigroviridis Barhl1.2 (ENSTNIP00000008179) Tetraodon nigroviridis Barhl1.3 (ENSTNIP00000008783) Takifugu rubripes Barhl1 (ENSTRUP00000013477) Takifugu rubripes Barhl2 (ENSTRUP00000032575) Oryzias latipes Barhl1 (Acc. No. ENSORLP00000015485) Ciona savignyi Barhl (ENSCSAVP00000019219) Drosophila melanogaster Bar1 (Acc. No. FBpp0074204) Drosophila melanogaster Bar2 (Acc. No. FBpp00742043) Sequences obtained from National Center for Biotechnology's (NCBI) GenBank database (Release 178): Xenopus laevis Barhl1 (Acc. No. AAG14451.1) Xenopus laevis Barhl2 (Acc. No. NP_001082021.1) Salmo salar Barhl1 (Acc. No. NP_001167081.1) Branchiostoma floridae Barhl1 (XP_002596391.1). The protein sequence of Oryzias latipes Barhl2 was obtained from translating the sequenced cDNA clone using the ExPASy online translate tool from Swiss Institute of Bioinformatics (SIB). Alignment of the sequences was produced using MUSCLE online at the European Bioinformatics Institute (EBI) 35. A phylogenetic tree was assembled using BioNJ online at phylogeny.fr performing 1000 bootstraps and using Jones-Taylor-Thornton matrix. This distance-based algorithm is claimed to be well suited for comparison of sequences with high substitution rates 3637. The tree was visualized using the open source software Dendroscope 38.

Chromosomal loci of barhl genes in human, zebrafish, stickleback, Tetraodon and medaka were compared by identifying all genes that occur in more than one of the loci. The position of each of these genes was then searched in all species using the EnsEMBL database search function (EnsEMBL release v58). Position and identity of genes were plotted schematically according to their order and orientation (Figure 9, not to scale). Exact position of each gene can be found in the Additional file 1: Table 1.