Lineage Tracing Reveals atoh7-Positive and Negative Retinal Ganglion Cell Populations in the Zebrafish Retina

Abstract

Purpose

Atoh7 is a transiently expressed developmental transcription factor that contributes to the generation of all seven major retinal cell types. Despite this broad lineage potential, Atoh7 is specifically required for retinal ganglion cell (RGC) formation and survival. In mice, a substantial proportion of RGCs arise from Atoh7-negative progenitors, suggesting potential nonautonomous roles for Atoh7 in RGC development. Although atoh7 function is conserved in zebrafish, the complete lineage, including the contribution to the RGC population, has not been fully defined. Here, we sought to determine the atoh7 retinal lineage in wild type and atoh7 mutant zebrafish.

Methods

We generated atoh7:iCre transgenic zebrafish and paired them with the established ubi:Switch reporter to permanently label the atoh7 lineage. We validated transgene expression and lineage labeling using in vivo live imaging and immunohistochemistry in embryonic, larval, and adult retinas, as well as in select regions of the adult brain.

Results

The atoh7:iCre;ubi:Switch system accurately recapitulated endogenous atoh7 onset, with reporter expression persisting into adulthood. We found that 79% of RGCs in wild-type retinas arise from atoh7-positive progenitors, a greater proportion than previously reported in mice. Additionally, mutant retinas displayed a significant increase in atoh7 lineage+/Pax6+ amacrine cells and an increased number of Prox1+ bipolar cells. We also identified atoh7 lineage positive cells in other central nervous system (CNS) regions.

Conclusions

Our findings reveal both atoh7 lineage positive and negative retinal cell types in zebrafish, including RGCs, providing a platform to study survival and cell fate mechanisms of atoh7 lineage negative RGCs during retinal development.

The vertebrate neural retina is composed of seven major cell types, including amacrine, bipolar, horizontal, retinal ganglion cells (RGCs), Müller glia (MG), and rod and cone photoreceptors, which arise from a common pool of multipotent naïve retinal progenitor cells (RPCs).^1–4 Retinal neurogenesis begins with naïve RPC expression of a combination of proneural transcription factors, transitioning RPCs into each of the post mitotic retinal cell types. One of these proneural factors is Atoh7 (atonal homolog 7), a basic-helix-loop-helix transcription factor that begins expressing prior to the onset of retinal neurogenesis, with its lineage representing all seven major retinal cell types.⁵ Despite its broad lineage, Atoh7 is only essential for the genesis and survival of RGCs, the first-born retinal neuron. In all species studied, loss of Atoh7 results in a ≥95% loss of all RGCs, whereas the other retinal cell types are either modestly or completely unaffected.^5–10 Interestingly, based on lineage tracing only 55% of total RGCs in the adult mouse retina come from the Atoh7+ lineage despite the severe loss of RGCs in Atoh7 mutant retinas.⁵ These data suggest the Atoh7+ RPC or RGC population may be required for the survival and genesis of all RGCs, although the exact mechanism remains unknown. Zebrafish are an ideal model to study this complex interaction, but it has not been established if they contain an atoh7-negative RGC lineage.

Atoh7 is transiently expressed during development, with its mRNA and protein being undetectable in fully differentiated retinal neurons, thus making it difficult to determine the full atoh7 lineage.^5,11–14 In fish, atoh7 (previously ath5) expression begins at approximately 25 hours post-fertilization (hpf) in the ventro-nasal retina. It spreads dorso-nasally and ventro-temporally, preceding the wave of retinal neurogenesis, and completes this wave by 48 hpf.^12,14 By 72 hpf, atoh7 mRNA is restricted to the ciliary margin.¹⁴ The first tracking of the zebrafish atoh7 retinal lineage was done using an atoh7 promoter driven transgene, ath5:GFP.^12,15,16 The stability of GFP allowed for successful tracking of the atoh7 lineage, revealing a faithful expression pattern during early neurogenesis and post mitotically in RGCs, photoreceptors, amacrine, and horizontal cells at 4 days post-fertilization (dpf).^12,15,17 Despite the ability to track the atoh7 lineage at these early timepoints, the ath5:GFP transgene is limited in its ability to determine the atoh7 lineage of the entire neural retina at late stages when all cell types are born. More recently, a zebrafish-specific atoh7:Cre BAC transgene was utilized to track RGC expression within the retina and optic tectum of larval zebrafish, but a full characterization was not completed.^18–20

To determine the zebrafish atoh7 RGC and full retinal lineage in the mature retina, we created a new transgenic line utilizing the zebrafish atoh7 promoter to drive expression of Cre recombinase, atoh7:iCre. atoh7 lineage+ cells were permanently labeled by combining the atoh7:iCre transgene with the ubi:Switch (LoxP-eGFP-LoxP-mCherry) transgene.^21,22 The mCherry expression matched the hallmark spatiotemporal expression pattern of endogenous atoh7 and was detected into adulthood.^14,17,21 We determined there are both atoh7 lineage-positive and -negative populations of RGCs, and all other major retinal cell types are represented in the atoh7 lineage, albeit in differing percentages compared with mice. We further characterized the atoh7 lineage in atoh7 (lakritz) mutants,⁶ discovering that the proportion of atoh7 lineage+ populations remain the same in all cell types, except RGCs and amacrine cells. Last, atoh7 is expressed in cell populations within the central nervous system (CNS) outside of the neural retina. The data presented indicate that our lineage tracing transgenes create an innovative way to study the possible genesis and survival mechanisms between the positive and negative atoh7 lineage cell populations of RGCs, other major retinal cell types, and a variety of cells in the CNS.

Methods

Zebrafish Husbandry

All adult zebrafish were housed in a controlled environment with a water temperature of 28°C and 14-hour light/10-hour dark cycle. Embryos were raised at 28°C in 1x InstantOcean (1xIO) Sea salt (0.06 g/L) upon collection and transferred at 24 hpf to 1xIO/1X PTU (0.003% 1-phenyl-2-thiourea; Acros Organics/Thermo Fisher Chemicals) to prevent pigmentation for transgenic screening, then returned to 1xIO until 5 dpf. For in vivo imaging, zebrafish larvae were anesthetized with 1x Tricaine (0.17 g/L; Tokyo Chemical Industry Co., Ltd.) diluted in 1xIO. Animals were euthanized with cold 1x Tricaine diluted in 1xIO/1X PTU. All animal work was performed in accordance with the Institutional Animal Care and Use Committee of the Medical College of Wisconsin, ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Plasmid and Transgenic Line Generation

The previously published atoh7 (ath5) zebrafish promoter includes 4772 nucleotides 5ʹ to the atoh7 transcriptional start site, confirmed by GENEWIZ Plasmid-EZ Sequencing. Based on sequence comparison to the zebrafish genome build GRCz11/danRer11, the atoh7 promoter plasmid sequence represents chr13: 25,444,901 to 25,449,692 in the University of California, Santa Cruz (UCSC) genome browser,²³ where genome build GRCz11/danRer11 contains an additional 20 nucleotides not found in the cloned atoh7 promoter plasmid (atoh7:iCre; Supplementary Fig. S1).^12,22 The final atoh7:iCre;cmlc:eGFP (atoh7:iCre) and atoh7:nd2GFP (nuclear destabilized GFP; Clonetech) plasmids were generated using the Tol2 Gateway Kit (Invitrogen).^24,25 Tol2 transposase mRNA (50 pg) was co-injected with the completed atoh7:iCre or atoh7:nd2GFP plasmid (50 pg) in 1 to 4 cell stage zebrafish embryos.²⁶ All experiments were conducted on the second generation (F2) to verify single insert inheritance and expression pattern consistency. The atoh7:iCre line is designated as mw800 and the atoh7:nd2GFP as mw801 in The Zebrafish Information Network database (zfin.org).

Immunohistochemistry

Larval zebrafish and adult heads were fixed in 4% paraformaldehyde (PFA)/1X phosphate buffered saline (PBS) overnight at 4°C, washed 3 times in 1X PBS, and processed through a sucrose gradient (5% or 15% sucrose/1X PBS/0.2% Na azide) prior to cryopreservation in optimal cutting temperature medium (O.C.T.; Thermo Scientific). Serial sections (10 µm) were collected and rehydrated with 1X PBS, treated with a MeOH series (33%, 66%, and 100%) to reduce endogenous eGFP and mCherry fluorescence from the ubi:Switch transgene followed by 3 washes in 1X PBS. Standard immunohistochemistry with 4% (w/v) nonfat dry milk and 0.1% (w/v) Tween-20 in Tris-buffered saline (TBST) was performed on cryosections.^27,28

Adult retinal flatmounts were performed by enucleating the right eye from fixed adult heads, isolating intact retinal tissue, and processing the retinas for immunohistochemistry, as previously described.²⁹ Primary antibodies included: goat α mCherry (AB0081-500; OriGene) 1:500, rabbit α Rbpms2 (AB181098-1001; Abcam) 1:1000 for sections and 1:500 for whole mount embryos, mIgG1 α Cre (MAB3120; Millipore) 1:500, rabbit α Pax6 (12323-1-AP; Proteintech) 1:500, rabbit α Prox1 (AB5475; Millipore) 1:500, mIgG1 α Zpr1 (ANZPR-1; Zirc) 1:100, and wheat germ agglutinin (WGA) conjugated to Alexa 488 (W11261; ThermoFisher) 1:500. Secondary antibodies were all used at 1:500 and included: donkey α rabbit 488 (A21206; Invitrogen), donkey α mIgG1 488 (A21121; Invitrogen), and donkey α goat Cy3 (705-165-147; Jackson ImmunoResearch Laboratories, Inc.). Nuclei were labeled with DAPI (4′,6-diamidino-2-phenylindole, 50 µg/mL) and slides cover slipped with Southern Biotech Fluoromount-G.

For cleared tissue, we utilized the Accu-OptiClearing strategy, a protocol based on the OPTIClear clearing solution.^30,31 The reagents used were as previously described.³¹ Adult wild-type zebrafish were euthanized with cold 1X Tricaine and decapitated. The heads were fixed in 4% PFA/1X PBS overnight at 4°C. Heads were washed 3 times in 1X PBS on a rotator at room temperature, then the whole brains, optic nerves, and retina were dissected out as intact pieces, submerged in 4% SDS OPTIClear, and incubated at 37°C for 2 days until the tissue appeared mostly transparent. Samples were then washed 3 times with 1X PBS on a rotator at room temperature and protected from light. Brains were mounted in OPTIClear (no SDS) overnight at 37°C in the dark, using an 18 Chambered Coverglass System (Cellvis, C18–1.5H).

Hybridization Chain Reaction RNA-Fluorescent In Situ Hybridization

Whole embryonic zebrafish (36 and 48 hpf) were fixed in 4% PFA/1X PBS overnight at 4°C. Embryos were washed 3 times for 5 minutes in sterile 1X PBS, then transferred to 100% MeOH overnight at −20°C. The embryos were rehydrated with a series of graded MeOH/PBST washes (75% MeOH, 50% MeOH, and 25% MeOH) followed by five washes in 1X PBST. Embryos were then processed according to the Molecular Instruments Zebrafish Embryos and Larvae Multiplexed hybridization chain reaction RNA-fluorescent in situ hybridization (HCR RNA-FISH) protocol version 3.0 revision 12, with probes designed by Molecular Instruments for atoh7 and pou4f2 zebrafish RNA coding sequences.^32–34 All solutions were made following the Molecular Instruments protocol.

Microscopy and Image Processing

Five dpf and adult cryosection images were taken on Zeiss AxioImagerZ.2, with the full adult retina sections stitched together using ZenBlue version 3.5 software. Adult flatmount samples, in vivo imaging, and HCR RNA-FISH whole mount samples were imaged on a Nikon Laser scanning confocal. Cleared adult brain samples were imaged on the Andor BC43 and Dragonfly 620SR (with an inverted Leica DMi8 microscope) spinning disk confocal microscopes (Andor; Oxford Instruments) at the Oxford Instruments Center for Advanced Microscopy - Electron Microscopy Core (OxCAM-EM, RRID:SCR_026315) at the Medical College of Wisconsin. Cleared brain images were acquired with Fusion software (Oxford Instruments) and processed using Imaris software (Oxford Instruments). Adult flatmount and brain images were processed in FiJi.³⁵ All images were processed using Adobe Photoshop. Cells from entire retinal sections were counted on one eye per fish (n = 5 or n = 3) at all time points analyzed.

Statistical Analyses

All data were analyzed on Graphpad Prism. Error bars on all figures represent mean ± standard deviation. Unpaired t-tests or 1-way ANOVA (analysis of variance) were run where appropriate (Prism) to determine statistical significance between experimental groups. P ≤ 0.05 was considered statistically significant.

Results

The atoh7:iCre Transgene Recapitulates Endogenous atoh7 Expression

To permanently label the zebrafish atoh7 lineage, transgenic zebrafish expressing Cre recombinase from the zebrafish atoh7 promoter were generated and crossed to the Cre inducible ubi:Switch transgenic line.²¹ In double transgenic offspring, cells containing Cre will switch their ubiquitous expression from eGFP to mCherry in a pattern consistent with atoh7 expression (Fig. 1; Supplementary Fig. S2). Previous analysis of atoh7 mRNA and transgene expression identified endogenous atoh7 expression begins at approximately 25 hpf in the ventral-nasal retina and spreads centrally/peripherally.^12,14 To confirm atoh7:iCre transgene expression follows the same dynamic expression pattern as endogenous atoh7, we performed time course confocal microscopy. As expected, mCherry expression was visible in the ventral-nasal retina at 29 hpf and spread centrally around the retina by 36 hpf (see Fig. 1B). At 48 hpf, mCherry+ cells reached the temporal retina and spread into the periphery (see Fig. 1B). Faithful atoh7 expression was also validated through comparison of the atoh7:iCre transgene with a newly generated atoh7:nd2GFP transgenic line, with Cre expression co-localizing with nd2GFP at 36 and 48 hpf (see Supplementary Fig. S2).

Figure 1.

Time course in vivo imaging demonstrates permanent recombination of the ubi:Switch transgene. (A) Schematic of the atoh7:iCre;cmlc:GFP and ubi:loxp-eGFP-STOP-loxp-mCherry-STOP transgene cross resulting in embryos which are ubi:loxp-mCherry-STOP positive. (B) In vivo z-stack and maximum intensity projection of atoh7:iCre;ubi:Switch F2 larvae eyes at 29, 36, and 48 hpf, respectively, demonstrating Cre recombination of ubi:Switch in a similar spatiotemporal pattern as endogenous atoh7 expression. The white arrow represents a cell with an active mitotic phenotype in the apical NBL of the developing retina. Scale bar = 50 µm. GCL, ganglion cell layer; INL, inner nuclear layer; PRL, photoreceptor layer; hpf, hours post-fertilization.

To validate atoh7:iCre expression was consistent and did not produce leaky Cre expression, we characterized three separate founders. Consistent recombination of the ubi:Switch transgene by all 3 founders was confirmed in 5 dpf larval retina sections (Supplementary Fig. S3). Each founder demonstrated atoh7:iCre;ubi:Switch+ retinas contained mCherry+ cells in each retinal layer (see Supplementary Fig. S3). ubi:Switch or atoh7:iCre single positive retinas showed no mCherry or eGFP within the retina, respectively, confirming successful recombination and no leaky mCherry expression without Cre recombination (Supplementary Fig. S4). Collectively, the atoh7:iCre transgene successfully recombined the ubi:Switch transgene in atoh7+ lineage cells within the developing zebrafish retina and showed a similar pattern to endogenous atoh7 expression.

Identification of atoh7 Positive and Negative RGC Lineages in Zebrafish

Lineage tracing analysis in mice found that approximately 55% of mature RGCs come from the atoh7+ lineage despite the loss of ≥95% of RGCs in atoh7 mutants.⁵ To investigate if zebrafish contain atoh7 lineage-positive and -negative populations of RGCs in wild-type retinas, we quantified the colocalization of Rbpms2+ cells, which labels RGCs in the ganglion cell layer (GCL), with atoh7 lineage cells (mCherry+) in the F2 generation of 3 atoh7:iCre founders.^36,37 At 5 dpf, the mCherry+ RGC population from each founder was 79% (±3.60%), 82% (±2.35%), and 87% (±3.25%), whereas the mCherry− RGC averages are 21% (±3.60%), 18% (±2.35%), and 13% (±3.25%), respectively (Fig. 2C; see Supplementary Fig. S3, Supplementary Table S1). Due to similar averages in mCherry+ and − RGCs for each founder we chose to analyze one for the remainder of the experiments (founder 1; see Fig. 2C; Supplementary Table S1).

Figure 2.

Five dpf and adult retinas show atoh7 lineage-positive and negative populations of RGCs. (A) Transverse sections (10 µm) of 5 dpf atoh7:iCre;ubi:Switch larval retina with Rbpms2+ RGCs and atoh7 lineage mCherry+ cells demonstrating the presence of atoh7 lineage+ (white arrows) and atoh7 lineage− RGCs (yellow arrows). (B) Transverse sections (10 µm) of 3 mpf F2 atoh7:iCre;ubi:Switch retina with Rbpms2+ RGCs and mCherry+ cells. Both atoh7 lineage+ (white arrows) and atoh7 lineage− RGCs can be visualized. (C) Bar graph demonstrating no significant difference in the average percent of both atoh7 lineage+ RGCs and atoh7 lineage− RGCs for 5 dpf (n = 5) and 3 mpf (n = 3) retinas (P = 0.1154, 2-tailed unpaired t-test). (D) Flatmount retina image from 3 mpf F2 atoh7:iCre;ubi:Switch retina focused on the GCL with Rbpms2+ RGCs and atoh7 lineage mCherry+ cells. Scale bars = 50 µm A, left and 20 µm A, right. dpf, days post-fertilization; mpf, months post-fertilization.

Because zebrafish eyes continue to grow throughout their life spans due to continued neurogenesis at the ciliary marginal zone (CMZ), we investigated if the proportion of atoh7+ versus atoh7− lineage RGCs remains consistent into adulthood. At 3 months post-fertilization (mpf), there was robust mCherry staining throughout all layers of the retina, similar to 5 dpf (Fig. 2B). In the GCL, 84% (±4.2%) of RGCs were mCherry+ and 16% (±4.2%) mCherry−, consistent with the percentages observed at 5 dpf (Fig. 2C). In addition to the RGC population, there are Rbpms2−/mCherry+ cells in the GCL at 5 dpf and 3 mpf, which we speculate are displaced amacrine cells (Figs. 2A, 2B, 2D). Collectively, we show atoh7 lineage-positive and -negative populations of RGCs are present as early as 5 dpf and their respective proportions remain consistent into adulthood.

The Zebrafish atoh7 Lineage Represents all Major Retinal Cell Types, Except Müller Glia

The mouse Atoh7 lineage includes all cell types derived from the multipotent RPC population, but it is unknown if this representation is conserved in the zebrafish atoh7 lineage. In addition, atoh7 mutant (lakritz) zebrafish retinas contain changes in the number or distribution of multiple cell types, including RGCs, amacrines, bipolars, and MG, but it is unclear if these changes are reflected in the atoh7 lineage population.⁶ To determine the overall retinal population of atoh7+ cells in wild type and atoh7 mutants, we assessed the mCherry+ and mCherry− populations using cell type specific markers.

As expected, atoh7 mutants maintain the presence of atoh7 mRNA but have a complete loss of RGCs as indicated by an absence of Rbpms2 staining at 36 hpf, 48 hpf, and 5 dpf (see Fig. 2A, Figs. 3A, 3C),⁶ despite only 79% of RGCs being mCherry+ in wild-type retinas (see Fig. 2C; Supplementary Table S1). Consistent with the lack of Rbpms2+ RGCs, pou4f2 mRNA, a marker of early RGC specification,^38–41 was undetectable in atoh7 mutant retinas at 36 and 48 hpf (Figs. 3B, 3D). Despite atoh7 mutants being devoid of RGC markers, we did see a consistent presence of mCherry+ cells within the GCL at 5 dpf, specifically in the dorsal region. To assess if these remaining mCherry+ cells are amacrine cells, we used Pax6, which labels amacrine cells and RGCs in the inner nuclear layer (INL) and GCL. These cells were Pax6+/Rbpms2−, suggesting they are displaced amacrine cells consistent with the previous analysis of atoh7 mutant retinas (see Fig. 2A, Fig. 4A).⁶ These data indicate both the atoh7+ and atoh7− RGC lineage populations are lost in atoh7 mutants.

Figure 3.

Early and late RGC markers are undetectable in atoh7 mutant zebrafish retinas. (A) Immunohistochemistry (IHC) staining of Rbpms2 and mCherry in atoh7:iCre;ubi:Switch atoh7+/− and atoh7−/− 36 hpf whole mount zebrafish larvae showing absence of Rbpms2 and RGCs at this timepoint in atoh7−/−. (B) HCR RNA-FISH of pou4f2 and atoh7 in atoh7+/− and atoh7−/− 36 hpf whole mount zebrafish larvae demonstrating lack of early RGC marker, pou4f2 mRNA, in the atoh7−/− retina. (C) IHC staining for Rbpms2 and mCherry in atoh7:iCre;ubi:Switch atoh7+/− and atoh7−/− at 48 hpf indicating complete absence of RGCs in atoh7−/− at this timepoint as well. (D) HCR RNA-FISH of pou4f2 and atoh7 in atoh7+/− and atoh7−/− retinas at 48 hpf showing no pou4f2 mRNA in atoh7−/− retinas. Scale bars = 50 µm. hpf, hours post-fertilization; D, dorsal; V, ventral; N, nasal; T, temporal.

Figure 4.

Amacrine cells, but not bipolar cells, have an increased atoh7 lineage in atoh7 mutants. (A) Transverse sections (10 µm) of wild type and atoh7−/− 5 dpf atoh7:iCre;ubi:Switch larval retina with Pax6 antibody-stained amacrine cells in the INL and RGCs/amacrines in the GCL. White arrows indicate Pax6+ amacrine INL cells which colocalize with atoh7 lineage/mCherry+, whereas yellow arrows show Pax6+ amacrine INL that are atoh7 lineage/mCherry−. (B) Bar graph demonstrating both atoh7 lineage+ and − Pax6+ amacrine cell populations in wild type (n = 5) and atoh7−/− (n = 5) backgrounds, showing a significant increase in the proportion of atoh7 lineage+ cells in atoh7−/− retinas compared to wild type (P = 0.0018, 2-tailed unpaired t-test). (C) Representative transverse section (10 µm) images of 5 dpf atoh7:iCre;ubi:Switch wild type and atoh7 mutant retinas stained with Prox1 to label bipolar cells. (D) Bar graph depicting percentage of atoh7 lineage+ and − Prox1+ bipolar cell populations within the wild type (n = 5) and atoh7 mutant (n = 5). Unpaired t-test analysis shows no significant difference in the atoh7 lineage+ Prox1+ bipolar cell populations between genotypes (P = 0.431, 2-tailed unpaired t-test). Scale bars = 50 µm A, left and 20 µm A, right. Dpf, days post-fertilization.

To determine the atoh7+ amacrine cell lineage in the INL, we counted Pax6+/mCherry+ and Pax6+/mCherry− cells in wild type and atoh7 mutants. In wild-type retinas, 40% (±2.2%) of Pax6+ INL located amacrine cells are mCherry+, whereas 60% (±2.2%) are mCherry− (see Figs. 4A, 4B; Supplementary Table S1). Interestingly, in the atoh7 mutants, 54% (±6.4%) of Pax6+ INL amacrine cells are mCherry+ and only 46% (±6.4%) are mCherry− (see Figs. 4A, 4B; Supplementary Table S1), representing a significant increase in the proportion of Pax6+/mCherry+ cells in atoh7 mutants compared to wild-type. However, this increase in the proportion of mCherry+ amacrine cells was not reflected in the total amacrine cell counts in wild type (149 ± 27.62) and atoh7 mutants (156 ± 29.38), indicating there was not an increase in the total INL amacrine cell population, but rather a shift from atoh7− to atoh7+ lineages (see Supplementary Table S1).

The atoh7+ bipolar lineage in mice is <0.1% and slightly increases in atoh7 mutants, accompanied by a decrease in the total bipolar cell population.^5,42 In the zebrafish retina the mCherry+ bipolar population is 8% ± 2.75%, marked by Prox1 and cellular location, a significantly greater proportion than mice (92% ± 2.75% were mCherry−). No change was observed in atoh7 mutants (9% ± 0.76% mCherry+ and 91% ± 0.76% mCherry−) despite the increase in the total bipolar cell population in atoh7 mutants (288 ± 8.96) compared with wild-type (134 ± 18.88; Figs. 4C, 4D; see Supplementary Table S1). These results are consistent with previous zebrafish atoh7 mutant characterizations, showing an increase in bipolar cells.⁶ Collectively, this shows that in the absence of atoh7, there is not a shift in Prox1+ bipolar cells born from the atoh7 lineage+ progenitor cell population and the wild type atoh7 lineage+ bipolar cell population in zebrafish is greater than in mice.

MG cells are another small population represented in the mouse atoh7+ lineage and differentially changed in atoh7 mutant mice versus zebrafish.^5–7,42 In addition to RGCs, Rbpms2 is lowly expressed in MG cells in the zebrafish retina, evidenced by the expression discrepancy between RGCs and MG and morphology of Rbpms2 labeled cells in the INL.^43,44 Cell counts of Rbpms2+ MG found that mCherry+ MG are extremely rare in wild type and atoh7 mutant retinas, with only 1 atoh7 lineage+ MG detected for each genotype (wild type; mCherry− 99.5% ± 1.7%, mCherry+ 0.5% ± 1.7%, atoh7 mutants; mCherry−, 99.4% ± 1.3% mCherry+, 0.6% ± 1.3%; Figs. 5A, 5B; see Supplementary Table S1). Although previous reports described an increase in MG cells in atoh7 mutant retinas, we did not observe an increase in total MG cells in atoh7 mutants (34 ± 2.28) compared to wild type (38 ± 7.3; see Supplementary Table S1).⁶ Both mice and zebrafish MG cells are lowly represented in the atoh7+ population and the increase in MG observed in atoh7 mutants are not from the atoh7+ lineage.

Figure 5.

Müller glia and horizontal cells mostly arise from an atoh7− RPC lineage. (A) Transverse sections (10 µm) of 5 dpf atoh7:iCre;ubi:Switch wild type and atoh7 mutant retinas stained with Rbpms2 and overexposed to visualize Müller glia cells which are colocalized with atoh7 lineage/mCherry+ and those which are atoh7 lineage/mCherry− (yellow arrows). (B) Percentage of atoh7 lineage+, Rbpms2+ Müller glia in both wild type (n = 5) and atoh7−/− (n = 5) show no significant difference in the proportion of atoh7+ Müller glia (P = 0.9842, 2-tailed unpaired t-test). (C) Five dpf atoh7:iCre;ubi:Switch wild type and atoh7 mutant retinas with DAPI nuclear stain to highlight cell morphology. White arrows indicate atoh7+ horizontal cells and yellow arrows indicate atoh7 lineage− cells. (D) Bar graph demonstrates atoh7 lineage+ and − average percentages of horizontal cell lineages are not significantly different in wild type (n = 5) compared with atoh7 (n = 5) retinas (P = 0.8526, 2-tailed unpaired t-test). Scale bars = 50 µm A, left and 20 µm A, right. dpf, days post-fertilization.

Next, we investigated the contribution of atoh7+ progenitors to the horizontal cell population within 5 dpf retinas in both wild type and atoh7 mutants. Utilizing DAPI to detect the distinct, oblong shape of horizontal cell nuclei at the apical portion of the INL, we identified 32% ± 10.8% of horizontal cells are born from the atoh7+ lineage, whereas 68% ± 10.8% are mCherry− within wild-type retinas (Figs. 5C, 5D; see Supplementary Table S1). Similar results were found in atoh7 mutants; 31% ± 6.2% of horizontal cells are mCherry+ and 69% ± 6.2% mCherry− (see Figs. 5C, 5D; Supplementary Table S1) with significant changes in the total number of horizontal cells in atoh7 mutant (45 ± 5.1) compared with wild type retinas (53 ± 2.95; see Supplementary Table S1). These results indicate there is no significant difference between atoh7 lineage+ horizontal cell populations at 5 dpf in both wild type and atoh7 mutants, although there is a significant decrease in total horizontal cell number in atoh7 mutants (see Figs. 5C, 5D; Supplementary Table S1).

In the rod dominant mouse retina, loss of Atoh7 results in increased cones and decreased rods, whereas in the cone rich zebrafish retina, cone numbers are reduced, which is similar to what occurs when ATOH7 is reduced or lost in human retinas.^7,42,45–47 To determine the atoh7+ photoreceptor lineage in zebrafish and how it changes upon loss of atoh7, we used WGA to label rods and Zpr1 for red and green double cones. In wild-type retinas, a majority of rods were mCherry− (77% ± 4.6%) compared to mCherry+ (23% ± 4.6%; Figs. 6A, 6B), whereas a large proportion of Zpr1+ cones came from the atoh7+ lineage (75% ± 3.2%) compared to the atoh7− lineage (25% ± 3.2%; Figs. 6C, 6D). Within atoh7 mutant retinas, neither the rod (21% ± 3.7%) or cone (76% ± 2.1%) atoh7+ lineage was significantly changed from wild type (see Fig. 6; Supplementary Table S1). Additionally, we did not observe any changes in the total number of cones in the atoh7 mutant retinas (121 ± 4.12, respectively) compared to wild type (112 ± 8.76, respectively) but did observe a statistically significant increase in the total number of rod photoreceptors in atoh7 mutants (90 ± 9.73) compared to wild type (70 ± 5.76; see Supplementary Table S1). Overall, the zebrafish atoh7+ photoreceptor lineages are higher than the reported percentages in mice.

Figure 6.

Most Zpr1+ cone photoreceptors are born from the atoh7 lineage, but few WGA+ rod photoreceptors are atoh7+. (A) Transverse sections (10 µm) of 5 dpf atoh7:iCre;ubi:Switch wild-type and atoh7−/− mutant retinas stained with WGA to identify the atoh7 lineage/mCherry+ (white arrows) and atoh7 lineage/mCherry− (yellow arrows) rod photoreceptor population. (B) Bar graph representing percentage of atoh7 lineage/mCherry+ (n = 5) and atoh7 lineage/mCherry− (n = 5) WGA+ rod photoreceptors within wild-type and atoh7−/− retinas, demonstrating no significant shift in the proportion of the atoh7 lineage in WGA+ rod photoreceptor cells (P = 0.3077, 2-tailed unpaired t-test). (C) Representative images of 5 dpf atoh7:iCre;ubi:Switch wild-type and atoh7 mutant transverse retinal sections stained for atoh7 lineage+ cells (mCherry) and red-green double cone photoreceptors (Zpr1). Atoh7+/Zpr1+ cone photoreceptors are identified by white arrows and the atoh7−/Zpr1+ cones are identified with yellow arrows. (D) Bar graph representing no significant difference in the proportion of atoh7 lineage/mCherry+ and atoh7 lineage/mCherry− populations of Zpr1+ photoreceptors between wild-type (n = 5) and lakritz (n = 5, P = 0.6938, 2-tailed unpaired t-test). Scale bars = 50 µm A, left and 20 µm A, right. dpf, days post-fertilization.

Atoh7 Lineage Cells Outside the Zebrafish Retina

Atoh7+ cells and projections have previously been visualized outside of the retina,^{5,7,16,47–50} so to thoroughly investigate the potential use of the atoh7:iCre transgene we analyzed cleared and immunostained adult wild type and atoh7 mutant brains at 3 months. As expected, we visualized cytoplasmic mCherry expression throughout the retina, optic nerve, optic chiasm, optic tract, and nerve fibers superficially traveling across the optic tectum (Figs. 7A, 7C). Outside the visual system, we detected small populations of mCherry+ (atoh7+ lineage) cells within the deep, central forebrain (Fig. 7B) and observed mCherry+ nuclei within the superficial ventral hindbrain (Fig. 7D). These regions of mCherry+ cells are likely part of the auditory or olfactory processing centers, as atoh7 has been shown to be expressed in the central auditory system of mice and olfactory placode in Xenopus.^48–50 These results demonstrate that our atoh7:iCre transgenic line alone or in combination with ubi:Switch can be used as a tool to investigate a variety of sensory system defects as they relate to the atoh7 lineage or different mutations.

Figure 7.

The atoh7:iCre transgene allows for identification of atoh7 lineage+ cells in the central nervous system beyond the retina. (A) Frontal view of an atoh7:iCre;ubi:Switch adult (3 mpf) brain stained with a GFP antibody and mCherry antibody to enhance the signal of the ubi:Switch transgene and the atoh7 lineage. (B) Transverse single plane maximum intensity projection within the middle forebrain region showing clusters of atoh7 lineage/mCherry+ cells bilaterally located within the deep, medial forebrain. (C) Ventral view of a whole mount atoh7:iCre;ubi:Switch cleared adult (3 mpf) brain stained with GFP and mCherry antibodies to enhance the signal of transgene expression. (D) Maximum intensity projection of hindbrain region showing populations of atoh7 lineage/mCherry+ cells present at the superficial, bilateral midline (white arrows). ON, optic nerve; Ch, optic chiasm; OT, optic tract; OTect, optic tectum. Scale bars = 300 µm A and 200 µm C. mpf, months post-fertilization.

Discussion

In combination with the ubi:Switch transgenic line, our atoh7:iCre transgenic zebrafish successfully labeled the atoh7+ RPC population, allowing us to trace the lineage of transient atoh7 expressing progenitors at time points when endogenous atoh7 is no longer expressed. In the retina, we observed the presence of atoh7 lineage+ and atoh7 lineage− populations of all major cell types, excluding MG. Although presence of each retinal cell type within the atoh7 lineage is consistent with the lineage tracing in mice, the atoh7+ percentages differ, except for horizontal cells and MG.^5,42 The differences observed between species may be due to the basic biology of each of these organisms. Mice are nocturnal and rely on navigating dimly lit environments and, thus, have a rod-dominant retina.⁵¹ Zebrafish are diurnal and rely heavily on color vision to catch prey with a more cone-dominant retina.⁵² The difference in photoreceptor cell composition is highlighted developmentally by their early production, with zebrafish cone photoreceptor precursors and RGCs being produced as daughter cells in the same mitotic division during the early stages of neurogenesis, a phenomenon not observed in mice.^5,53 This may explain why the proportion of atoh7 lineage+ RGCs and cones were very similar (79% and 75%, respectively). Outside of biology, technical differences such as Cre efficiency or transgene mosaicism, especially in fast- versus slower-developing organisms, may account for the atoh7+ lineage differences between mice and zebrafish, although our ability to detect Cre and mCherry expression close to the onset of atoh7 expression suggests developmental timing may not be an issue.

It was previously unknown if zebrafish contained an atoh7 lineage− RGC population due to the lack of a long-term lineage trace paradigm. Our interest in unveiling the zebrafish atoh7 RGC lineage is due to the requirement of Atoh7 for the genesis and survival of a greater number of RGCs than is represented in the Atoh7 lineage in mice, which creates two interesting questions: (1) how do Atoh7+ cells influence atoh7 lineage− RGCs, and (2) what transcription factors and/or signaling pathways are responsible for atoh7 lineage− RGC genesis and survival. The data presented here reveals zebrafish contain an atoh7 lineage− RGC population providing support for the use of zebrafish to study the intricate mechanisms important for nonautonomous RGC genesis and survival. Based on the lack of detectable Rbpms2 and pou4f2 expression in atoh7−/− fish, RGCs may never be born, unlike in mice, where RGC-specific mRNA and protein are still detectable in Atoh7 mutant retinas.^29,54 It is possible the atoh7+ RGC lineage is higher than we detected with our atoh7:iCre transgene or the rapid development of zebrafish may make it difficult to detect short lived RGCs in atoh7 mutants using in situ and immunohistologic techniques. Each possibility warrants future studies to validate the absence of immature RGCs in atoh7 mutant zebrafish and the contribution of atoh7 lineage+ cells on RGC genesis and survival. In addition, our data demonstrate there was no change in the proportion of atoh7 lineage+ RGCs in both the larval and adult retina, showing that as new cells are made at the CMZ, the atoh7+ tRPCs generate RGCs in the same proportion as during development.

The atoh7 lineage trace comparison between wild type and atoh7 mutants validated previous results and revealed new information (Fig. 8). In atoh7 mutants, the proportions of atoh7 lineage+ cells for each cell type were not significantly changed, except in the Pax6+ amacrine cell population, which contains significantly more cells born from the atoh7 lineage. Despite the increase in Pax6+/atoh7 lineage+ amacrine cells, the total number of Pax6+ amacrine cells within the INL did not change, consistent with previous findings in atoh7 mutants (see Figs. 4A, 4B; Supplementary Table S1).⁶ Our observation of an increased atoh7 lineage+ amacrine cell population could be due to the chosen founder line, but because the same founder was bred into both the wild type and atoh7 mutant background the increase in amacrine cells from the atoh7+ lineage is more likely to be a cell fate change rather than a technical artifact. The atoh7 mutant cell counts also confirmed an increase in bipolar cells,⁶ although the increase was not associated with the atoh7+ lineage, suggesting the additional bipolar cells arise from an atoh7− progenitor population. Our total MG cell count data in atoh7 mutants was not consistent with previous reports but the labeling method used differed possibly leading to this difference (see Supplementary Table S1).⁶ Together, these results support the hypothesis that there is a fate switch of RPCs in atoh7 mutants to later born cell types without the presence of Atoh7 protein to specify the RGC fate.^5,6,42

Figure 8.

The atoh7 lineage of each major cell type in the mature wild type zebrafish retina. A diagram depicting 5 dpf zebrafish retinal cell types and the respective proportions of each that are from the atoh7 lineage in the retina of wild type and atoh7 mutants. Quantified by colocalization quantifications of cell-specific stains and mCherry expression from the lineage tracing transgenes. G = retinal ganglion cell; A = amacrine cell; B = bipolar cell, M = Müller glia cell; H = horizontal cell; R = rod photoreceptor cell; C = cone photoreceptor cell. dpf, days post-fertilization.

The influence of Atoh7 function and its lineage outside the eye in zebrafish has been limited due to the extra care needed to achieve adult survival of atoh7 mutants. The lone publication investigating adult atoh7 mutants reported a decrease in total brain area due to the loss of optic tract input to the optic tectum with no significant difference detected in the forebrain and hindbrain.⁴⁵ Our cleared, whole brain imaging of adult atoh7:iCre;ubi:Switch fish confirmed a high concentration of atoh7 lineage+ axons in the optic tectum and identified novel populations of atoh7+ cells located within the deep, central forebrain, and superficial, ventral hindbrain (see Figs. 7A–D). The expression of atoh7 we observed in the developing olfactory bulb and adult brain complements previous reports of atoh7 being linked to the olfactory and auditory systems within Xenopus and mice.^48–50 The small populations of mCherry+ cells in the forebrain and hindbrain are either not dependent on atoh7 expression for their genesis or too small a population to affect changes to brain size in atoh7 mutants.

Once championed as mainly a developmental model, the use of zebrafish to study adult-onset human disease and regeneration of adult tissue has significantly increased over time. This change has prompted the need to extend our knowledge of zebrafish development, including the influence of the developmentally expressed transcription factor, Atoh7. The generation of the atoh7:iCre lineage reporter, which recapitulates the timing of embryonic atoh7 expression and can be detected into adulthood, confirmed previous reports and revealed new information. The existence of atoh7 lineage-positive and -negative populations for each major retinal cell type (except for MG), including RGCs, provides opportunities to further explore RGC-specific genesis and survival mechanisms, and highlights the atoh7:iCre line as a unique tool to be utilized for lineage specific questions or targeted gene manipulations within the retina or CNS.