Gsx2, but not Gsx1, is necessary for early forebrain patterning and long-term survival in zebrafish

Abstract

Background:

Homeobox transcription factor encoding genes, genomic screen homeobox 1 and 2 (gsx1 and gsx2), are expressed during neurodevelopment in multiple vertebrates. However, we have limited knowledge of the dynamic expression of these genes through developmental time and the gene networks that they regulate in zebrafish.

Results:

We confirmed that gsx1 is expressed initially in the hindbrain and diencephalon and later in the optic tectum, pretectum, and cerebellar plate. gsx2 is expressed in the early telencephalon and later in the pallium and olfactory bulb. gsx1 and gsx2 are co-expressed in the hypothalamus, preoptic area, and hindbrain, however, rarely co-localize in the same cells. gsx1 and gsx2 mutant zebrafish were made with TALENs. gsx1 mutants exhibit stunted growth, however, they survive to adulthood and are fertile. gsx2 mutants experience swim bladder inflation failure that prevents survival. We also observed significantly reduced expression of multiple forebrain patterning distal-less homeobox genes in mutants, and expression of foxp2 was not significantly affected.

Conclusions:

This work provides novel tools with which other target genes and functions of Gsx1 and Gsx2 can be characterized across the central nervous system to better understand the unique and overlapping roles of these highly conserved transcription factors.

1 |. INTRODUCTION

Central nervous system (CNS) development is a complex process wherein regionally expressed transcription factors contribute significantly in determining initial neuronal cell identity, connectivity, and function.^1–3 Transcription factors act coordinately to activate or repress target gene expression in progenitor cell domains.^4,5 Differential gene expression among neural progenitors generates distinct cell types and specifies neuronal properties, such as cell neurotransmitter content as seen in mouse,^6–8 chicken,⁹ and zebrafish.^10–12 This process ultimately imparts initial identity to mature neuronal cells and forms the basis for neural circuit assembly and function. Thus, defining the spatiotemporal expression patterns and essential roles of vertebrate transcription factors is important for elucidating the molecular mechanisms governing neurodevelopment. More importantly, these studies can provide fundamental insights about molecular genetic contributions to the diverse neuroanatomical and behavioral phenotypes that are associated with neurodevelopmental disorders (NDDs).

Genomic screen homeobox 1 and 2 (gsx1 and gsx2, previously gsh1 and gsh2) are closely related genes encoding homeobox transcription factors expressed in the CNS that were discovered in a screen for novel, nonclustered homeobox genes in mouse.¹³ Homeobox genes characteristically encode transcription factors with a conserved 60-amino acid DNA-binding homeodomain.^14,15 Genes such as the hox genes specify cell types and body structures along the anterior-posterior axis in many species in patterns collinear with their 5^’ to 3^’ chromosomal positions within gene clusters.^15–18 As nonclustered and pseudo-clustered genes, gsx1 and gsx2 encode homeodomains with high (>80%) similarity to the hox genes.^4,5,19 gsx1 and gsx2 are the vertebrate homologs of Drosophila melanogaster intermediate neuroblasts defective (ind). ind and the gsx genes similarly regulate dorsoventral (DV) patterning,^20–22 and Ind and murine GSX2 elicit similar regulatory outcomes based on monomer vs homodimer DNA binding.²³ Interestingly, ind and the gsx genes are expressed in similar patterns in the fly neuroectoderm,²⁰ mouse neural tube,²⁴ and Xenopus neural plate,²⁵ supporting models for conserved neuroaxis domain specification across species.

Expression of gsx1 and gsx2 has been described in several vertebrates in varied detail. gsx1 expression patterns are highly conserved across species, beginning in the hindbrain during somitogenesis in mouse,²⁴ Xenopus,²⁵ medaka,²⁶ and zebrafish.²⁷ During early embryonic stages in mouse Gsx1 is expressed in the diencephalon and telencephalon and expands to the hypothalamus, thalamus, optic stalk, medulla, pons, and cerebellum.²⁴ Early expression in Xenopus, medaka,²⁶ and zebrafish^27,28 occurs in similar regions such as the hypothalamus, olfactory bulb, optic tectum, and cerebellum. gsx1 is also expressed as two dorsolateral stripes in the hindbrain and in the intermediate spinal cord in mice,²⁴ medaka,²⁶ and zebrafish.¹⁰ gsx2 is first detected slightly later than gsx1 in the telencephalon and mesencephalon in mice and in the hindbrain in Xenopus.^25,29 Throughout neurodevelopment gsx2 is expressed in the telencephalon, thalamus, hypothalamus, and cerebellum in mouse,²⁹ Xenopus,²⁵ and zebrafish.³⁰ Like gsx1, expression of gsx2 appears similarly across species as two dorsolateral stripes in the hindbrain. Gsx2 is expressed dorsal to Gsx1 in the hindbrain in Xenopus²⁵ and in the spinal cord of zebrafish,¹⁰ consistent with their roles in DV patterning. Outside of the aforementioned reports,^10,27,28,30 expression of zebrafish gsx1 and gsx2 has not been comprehensively characterized and compared across all embryonic and early larval stages. Here, we capitalize on the zebrafish model, which allows whole brain examination of expression to rigorously define the gsx1 and gsx2 expression profile. Defining a more complete expression profile of the gsx genes in zebrafish is an important step forward in elucidating critical Gsx1 and Gsx2 functions.

GSX1 and GSX2 promote regional neuronal identity in the ventral telencephalon and regulate the development of cortical, striatal, and olfactory bulb interneurons in mice.^21,31–40 Despite having similar roles in progenitor specification, GSX1 and GSX2 differentially regulate progenitor maturation; Gsx2 maintains progenitors in an undifferentiated state while Gsx1 promotes maturation by down-regulating Gsx2.^22,34 Gsx1 is implicated in hypothalamic and pituitary development, as knockout (KO) mice display a dwarf phenotype, reduced pituitary size, hormonal imbalances, and only survive a few weeks post-birth.⁴¹ Consistently, Gsx1 specifies multiple types of neuropeptidergic neurons in the arcuate nucleus of the hypothalamus.⁴² Gsx2 mouse KOs do not survive more than 1 day following birth, exhibit disturbed forebrain and hindbrain morphology,⁴³ and have expanded Gsx1 expression in the ventral telencephalon.⁴⁴ Interestingly, Gsx1 and Gsx2 double KO mice display more severe forebrain phenotypes than Gsx2 single KOs, supporting a model in which GSX1 partially compensates for loss of GSX2 function.^21,44,45 While much is known in the mouse forebrain, many key neurodevelopmental roles for gsx1 and gsx2 remain understudied, and roles for these transcription factors across the CNS have yet to be fully characterized in any vertebrate.

Outside of the forebrain, limited functional roles are reported for GSX1 and GSX2 in mammalian and non-mammalian model systems. Gsx1 regulates an identity switch in mouse cerebellar neuronal progenitors in part through BMP/SMAD signaling.^46,47 Through a Notch signaling dependent mechanism, Gsx1 and Gsx2 regulate the temporal specification of glutamatergic and GABAergic interneurons in the mouse spinal cord.⁷ In this region, Gsx1 also promotes neural stem and progenitor cell generation and decreases reactive glial scar formation to facilitate recovery from injury.⁴⁸ gsx1 is a molecular marker of glutamatergic neurons in the dorsal brainstem in zebrafish that regulate the acoustic startle response, and zebrafish with ablated gsx1-expressing neurons and mouse Gsx1 KOs similarly exhibit disrupted responsiveness to paired pulse acoustic-vibrational stimuli.^28,49 In zebrafish, gsx2 is required for specification of neurons in the inferior olivary nuclei of the medulla.³⁰ gsx1 and gsx2 mark specific progenitor domains in the spinal cord of transgenic zebrafish similarly to mouse, with gsx1 domains specifying glutamatergic, GABAergic, and glycinergic fates, and gsx2 domains specifying glutamatergic fates only.¹⁰

Some GSX1 and GSX2 transcriptional target genes have been reported in the mouse forebrain and other brain regions.^{24,29,45,50,51} However, target gene regulation by Gsx1 and Gsx2 across many brain regions has been understudied across vertebrates, including zebrafish. Several zebrafish orthologs for mouse GSX1 and GSX2 target genes exist, one example being Distal-less homeobox 2 (Dlx2). Two paralogs, dlx2a and dlx2b, are found in the zebrafish genome, with dlx2a predicted to be the ortholog of mammalian Dlx2.⁵² In mouse, GSX2 promotes Dlx2 expression in the ventral telencephalon, while DLX2 in turn represses Gsx1 and Gsx2,⁵¹ and collectively this promotes ventral identity and mediates proliferative characteristics. Removal of Gsx1 or Gsx2 from a Dlx1 and Dlx2 double mutant background rescues some phenotypes observed, demonstrating that GSX/DLX inter-regulation is required for appropriate forebrain patterning. Two other Dlx genes, Dlx5 and Dlx6, are similarly expressed in the forebrain of mice and zebrafish.^52,53 The Dlx genes coordinately regulate patterning of inhibitory neurons in the forebrain,^52–54 and importantly, the DLX, FOX, and other families of forebrain transcription factor encoding genes are implicated in aberrant neuronal signaling observed in patients with various NDDs.^1,40,55–58 As such, it is important to investigate putative target genes for Gsx1 and Gsx2 to better understand their roles across brain regions during vertebrate neurodevelopment. In fact, the zebrafish model provides a tool with which this can be done readily and from the earliest neurodevelopmental time point possible.

In this study, we comprehensively resolve the neurodevelopmental expression of gsx1 and gsx2 in the zebrafish CNS from early embryonic to late larval stages. Using gsx1 and gsx2 zebrafish mutants made using TALENs, we also demonstrate that dlx2a, dlx2b, dlx5a, and dlx6a are differentially regulated by Gsx1 and Gsx2. We further demonstrate that forkhead box P2 (foxp2), a gene that is expressed in the mammalian and zebrafish CNS^59,60 and is implicated in language deficits,⁵⁸ is not significantly regulated by Gsx1 and Gsx2 in the zebrafish telencephalon. These studies establish and validate novel tools for investigating Gsx1 and Gsx2 function during neurodevelopment and beyond in zebrafish across CNS regions.

2 |. RESULTS

2.1 |. gsx1 and gsx2 expression in zebrafish embryos and larvae

To assess similarity between the Gsx1 and Gsx2 protein sequences in zebrafish, mouse, and human, we used a bioinformatics approach. We found that zebrafish Gsx1 shares 57/60 (95%) amino acids in the homeodomain with human and mouse GSX1 (Figure 1A). Zebrafish Gsx1 also shares 57/60 (93%) amino acids in the homeodomain with zebrafish, mouse, and human Gsx2. Interestingly, the homeodomain sequence is 100% identical between zebrafish, mouse, and human Gsx2. A rooted phylogenetic tree containing published Gsx1 and Gsx2 protein sequences reveals that zebrafish Gsx1 and Gsx2 cluster with their mammalian orthologs and also displays evolutionary divergence from Drosophila ortholog Ind (Figure 1B).

FIGURE 1.

Gsx1 and Gsx2 are conserved across vertebrates and expressed in zebrafish from embryonic stages through adulthood. (A) Amino acid sequence alignment of zebrafish, mouse, and human Gsx1 and Gsx2. Dr, Danio rerio; Mm, Mus musculus; Hs, Homo sapiens. Identical amino acids are shaded in black, similar amino acids are shaded in gray, and the region encoding the conserved DNA-binding homeodomain is underlined in orange. (B) Rooted phylogenetic tree displaying the clustered relationship of published Gsx1 and Gsx2 protein sequences as well as a divergence from the Drosophila ortholog Ind. Values represent distance scores. Dm, Drosophila melanogaster; Lo, Lepisosteus oculatus; Ol, Oryzias latipes; Xt, Xenopus tropicalis. (C) Schematic of the zebrafish gsx1 and gsx2 gene bodies. Blue boxes represent exons, black lines represent introns, and orange boxes represent the region encoding the homeodomain. Red arrowheads represent RT-PCR primer annealing sites. (D) Agarose gels showing full-length cDNA transcripts of zebrafish gsx1 (top), gsx2 (middle), and ef1a (bottom) at specific ages. Red text indicates ladder sizes. Upper bands in the gsx1 gel image represents trace genomic DNA that includes the 89 bp intron

To document the neurodevelopmental time-course of gsx1 and gsx2 expression in zebrafish, we extracted total RNA from zebrafish embryos and larvae for use in RT-PCR (Figure 1C,D). gsx1 expression was identified at 10 hours postfertilization (hpf), consistent with a previous report,²⁷ and persisted through our latest time point tested, 120 hpf (Figure 1D). Expression of zebrafish gsx2 was first detected at 12 hpf and also persisted through 120 hpf. Interestingly, gsx1 expression was observed at 3.5 hpf, suggestive of maternal contributions of gsx1 to early embryonic development. However, analysis of maternal zygotic gsx1 TALEN-generated mutants obtained through in vitro fertilization revealed that gsx1 is not an essential maternal factor as maternal zygotic mutants are indistinguishable from zygotic mutants and develop identically (data provided by request). From these results, we can confirm that zebrafish gsx1 and gsx2 are expressed in zebrafish from embryonic to larval stages, suggesting an importance in early and later brain development and function.

2.2 |. Expression of gsx1 and gsx2 during early development is complementary yet distinct

Known expression of gsx1 in zebrafish is limited to select CNS regions and ages,^10,27 and expression of zebrafish gsx2 is minimally reported from 48 to 72 hpf.³⁰ Outside of a transgenic analysis documenting gsx1 and gsx2 expression together in the 36 to 48 hpf spinal cord,¹⁰ expression of gsx1 and gsx2 during neurodevelopment in zebrafish has not been comprehensively analyzed. We first used whole-mount in situ hybridization (WISH) to characterize and compare gsx1 and gsx2 expression in zebrafish embryos and larvae. Consistent with RT-PCR results, gsx2 expression was detected at 12 hpf in the presumptive forebrain in the anterior neural plate (Figure 2H). At 12 hpf gsx1 is expressed in the presumptive hindbrain in rhombomere 3 (Figure 2A), consistent with a previous report.²⁷ From 16 to 24 hpf gsx2 expression is present in the diencephalon and telencephalon (Figure 2I,J), with 24 hpf marking the first appearance of gsx2 expression in the caudal hindbrain (Figure 2K). Conversely, gsx1 is expressed across the rostral to caudal extent of the hindbrain at 24 hpf (Figure 2D). Expression of gsx2 in the zebrafish spinal cord is seen clearly in transgenic reporter lines in early embryos,¹⁰ but we had difficulty detecting it by WISH. From 16 to 24 hpf gsx1 expression is observed in the forebrain, midbrain, hindbrain, and spinal cord (Figure 2B,C), consistent with other reports. gsx2 expression persists in the diencephalon, telencephalon, hindbrain, and spinal cord through 30 hpf (Figure 2L,M), and at this age gsx1 is expressed in the diencephalon, midbrain, hindbrain, and spinal cord (Figure 2E,F). Dorsal views at this age reveal that gsx1 and gsx2 exist in two dorsolateral columns in the hindbrain (Figure 2O,S).

FIGURE 2.

Expression of gsx1 and gsx2 in embryonic zebrafish is dynamic and unique. (A-G) Lateral mounts showing expression of gsx1 from 12 to 48 hpf. (H-N) Lateral mounts showing expression of gsx2 from 12 to 48 hpf. (O-R) Dorsal (O and Q) and ventral (P and R) mounts showing expression of gsx1 from 30 to 48 hpf. (S-V) Dorsal (S and U) and ventral (T and V) mounts showing expression of gsx2 from 30 to 48 hpf. A, B, H, and I are dissecting scope images and scale bar represents 500 μm. Remaining images are compound scope images taken at ×20 with samples mounted under cover glass and anterior facing left, eyes removed in lateral views. Scale bars represent 100 μm. CeP, cerebellar plate; Di, diencephalon; Fb, forebrain; Hb, hindbrain; Hyp, hypothalamus; Mb, midbrain; OB, olfactory bulb; P, pallium; Po, preoptic area; Pr, pretectum; Sc, spinal cord; Tel, telencephalon; TeO, optic tectum

Our observed gsx1 expression through 30 hpf confirms previous findings²⁷; however, we continued characterizing gsx1 expression through late embryonic and larval development along with gsx2. By 48 hpf, gsx2 expression is restricted to the olfactory bulb, preoptic area, hypothalamus, pallium, and hindbrain (Figure 2N,U,V). At this age, gsx1 expression is seen in the preoptic region, hypothalamus, pretectum, optic tectum, cerebellar plate, hindbrain, and spinal cord (Figure 2G,Q,R). At 72 hpf, gsx2 expression is faintly present in the pallium and olfactory bulb (Figure 3J,K), while gsx1 expression persists in the pretectum, optic tectum, hypothalamus, and hindbrain (Figure 3A–C). Expression of gsx2 through 4 to 5 days postfertilization (dpf) persists faintly in the pallium and hindbrain (Figure 3L–O); however, gsx1 expression strongly persists in the pretectum, optic tectum, hypothalamus, and hindbrain (Figure 3D–I). Collectively, these WISH analyses provide insight to the dynamic expression patterns of gsx1 and gsx2 during embryonic and larval stages in zebrafish.

FIGURE 3.

Expression of gsx1 and gsx2 is restricted in late embryonic and early larval stages. (A-I) Expression of gsx1 from 72 to 120 hpf. Leftmost column are dorsal views, middle column are ventral views, and rightmost column are lateral views. (J-O) Expression of gsx2 from 72 to 120 hpf. Left column are dorsal views and rightmost column are lateral views. All images are compound scope images taken at ×20 with samples mounted under cover glass and anterior facing left, eyes removed in lateral views. Scale bars represent 100 μm. Insets are whole brain dissections at the same age mounted dorsally. Hb, hindbrain; Hyp, hypothalamus; OB, olfactory bulb; P, pallium; Pr, pretectum; TeO, optic tectum

2.3 |. Co-localization of gsx1 and gsx2 is minimal

To be more precise in assessing co-localization of gsx1 and gsx2 in cells, we turned to fluorescence in situ hybridization (FISH) at embryonic and larval stages. At 24 hpf, gsx1 and gsx2 are regionally co-expressed at the border of the dorsal diencephalon and ventral telencephalon (Figure 4A^i-iii and 4B^i-iii; max z-projections); however, they very minimally co-localize in the same cells (insets in Figure 4Aⁱⁱⁱ and 4Bⁱⁱⁱ; single z-stack plane). At this age, gsx1 and gsx2 are also regionally co-expressed in the hindbrain, with gsx2 expressed dorsal to gsx1 (Figure 4A^iv-vi and 4B^iv-vi). At 30 hpf, gsx1 and gsx2 are regionally co-expressed in the ventral diencephalon (Figure 4C^i-iii and 4D^i-iii; max z-projections), however, rarely co-localize in the same cells (insets in Figure 4Cⁱⁱⁱ and 4Dⁱⁱⁱ; single z-stack plane). In the hindbrain at this age gsx1 and gsx2 remain segregated dorsoventrally (Figure 4C^iv-vi and 4D^iv-vi; max z-projections) and rarely co-localize in the same cells (inset in Figure 4D^vi; single z-stack plane).

FIGURE 4.

Fluorescence in situ hybridization confirms minimal co-localization of gsx1 and gsx2 during embryonic development. (Aⁱ-A^vi) Lateral views of gsx1 and gsx2 expression at 24 hpf. (Bⁱ-Bⁱⁱⁱ) Ventral view of gsx1 and gsx2 expression at 24 hpf. (B^iv-B^vi) Cross-sectional view taken at the dashed line in A^vi. (Cⁱ-C^vi) Lateral views of gsx1 and gsx2 expression at 30 hpf. (Dⁱ-Dⁱⁱⁱ) Ventral view of gsx1 and gsx2 expression at 30 hpf. (D^iv-D^vi) Cross-sectional view taken at the dashed line in C^vi. Lateral views were taken at ×20 and ventral/Cross-sectional views were taken at ×40, all with anterior facing left. All were pseudocolored using FIJI ImageJ and scale bars represent 100 μm. For lateral views, eyes were dissected off. Main images are max z-projections and insets are single z-stack slices zoomed into the boxed region shown in the main image. Di, diencephalon; Hb, hindbrain; Mb, midbrain; Sc, spinal cord; Tel, telencephalon

By 48 hpf, gsx1 and gsx2 are regionally co-expressed in the hypothalamus and preoptic area (Figure 5A^i-iii and 5B^i-iii; max z-projections); however, rarely co-localize in the same cells (insets in Figure 5Aⁱⁱⁱ and 5Bⁱⁱⁱ; single z-stack plane). Distinct segregation of gsx1 ventrally and gsx2 dorsally in the hindbrain is still apparent at 48 hpf (Figure 5A^iv-vi and 5B^iv-vi; max z-projections); however, they rarely co-localize in the same cells (inset in Figure 5A^vi; single z-stack plane). Interestingly, by this age gsx1 expression appears to extend ventrally while gsx2 expression remains isolated dorsally (Figure 5A^iv-vi and 5B^iv-vi). This finding is reminiscent of reported roles for Gsx2 and Gsx1 to regulate neuronal progenitor proliferation vs differentiation, respectively,²² and we believe this ventral extension represents the outgrowth of projections from maturing neuronal progenitors. At 72 hpf, regional co-expression of gsx1 and gsx2 is restricted to the preoptic area (Figure 5C^i-vi; max z-projections), however again, co-localization in the same cells is minimal (Figure 5Cⁱⁱⁱ; single z-stack plane).

FIGURE 5.

Fluorescence in situ hybridization confirms minimal co-localization of gsx1 and gsx2 during late embryonic and early larval development. (Aⁱ-A^vi) Lateral views showing expression of gsx1 and gsx2 at 48 hpf. (Bⁱ-Bⁱⁱ) Ventral views showing expression of gsx1 and gsx2 at 48 hpf. (B^iv-B^vi) Cross-sectional view taken at the dashed line in A^vi. (Cⁱ-Cⁱⁱⁱ) Lateral view showing gsx1 and gsx2 expression at 72 hpf. (C^iv-C^vi) Ventral view showing gsx1 and gsx2 expression at 72 hpf. Lateral views were taken at ×20 and ventral/Cross-sectional views were taken at ×40, all with anterior facing left. All were pseudocolored using FIJI ImageJ and scale bars represent 100 μm. For lateral views, eyes were dissected off. Main images are max z-projections and insets are single z-stack slices zoomed into the boxed region shown in the main image. CeP, cerebellar plate; Di, diencephalon; Hb, hindbrain; Hyp, hypothalamus; Mb, midbrain; OB, olfactory bulb; P, pallium; Po, preoptic area; Pr, pretectum; Sc, spinal cord; Tel, telencephalon; TeO, optic tectum

In brains dissected from 6 dpf larvae, we observed gsx2 expression patterns through FISH analysis that were not directly apparent through colorimetric WISH. gsx1 expression appears in regions reminiscent of our WISH analyses including the pretectum, hypothalamus, optic tectum, preoptic area, cerebellar plate, and hindbrain (Figure 6A–F, max z-projections). Dorsal views of the FISH stained brain confirmed that gsx2 is expressed in the pallium at 6 dpf (Figure 6AC), and it also revealed distinct expression of gsx2 in the hindbrain that was not clearly observed through WISH. Additionally, ventral views of the brain at 6 dpf also revealed that gsx2 is regionally co-expressed with gsx1 in the hypothalamus (Figure 6D–F); however, they rarely co-localize in the same cells (Figure 6F; single z-stack plane). Combined, our WISH analyses reveal that gsx1 and gsx2 expression is dynamic throughout neurodevelopment, and FISH demonstrates for the first time that they largely exist in distinct cellular populations.

FIGURE 6.

gsx1 and gsx2 are expressed through 6 dpf in the brain. (A-C) Dorsal views showing expression of gsx1 and gsx2 at 6 dpf. (D-F) Ventral views showing expression of gsx1 and gsx2 at 6 dpf. Images taken at ×20 with anterior facing left. All were pseudocolored using FIJI ImageJ and scale bars represent 100 μm. Main images are max z-projections and insets are single z-stack slices zoomed into the boxed region shown in the main image. Hb, hindbrain; Hyp, hypothalamus; OB, olfactory bulb; P, pallium; Po, preoptic area; Pr, pretectum; Tel, telencephalon; TeO, optic tectum

2.4 |. gsx1 and gsx2 TALEN mutants exhibit unique phenotypes

There is limited knowledge about how Gsx1 and Gsx2 function across several developing brain regions where they are expressed in vertebrates and that we report by WISH and FISH in zebrafish. In mouse, loss of Gsx1 leads to abnormal hypothalamic-pituitary signaling,⁴¹ and mutations in Gsx2 leads to disturbed forebrain morphology.⁴³ To further examine the roles of gsx1 and gsx2 in neurodevelopment, we generated zebrafish mutants using TALENs (Figure 7A). For gsx1, we generated alleles with an 11 base-pair (bp) deletion (y689) and a 5 bp deletion (y690). For gsx2, we generated alleles with a 13 bp deletion (y691) and a 5 bp deletion (y692). All mutations occur in the first exon of the zebrafish gsx1 and gsx2 genes and should result in premature stop codons and immature transcripts lacking the homeobox DNA binding domain encoding region.

FIGURE 7.

gsx1 TALEN mutants experience stunted growth and gsx2 TALEN mutants experience swim bladder inflation failure. (A) Schematic of the gsx1 and gsx2 gene bodies and targeted TALEN mutation site (red *). All mutations should result in a premature stop codon. Text color corresponds with gene body structure (blue = exon, black = intron, orange = homeodomain, red = mutant sequence). Inset gel shows restriction digest of gsx1 and gsx2 amplicons in wild type (W), heterozygous (H), and mutant (M) individuals; endonuclease cutting is disrupted in mutants. (B) Comparison of the standard length (SL, red line) of gsx1^+/+, gsx1^+/y689, and gsx1^y689/y689 siblings at 30 dpf. Images are dissecting scope images and distances between black tick marks represent 1 mm. Graph shows the quantification of long-term growth study data from 4 to 99 dpf. Asterisks indicate significant differences between that group and the genotype with the same color asterisk (blue bar, wild type; red bar, heterozygote; green bar, mutant). (C) Top, comparisons of gsx1^+/+, gsx1^+/y689, and gsx1^y689/y689 siblings and gsx2^+/+, gsx2^+/y691, and gsx2^y691/y691 siblings with (+SB) and without (-SB) swim bladders, respectively; bottom, quantification of the percentage of larvae of each genotype with and without swim bladders (gsx1 n = 73 siblings, gsx2 n = 60 siblings)

Through assessing our gsx1 mutant zebrafish, we found that these fish experience stunted growth starting at 14 dpf. No significant differences in standard length were found across genotypes in 4 dpf larvae (Figure 7B); however, by 14 dpf standard length of gsx1^y689 larvae was significantly smaller than gsx1^+/y689 siblings (P = .002). By 1 month gsx1^y689 larvae were significantly smaller than both gsx1⁺ and gsx1^+/y689 siblings (P < .001 for both), and this difference persisted through 2 months (P < .001 for both) and 3 months of age (P < .001 for both). These analyses reveal a growth-related phenotype in gsx1^y689 zebrafish similar to reports in mouse Gsx1 KOs.⁴¹ However, unlike Gsx1 KO mice, our gsx1 mutant zebrafish survive to adulthood, allowing investigations of early and later Gsx1 function across brain regions.

Embryos derived from crossing gsx2^+/y691 parents are initially indistinct from gsx1^+/y689 cross embryos (Figure 7C, top). However, gsx2^y691 larvae largely fail to inflate their swim bladders by 6 dpf under standard rearing conditions, preventing their survival. There was a significant association between swim bladder inflation and genotype in offspring from gsx2^+/y691 crosses, as less gsx2^y691 larvae had inflated swim bladders compared to gsx2⁺ and gsx2^+/y691 larvae (Figure 7C, bottom; X² = 22.8, P < .001). Swim bladder inflation did not differ between genotypes in gsx1^+/y689 crosses (X² = .32, P = .851). We observed the same results in offspring from gsx1^+/y689 and gsx1^+/y690 crosses (no association between genotype and swim bladder state, X² = 0.25, P = .882) and gsx2^+/y691 and gsx2^+/y692 crosses (significant association between genotype and swim bladder state, X² = 8.6, P = .013). These results demonstrate that swim bladder inflation failure is a result of a mutation in gsx2 and supports the important developmental role for Gsx2 in vertebrates, including zebrafish.

2.5 |. Gsx1 and Gsx2 differentially regulate Dlx2a and Dlx2b

Putative binding sites have been reported for murine GSX1²⁴ and GSX2,²⁹ and previous studies report Dlx2 as a target gene of GSX1 and GSX2 in the mouse forebrain.⁵¹ Dlx2 expression overlaps with Gsx1 and Gsx2 in the medial, caudal, and lateral ganglionic eminences (MGE, CGE, and LGE, respectively) of the mouse telencephalon where they coordinately regulate early neuronal progenitor patterning.^40,61,62 This work shows that GSX1 and GSX2 up-regulate Dlx2.⁵¹ Therefore, we sought to determine if the zebrafish ortholog dlx2a or its paralog dlx2b are Gsx1 and Gsx2 target genes. Published gene sequences for human, mouse, and zebrafish Dlx2 were analyzed in silico for putative Gsx1 and Gsx2 binding sites, which we assume are conserved in zebrafish. We found that human DLX2, mouse Dlx2, and zebrafish dlx2b possess putative Gsx1 and Gsx2 binding sites upstream of their 5^’UTRs (Figure 8A). Human DLX2 and zebrafish dlx2b possess putative binding sites for both Gsx1 and Gsx2. Zebrafish dlx2a possesses putative Gsx2 binding sites only.

FIGURE 8.

Gsx1 and Gsx2 differentially regulate dlx2a and dlx2b expression. (A) Schematic of Gsx1 and 2 putative binding sites (putative bs) in the Dlx2 orthologs. (B) Left, dlx2a expression at 30 hpf in wild type, gsx1^y689, gsx2^y691, and gsx1^y689;gsx2^y691 zebrafish. Images are ×20 compound scope images with samples mounted under cover glass, eyes dissected, and anterior facing left. Scale bar = 50 μm. Right, FIJI-ImageJ quantification of dlx2a expression in the telencephalon (gray bars) and diencephalon (white bars). Genotypes and sample size are listed under the x-axis. Different letters represent significant differences. (C) Left, dlx2b expression at 30 hpf in wild type, gsx1^y689, gsx2^y691, and gsx1^y689;gsx2^y691 zebrafish; right, quantification of dlx2b expression. (D) RT-qPCR data showing relative expression of dlx2a (blue bars) and dlx2b (red bars) in wild type, gsx1^y689, gsx2^y691, and gsx1^y689;gsx2^y691 zebrafish compared to the reference gene ef1a. Different letters indicate significant differences within each target gene. Asterisks correspond with P-values; *P < .05, **P < .01, ***P < .005, ****P < .001. (E) Schematics of gsx1, gsx2, dlx2a, and dlx2b expression at 30 hpf

To determine if Gsx1 or Gsx2 regulate dlx2a and/or dlx2b in zebrafish, we quantified dlx2a and dlx2b gene expression area in 30 hpf embryos yielded from gsx1^+/y689; gsx2^+/y691 crosses using WISH and RT-qPCR. We found that dlx2a expression is not significantly different between wild-type and gsx1^y689 embryos (Figure 8B^I-II); however, it is significantly reduced in both gsx2^y691 and gsx1^y689; gsx2^y691 embryos in the diencephalon (P < .001 for both) and telencephalon (P < .001 for both; Figure 8B^III-IV and graphs). Expression of dlx2b is not different between wild-type and gsx1^y689 embryos (Figure 8C^I-II); however, it is significantly reduced in both gsx2^y691 and gsx1^y689; gsx2^y691 embryos in the diencephalon (P < .001 for both) and telencephalon (P < .001 for both; Figure 8C^III-IV and graphs).

Consistent with WISH, RT-qPCR revealed that dlx2b expression is significantly reduced in gsx2^y691 and gsx1^y689;gsx2^y691 embryos compared to wild types (P = .005 and .002, respectively; Figure 8D). Furthermore, we also observed that dlx2b expression is significantly reduced in gsx1^y689;gsx2^y691 embryos compared to gsx2^y691 embryos (P = .012), suggesting that Gsx1 partially sustains dlx2b expression which becomes further reduced upon loss of both gsx1 and gsx2. Unlike WISH, RT-qPCR showed that dlx2a expression is only significantly reduced in gsx1^y689;gsx2^y691 embryos (P = <.001) and not in gsx2^y691 embryos (P = .225; Figure 8D). WISH shows that in gsx2^y691 embryos, dlx2a and dlx2b expression is lost in the telencephalon, where gsx2 is expressed, yet sustained in the diencephalon, where gsx1 is expressed. This suggests that zebrafish Gsx1 and Gsx2 differentially regulate dlx2a and dlx2b expression in the telencephalon and diencephalon and that visible changes in dlx2a expression cannot be detected by RT-qPCR of whole heads, including the brain at 30 hpf.

2.6 |. Gsx1 and Gsx2 differentially regulate dlx5a, dlx6a, and foxp2

To identify additional Gsx1 and Gsx2 target genes in the zebrafish forebrain, we applied our in silico and WISH approaches to assess regulation of dlx5a and dlx6a, which are closely related to dlx2a and dlx2b. Both dlx5a and dlx6a are expressed in overlapping patterns with dlx2a and dlx2b in the zebrafish forebrain^52,57 and coordinately regulate inhibitory neuron patterning in subpallial regions with the other dlx orthologs.⁶³ Published gene sequences for zebrafish dlx5a and dlx6a were analyzed in silico for putative Gsx1 and Gsx2 binding sites. Zebrafish dlx5a possesses putative binding sites for both Gsx1 and Gsx2 in the 25 kb region upstream of the 5^’UTR, while zebrafish dlx6a only possesses some for Gsx2 (Figure 9A).

FIGURE 9.

Regulation of dlx5a, dlx6a, and foxp2 by Gsx1 and Gsx2. (A) Schematic of Gsx1 and 2 putative binding sites (putative bs) upstream of the zebrafish dlx5a, dlx6a, and foxp2 gene bodies. (Bⁱ-B^iv) dlx5a expression at 30 hpf in wild type, gsx1^y689, gsx2^y691, and gsx1^y689; gsx2^y691 zebrafish. Images are ×20 compound scope images with samples mounted under cover glass, eyes dissected, and anterior facing left. Scale bar = 50 μm. (C) Quantification of dlx5a expression from experiment shown in (B) panels. Sample sizes for all FIJI-ImageJ analyses are indicated. Different letters represent significant differences and asterisks correspond with P-values; *P < .05, **P < .01, ***P < .005, ****P < .001. (Dⁱ-D^iv) dlx5a expression at 30 hpf in wild type, gsx1^y689/y690 mutants and gsx2^y691/y692 mutants. (E). Quantification of dlx5a expression in gsx1^y689/y690 mutants experiment in (D) panels. (F) FIJI-ImageJ quantification of dlx5a expression in gsx2^y691/y692 mutants experiment shown in (D) panels. (G) dlx6a expression at 30 hpf in wild type, gsx1^y689, gsx2^y691, and gsx1^y689;gsx2^y691 zebrafish. (H) Quantification of dlx6a expression in (G) panels. (Iⁱ-I^iv) foxp2 expression at 30 hpf in wild type, gsx1^y689, gsx2^y691, and gsx1^y689;gsx2^y691 zebrafish. (J) Quantification of foxp2 expression in (I) panels experiment

To determine if Gsx1 or Gsx2 regulate dlx5a and dlx6a in the zebrafish forebrain, we again quantified gene expression area in 30 hpf embryos yielded from gsx1^+/y689;gsx2^+/y691 crosses. We found that dlx5a is significantly reduced in both gsx2^y691 and gsx1^y689;gsx2^y691 embryos in the telencephalon (P = <.001), and only significantly reduced in the diencephalon in gsx1^y689;gsx2^y691 embryos (P = <.001, Figure 9B^i-iv,C). These results suggest that in the telencephalon, dlx5a is regulated by Gsx2; however, in the diencephalon, both Gsx1 and Gsx2 are required for normal expression. Complementation testing between gsx1^y689 and gsx1^y690 or gsx2^y691 and gsx2^y692 alleles confirmed these results for the function of gsx2 in the telencephalon (Figure 9D–F). For dlx6a, we observed significant reductions in expression in both gsx2^y691 and gsx1^y689;gsx2^y691 embryos in the telencephalon only (P = <.001, Figure 9G^i-iv,H). These results suggest that Gsx2 is the main regulator of dlx6a in the telencephalon, and in the diencephalon neither Gsx1 nor Gsx2 are essential for dlx6a expression.

We also assessed whether Gsx1 or Gsx2 regulate expression of forkhead box P2 (foxp2) in zebrafish. foxp2 is a gene belonging to the forkhead domain transcription factors, which are an evolutionarily conserved group of proteins that have roles in early developmental patterning.⁶⁴ In humans, FOXP2 is critical for speech and language development⁵⁸; mutations in FOXP2 lead to poor linguistic and grammatical skill development and abnormal control of facial movements.⁶⁵ foxp2 is expressed in the nervous system in zebrafish in many overlapping brain regions which we report gsx1 and gsx2 expression in, including the telencephalon, diencephalon, optic tectum, hindbrain, and spinal cord.⁶⁰ Thus, we were interested in determining if foxp2 is regulated by either Gsx1 or Gsx2, particularly in the forebrain. Zebrafish foxp2 possesses putative Gsx1 and Gsx2 binding sites, as well as a potential binding sites for both Gsx1 and Gsx2 (Figure 9A). foxp2 expression appears to be reduced in the gsx2 mutant and gsx1 and gsx2 double mutant telencephalon; however, this reduction is not significant among genotypes (P = .312, Figure 9I^i-iv,J) by our measures. This indicates that Gsx2 does not significantly regulate foxp2 expression at this age unlike other forebrain patterning genes like the dlx group.

2.7 |. gsx1 expression changes in gsx2 mutants

We used WISH and FISH to examine gsx1 expression in gsx2^y691 mutant embryos and gsx2 expression in gsx1^y689 mutant embryos (Figure 10A,B^i-iv) at 30 hpf. Quantification of expression in WISH images suggests that gsx1 expression increases in gsx2^y691embryos compared to wild-type siblings (P = .007), and that gsx2 expression does not change between wild-type embryos and gsx1^y689 embryos (P = .766; Figure 10C). FISH produced consistent results, suggesting increased gsx1 expression in gsx2^y691 embryos compared to wild-type siblings (P = .031), and no change in gsx2 expression between wild-type and gsx1^y689 embryos (P = .847; Figure 10D).

FIGURE 10.

gsx1 and gsx2 expression in gsx1 and gsx2 mutant zebrafish. (Aⁱ-A^iv) whole-mount in situ hybridization (WISH) images at 30 hpf showing gsx1 expression in wild-type and gsx2^y691 embryos (top) and gsx2 expression in wild-type and gsx1^y689 embryos. Images are ×20 compound scope images with embryos mounted under cover glass, eyes dissected, and anterior facing left. (Bⁱ-B^iv) FISH images at 30 hpf. Images are ×20 confocal images with embryos mounted under cover glass, eyes dissected, and anterior facing left. (C) FIJI-ImageJ quantification of images in (Aⁱ-A^iv). (D) FIJI-ImageJ quantification of images in (Bⁱ-B^iv). Sample sizes for all FIJI-ImageJ analyses are indicated. Asterisks indicate significant differences and correspond with P-values; *P < .05, **P < .01, ***P < .005, ****P < .001. (E) RT-qPCR quantification of gsx1 and gsx2 expression in 30 hpf embryos across genotypes. The 2^-ΔΔCt method was used to calculate relative expression to the reference gene ef1a. Data are representative of two biological replicates (parent groups)

To determine the extent of the expansion of gsx1 expression further, we used RT-qPCR. Unlike WISH and FISH, RT-qPCR suggests that no change in gsx1 expression occurs in gsx2^y691 embryos (P = .819; Figure 10E). However, RT-qPCR does confirm that no change in gsx2 expression occurs in gsx1^y689 embryos (P = .456; Figure 10E). Given that RT-qPCR was done on whole head tissue samples and not by brain region or single cells, we can deduce that the changes in gsx1 expression are occurring at 30 hpf in our gsx2 mutant zebrafish in a very select regional way, an effect that gets diluted out by whole tissue analysis which can only be detected by examining samples visually using WISH and FISH.

3 |. DISCUSSION

3.1 |. gsx1 and gsx2 expression during neurodevelopment in zebrafish embryos and larvae

In this study, we comprehensively document gsx1 and gsx2 expression in embryonic and larval zebrafish using multiple strategies, and our analysis presents a time-course for their co-expression during neurodevelopment. In embryonic and larval stages in zebrafish, gsx1 is expressed in the diencephalon, hypothalamus, preoptic region, hindbrain, cerebellar plate, spinal cord, optic tectum, and pretectum. Across these ages, gsx2 is expressed in the telencephalon, hypothalamus, pallium, olfactory bulb, and hindbrain. These patterns are largely consistent with expression of Gsx1 and Gsx2 in mouse,^24,29 medaka,²⁶ Xenopus,²⁵ and previous reports in zebrafish^10,27,30 with minor exceptions. In Xenopus, Gsx2 is first detected slightly earlier than Gsx1; however, we report in zebrafish that gsx1 is expressed at 10 hpf slightly earlier than gsx2 at 12 hpf. Furthermore, we report that gsx2 and not gsx1 is expressed in the olfactory bulb in zebrafish; however, in Xenopus Gsx1 and not Gsx2 is expressed in this region.

Prior to our study, a comprehensive knowledge of the unique and overlapping roles for Gsx1 and Gsx2 across the vertebrate brain was not attainable due to the lack of gene expression data. FISH revealed that regional co-expression of gsx1 and gsx2 occurs in the hindbrain, hypothalamus, and preoptic area in zebrafish; however, they rarely co-localize in the same cells. In the hindbrain, gsx1 and gsx2 exist in two adjacent dorsolateral columns, with gsx2 dorsal to gsx1, consistent with previous reports and their roles in DV patterning.^24,25 In mouse, Gsx1 regulates cerebellar neuronal progenitor identity through a temporally-regulated BMP/SMAD signaling gradient,^46,47 and in zebrafish gsx2 is reported to specify neuronal fate in the inferior olivary nuclei of the medulla.³⁰ Outside of these studies, the coordinate roles for Gsx1 and Gsx2 in the hindbrain remain under studied. Our findings demonstrate that expression of gsx1 and gsx2 remain distinct from each other dorsoventrally, and starting at 48 hpf, expression of gsx1 begins to extend ventrally while gsx2 is restricted dorsally. These patterns may represent the outgrowth of axons from maturing neuronal progenitors or their initial migration, which would agree with previously reported roles for Gsx2 and Gsx1 in regulating progenitor proliferation and differentiation, respectively.²² Thus, this work provides an essential foundation for future studies to interrogate the functional roles of Gsx1 and Gsx2 in the hindbrain.

gsx1 expression in the hypothalamus has been shown in medaka, Xenopus, mice, and zebrafish^24–27,41; however, no roles for gsx2 in the hypothalamus have been reported. We show that gsx1 and gsx2 are regionally co-expressed in the hypothalamus in zebrafish, which necessitates further studies of Gsx2 function in this region. Zebrafish gsx2 expression in the hypothalamus begins between 24 and 30 hpf (Figure 2J,L), slightly earlier than the onset of gsx1 in this region between 30 and 48 hpf (Figure 2E,G). Expression of both gsx1 and gsx2 is sustained in the hypothalamus through 6 dpf (Figure 6).

Functions for Gsx1 and Gsx2 in this region in zebrafish could be similar to reports in mouse forebrain showing that Gsx2 maintains neuronal progenitor pools and Gsx1 drives neuronal differentiation.²² Interestingly, one single-cell sequencing report conducted in the adult (1–2 years) zebrafish hypothalamus reported Gsx1 as a transcription factor significantly associated with 13 genes categorized as either neuropeptide, neurotransmitter, ion channel, or synaptic genes.⁶⁶ Identification of Gsx1 as an important regulatory factor in the mature zebrafish hypothalamus suggests prolonged requirements for Gsx1 in hypothalamic function. As our gsx1 mutant zebrafish survive through adulthood, roles for Gsx1 in the development and function of the hypothalamus along with associated growth, behavioral, and metabolic changes can be investigated in the future.

3.2 |. Mutations in gsx1 and gsx2 in zebrafish disturbs early growth and development

We observed a reduced growth phenotype in gsx1 mutant zebrafish through adulthood. These studies provide a detailed description of the onset of significant growth deficits as well as the basic trend and continuation of these deficits. We observed that significant deficits were not present at 4 dpf, but appeared by 14 dpf, allowing us to determine the relative window under which these deficits begin. These data are consistent with work in Gsx1 mutant mice, which were the same size as their wild-type siblings at birth but began to show growth deficits as development progressed.⁴¹ Unlike Gsx1 mutant mice, our gsx1 mutant zebrafish survive to adulthood, permitting investigations of later Gsx1 function. The premature death of Gsx1 mutant mice is largely attributed to defects in forebrain neurogenesis and disruptions in ascending cortical interneuron migration,^22,44,51 thus continued examination of the impact of mutations in gsx1 in zebrafish will further elucidate its important neurodevelopmental and later roles in vertebrates.

We have additionally identified a unique swim bladder inflation failure phenotype in gsx2 mutant zebrafish that prevents their survival under standard rearing conditions, supporting the critical role for Gsx2 in growth and development among vertebrates. Gsx2 mutant mice fail to survive more than a day following birth, however, also exhibit severely disrupted forebrain and hindbrain morphology. Comprehensive knowledge of GSX1 and GSX2 function together and separately is minimal outside of the mouse forebrain^34,44,51 and few reports in the cerebellum^30,46,67 and spinal cord.^7,10,48 As such, analysis of Gsx1 and Gsx2 function in our zebrafish mutants in these and other CNS regions can supplement these reports.

3.3 |. Identifying Gsx1 and Gsx2 target genes in the zebrafish forebrain

We demonstrated differential regulation of dlx2a, dlx2b, dlx5a, dlx6a, and foxp2 by Gsx1 and Gsx2 in the forebrain of our gsx1 and gsx2 mutant zebrafish. A complex relationship between the Gsx and Dlx genes has been reported in the mouse forebrain⁵¹ that facilitates regulation of a major transcriptional control program dictating the expression of diverse target genes. The Dlx pathways in the forebrain also serve to regulate the differentiation of inhibitory projection neurons and interneurons that migrate to mature regions like the cortex and olfactory bulb.^53–55 Conservation of the Gsx/Dlx regulatory network in zebrafish is significant in that it establishes initial understanding of Gsx function in neurodevelopment in zebrafish. Our embryonic and larval stage whole brain expression analyses also justify continued investigations of Gsx function together and separately across brain regions to add to our knowledge of their role in neurodevelopment across vertebrates.

Our data suggest that in zebrafish, Gsx2 is largely responsible for regulating expression of dlx2a and dlx2b. Through WISH, we identified significant reductions in dlx2a and dlx2b expression in the telencephalon of gsx2 mutant and gsx1 and gsx2 double mutant embryos (Figure 8). In the diencephalon, dlx2a expression was reduced in both gsx2 mutant and gsx1 and gsx2 double mutant embryos; however, dlx2b expression was only reduced in gsx1 and gsx2 double mutant embryos, suggesting that Gsx1 may be compensating for Gsx2 and sustaining dlx2b expression in the diencephalon specifically. RT-qPCR analysis of dlx2b expression was consistent with these results, revealing significant reductions in gsx2 mutant embryos and more significant reductions in gsx1 and gsx2 double mutant embryos. Unlike our WISH analysis, RT-qPCR shows that dlx2a expression is only significantly reduced in gsx1 and gsx2 double mutant embryos and not gsx2 mutant embryos. One potential explanation for this variability is the pattern in which dlx2a expression is reduced by WISH. In gsx1 and gsx2 double mutant embryos, dlx2a expression is lost in the telencephalon, where gsx2 is expressed; however, not in the diencephalon, where gsx1 is expressed. This variability could also be related to alternative transcript detection through RT-qPCR. However, our in situ probe for dlx2a detects a product that overlaps completely with the transcript amplified by our dlx2a RT-qPCR primers, and we predict these targets are identical for both dlx2a splice variants that exist in zebrafish.

Expression of zebrafish dlx5a is largely regulated by Gsx2 in the telencephalon, as we identified significant reductions in gsx2 mutant and gsx1 and gsx2 double mutant embryos (Figure 9B,C). However, in the diencephalon, it appears that both Gsx1 and Gsx2 regulate dlx5a expression, as significant reductions were only identified in gsx1 and gsx2 double mutant embryos. In turn, this suggests that Gsx1 is in part compensating for loss of Gsx2 function in gsx2 mutant embryos, which were not significantly reduced compared to wild types. This compensation can only be explained by a very small expansion of gsx1 expression in the gsx2 mutant diencephalon that was only detectable by in situ hybridization and not by RT-qPCR (Figure 10). Interestingly, expression of dlx6a appears to be regulated most strongly by Gsx2 in the telencephalon only, agreeing with initial predictions based on putative binding site presence for Gsx2 and not Gsx1 25 kb upstream of the dlx6a gene (Figure 9A,G,H). Collectively, these results demonstrate that a complex relationship between the gsx and dlx genes exists in zebrafish that is reminiscent of reports in other vertebrates.⁵¹ Future studies will focus on confirming more Gsx1 and Gsx2 target genes in zebrafish in order to elucidate their unique and overlapping roles during CNS development.

Outside of confirming that several dlx paralogs are regulated by Gsx1 and Gsx2 in the zebrafish forebrain, we also found that foxp2 is not significantly regulated by Gsx1 or Gsx2 in the telencephalon. It is important to note that this study was conducted at 30 hpf only, and foxp2 expression begins in zebrafish as early as 10 hpf in the presumptive forebrain⁶⁰ and is documented through 3 months of age.⁶⁸ The onset of foxp2 expression is similar to the onset of gsx2 in the presumptive forebrain at 12 hpf (Figure 2H); however, during neurodevelopment foxp2 is expressed in several overlapping regions with the gsx genes, such as the optic tectum, hindbrain, and spinal cord. It is interesting to note that expression of foxp2 and dlx6a only minimally overlaps in dorsal subpallial regions in zebrafish,⁶⁰ and dlx6a is regulated by Gsx2 in this dorsal telencephalic region only. Collectively, our approaches for identifying and validating target genes for Gsx1 and Gsx2 during neurodevelopment provide a new in vivo model for gaining even greater insight into regulatory roles of these and other transcription factors across CNS gene networks.

4 |. EXPERIMENTAL PROCEDURES

4.1 |. Zebrafish husbandry

All aspects of this study were approved by the West Virginia University IACUC. Adult zebrafish were maintained on a 14 hours/10 hours light/dark cycle at water temperature at 28°C to 29°C. Breeding was performed using 1-L breeding chambers with dividers (Aquaneering). Embryos were raised in 90 × 15 mm petri dishes at 28.5°C in E3 media (pH 7.4; 0.005M NaCl, 0.00017M KCl, 0.00033M CaCl, 0.00033M MgSO4.7H₂0, 1.5 mM HEPES) in an incubator operating on a 14 hours/10 hours light/dark cycle. Staging of embryos was performed using standard procedures.⁶⁹ The following strain was used: TL (Tupfel long fin).

4.2 |. Bioinformatics

Gene and protein sequences for all genes were obtained from the NCBI database (https://www.ncbi.nlm.nih.gov; see Table 1 for accession numbers) and aligned using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). Geneious was used to construct the rooted phylogenetic tree (https://www.geneious.com/academic/). The UCSC Genome Browser (http://genome.ucsc.edu/) and Ensembl database (http://uswest.ensembl.org/index.html ) were used to evaluate exon and intron structures.

TABLE 1.

List of NCBI sequences for gene and protein alignments and analyses

Gene sequences
Species	Gene	Accession number
Danio rerio	gsx1	NM_001012251.1
D. rerio	gsx2	NM_001025512.2
D. rerio	dlx2a	AF349437.2
D. rerio	dlx2b	NM_131297.2
Mus musculus	Dlx2	NM_010054.2
Homo sapiens	Dlx2	NM_004405.4
D. rerio	dlx5a	NM_131306.2
D. rerio	dlx6a	NM_131323.1
D. rerio	foxp2	NM_001030082.2
Protein sequences
Species	Protein	Accession number
D. rerio	Gsx1	AAI65050.1
M. musculus	Gsx1	AAI37770.1
Rattus norvegicus	Gsx1	NP_001178592.1
H. sapiens	Gsx1	NP_663632.1
Xenopus tropicalis	Gsx1	NP_001039254
Oryzias latipes	Gsx1	NP_001098303
Lepisosteus oculatus	Gsx1	XP_006627824
D. rerio	Gsx2	AAI64330.1
M. musculus	Gsx2	NP_573555.1
R. norvegicus	Gsx2	NP_001131035.1
H. sapiens	Gsx2	NP_573574.2
X. tropicalis	Gsx2	AAI58504.1
O. latipes	Gsx2	NP_001116381
L. oculatus	Gsx2	XP_006630061
Drosophila melanogaster	Ind	NP_996087.2

4.3 |. Identification of zebrafish gsx1 and gsx2 mRNA transcripts using RT-PCR

Embryos and larvae obtained from TL crosses were raised to the desired ages (3.5–120 hpf), euthanized, frozen in liquid nitrogen, and stored at −80°C. Total RNA was extracted from 30 embryos and larvae at each age using a phenol chloroform extraction method with TRI-Reagent (Invitrogen). Then, 1 μg of total RNA was used with oligoDT to synthesize cDNA libraries (Superscript II First-Strand Synthesis kit, Invitrogen). In addition, 2 μg of cDNA was used in 28 cycles of PCR with PlatinumTaq (Invitrogen) and intron-spanning gene-specific primers (see Table 2). Amplicons were visualized and imaged using a FluorChemQ imager (ProteinSimple) on a 2% agarose gel with SYBR Safe DNA gel stain (Invitrogen) and excised using a blue light transilluminator (Clare Chemical Research). Sanger sequencing was used to confirm identity with NCBI sequences.

TABLE 2.

Table of primers, plasmids, and antibodies

Primers and plasmids
Gene	Primer sequence	Used for	Additional information
gsx1	FW: 5’-AGCATTTGGTACACGAGCGA-3’ RV: 5’-GGTGTGGCGTACAGAGTCTT-3’	Semiquantitative RT-PCR
gsx2	FW: 5’-CAAGTTCTTGGAGCATCGCC-3’ RV: 5’-TCCGTTTAAAAGTGCCACGT-3’
efla	FW: 5’-TACAAATGCGGTGGAATCGAC-3’ RV: 5’-TGTGCAGACTTTGTGACCTTG-3’
gsx1	Cheesman and Eisen27	Antisense mRNA in situ probes	Plasmid linearized with ClaI, in vitro transcribed with T3 polymerase
gsx2	FW: 5’-ACAACAGCCACATACAGAACG-3’ RV: 5’-CACAGCTTCTCAGTAGTCTAGGA-3’		Plasmid linearized with EcoRI, in vitro transcribed with SP6 polymerase
dlx2a	Akimenko et al.70		Plasmid linearized with NcoI, in vitro transcribed with SP6 polymerase
dlx2b	FW: 5’-GCGCAGATTCCAGAAGACC-3’ RV: 5’-ACCCGTTTGTACTTGGAATGTG-3’		Plasmid linearized with Not1, in vitro transcribed with SP6 polymerase
dlx5a	FW: 5’-ATTTAGGTGACACTATAGCC GAAGTAAGGA TGGTCAAC-3’ RV: 5’-TAATACGACTCACTATAGCAGTACAACGTTCCTGATCC-3’		In vitro transcribed with T7 polymerase
dlx6a	FW: 5’-ATTTAGGTGACACTATAGACAGCAGAAAACAACAGTGA-3’ RV: 5’-TAATACGACTCACTATAGGTGACGAGT ACCAGTGTGAAT-3’		In vitro transcribed with T7 polymerase
foxp2	FW: 5’-GCCACACCGACAAATACTCC-3’ RV:5’-ATTTAGGTGACACTATAGCTGCTG TGTCCATTGGTGTC-3’		In vitro transcribed with SP6 polymerase
gsx1Δ11 y689	FW: 5’-TCCAGATCCACGACAGTTCC-3’ RV: 5’-TGACTGCTGCTATTTTCTGTTGA-3’	Genotyping by amplicon size
gsx2Δ13 y691	FW: 5’-TGCGTATCCTCACACATCCA-3’ RV: 5’-TGTCCAGGGTGCGCTAAC-3’
gsx1Δ05 y690	FW: 5’-AGCCCTCCGTTATTTCCGTA-3’ RV: 5’-CGTTTGCTGCTCTGAAGTT-3’	Confirmation of TALEN efficacy
gsx2Δ05 y692	FW: 5’-AGCAATCATGTCGAGGTCTT-3’ RV: 5’-GCGCACTCACTCACCTAGAGA-3’	Genotyping by RFLP
dlx2a	FW: 5’-CCTGCAGAGGAGGTTTCAGA-3’ RV: 5’-GGGTGGGATCTCTCCACTTT-3’	qPCR
dlx2b	FW: 5’-TCCTATGGCGCTTATGGAAC-3’ RV: 5’-GAGTAGATGGTTCGCGGTTT-3’
ef1a	FW: 5’-TGATCTACAAATGCGGTGGA-3’ RV: 5’-CAATGGTGATACCACGCTCA-3’
Antibodies
Name		Manufacturer	Item #
Anti-Digoxigenin-AP, Fab fragments		Roche	#11093274910

4.4 |. Whole-mount in situ hybridization

Embryos and larvae were raised to the desired ages and supplemented with 0.003% phenylthiourea in E3 after 6 hpf to prevent pigmentation. For embryos younger than 48 hpf, chorions were removed by incubating in 50 μg/mL Pronase (Sigma) at 28.5°C for 15 minutes. Embryos and larvae were anesthetized and fixed in cold 4% paraformaldehyde (PFA) overnight at 4°C. Fixed embryos were dehydrated using an increasing methanol wash series in 1xPBS (0%, 50%, 100% vol/vol methanol) and stored at −20°C in 100% MeOH for at least 24 hours and up to 1 year. The gsx1²⁷ and dlx2a⁷⁰ probes have been previously reported and were kind gifts of the Eisen and Karlstrom zebrafish labs. The probes for gsx2, dlx2b, dlx5a, dlx6a, and foxp2 were designed in our lab. To generate antisense mRNA probes for gsx2 and dlx2b, 1 μg of age-specific total RNA was used with Invitrogen’s Super-Script III One-Step RT-PCR kit and gene-specific primers (see Table 2) to amplify cDNA in 35 cycles of PCR. Amplicons were separated by 1% agarose gel electrophoresis, extracted using a QIAquick Gel Extraction kit (Qiagen), and subcloned into a pCR2.1_TOPO 4.0 kb vector (Invitrogen). Sanger sequencing was used to confirm insert identity and directionality. Then, 5 μg of each plasmid was linearized with EcoR1 (gsx2) and Not1 (dlx2b) and probes were transcribed in vitro using SP6 polymerase (mMESSAGE mMACHINE kit, Ambion). To generate antisense mRNA probes for dlx5a, dlx6a, and foxp2, 1 μg of age-specific total RNA was used to synthesize cDNA libraries with oligoDT (Superscript II First-Strand Synthesis kit, Invitrogen). Also, 1 μg of cDNA was then used with gene-specific primers (see Table 2) in 36 cycles of PCR. Amplicons were visualized using a 1% agarose gel with SYBR Safe DNA gel stain (Invitrogen) and purified using Qiagen’s QIAquick Gel Extraction kit. Probes were transcribed in vitro directly from purified amplicons using T7 polymerase (dlx5a and dlx6a) and SP6 polymerase (foxp2) (mMESSAGE mMACHINE kit, Ambion). The protocol for colorimetric WISH was adapted from Thisse and Thisse⁷¹ and performed essentially as in Bergeron et al.²⁸ Embryos were hybridized with a digoxigenin (DIG)-tagged antisense mRNA probes detected by an anti-DIG antibody (Roche) and developed in NBT/BCIP (Roche). Staining was stopped by postfixation in cold 4% PFA overnight at 4°C. Stained embryos were cleared in 75% glycerol and stored at 4°C protected from light.

4.5 |. Fluorescence in situ hybridization

FISH procedures were performed according to the In Situ Hybridization Chain Reaction v3.0 protocol (Molecular Instruments, Los Angeles, California).⁷² Embryos were simultaneously hybridized with gsx1 and gsx2 probes (designed by Molecular Instruments) diluted in probe hybridization buffer overnight at 37°C. Excess probe was washed off the following day using probe wash buffer. Embryos were incubated for 30 minutes at room temperature in amplification buffer before adding the provided Alexa hairpins specific to the gsx1 (Alexa Fluor 488) and gsx2 (Alexa Fluor 546) mRNA sequences and incubating overnight at room temperature. Embryos were washed using 5x SSCT (5x SSC + 0.1% Tween 20) and stored in 5x SSCT at 4°C protected from light.

4.6 |. Generation of gsx1 and gsx2 TALEN mutants

TALEN were designed using the freely available TALENT website that was created and is maintained by labs at Cornell University.^73,74 TALEN assembly was carried out using the Golden Gate vector system⁷⁴ and separate destination vectors containing a modified FokI domain.⁷⁵ Then, 100 pg of in vitro transcribed mRNA (mMESSAGE mMACHINE kit, Ambion) was injected into the cell of each 1-cell stage zebrafish embryo to create G0. TALEN efficacy was checked by amplifying a 436 bp fragment around the gsx1 target site and a 409 bp fragment around the gsx2 target site using gene-specific primers (see Table 2), followed by restriction digest of these amplicons using BtsI and EcoRI enzymes (NEB) respectively. Disruption of endonuclease cutting as evidenced by the presence of a full-length amplicon was considered effective, and siblings of these embryos were raised to adulthood and screened by crossing to wild-type (TL strain) adults to generate F1 offspring with single gsx1 and gsx2 mutant alleles. These alleles were sequence confirmed by DNA extraction from a subset of pooled sibling F1 embryos and PCR using gene-specific primers (see Table 2), TOPO-TA subcloning (Invitrogen), DH5α transformation, and Sanger sequencing of individual clones. F1 siblings carrying predicted loss of function mutations were raised to adulthood, genotyped, and crossed together to produce homozygous F2 for each new allele as a first pass mutant screen. Mutant lines are maintained by continuously crossing carriers to TL to eliminate possible off-target mutations over time; however, no off-target sites were predicted by TALE-NT.

4.7 |. Genotyping for gsx1 and gsx2 alleles

When genotyping was required to distinguish isolated gsx1 and gsx2 alleles, tissue was dissected from the most posterior end of the tail to use in DNA preparation. Tail tissue was denatured at 95°C for 10 minutes in DNA lysis buffer (10 mM Tris pH 7.5, 50 mM KCl, 0.3% Tween20, 0.3% Triton X, 1 mM EDTA) and digested using 2 mg/mL proteinase-K (Omega) at 55°C for at least 2 hours to overnight. Proteinase-K was heat-inactivated at 95°C for 10 minutes before the DNA was used in a standard DreamTaq (Thermo) PCR reaction with gene-specific primers (see Table 2). Amplicons were visualized and imaged using a Syngene NuGenius imager with a blue light transilluminator (Clare Chemical Research) on a 4% agarose gel with SYBR Safe DNA gel stain (Invitrogen). gsx1 wild-type individuals have one band (140 bp), y689 mutants have one band (129 bp), and y689 heterozygotes have two bands at both sizes. For gsx2 wild-type individuals have one band (134 bp), y691 mutants have one band (121 bp), and y691 heterozygotes have two bands at both sizes. y690 and y692 alleles were genotyped as described in the previous section with the restriction digest method.

4.8 |. In silico analyses and expression quantification

Sequences for dlx2a, dlx2b, dlx5a, dlx6a, and foxp2 were identified in the NCBI database (see Table 1 for accession numbers). The 25 kb region upstream of the 5^’UTR of each gene was collected using Ensembl (https://useast.ensembl.org/index.html) and entered into ApE (http://jorgensen.biology.utah.edu/wayned/ape/). Assuming conservation of the GSX1²⁴ and GSX2²⁹ enhancer sequences identified in mouse, a 25 kb region upstream of each gene 5^’UTR was scanned for enhancer sequence variants. Annotated gene body schematics for the Dlx2 orthologs were designed in Inkscape (https://inkscape.org/) and drawn using sequence information from Ensembl and ApE. Gene expression area was measured using FIJI-ImageJ by tracing the area of staining in the diencephalon or telencephalon. The telencephalon area was measured and used as a proxy for head size to correct for embryo size differences.

4.9 |. Quantification of gene expression using RT-qPCR

Embryos derived from double heterozygous gsx1^+/y689; gsx2^+/y691 adults were euthanized and dissected at 30 hpf in cold RNAlater (Sigma) chilled by housing a 60 × 15 mm petri dish on ice. A dissection anterior to the spinal cord was made to separate the head from the tail. Heads were stored in RNAlater at 4°C for up to 1 month and tails were used for DNA extraction and genotyping as previously described. Ten to twelve embryo heads of the same genotype were combined in a single 1.5 mL snap tube and total RNA was extracted using a phenol chloroform extraction method with TRI-Reagent. Then, 0.5 μg of total RNA was used to synthesize cDNA libraries using oligoDT (Superscript II First-Strand Synthesis kit, Invitrogen). Also, 1 μL of cDNA was then used in a standard SYBR Green (Bio-Rad) qPCR reaction using gene-specific primers for ef1a, dlx2a, and dlx2b (see Table 2). Samples were run on a Bio-Rad CFX Connect Real Time System using Bio-Rad CFX Maestro 1.1 Software. The 2^-ΔΔCt method⁷⁶ was used to analyze raw Ct values and calculate gene expression changes relative to the housekeeping gene ef1a.⁷⁷

4.10 |. Microscopy and imaging

For WISH, embryos at 12 hpf were imaged at ×6.3 on a Zeiss Stereo Discovery V.8 dissecting scope with an Axiocam 105 Color camera and analyzed using the ZEN 2.3 Lite software. Embryos of the remaining ages (24–144 hpf) were dissected and mounted in 75% glycerol under glass coverslips and imaged at ×20 on a compound Zeiss Observer.Z1 with an Axiocam 503 Color camera. Imaging of genotyped samples for mutant studies was done blind by using a numeric code that could be aligned with genotype afterward.

For FISH, embryos were dissected and mounted in 1x PBS under glass coverslips and imaged on an Olympus BX61 confocal microscope with Fluoview FV100 software. Imaging objectives were interchanged depending upon the area being investigated (Olympus UPlanApo, ×20 or ×40 oil immersion objectives with Olympus Immoil F30CC). Fluorophores used were Alexa Fluor 488 (gsx1) and Alexa Fluor 546 (gsx2).

For standard length measurements, embryos from crosses of gsx1^+/y689 adults were raised and imaged at 4 dpf, 14 dpf, 1 month, 2 months, and 3 months old. Fish were anesthetized using MS-222, embedded in 1.5% low melt agarose in E3, and imaged next to a ruler on a Zeiss Stereo Discovery V.8 dissecting scope with an Axiocam 105 Color camera. FIJI-ImageJ was used to measure standard length.^78,79

For swim bladder inflation studies, embryos derived from heterozygous gsx1^+/y689 or gsx2^+/y691 crosses were raised under standard rearing conditions in 60 × 15 mm petri dishes with 10 mL of E3 media at a density of 30 embryos per dish. The number of larvae with and without inflated swim bladders was counted from days 3 to 6, and E3 was refreshed daily. Larvae were imaged on a Zeiss Stereo Discovery V.8 dissecting scope with an Axiocam 105 Color camera.

4.11 |. Statistics

One-way analysis of variance with multiple comparisons and post hoc Tukey tests at α = .05 were performed in SPSS to evaluate significant differences between genotypes for the gsx1 growth study and WISH expression analyses. For RT-qPCR studies, independent two-tailed t tests at α= .05 were performed in SPSS to evaluate significant differences between 2^-ΔΔCt values calculated from raw Ct values. A chi-square test (Pearson’s test) was performed using GraphPad to evaluate the association of genotype and swim bladder inflation at α= .05. Outliers for the growth study and WISH expression analyses were identified using GraphPad (Grubb’s test) and removed.