A transgene targeted to the zebrafish nkx2.4b locus drives specific GFP expression and disrupts thyroid development

Abstract

Background:

With the goal of labeling and manipulating the zebrafish hypothalamus, we sought to target a gfp transgene to the expression domains of nkx2.4b, a gene expressed during hypothalamic and thyroid development. We combined TALENs mediated mutagenesis with a targeting construct to enable insertion of a gfp transgene into the endogenous nkx2.4b genomic locus.

Results:

Injection of TALENs targeted to the first exon of nkx2.4b created a predicted null allele, and homozygous mutant embryos displayed loss of thyroid markers. From embryos injected with both TALENs and a targeting construct carrying a gfp transgene, we recovered a line in which GFP was expressed specifically in the hypothalamus and thyroid. Fish homozygous for this allele lacked exon 1 of nkx2.4b and exhibited hypothyroid phenotypes.

Conclusions:

By combining TALENs injections with a targeting construct that contained a gfp transgene, we were able to recover an allele in which GFP is expressed in the nkx2.4b expression domains, with homozygous phenotypes suggesting the creation of a loss-of-function transgenic line. These results demonstrate the creation of a useful tool for studying hypothalamus and thyroid development.

Introduction

The hypothalamus is a highly conserved brain region in the ventral diencephalon of vertebrates that regulates a wide range of basic physiological functions. It impacts innate behaviors such as feeding and drinking, mating, and sleep, as well as homeostatic outputs such as metabolism, body temperature, and growth, through the autonomic nervous system and neuroendocrine hormone system ¹. Zebrafish have emerged as an important animal model to study the development and function of the vertebrate hypothalamus, but genetic tools to specifically label and manipulate hypothalamic cells are lacking. Multiple studies have shown that transcription factor-encoding Nkx2.1/4 gene family members are expressed in the developing hypothalamus, and thus these genes may represent good candidates for transgene targeting.

Loss of function studies in multiple species have shown that the expression of Nkx2.1/4 family members is necessary for hypothalamic development ^2–4. Nkx2.1/4 orthologs are expressed in the developing ventral diencephalon of many species including nonmammalian vertebrates and chordates such as Xenopus laevis and Amphioxus ^{5, 6}. In zebrafish, nkx2.4a and nkx2.4b (previously named nkx2.1a) are expressed in the developing hypothalamus, while nkx2.1 (previously named nxk2.1b) is expressed briefly in the ventral diencephalon, then later in the ventral telencephalon ^{3, 7, 8}. Because nkx2.4b expression in the brain is limited to the hypothalamus, we chose to target this locus for generating transgenic zebrafish lines.

The only other identified domain of nkx2.4b expression in the zebrafish embryo is in the developing thyroid gland ³. The thyroid develops from an endodermal evagination in the midline of the developing pharynx, and in zebrafish it consists of loosely organized follicular tissue arranged along the midline ⁹. Once fully developed into thyroid follicular cells, expression of functional genes such as Thyroglobulin (Tg), Thyroperoxidase (Tpo), and Thyrotropin receptor (Tshr) is initiated. As with the hypothalamus, the molecular and genetic control of thyroid development is highly conserved in vertebrates ^9–11. In human, mouse, and zebrafish, nkx2.1/4 family members, hhex, and either pax2 or pax8 are expressed in the developing thyroid anlage ^{5, 6, 12}. These genes are required for the differentiation and survival of the thyroid primordium, although not its initial specification ⁹. Nkx2.1 is also known as T/ebp or TTF1 due to its function regulating thyroid gland development and thyroid specific proteins ^{2, 13, 14}. T/ebp was originally shown to regulate the expression of Tg ¹⁵, Tpo ^{14, 16–18} and Tshr ^{19, 20}. Thyroid dysgenesis, including hypothyroidism and athyreosis, can cause cognitive deficits due to growth and metabolic effects on neural development as well as generalized growth defects ¹².

Here we first report mutagenesis with Transcription activator-like effector nucleases (TALENs) to create a nkx2.4b loss-of-function allele, which exhibits a homozygous phenotype of thyroid marker loss. Co-injection of TALENs with a targeting vector, in which a gfp transgene was flanked by genomic nkx2.4b sequence, was then used to create a targeted transgenic allele that expresses GFP in nkx2.4b expression domains, and also appears to be a genetic hypothyroid model.

Results and Discussion

As a first step toward inserting a transgene into the nkx2.4b locus, TALENs were designed to target the first exon of the gene (Figure 1A), upstream of the DNA binding homeodomain that functions in transcriptional regulation and is encoded by exon 2. The homeodomain represents the only known functional domain in transcription factors of the Nkx2 family, and loss of this domain is predicted to result in a null mutation. Wild-type fish were injected with TALEN mRNAs, raised, and outcrossed to screen for germline transmission of nkx2.4b mutations. A line carrying a 5bp deletion that produces a frameshift (Figure 1A – underlined, 1B – red bar) resulting in a premature stop codon 43 residues later, before the DNA binding homeodomain (Figure 1C, D), was assigned the zd20 allele designation.

Figure 1: Generation of the nkx2.4b zd20 null allele.

A) TALENs were targeted to exon 1 of the nkx2.4b gene, upstream of exon 2 which encodes the DNA binding homeodomain (HD). The 5bp deletion in zd20 is underlined in the spacer sequence located between the two TALENs binding sites. B) DNA alignment of nkx2.4b in wild type and zd20 mutant fish showing the 5bp deletion, identified with a red bar. C) Wild type Nkx2.4b amino acid sequence - serine 71 is labeled in magenta. An arrowhead marks the exon 1 – exon 2 boundary, and the homeodomain in exon 2 is underlined. D) The zd20 allele translation showing the predicted amino acid sequence resulting from the frameshift in grey, with a stop codon at position 114.

The zebrafish nkx2.4b gene has previously been demonstrated to play a redundant role with nkx2.4a and nkx2.1 in zebrafish hypothalamus development ⁴, and we did not observe any obvious anatomical or behavioral phenotypes in homozygous zd20 mutant embryos. However, since nkx2.4a and nkx2.1 are not expressed in the thyroid ³, and based on the phenotypes of Nkx2.1 knockout in mouse ² and nkx2.4b morpholino knockdown in zebrafish ⁹, we predicted that embryos homozygous for the zd20 allele would exhibit disrupted thyroid formation. Indeed, incrosses of nkx2.4b^zd20/+ fish produced embryos at mendelian frequencies exhibiting a complete loss of two markers of thyroid development (Figure 2A–H, hhex: 10/28, and pax8: 7/29) and two markers of thyroid function (Figure 2 I–P, tg: 10/31 and tpo: 9/27) at 2 dpf, suggesting that the zd20 allele is an nkx2.4b loss-of-function mutation and that nkx2.4b is required for thyroid development.

Figure 2: nkx2.4b^zd20/zd20 embryos fail to express markers of thyroid differentiation.

Lateral views of 2 dpf embryos stained by in situ hybridization for thyroid primordium markers (hhex and pax8), and markers of thyroid function (tg and tpo). Yellow boxes indicate magnified regions.

To target the nkx2.4b locus for reporter transgene expression, we co-injected 1-cell stage embryos with both of the previously validated TALEN mRNAs and a DNA plasmid containing a gfp cDNA flanked by endogenous genomic sequences 5’ and 3’ to the nkx2.4b first exon (Figure 3A). Following recombination into the nkx2.4b genomic locus, the gfp transgene was expected to be expressed under the control of endogenous nkx2.4b regulatory elements, while simultaneously deleting the first exon of nkx2.4b (Figure 3A). Injected F0 fish were raised and outcrossed to generate heterozygous F1 progeny, which were examined from 1-7 dpf for expression of GFP. Embryos were identified that expressed GFP in the hypothalamus beginning at 1 dpf, and in the developing thyroid beginning at 1.5 dpf, continuing through 7 dpf. No expression of GFP in other tissues was observed at any time. GFP-positive offspring from a single F1 founder, designated as allele zd21, were raised and outcrossed to generate heterozygous F2 fish. Incrosses of F2 fish resulted in clutches containing approximately 25% GFP-negative embryos, suggesting mendelian inheritance of the gfp transgene. From 1-3 dpf, 100% (20/20) of GFP-positive F2 incross progeny showed transgene expression in the hypothalamus and thyroid primordium (Figure 3B–E). However, at 5 dpf we observed that 25% of F2 incross progeny (5/20) lacked GFP-expressing thyroid follicular cells in the jaw midline, while maintaining GFP expression in the hypothalamus (Figure 3F,G). These data suggest that nkx2.4b is required for the survival, but not the initial induction, of thyroid progenitors. At 1 month post-fertilization approximately 25% of F2 incross progeny (22/80) exhibited increased melanophore density and erythema (Figure 3H,I), similar to previously reported hypothyroid phenotypes in zebrafish ^{21, 22}.

Figure 3: Targeting a gfp transgene to the nkx2.4b locus.

A) Schematic showing the strategy for targeted insertion of a gfp transgene to replace the first exon of nkx2.4b. Genotyping primers for panel (G) are indicated by arrows. B,C) Whole mount image of a live embryo at 1 dpf (B) and a confocal maximum-Z projection of a dissected brain (ventral view) at 7 dpf (C), showing GFP expression in the hypothalamus of nkx2.4b^zd21 fish. D,E) Live nkx2.4b^zd21 embryos at 2 dpf (D), and 3 dpf (E), showing GFP expression in the thyroid primordium (arrows). Expression of GFP in the hypothalamus is obscured by the retinal pigment epithelium in panel (D). F,G) Ventral views of live larvae at 5 dpf, showing GFP-expressing thyroid follicular cells present in the jaw midline of nkx2.4b^zd21/+ larvae (arrow in F), but absent in nkx2.4b^zd21/zd21 larvae (G). In both genotypes, hypothalamic GFP expression is visible out of the plane of focus. H,I) Gross morphological comparison at 1 month post-fertilization showing nkx2.4b^zd21/zd21 fish with increased melanophore density and erythema, similar to previously described hypothyroid phenotypes. Yellow boxes indicate magnified regions. J) Genomic PCR analysis of GFP-negative progeny from a nkx2.4b^zd21/+ incross shows the presence of an amplicon for nxk2.4b exon 1 (137bp product), and no amplicon for gfp (356bp product, a smaller non-specific amplicon is marked with asterisk). Analysis of GFP-positive progeny with hypothyroid morphology shows the absence of the nxk2.4b exon 1 amplicon and the presence of the gfp amplicon.

PCR analysis of 4 GFP-negative F2 incross progeny produced an amplicon for exon 1 of nkx2.4b, while in 4 GFP-positive F2 incross progeny with hypothyroid phenotypes the exon 1 amplicon was absent (Figure 3J). These results, together with observed GFP expression in 22/22 F2 incross progeny exhibiting hypothyroid phenotypes, suggest that the gfp transgene was successfully targeted to the endogenous nkx2.4b locus, and replaced exon 1. However, we were unable to directly amplify and sequence genomic DNA spanning the homology arms of the targeting construct from transgenic fish, and thus could not confirm whether the zd21 allele resulted from precise homologous recombination of the transgene.

Here we have described the generation of two zebrafish lines carrying predicted nkx2.4b loss-of-function mutations, both exhibiting hypothyroid phenotypes. Our data indicate that insertion of a gfp transgene into the nkx2.4b locus, achieved by co-injection of TALEN mRNAs and a targeting vector, created a line expressing GFP specifically in the hypothalamus and thyroid. Future targeted insertions using constructs expressing Cre recombinase or other effector molecules will allow genetic manipulation of both tissues.

Experimental procedures

Injections and establishment of lines

All fish were used and handled in accordance with University of Utah IACUC guidelines. Wild-type Tübingen strain fish were crossed to generate embryos. Injections were performed at the 1-cell stage by injecting each TALENs mRNA at a concentration of 50pg/nl. Forward and reverse primers (5’-CTAATTTGCAGCAGCACAGCAT-3’ and 5’-CATGCTCTCCTGGTAGGATGG-3’, respectively) were optimized for High Resolution Melt Analysis (HRMA) using an Idaho Technologies LightScanner to assay for in/del mutations. Establishment of lines carrying mutations including the zd20 allele was carried out by raising injected F0 fish to maturity, outcrossing them to WT fish, then using HRMA and direct sequencing to verify and characterize the presence of mutations in F1 offspring.

Generation and establishment of the zd21 line in which GFP was targeted to the nkx2.4b locus was performed in the same manner as described above, except both TALENs (50pg/nl of each) and a targeting vector (25pg/nl uncut plasmid DNA) containing a gfp transgene flanked by nkx2.4b non-coding genomic sequences were co-injected at the one cell stage. Injected F0 fish were outcrossed and embryos were visually inspected for the expression of GFP in the hypothalamus and thyroid. PCR amplification for genotyping was performed on fin clip DNA with an annealing temperature of 59ºC using the following primers:

nkx2.4b exon 1:	5’-CTAATTTGCAGCAGCACAGCAT-3’
	5’-CATGCTCTCCTGGTAGGATGG-3’
gfp:	5’-TGAGGGATACGTGCAGGAGAG-3’
	5’-AAAGGGCAGATTGTGTGGAC-3’

Imaging and analysis

Whole mount imaging of embryos was performed on an Olympus SZX16 dissecting microscope, with fluorescence used to validate GFP expression in the zd21 allele. Live embryos were imaged at 4X-20X magnification on an Olympus BX51WI fluorescent compound microscope at 1 day post-fertilization (dpf), 2 dpf, 3 dpf, and 7 dpf to document GFP expression patterns. Confocal imaging was performed with an Olympus FV1000 microscope using a 60x oil objective. Images were generated from maximum intensity z-projections of 3 μm slices.

TALENs

TALENs constructs were designed by the University of Utah Genomic Mutation Generation and Detection Core facility using the nkx2.4b coding sequence from the Zv9 genomic build. pCS2TAL3-nkx-TALEN-DDD was targeted to the 21 base pair (bp) DNA sequence: 5’-ACCACATGCCGCACTCGCAGT-3’ (GRCz11 Chromosome 20: 48,803,463-48,803,483) and pCS2TAL3-nkx-TALEN-RRR was targeted to the 21bp DNA sequence: 5’-CCCGTTGCAATATCCGCCCAT-3’ (CRCz11 Chromosome 20: 48,803,498-48,803,518). These flanked a 14bp spacer sequence in the nkx2.4b gene: 5’-GGTGCTTGCGAGA-3’ (CRCz11 Chromosome 20: 48,803,484-48,803,497). The pCS2 vector was linearized using Not1 restriction enzyme (NEB R0189S) and mRNA for injection was prepared using a mMessage mMachine SP6 transcription kit (Ambion, AM1340).

Construction of GFP transgene targeting vector

Genomic sequences were PCR amplified from the same family of Tübingen fish used to establish the zd21 allele.

A 5’ Gateway entry clone was created containing 2672bp of genomic DNA ending just upstream of the nkx2.4b start codon, spanning the region from CRCz11 Chromosome 20: 48,800,609-48,803,281. PCR amplification was performed from Tübingen genomic DNA using the following primer sequences, with the addition of att sites for BP cloning:

Forward primer: 5’-CCGATAGGCCTACAGCTCAG-3’

Reverse primer: 5’-GGTTCAGCCCGCGAGCAGGT-3’

A middle Gateway entry clone was created containing the BamHI restriction site, Kozak consensus sequence, and mmgfp5 cDNA from the pG1 plasmid ²³. PCR amplification was performed from plasmid DNA using the following primer sequences, with the addition of att sites for BP cloning:

Forward primer: 5’-AGGATCCACAGCCACCATGAGTAA-3’

Reverse primer: 5’-TTATTTGTATAGTTCATCCA-3’

A 3’ Gateway entry clone was created containing 1733bp of nkx2.4b intron 1 sequence, spanning the region from CRCz11 Chromosome 20: 48,803,982-48,805,715. PCR amplification was performed from Tübingen genomic DNA using the following primer sequences, with the addition of att sites for BP cloning:

Forward primer: 5’-GTTTAGCGCTTGAGATTGGA-3’

Reverse primer: 5’-ACAATCATCAGTTGCGTTGC-3’

A Gateway LR reaction was then performed with the 3 entry clones and the pDestTol2CG3 destination vector ²⁴ to create a targeting plasmid.

In Situ Hybridization

In situ hybridization was carried out as previously described ²⁵. Probes used for thyroid specific mRNA expression were hhex, pax8, tg, and tpo, designed using primer sets from Opitz et al., ²⁶.