Loss of the epithelial transcription factor grhl3 leads to variably penetrant developmental phenotypes in zebrafish

Abstract

Background

Environmental influence is critical for embryogenesis but is significantly under‐appreciated under lab conditions, which are not typically designed to robustly test environmental variability. Here, we report environmental effects on the developmental phenotype of zebrafish lacking the transcription factor Grainyhead‐like 3 (grhl3), a highly conserved gene that is pivotal in epithelial barrier formation, neurulation, craniofacial development, and convergence‐extension.

Results

We had previously reported that deletion of grhl3 led to embryonic lethality by 11 h post‐fertilization (hpf); however, housing these grhl3‐lines in a different aquatic facility led to substantial differences in phenotypic presentation in grhl3‐nullizygous (grhl3 ^−/− ) embryos. We found that grhl3 ^−/− embryos presented with three distinct phenotypes, characterized by significant reductions in body length, aberrant orofacial cavity formation and craniofacial morphogenesis and impaired intestinal barrier maintenance.

Conclusions

Our study describes a new model of partial phenotypic penetrance in genetically identical embryos. This may serve as a valuable model system in which to understand gene–environment interactions in developmental and epithelial homeostasis.

Key Findings

• Deletion of Grainyhead‐like 3 (grhl3) in zebrafish leads to 3 distinct axialbody development phenotypes, of varying severity.

• These phenotypes vary depending on the housing environment of the grhl3^+/− adults.

• Further phenotypes include malformations of the orofacial cavity, craniofacial skeleton, and intestinal barrier.

1. INTRODUCTION

The grainyhead (grh) transcription factor was first identified in Drosophila, ¹ and experimentally validated orthologues have since been identified in C. elegans, ² Xenopus laevis, ³ zebrafish, ⁴ , ⁵ , ⁶ , ⁷ , ⁸ mice, ⁹ humans, ¹⁰ and fungi. ¹¹ The sub‐functionalized mammalian orthologues, namely Grhl1, Grhl2, and Grhl3, have been well described as key epithelial regulators in murine development, wound healing, neural tube closure, craniofacial formation, epidermal cancer, and skin barrier homeostasis. ⁹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ Moreover, GRHL function is also highly conserved in human craniofacial and neurological development and adult‐onset epithelial disease. ¹⁰ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Together, these studies have highlighted an incredible functional genetic conservation over more than 700 million years of evolution.

Zebrafish comprise four sub‐functionalized GRHL orthologues, namely grhl1, grhl2a, grhl2b, and grhl3. ²⁸ Morpholino (MO) mediated knockdown of zebrafish grhl3 leads to multiple developmental defects, including craniofacial hypoplasia, ⁸ shortening of body axis and curved tail at 96 hpf, ⁵ , ⁸ and aberrant folding and morphogenesis of the midbrain‐hindbrain boundary. ⁵ Zebrafish grhl3 is also strongly and specifically expressed in all cells of the enveloping layer (EVL) at the onset of the maternal‐zygotic transition. ⁵ Consistent with this expression pattern, we had previously shown that CRISPR/Cas9‐mediated germline deletion of grhl3 led to fully penetrant premature contraction of the actin ring at 75% epiboly, delayed ventral epithelial migration and decreased cell size in the grhl3 ^−/− embryos and defective EVL fidelity, resulting in embryonic rupture and death by 11hpf. ⁵

Here, we report an unexpected change in the zebrafish grhl3‐deletion phenotype upon transferral of this line to a new zebrafish housing facility. We further substantiate these findings by generating and characterizing two additional grhl3 ^−/− lines (grhl3 ^−31bp and grhl3 ^Δ1–4) that present with a highly concordant phenotypic spectrum. Lastly, we identify a putative novel downstream target gene that significantly reduces grhl3 ^−/− embryo mortality and substantially ameliorates overall phenotypic severity in this model.

2. RESULTS

2.1. grhl3 ^{−/−(+14bp)} embryos survive past gastrulation and display severe body development defects when transferred to a new environment

We had previously reported that grhl3 ^−/− embryos generated from the grhl3 ^+/−(+14bp) zebrafish line presented with defective EVL integrity, culminating in periderm rupture and death by 11 hpf. ⁵ In our initial study, embryos likely represent heterozygous grhl3 ^+/−(+14bp) adults, or grhl3 ^+/? (control) embryos did not present with any observable phenotypes or noticeable survival impairment, consistent with previous reports indicating that loss of a single Grhl3 allele in heterozygous mammalian models also did not lead to observable phenotypes. ²⁹ However, when grhl3 ^+/−(+14bp) adult zebrafish were transferred and maintained at a new facility in Melbourne, Australia, we noticed a definitive change in the phenotypic presentation of grhl3 ^{−/−(+14bp)} embryos. Surprisingly, we found that grhl3 ^{−/−(+14bp)} embryos now progressed through the previously described 11 hpf “gastrulation block,” with no evidence of impaired actin ring contractility or yolk bulging. Moreover, these embryos did not rupture and were largely phenotypically indistinct from grhl3 ^+/? sibling embryos until 19hpf, with no appreciable decrease in survival.

From 19 hpf, grhl3‐nullizygous embryos exhibited a slight bend in the vertebral axis, toward the tail, and by 48hpf, three phenotypes showing varying degrees of body development defects were seen in embryos subsequently genotyped as grhl3 ^−/− (Figure 1A). Based on the severity of phenotypes, these embryos were qualitatively categorized into three distinct (discrete) phenotypic classes, which we have termed class 1, class 2, and class 3. Class 1 embryos (199/2048; 9.72%) exhibited relatively minor fin fold defects and tail bending at the most distal point. Class 2 mutants (55/2048; 2.69%) displayed trunk/tail curvature along with marked shortening of body length. Class 3 grhl3 ^−/− embryos (26/2048; 1.27%) exhibited the most severe phenotypes, displaying severe body‐extension defects (tail rarely extended beyond the yolk sac) alongside substantially impaired development of the head and craniofacial skeleton (Figure 1A). Importantly, these defects (and their incidence) were consistent across multiple clutches from numerous breeding adult pairs. These data indicate that a phenotypic spectrum exists following the loss of grhl3 when fish are raised within environments that may be more permissive for embryonic survival.

FIGURE 1.

Characterization of grhl3 ^{−/−(+14bp)} mutants. (A) Three distinct phenotypes of grhl3 ^{−/−(+14bp)} embryos are present at 72 hpf (classified as class 1, class 2, and class 3 based on the severity and incidence of phenotypes). (B) Structure of a functional Grhl protein comprising 557 amino acids, showing transactivation domain (orange), DNA binding domain (blue), and dimerization domain (yellow) relative to predicted non‐functional proteins produced following CRISPR/Cas 9‐mediated deletion in grhl3 ^{−/−(+14bp)}, grhl3 ^{−/−(−31bp)}, and grhl3 ^{−/−(Δ1‐4)} zebrafish. The grhl3 mutants were generated by targeting exon 4 at a site lying 3′ to seven putative in‐frame ATG‐initiation codons (indicated by red arrow) via CRISPR/Cas 9 mediated deletion and homologous integration. (C) The three distinct phenotypes (class1‐3) of grhl3 ^{−/−(−31bp)} and grhl3 ^{−/−(Δ1‐4)} embryos at 72 hpf. Scale bars: 200 μm.

2.2. grhl3 ^{−/−(−31bp)} and grhl3 ^{−/−(Δ1‐4)} embryos display phenotypes identical to those of grhl3 ^{−/−(+14bp)} embryos

To determine whether these altered phenotypes were unique to the grhl3 ^{−/−(+14bp)} line, ⁵ we generated two additional CRISPR knockout lines, grhl3 ^(−31bp) and grhl3 ^(Δ1‐4). These lines were created to determine whether the facility‐specific differences in phenotypic penetrance were an artifact of the grhl3 ^(+14bp) line, or whether they were consistent with other disruptions to the grhl3 coding region. As the ATG‐initiation codon is present in exon 4 of the grhl3 gene, the zebrafish grhl3 mutants were generated by targeting exon 4 at a site lying 3′ to seven putative in‐frame ATG‐initiation codons (Figure 1B). grhl3 ^{−/−(−31bp)} mutants contain a 31 bp deletion in exon 4, which results in a frameshift mutation to produce a truncated peptide of 224 amino acids lacking the conserved DNA binding domain. The grhl3 ^(Δ1‐4) mutant line was generated using two sets of guide RNAs, one targeting exon 1 and the other targeting exon 4, which resulted in the deletion of exons 1–4, generating a predicted non‐functional grhl3 protein of 118 amino acids. Sequencing of the locus confirmed that we had generated three distinct genomic deletions within the grhl3 coding region.

Following inter‐crossing heterozygous adults of both lines and subsequent phenotypic characterization of resultant embryos, we noted the appearance of the same three classes of body development defect we noted in the grhl3 ^{−/−(+14bp)} embryos (Figure 1C). Quantitation of the phenotypic penetrance revealed no significant difference in the proportions of body development defects between embryos derived from either grhl3 ^{+/−(−31bp)} adults (class 1–9.41%; class 2–3.41%; and class 3–2.18%) or grhl3 ^{+/−(Δ1‐4)} adults (class 1–7.31%; class 2–3.41%; and class 3–1.46%) when compared to the grhl3 ^(+14bp) adults. These data indicated that there were no significant qualitative or quantitative differences in the phenotypic presentation of embryos following grhl3‐deletion across any of the three lines, supporting the hypothesis that the development defects we observed were not an artifact unique to the grhl3 ^{−/−(+14bp)} mutation.

Next, we quantified the overall survival rate of embryos from 0 to 72 h post‐fertilization (hpf; Figure 2A) and found that embryos derived from both grhl3 ^{+/−(−31bp)} and grhl3 ^{+/−(Δ1‐4)} lines presented with no significant differences in the survival rate compared to grhl3 ^{−/−(+14bp)} embryos. We also quantified the incidence of class 1, class 2, and class 3 embryos across all three lines, and again found no significant differences in the percentage of embryos with each class of phenotype across all three lines (Figure 2B).

FIGURE 2.

Survival rates and genotypes of embryos derived from inter‐crossing grhl3 ^+/− zebrafish of all three lines. (A) Comparison of survival rate of embryos derived from inter‐crossing the grhl3 ^+/−(+14bp), grhl3 ^{+/−(−31bp)}, and grhl3 ^{+/−(Δ1‐4)} lines. (B) Comparison of the phenotypic incidence of class 1, class 2, and class 3 phenotypes in grhl3 ^{−/−(+14bp)}, grhl3 ^{−/−(−31bp)}, and grhl3 ^{−/−(Δ1‐4)} embryos at 72 hpf. (C) Percentage of heterozygous, homozygous null, and WT larvae present at 5 dpf in a representative sample of 39–40 embryos. (D–F) Representative genotyping experiments showing the presence of wild type (+/+), heterozygous (+/−), and homozygous null (−/−) embryos of grhl3 ^(+14bp), grhl3 ^(−31bp), and grhl3 ^(Δ1‐4) lines.

Given we did not see elevated death rates in clutches derived from inter‐crossing grhl3 ^+/− fish, yet only ~15% of embryos presented with phenotypes (below the expected Mendelian ratio of 25%), we sought to determine whether all the phenotypic embryos were indeed nullizygous for grhl3, and conversely, also whether any nullizygous embryos were non‐phenotypic. We therefore genotyped 39–40 embryos from each line at 5 dpf, comprising all phenotypic embryos (6/40 for the grhl3 ^(+14bp), 7/39 for the grhl3 ^(−31bp) line, and 6/39 for the grhl3 ^(Δ1‐4) embryos), alongside non‐phenotypic siblings. As expected, all phenotypic embryos were nullizygous for grhl3 (Figure 2C). Unexpectedly, however, we found that numerous non‐phenotypic embryos (7/40 for the grhl3 ^(+14bp) and 5/39 for the grhl3 ^(−31bp) line) were also grhl3 ^−/− , indicating non‐presentation of phenotypic defects despite grhl3‐nullizygosity (Figure 2D–F). Notably, we did not see any non‐phenotypic grhl3 ^−/− embryos derived from inter‐crossing grhl3 ^{+/−(Δ1‐4)} fish at 5 dpf. This suggests that the phenotypic spectrum due to grhl3 ^(Δ1‐4) deletion is more severe than that observed for the other two lines.

2.3. grhl3 ^−/− embryos present with defects in trunk/tail extension

To determine the consequences of grhl3 deletion on the growth of zebrafish embryos, the body length of class 1, class 2, and class 3 grhl3 ^−/− mutant embryos was quantified (n = 10), relative to non‐phenotypic siblings at 5 dpf. The mean body lengths of class (2.5 ± 0.3 mm), class 2 (1.6 ± 0.2 mm), and class 3 (1.2 ± 0.1 mm) embryos, respectively, were significantly shorter than grhl3 ^+/? control siblings (3.5 ± 0.15). These measurements were highly consistent across all three lines, indicating a significant reduction in the body length of class 1 (p <.0001), class 2 (p <.0001), and class 3 (p <.0001) grhl3 ^−/− embryos (Figure 3A–C) relative to non‐phenotypic controls. This near‐identical presentation of the phenotypic spectrum across all three grhl3‐lines further supports the notion that these phenotypes were true biological consequences of grhl3‐loss rather than artifacts or off‐target effects of CRISPR‐mediated deletion, likely due to stochastic differences in target gene regulation.

FIGURE 3.

Maximum body length and survival age quantitation of grhl3 ^−/− embryos. (A–C) Quantitation of body length of class 1, class 2, and class 3 grhl3 ^{−/−(+14 bp)}, grhl3 ^{−/−(−31bp)}, and grhl3 ^{−/−(Δ1‐4)} embryos. Data are represented as mean ± SEM; n = 10. (D–F) Maximum survival age of grhl3 ^{−/−(+14bp)}, grhl3 ^{−/−(−31 bp)}, and grhl3 ^{−/−(Δ1‐4)} embryos. ***p <.005; ****p <.001; ns, non‐significant. Statistical significance determined by Student's t‐tests.

2.4. grhl3 ^−/− embryos do not survive beyond 10 dpf

As we had previously noted that grhl3‐deletion was embryonic lethal by 11 hpf, ⁵ yet we now saw grhl3 ^−/− embryos survive until at least 72 hpf, we sought to determine the consequences of grhl3 loss on maximum survival age in this new housing environment (Figure 3D–F). Therefore, we monitored grhl3 ^−/− phenotypic mutants and non‐phenotypic siblings alongside control embryos derived from WT adults. Phenotypic embryos of all three lines began to show increased mortality from 4 dpf, with the most substantial increase in embryo lethality seen between 6 and 7 dpf in the grhl3 ^(+14bp) (84.5% alive at 6 dpf; 11.26% at 7 dpf) and grhl3 ^(−31bp) (73.27% alive at 6 dpf; 4.95% at 7 dpf) embryos and between 7 and 8 dpf in the grhl3 ^(Δ1‐4) line (73.21% alive at 7 dpf; 14.28% at 8 dpf). Non‐phenotypic siblings from the grhl3 ^(+14bp), grhl3 ^(Δ1‐4), and WT lines did not present with any significant differences in lethality by 10 dpf. However, we did note a subtle difference in the non‐phenotypic embryos of the grhl3 ^(−31bp) line, which presented with reduced survival from 8 dpf (Figure 3D–F). Taken together, these data demonstrate subtle differences in embryonic survival depending on the nature of the germline grhl3 deletion; however, overall, the high consistency of phenotype across all three of our lines demonstrates that grhl3 is essential for embryo survival and larval development in zebrafish beyond 10 dpf.

2.5. Loss of grhl3 causes defects in craniofacial skeleton development

We had previously reported that transient, morpholino‐mediated knockdown of grhl3 in zebrafish embryos led to defects in the formation of the craniofacial skeleton, particularly hypoplasia of the Meckel's cartilage (MC), ceratohyal, palatoquadrate, and ceratobranchials. ⁸ However, we had not previously been able to analyze craniofacial skeleton formation in the grhl3 ^−/− embryos of the grhl3 ^(+14bp) line, owing to the gastrulation‐stage lethality. ⁵ Therefore, to determine the consequences of germline grhl3 deletion on craniofacial development, we examined the craniofacial cartilage of phenotypic grhl3 ^{−/−(+14bp)} and grhl3 non‐phenotypic sibling embryos at 4 dpf (Figure 4A–H). Class 2 (Figure 4C,G) and class 3 (Figure 4D,H) embryos presented with substantial pronounced reductions in the formation of MC, palatoquadrate, ceratohyal, and ceratobranchials, alongside further defects in the formation of the ethmoid plate and hyosymplectic. The jaw phenotypes observed in class 2 and class 3 embryos of all three lines were much more severe than the craniofacial defects observed in class 1 grhl3 ^−/− embryos, likely a secondary consequence of impaired development of the entire embryo.

FIGURE 4.

Craniofacial development in grhl3 ^{−/−(+14 bp)} embryos at 4 dpf. (A–H) Class 1, class 2, and class 3 grhl3 ^{−/−(+14bp)} embryos and non‐phenotypic siblings (4 dpf) stained with Alcian blue, showing lateral (A–D) and ventral (E–H) views of the craniofacial cartilage (palatoquadrate [pq], ceratohyal [ch] Meckel's cartilage [mc], ethmoid plate [ep], and hyosymplectic [hs]). (I‐I') The height between Meckel's cartilage and ethmoid plate (indicated by vertical double‐headed red arrow) and (J‐J') the distance between Meckel's cartilage and ceratohyal (indicated by horizontal double‐headed red arrow) were measured and analyzed in grhl3 ^{−/−(+14bp)} embryos and non‐phenotypic siblings. (K, L) H&E staining of transverse sections across the rostral region of 7 dpf class 1 grhl3 ^−/− mutant and control embryos (ceratohyal [ch] trabecular bar, orofacial cavity [ofc]). (M) Orofacial cavity area of grhl3 ^−/− and control embryos. Data are shown as mean ± SEM; n = 5. Statistical significance determined by Student's t‐test ****p <.001. Scale bars: 100 μm (A‐J'), 50 μm (K, L).

We, therefore, quantified only the relatively more subtle defects seen in class 1 grhl3 ^{−/−(+14bp)} embryos, which confirmed a complete absence of ceratobranchials, reduction in the height between lower and upper jaw, and diminished distance between ceratohyal and MC. The vertical separation between lower and upper jaw in grhl3 ^{−/−(+14bp)} class 1 embryos (26.1 ± 7.6 μm) was significantly lower than in grhl3 ^+/?(+14bp) non‐phenotypic siblings (82.9 ± 7.1 μm; p <.0001; Figure 4I') and the lateral distance between the anterior‐most points of the ceratohyal and MC in grhl3 ^{−/−(+14bp)} embryos (83.6 ± 9.6 μm) was significantly shorter than in grhl3 ^+/?(+14bp) siblings (158.3 ± 12.4 μm; p <.0001; Figure 4J'). Similar craniofacial phenotypes were seen in grhl3 ^{−/−(−31bp)} (Figure 5) and grhl3 ^{−/−(Δ1‐4)} embryos (Figure 6). Together, our data from class 1 germline grhl3‐mutants are consistent with craniofacial cartilage hypoplasia demonstrated previously following transient grhl3 knockdown.

FIGURE 5.

Craniofacial development in grhl3 ^{−/−(−31 bp)} embryos at 4 dpf. (A–H) Class 1, class 2, ad class 3 grhl3^{−/−(−31bp)} embryos and non‐phenotypic siblings (4 dpf) stained with Alcian blue, showing lateral (A–D) and ventral (E–H) views of the craniofacial cartilage (palatoquadrate [pq], ceratohyal [ch] Meckel's cartilage [mc], ethmoid plate [ep], hyosymplectic [hs]). (I‐I') The height between Meckel's cartilage and ethmoid plate (indicated by vertical double‐headed red arrow) and (J‐J') the distance between Meckel's cartilage and ceratohyal (indicated by horizontal double‐headed red arrow) were measured and analyzed in grhl3^{−/−(−31bp)} embryos and non‐phenotypic siblings. Data are shown as mean ± SEM; n = 5. Statistical significance determined by Student's t‐test ****p <.001. Scale bars: 100 μm (A‐J').

FIGURE 6.

Craniofacial development in grhl3^{−/−(Δ1‐4)} embryos at 4 dpf. (A–H) Class 1, class 2, and class 3 grhl3 ^{−/−(Δ1‐4)} embryos and non‐phenotypic siblings (4 dpf) stained with Alcian blue, showing lateral (A–D) and ventral (E–H) views of the craniofacial cartilage (palatoquadrate [pq], ceratohyal [ch] Meckel's cartilage [mc], ethmoid plate [ep], hyosymplectic [hs]). (G') The height between Meckel's cartilage and ethmoid plate (indicated by vertical double‐headed red arrow) and (H') the distance between Meckel's cartilage and ceratohyal (indicated by horizontal double‐headed red arrow) were measured and analyzed in grhl3 ^{−/(Δ1‐4)} embryos and non‐phenotypic siblings. Data are shown as mean ± SEM; n = 5. Statistical significance determined by Student's t‐test ****p <.001. Scale bars: 100 μm (A‐J').

To further explore the role of grhl3 in orofacial development, the orofacial cavity of class 1 grhl3 ^−/− larvae and grhl3 non‐phenotypic siblings was examined histologically at 7 dpf (Figure 4K,L). Although formation of the ceratohyal was largely normal in grhl3 ^−/− embryos, the area surrounding the ceratohyal was significantly reduced (5802.4 ± 1586.85 μm²) relative to non‐phenotypic siblings (12,720.7 ± 1221.8 μm²) indicating a significant reduction in the mean area of the orofacial cavity (p >.001; n = 5) in grhl3 ^−/− embryos (Figure 4M). Overall, we found that survival rates, body development defects, and craniofacial anomalies seen in grhl3 ^{−/−(+14bp)}, grhl3 ^{−/−(Δ1‐4)}, and grhl3 ^{−/−(−31bp)} embryos were highly consistent, with no substantive unique differences observed in grhl3‐nullizygous embryos of any of these three lines. Taken together, these data indicated that the phenotypes observed were due to abrogated grhl3‐signaling rather than off‐target or random effects unique to the grhl3 ^(+14bp) line.

2.6. Loss of grhl3 leads to intestinal morphology defects and reduction in swim bladder size and inflation

Common reasons why zebrafish larvae die around 6–7 dpf include aberrations in correct intestinal formation and a failure of swim bladder inflation as a secondary issue caused by craniofacial defects, which can impact air “gulping” to maintain buoyancy. ³⁰ , ³¹ We therefore investigated whether the increased embryonic lethality of grhl3‐deficient embryos was due to either of these problems. We first examined the morphology of the esophageal‐intestinal (OI) junction and the medial intestine epithelium of 5 dpf grhl3 ^{−/−(−31bp)} embryos and non‐phenotypic siblings and found no aberrant morphology in gut formation (Figure 7A–D'). Next, we examined the expression of E‐cadherin (Figure 7E,F), an epithelial‐specific cell‐adhesion molecule and known direct grhl3‐target gene, ³² , ³³ and similarly detected no differences in expression, suggesting that overall epithelial establishment in the gut of embryos lacking grhl3 was not compromised. Lastly, the intestinal epithelial morphology was investigated in grhl3 ^{−/−(−31bp)} larvae at 7 dpf, as this was the timepoint consistent with maximum embryonic lethality in grhl3 ^{−/−(−31bp)} fish. We found that grhl3 ^{−/−(−31bp)} mutants displayed hemorrhage in the intestinal lumen, and the basement and apical membranes appeared to be poorly established and failed to form a distinct boundary both on the external and internal surfaces of the intestinal epithelium (Figure 7G,H). The epithelial cells lining the intestine also appeared disorganized, and qualitative observation suggested the cell shape was altered relative to controls (Figure 7G,H). These results suggest that failure to maintain intestinal integrity following the transition to self‐feeding during zebrafish development may underpin some of the embryonic lethality seen following the loss of grhl3.

FIGURE 7.

Intestinal epithelial and swim bladder morphology in grhl3 ^{−/−(−31)} mutants. (A–D) Transverse‐sections of 5 dpf intestine of control (A, B, B') and class 1 grhl3 ^{−/−(−31)} (C, D, D') embryos highlighting the morphology of the intestinal region. (E, F) E‐cadherin (green) and rhodamine phalloidin (red) expression in the intestinal lumen of control (E) and grhl3 ^{−/−(−31)} (F) 5 dpf embryos. (G, H) Transverse‐sections of 7 dpf intestine of control (G) and class 1 grhl3 ^−/− mutants (H); black arrows show extensive red blood cell infiltration into the intestinal lumen. (I, J) Swim bladder morphology and size in grhl3 ^−/− mutants and non‐phenotypic siblings. (K) Area of swim bladder in grhl3 ^−/− embryos and non‐phenotypic siblings. Data represented as mean ± SEM; n = 3. Statistical significance determined by Student's t‐test *p <.05. Scale bars: 20 μm (A, C, B', D', G, H), 50 μm (B, D, E, F) and 100 μm (I, J). Esophageal‐Intestinal Junction [oi]; Caudal Intestine [C.I.]; Swim bladder [sb]; intestine[i]; yolk sac [ys]; notochord [nc]; somatic muscle [sm].

Lastly, we examined swim bladder formation in embryos lacking grhl3. Although we detected no qualitative differences in the formation of the swim bladder epithelia (Figure 7I,J), we found that the average swim bladder area in grhl3 ^{−/−(−31bp)} mutants was 12,022 ± 4072.8 μm² while in non‐phenotypic siblings it was 26,336.17 ± 10,749.61 μm² (n = 5; p = .024; Figure 7I–K), indicative of greatly reduced volume. Whether this is a primary defect in swim bladder establishment, or whether it is secondary to the craniofacial defects we observed, remains to be determined; however, the lack of inflation and subsequent defects in larval buoyancy may also contribute to lethality at 6–7 dpf in embryos lacking grhl3.

3. DISCUSSION

Previously, we had reported that deletion of grhl3 in zebrafish led to embryonic lethality by ~11 hpf, due to complete embryo rupture during gastrulation. Here, however, we report a novel case of environmental (housing) influence on phenotypic penetrance. We have observed a striking improvement in embryonic survival and development when this line was housed and maintained at a new facility. Our work suggests that environmental factors can strongly influence the phenotypic outcome upon grhl3‐loss in zebrafish. Although the precise conditions responsible for this improvement are currently under investigation, this provides an impetus to discover whether bioactive factors may similarly improve clinical outcomes for patients with GRHL3‐mediated developmental defects such as spina bifida or cleft palate.

Phenotypic analysis of grhl3 ^−/− embryos revealed the presence of reduced body length, craniofacial hypoplasia, and impaired swim bladder inflation, very similar to those we described previously in grhl3 morphants. ⁸ However, grhl3 null embryos displayed three classes of body development defect severity, as opposed to the more consistent phenotype seen in grhl3 morphants. Although it is possible that the mechanisms behind each of the three defects are distinct, the precise genetic, biochemical and cellular reasons for the occurrence of three discrete axial phenotypes in grhl3 ^−/− embryos is currently unknown. That being said, transcription factors (TFs) such as the grh/Grhl proteins bind to the promoters of target genes, where they homo‐ and heterodimerize to regulate transcription through recruitment of transcriptional coactivators. Many of these Grainyhead‐like target genes have been experimentally identified and validated using various methodologies, such as RNA‐SEQ and ChIP‐SEQ approaches. ³³ , ³⁷ Previously, through in‐silico analysis, we have identified 305 novel predicted targets (a database that we have termed PHYLO‐PROM), ³⁷ many of which we have subsequently validated biologically (e.g., Tgm1/5, ³⁴ ARHGEF19, ¹⁶ Dsg1, ³⁵ eng2a and spec1, ²⁸ edn1, ⁸ PTEN, ³⁶ and Noggin ¹⁹ ). From our PHYLO‐PROM dataset, we had previously identified putative target genes that are also downregulated in grhl3 ^−/− zebrafish, including cldn23a, periplakin, prominin 2, occludin a, and aldh1a3. ³⁷ Many of these genes are involved in maintaining epithelial and/or epidermal integrity and, as such, may mediate body extension during gastrulation and subsequent early embryonic development. Our future work will determine whether any of these genetic mechanisms may be involved in mediating the grhl3‐phenotype in our model, alongside an unbiased RNA‐SEQ approach to characterize the transcriptome in class 1, class 2, class 3, and non‐phenotypic grhl3‐mutants to investigate whether a “resistance” or “susceptibility” transcriptional signature correlates with grhl3‐dependent embryogenesis. Of particular interest will be our group of non‐phenotypic grhl3 ^−/− embryos, as comparison of the transcriptome of these to that of WT controls will presumably eliminate a subset of genes that are differentially regulated yet do not contribute to the body‐extension phenotype.

Our findings are completely consistent with partially penetrant phenotypes observed in two mouse models with abrogated Grhl3 signaling, namely the curly‐tail and Grhl3 ^−/− mouse lines. ²⁹ The “curly‐tail” (ct) mice, harboring a single nucleotide polymorphism in a putative Grhl3 enhancer located ~21 kb upstream of the Grhl3 transcriptional start‐site ³⁸ retain ~30% Grhl3 activity and present with exencephaly (~3% of embryos), spina bifida (10%), and a curled tail (50%). ³⁹ The penetrance of these defects in the ct/ct mouse embryos is significantly influenced by environmental factors. Administration of retinoic acid, hydroxyurea, and mitomycin C reduces spina bifida but increases exencephaly, depending on the embryonic timepoint of administration. ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ Alternatively, treatment of ct mice with trypan blue, methotrexate, cortisone, progesterone, vitamin E, and, interestingly, folate had no effect on neural tube closure. Subsequently, inositol deficiency was found to exacerbate these defects, and injection of pregnant ct dams with inositol significantly reduced the incidence of spina bifida. ⁴⁶ Interestingly, inositol supplementation could not rescue defects in Grhl3 ^−/− mice. ⁴⁷ These data point to as yet undiscovered epigenetic influences in the regulation of Grhl3/grhl3 dependent signaling, and excitingly, these appear to be conserved between zebrafish and higher vertebrates. Our current work is focused on characterizing a range of bioactive supplements to determine whether they may also influence phenotypic severity, as seen in the ct mice.

The reduced body length and malformed craniofacial skeletons in grhl3 ^−/− embryos larvae hinder their ability to both reach and consume food, ⁴⁸ and also their ability to swim toward the surface to swallow the surface air effectively. ³⁰ , ⁴⁸ This contributes to the failure of swim bladder inflation and leads to poor buoyancy and impaired depth regulation. Together, these defects are likely to contribute to the lethality of grhl3 ^−/− larvae by 9–10 dpf.

Intestinal defects could also contribute to early larval death. Although we did not observe significant intestinal defects at 5 dpf, this is likely due to the fact that embryos draw nourishment from the yolk sac until 5 dpf, and only after this timepoint do they begin to feed exogenously; hence, little functional stress is placed on the developing gut. As differentiation of intestinal cells and barrier formation is reliant on the presence of microflora obtained from the environment, particularly food, ⁴⁹ subsequent poor colonization of the intestine with microflora may contribute to arrest in cell differentiation, ⁴⁹ concomitant with poor uptake of proteins and essential nutrients, ⁵⁰ , ⁵¹ as a secondary consequence. An alternate, and not mutually exclusive, hypothesis is that grhl3‐loss itself contributes to defective intestinogenesis or maintenance. Although not yet described in zebrafish, Grhl3 is expressed throughout the mouse intestinal epithelium during development, ¹² , ⁵² and Grhl3‐deletion in mice leads to a shorter intestine with blood in the lumen, ⁵³ suggesting a conserved role for Grhl3 in establishment and/or homeostasis of the intestinal epithelium. Therefore, it is likely that a combination of poor swimming motions due to body development defects, failure of swim bladder inflation, and intestinal epithelial defects secondary to craniofacial deformities contribute to lethality of grhl3 ^−/− embryos by 10 dpf, rather than one single factor being causative.

Taken together, our study has generated an intriguing new model of phenotypic penetrance in genetically identical animals and provides an excellent opportunity to interrogate genetic and non‐genetic factors that may exacerbate or ameliorate craniofacial, trunk/tail extension, and intestinal development and maintenance. Our current focus is on robustly characterizing the environmental differences present within zebrafish housing facilities that may have contributed to the differences in phenotypes seen in grhl3 ^−/− embryos, alongside a more in‐depth and robust characterization of intestinal development, homeostasis, and maintenance.

4. EXPERIMENTAL PROCEDURES

4.1. Fish husbandry and collection of embryos

Adult zebrafish were maintained in accordance with procedures approved at La Trobe University under Animal Ethics Committee project number AEC16‐91.

4.2. Genotyping

Adult zebrafish were fin‐clipped, and DNA was extracted using the Hotshot Genomic Protocol. One microliter of extracted genomic DNA was combined with 10 μL GoTaq® Green Master Mix (Promega), 0.5 μL forward primer, and 0.5 μL reverse primer in a final volume of 20 μL (made up of sterile water). PCR was performed using Bio‐Rad C1000™ Thermal Cycler with an initial denaturation step of 95°C for 3 min, followed by 34 cycles of 95°C for 30 s, 60°C/56°C for 30 s, and 72°C for 20 s, before a final extension step of 72°C for 5 min. Finally, the PCR samples were run on a 3% gel and the gel was electrophoresed at 90 V for 1 h. Gel was imaged using the Bio‐Rad Molecular Imager® ChemiDoc™. Primers for genotyping grhl3 ^(+14bp) embryos were: grhl3e4F (5′‐CAGCTCTTCCCCTGAACTTG‐3′), grhl3e4R (5′‐ACATAAATGCGGACCTCAGG‐3′), and grhl3 ^(+14bp)R (5′‐TGCCATGATTTCTGCTACAACA‐3′). Primers for genotyping grhl3 ^(−31bp) embryos were grhl3e4F and grhl3e4R. Primers for genotyping grhl3 ^(Δ1‐4) embryos were grhl3 ^(Δ1‐4)F (5′‐GCCAACCTGCCAGATGGATTA‐3′), grhl3e4F, and grhl3 ^(Δ1‐4)R (5′‐TAAATGCGGACCTCAGGTGT‐3′).

4.3. Alcian blue staining

Four days post‐fertilization of grhl3 ^−/− and grhl3 ^+/? (control) embryos were anesthetized in 0.04% tricaine solution and fixed in 4% PFA overnight at 4°C. The embryos were next rinsed with acid alcohol (70% ethanol:1% HCl) and immersed in 0.1% Alcian blue solution overnight at room temperature. The embryos were washed thoroughly with acid alcohol, followed by treatment with hydrogen peroxide solution (3% H₂O₂ in 1% KOH) until the eyes became transparent. The embryos were washed with 1 x PBST and stored in 70% glycerol at 4°C.

4.4. Histological analysis

grhl3 ^−/− and grhl3 ^+/? (control) siblings were fixed at 5 and 7 dpf in 4% PFA overnight at 4°C. This was then followed by washing in PBST and sequential immersion of larvae in 25% sucrose in PBS and 35% sucrose (dissolved in 1x PBS). The embryos were washed thoroughly and embedded in a cryomold filled with OCT (optimal cutting temperature) medium. The cryomold was gently submerged into cold 2‐methyl butane (Sigma Aldrich), followed by immersion into liquid nitrogen. Cryoblocks were stored at −80°C. Ten micromolar of cryosections were cut along the gut region in the cryostat (Leica CM1860), and the slides were counterstained with hematoxylin and eosin.

4.5. Histology and immunofluorescence

Larvae were anesthetized in 0.04% tricaine solution, fixed in 4% PFA, washed repeatedly in PBST, immersed in sucrose at 4°C and were cryopreserved in OCT and 10 μM cryosections were used for subsequent immunofluorescent histochemistry using standard techniques. Antibodies used in this study were rabbit anti‐zebrafish E‐cadherin (used 1:200; GenTex; Cat# GTX125890) and Rhodamine Phalloidin (used 1:1000; Thermo Fisher Cat# R415).

4.6. Microscopy and imaging

Live embryos were visualized using SZM 18 Fluorescence microscope/Nikon SMZ 745T dissecting microscope, and images were captured using the Nikon DS‐Ri2 camera. Histological sections were captured using the Olympus DP26 camera connected to an Olympus BX53 Fluorescence microscope or a Zeiss Confocal LSM 780 PicoQuant FLIM microscope. Body length quantitation of phenotypically distinct embryos was performed using NIS‐Elements® Imaging software and ImageJ.

4.7. Statistical analysis

Statistical analyses for all experiments were performed using GraphPad Prism 8. Unpaired Student's T‐tests and multiple T‐tests were used to compare the degree of significance between the control and experimental variable, where p values are as follows: *p ≤.05, **p ≤.01, ***p ≤.005, ****p ≤.001. Comparison of survival curves for maximum embryo survival was done using the Log‐rank (Mantel‐Cox) test; *p ≤.05, **p ≤.01, ***p ≤.005, ****p ≤.001. Correction for multiple comparisons was performed using the Holm–Sidak method *p <.05; **p <.01; ****p <.001.

Mathiyalagan N, Johnson TK, Di Pastena Z, Fuller JN, Miles LB, Dworkin S. Loss of the epithelial transcription factor grhl3 leads to variably penetrant developmental phenotypes in zebrafish. Developmental Dynamics. 2025;254(10):1133‐1147. doi: 10.1002/dvdy.70003