SIX1 BOR variants have different effects on embryonic craniofacial gene expression and cartilage formation

Abstract

SIX1 variants underlying Branchio-oto-renal syndrome occur in the SIX domain (SD) or homeodomain (HD). We tested whether different variants - V17E (SD), Y129C (HD) - cause distinct developmental phenotypes in embryos with reduced Six1 in comparison to wildtype Six1 (Six1WT). In Six1-morphants, Six1WT restored neural crest and preplacodal gene expression; V17E restored foxd3 and irx1 better than Y129C and Y129C restored sox11 better. In six1-null otic vesicles, Six1WT partially restored tbx1 and sobp, V17E was less effective and Y129C was least effective; all three restored dlx5. In six1-heterozygotes, Six1WT and Y129C had similar pleiotropic effects on tbx1, whereas V17E had no effect; Six1WT restored dlx5 expression, V17E was less effective and Y129C was most deficient. In six1-null tadpoles, reduced cranial cartilage volume and individual cartilage abnormalities were rescued by Six1WT, less so by V17E and not by Y129C. In heterozygotes and wildtypes, Y129C caused a higher frequency of abnormal cartilages compared to Six1WT or V17E. Thus, variants with different functional deficits have distinguishable effects in both nulls and heterozygotes on the formation of the tissues affected in Branchio-oto-renal syndrome.

SUMMARY STATEMENT

SIX1 variants underlying Branchio-oto-renal syndrome were expressed in Six1-depleted embryos. These variants differentially affected neural crest, preplacodal ectoderm, and otic vesicle gene expression, and cranial cartilage volume and morphology.

INTRODUCTION

Branchio-oto-renal syndrome (BOR) is an autosomal dominant hearing loss syndrome. Affected individuals present with highly variable combinations of malformations in neural crest-derived structures (hyoid region, external ear, middle ear), cranial placode-derived structures (inner ear) and in some cases kidney, and variants in SIX1 and EYA1 are causative in about half of the patients (Neal et al., 2024; Smith and Azaiez, 2025). Patients diagnosed as BOS3 (Branchio-otic syndrome-3; OMIM #608389) carry single nucleotide missense mutations or deletions in the gene encoding the SIX1 transcription factor, whose activity is modulated by cofactors such as EYA1 (Ikeda et al., 2002; Li et al., 2003; Silver et al., 2003; Tavares et al., 2021; Neal et al., 2024).

Studies of the developmental function of Six1 demonstrate that it plays a central role in the formation of structures derived from cranial neural crest (NC) and preplacodal ectoderm (PPE). Six1 loss-of-function in Xenopus, zebrafish, chick and mouse results in reduced expression of several PPE genes and defects in otic and craniofacial development (Laclef et al., 2003; Zheng et al., 2003; Brugmann et al., 2004; Ozaki et al., 2004; Zou et al., 2004; Bricaud and Collazo, 2006; Konishi et al., 2006; Chen et al., 2009; Christophorou et al., 2009; Ikeda et al., 2010; Coppenrath et al., 2021). Six1 contains an N-terminal protein-protein interaction domain (SD) and a homeodomain (HD) (Pignoni et al., 1997; Kawakami et al., 2000). The majority of causative SIX1 BOR variants are missense mutations that result in single amino acid substitutions or deletions in either the SD or HD (Fig. 1A; Neal et al., 2024). In cultured mammalian cells, V17E (SD) abolishes the SIX1-EYA1 interaction and prevents the translocation of EYA1 into the nucleus and Y129C (HD) significantly reduces DNA binding and transcription via a luciferase reporter construct (Ruf et al., 2004; Sanggaard et al., 2007; Kochhar et al., 2008; Patrick et al., 2009; Patrick et al., 2013; Shah et al., 2020; Lee et al., 2023).

Figure 1: Six1 protein domains and experimental approaches.

A. Xenopus Six1 protein indicating the Six Domain (SD), homeodomain (HD) and two BOR variants (V17E, Y129C). K* indicates the stop codon in Xenopus tropicalis mutants.

B. At the 4-cell stage (4CS), the animal poles of the left blastomeres of Xenopus laevis embryos were injected with translation-blocking Six1 morpholinos (MO). At the next cell cycle (8CS), the animal pole daughters were injected with mRNA encoding either wildtype or variant Six1. At neural plate stages, embryos were fixed and processed for neural crest and PPE gene expression by in situ hybridization (ISH).

C. At the 2-cell stage (2CS), one blastomere of Xenopus tropicalis embryos from the six1-mutant line was injected with mRNAs. At stages 28–32, larvae were fixed, genotyped and processed for otic vesicle gene expression by ISH.

D. At the 1-cell stage (1CS), Xenopus tropicalis zygotes from the six1-mutant line were injected with mRNAs. Tadpoles were fixed, genotyped and processed for changes in cartilage volume or morphology.

Previously we found that V17E and Y129C have different effects on gene expression when overexpressed in wildtype Xenopus embryos that carry a normal endogenous level of Six1 (Shah et al., 2020; Mehdizadeh et al., 2021). In these embryos, V17E altered the expression of NC, PPE and OV genes at frequencies very similar to an equivalent dose of wildtype Six1, whereas Y129C had notably weaker effects. To make those findings more relevant to the heterozygous BOR genotype, herein we expressed the variants on reduced Six1 backgrounds and assessed how they altered the expression of genes critical for the formation of the NC, PPE and OV, embryonic precursor populations highly relevant to BOR phenotypes. We also assessed the formation of tadpole cranial cartilage, derived from cranial NC, by quantifying cartilage volume using a deep learning–based three-dimensional reconstruction approach (Naert et al., 2021), and analyzing dysmorphologies in individual cartilages. These analyses demonstrate that variants with different functional deficits have distinguishable effects in both nulls and heterozygotes that impact the early development and morphogenesis of tissues affected in Branchio-oto-renal syndrome.

RESULTS

BOR variants differentially alter neural crest and preplacodal gene expression

BOR patients carry one copy of wildtype SIX1 and one copy of a causative variant. To approximate a heterozygous level of endogenous Six1 protein on one side of wildtype Xenopus laevis embryos, Six1 translation-blocking MOs were injected into the animal pole of the blastomeres on the left side of the 4-cell embryo (Fig. 1B). These MOs were previously demonstrated to be specific and efficiently knockdown translation of endogenous protein (Brugmann et al., 2004; Sullivan et al., 2019). At neural plate stages, embryos were processed by in situ hybridization (ISH) for the expression of genes involved in NC and PPE formation (Fig. 1B), precursor populations of the cranial cartilages and inner ear that are dysmorphic in BOR. foxd3 is essential for determining the cranial NC (Dottori et al., 2001; Sasai et al., 2001), irx1 is required for the formation of both NC and PPE (Glavic et al., 2004), and sox11 is required for PPE formation and neurogenesis (Brugmann et al., 2004; Chen et al., 2016). In each MO-injected embryo, the intensity of ISH staining was compared between injected (left) and uninjected (right) sides and qualitatively scored in categories: “fainter”, “larger” or “same”, (Table 1; see Methods for details). Bilateral comparison within the same embryo is essential because there is up to a 38% variation in diameter and 2.6-fold difference in volume in Xenopus zygotes, even in clutches derived from the same female (Leibovich et al., 2020). These size differences, as well as differences in the rates at which siblings grow, result in wide variations in gene expression levels and domain sizes between embryos, making intra-embryo comparisons essential.

Table 1:

Percentages of embryos with the noted intensity of ISH staining on the injected side compared to the uninjected side of the same embryo. Differences were scored as fainter, larger or the same (see Methods). Red font indicates effects of: 1) MO-only; 2) Six1WT differences from MO-only; and 3) corresponding effects caused by the variants. (n), indicates number of embryos examined.

	MO-only	MO+Six1WT	MO+V17E	MO+Y129C
foxd3	(37)	(42)	(45)	(42)
% fainter	2.7	83.3	73.3	23.9
% larger	78.4	2.40	15.6	38.1
% same	18.9	14.3	11.1	38.0
irx1	(47)	(40)	(41)	(78)
% fainter	0	82.5	80.5	59.0
% larger	66.0	0	4.9	11.5
% same	34.0	17.5	14.6	29.5
sox11	(34)	(33)	(44)	(48)
% fainter	82.4	36.4	68.2	10.4
% larger	2.9	27.3	20.5	2.1
% same	14.7	36.4	11.4	87.5
larger + same	17.6	63.7	31.9	89.6

The majority of MO-injected embryos (MO-only) had larger foxd3 NC domains, larger irx1 PPE domains and fainter sox11 PPE domains on the injected morphant side (Fig. 2A-C; Table 1), indicating that Six1 normally restricts foxd3 NC and irx1 PPE domains, and expands sox11 PPE domains. To determine the ability of wildtype Six1 to rescue the effect of reduced endogenous protein, we injected its mRNA (Six1WT) into the two 8-cell daughters of the MO-injected blastomere precursors of the NC and PPE (Fig. 1B). Supplying Six1WT to the morphant side caused foxd3 staining to be fainter on the injected side in the majority of embryos (Fig. 2D; Table 1), in significant contrast to MO-only (p<0.0001). Thus, Six1WT reversed the effects of endogenous Six1 knockdown. To confirm the staining categories, we also measured the sizes of foxd3 domains on both sides of a subset of embryos scored as fainter (Supplemental Fig. 1A); they were significantly smaller on the Six1WT-injected side (Fig. 2N), consistent with the scoring categories.

Figure 2: BOR variants differentially affected NC and PPE expression domains.

A-C. Morphants injected only with Six1MO (MO-only) on the left side and processed by ISH for foxd3 (A), irx1 (B) or sox11 (C). Asterisk denotes injected morphant side, black arrows denote expression domains on uninjected side, and red arrows denote expression domains on morphant side of the same embryo.

D-F. Examples of morphants injected with wildtype six1 mRNA (Six1MO+WT) that have fainter staining of foxd3 (D), irx1 (E) and sox11 (F) on the injected (red arrows) compared to uninjected (black arrows) sides.

G-I. Examples of morphants injected with V17E mRNA (Six1MO+V17E) that have fainter staining of foxd3 (G), irx1 (H) and sox11 (I) on the injected side.

J-M. Examples of morphants injected with Y129C mRNA (Six1MO+Y129C) that have a larger foxd3 domain (J), a fainter (K) or larger (L) irx1 domain and (M) the same sox11 domain on the injected side.

N. The perimeter of the foxd3 NC domain was measured in a subset of embryos. For Six1WT- and V17E-injected morphants with fainter staining on the injected side (inj), the domains were significantly smaller compared to uninjected side (uninj) of the same embryo. For Y129C-injected morphants, those visually scored with a larger staining domain measured significantly larger. ***, p<0.0001.

O. For all three proteins, the size of the fainter irx1 PPE domain was significantly smaller on the injected side. ***, p<0.0001.

P. For Six1WT and V17E, the sox11 domain was significantly smaller on the injected side. ***, p<0.0001. For Y129C, it was the same on both sides (ns, p>0.05).

Supplying Six1WT to morphants also reduced the effects of MO-injection on genes expressed in the PPE. In significant contrast to MO-only embryos (Fig. 2B), more Six1WT-injected morphants had fainter irx1 PPE staining and none had larger domains (Fig. 2E; Table 1; p<0.0001). Supporting this scoring, the sizes of the “fainter” irx1 domains were significantly smaller on the Six1WT-injected side (Fig. 2O). Although some Six1WT-injected morphants also had fainter sox11 domains (Fig. 2F), compared to MO-only proportionally more had either larger domains or the same staining intensity (Table 1; p=0.0003). Measuring the sox11 domains confirmed these qualitative categories (Fig. 2P; Supplemental Fig. 1C,D). Thus, Six1WT partially reversed the effects of endogenous Six1 knockdown on both PPE genes.

To determine whether BOR variants restored gene expression as effectively as Six1WT, we injected their mRNAs into the daughters of the MO-injected blastomeres (Fig. 1B). For foxd3, most V17E-injected morphants had fainter staining and significantly smaller domains on the injected side (Fig. 2G, N), similar to Six1WT-injected morphants (p=0.1182; Table 1). Since V17E and Six1WT each caused smaller foxd3 domains, we assessed whether the reductions were of the same magnitude by converting the size of the domain on the injected side to a percent of that on the uninjected side of the same embryo; this normalizes inter-embryo size variation. The mean percent reduction of V17E-injected foxd3 domains (13.2%) was significantly less than Six1WT (23.1%; p=0.0073), indicating that V17E is less effective than Six1WT at correcting MO-caused changes. In comparison, only a small percentage of Y129C-injected morphants had fainter foxd3 staining and many more had “larger” or “same” domains (Fig. 2J; Table 1), in proportions that significantly differed from Six1WT (p<0.0001) and V17E (p=0.004). Measuring the domains scored as larger (Supplemental Fig. 1B) showed they were significantly larger on the Y129C-injected side (Fig. 2N), supporting the scoring categories and revealing that Y129C was less effective at restricting foxd3 expression compared to Six1WT or V17E.

The majority of V17E-injected morphants had fainter irx1 staining on the injected side (Fig. 2H; Table 1), similar to Six1WT (p=0.6076), indicating that it also restricted irx1 expression. In comparison to Six1WT, Y129C-injected morphants had a smaller proportion of fainter irx1 staining (Fig. 2K) and higher proportion of “larger” and “same” domains (Fig. 2L; Table 1; p=0.0113), indicating it is less effective than Six1WT at restricting irx1 expression. Interestingly, its effects did not reach significance when compared to V17E (p=0.0621). For Six1WT, V17E and Y129C the irx1 domains scored as fainter also were significantly smaller (Fig. 2O), consistent with the scoring categories. The mean percent differences in size for each variant (V17E, 16.2%; Y129C 14.0%) was significantly smaller compared to Six1WT (35.2%; p<0.0001 each), but were similar to each other (p=0.5443). These results demonstrate that neither variant was as effective as Six1WT in restricting irx1 expression.

Most V17E-injected morphants had fainter sox11 staining on the injected side (Fig. 2I; Table 1), in proportions similar to MO-only (p>0.05) and significantly different from Six1WT (p<0.0001). The fainter sox11 domains also were significantly smaller on the injected side (Fig. 2P), and the mean percent reduction in domain size was significantly less for V17E (18.0%) compared to Six1WT (31.6%; p=0.0003). These data indicate that V17E was less effective at altering sox11 expression than in morphants. In contrast, in the majority of Y129C-injected morphants sox11 staining was restored to the same staining intensity as the uninjected side (Fig. 2M; Table 1), a significant difference compared to Six1WT and V17E (p<0.0001 each). Consistent with the scoring categories, the sizes of the sox11 domains on the injected and uninjected sides of Y129C-injected morphants were indistinguishable (Fig. 2P; p=0.4648).

In summary, these experiments reveal that in morphants Six1WT can partially restore the effects of reduced endogenous Six1 on NC and PPE genes. V17E was similar to Six1WT in that it partially restored foxd3 and irx1 domains, but was considerably less effective; Y129C was even less able to restore their expression. In contrast, V17E only weakly restored sox11 expression, whereas Y129C was significantly more effective, even compared to Six1WT.

BOR variants differentially alter otic vesicle gene expression

Since BOR variants differentially affected PPE gene expression, we asked whether they also have different effects on otic vesicle (OV) development. We addressed this question in a Xenopus tropicalis line that carries genetically reduced levels of Six1; this line was not used for experiments above due to difficulties in dissecting sufficient material for genotyping at neural plate stages. We first assessed whether the size of the OV was different between genotypes by measuring its diameter in the dorsal-ventral axis (Fig. 3A). The six1-null OVs were significantly smaller compared to heterozygotes or wildtypes (Fig. 3B), similar to mouse (Laclef et al., 2003; Zou et al., 2004). Next, we tested whether Six1WT or the variants could restore the effects of reduced Six1 by injecting their mRNAs on one side at the 2-cell stage and examining larvae for the expression of three genes important for OV development: tbx1, dlx5, sobp (Raft et al., 2004; Merlo et al., 2002; Tavares et al., 2021) (Fig. 1C). Note that although Six1WT and variant mRNAs were synthesized from X. laevis plasmids, the X. tropicalis protein is identical in the SD and HD, the functional domains affected in BOR (Supplemental Fig. 2).

Figure 3: Measurements of OV diameters in six1-nulls and heterozygotes.

A. Left side: tbx1 OV expression (arrow) in a wildtype Xenopus tropicalis larva. ba, branchial arches; e, eye; nt, neural tube; arrow, OV. Right side: the same image showing the length of the OV dorsal-ventral axis (black line) and the length of the tbx1 OV domain (red line).

B. The OV diameters (in microns) in the dorsal-ventral axis of uninjected nulls, heterozygotes (hets) and wildtypes (Wts) compared by unpaired t-test. ***, p<0.0001; ns, not significant.

C. The OV diameters in six1-nulls processed for tbx1 expression and injected with the indicated mRNAs did not show significant differences between the uninjected (uninj) and injected (inj) sides of the same embryo.

D. The OV diameters in six1 heterozygotes processed for tbx1 expression and injected with the indicated mRNAs did not show significant differences between the uninjected and injected sides of the same embryo.

Tbx1 is expressed in six1-null OVs, but the domain is visually fainter or smaller compared to wildtypes (Fig. 4A). Because a smaller domain might simply result from a smaller OV diameter, we first compared the OV diameters on both sides of Six1WT-injected six1-nulls; they were statistically indistinguishable (Fig. 3C). To further control for variation in OV diameters between larvae, we measured the length of the tbx1 domain (Fig. 3A) and expressed it as a fraction of the diameter of the same OV. In most Six1WT-injected six1-nulls, the tbx1 domain visually appeared larger (Fig 4B; Table 2), which was confirmed by calculating length/diameter (L/D; Fig. 4C), indicating that Six1WT can restore tbx1 OV expression.

Figure 4: In six1-nulls, BOR variants differentially alter OV gene expression.

A. tbx1 expression was reduced in uninjected six1^−/− compared to six1^+/+. Numbers = larvae showing the indicated staining pattern.

B. Most Six1WT-injected six1-nulls had a visually larger tbx1 domain on the injected (inj) compared to uninjected (uninj) side of the same embryo.

C. The lengths of the tbx1 domain/OV diameter (L/D) on each side of Six1WT-injected six1-nulls were significantly larger. *, p<0.05

D. In the majority of V17E-injected six1-nulls, there was no visual change in tbx1 OV staining on the injected side.

E. The L/Ds on each side of V17E-injected six1-nulls were not significantly different (ns).

F. In the majority of Y129C-injected six1-nulls, the tbx1 OV staining was visually smaller on the injected side.

G. The L/Ds on each side of Y129C-injected six1-nulls were significantly smaller. *, p<0.05

H. The LDs of Six1WT were significantly different from each variant (***, p<0.001). The LDs of the variants were not significantly different from each other (ns).

I. dlx5 OV expression was fainter in uninjected six1-nulls compared to wildtypes.

J. Most Six1WT-injected six1-nulls had a fainter dlx5 domain on the injected side.

K. The OV diameter was not significantly different between injected and uninjected sides of Six1WT-injected six1-nulls stained for dlx5.

L. In some V17E-injected six1-nulls, dlx5 OV staining was darker on the injected side.

M. The OV diameter was not significantly different between injected and uninjected sides of V17E-injected six1-nulls stained for dlx5.

N. In some Y129C-injected six1-nulls, dlx5 OV staining was the same on both the injected and uninjected sides.

O. The OV diameter was not significantly different between injected and uninjected sides of Y129C-injected six1-nulls stained for dlx5.

P. sobp OV staining was not detected in uninjected six1-nulls, but strong in wildtypes.

Q. Many Six1WT-injected six1-nulls had darker sobp staining on the injected side.

R. The OV diameter was not significantly different between injected and uninjected sides of Six1WT-injected six1-nulls stained for sobp.

S. In most V17E-injected six1-nulls, sobp staining was not detected on the injected side.

T. The OV diameter was not significantly different between injected and uninjected sides of V17E-injected six1-nulls stained for sobp.

U. sobp staining was not detected in any Y129C-injected six1-nulls.

V. The OV diameter was not significantly different between injected and uninjected sides of Y129C-injected six1-nulls stained for sobp.

Table 2:

Percentages of embryos with the noted intensity of ISH staining scored as fainter, larger/darker or the same (see Methods). For mRNA-injected embryos, staining on the injected side was compared to the uninjected side of the same embryo.

	Uninj Nulls ‡	Nulls Six1WT	Nulls V17E	Nulls Y129C		Uninj Hets ‡	Hets Six1WT	Hets V17E	Hets Y129C
tbx1	(14)	(11)	(18)	(22)		(36)	(18)	(35)	(45)
% smaller	100	0	16.7	50.0		33.0	44.4	25.7	33.3
% larger	0	63.6	5.6	9.1		0	16.7	2.9	11.1
% same	0	36.4	77.7	40.9		67.0	38.9	71.4	55.6
% larger + same	0	100	83.3	50.0		67.0	55.6	74.3	66.7
dlx5	(11)	(9)	(9)	(20)		(52)	(17)	(19)	(45)
% fainter	100	44.5	11.1	25.0		40.4	0	26.3	26.7
% darker	0	33.3	33.3	15.0		0	29.4	15.8	8.9
% same	0	22.2	55.6	60.0		59.6	70.6	57.9	64.4
% darker + same	0	55.5	88.9	75.0		59.6	100	73.7	73.3
sobp	(6)	(9)	(19)	(20)		(18)	(11)	(33)	(43)
% fainter	100 *	33.3*	78.9*	100*		11.1	27.3	21.2	16.3
% darker	0	66.7	21.1	0		0	18.1	12.1	7.0
% same	0	0	0	0		88.9	54.6	66.7	76.7
% darker + same	0	66.7	21.1	0		88.9	72.7	78.8	83.7

^‡, indicates that for uninjected six1-nulls and uninjected six1-heterozygotes, staining was compared to uninjected wildtypes.

^*, indicates that for sobp stained six1-nulls, “fainter” means no stain detected.

Red font indicates effects: 1) in uninjected nulls/heterozygotes; 2) Six1WT differences from uninjected nulls/heterozygotes; and 3) corresponding effects cause by the variants.

(n), indicates number of embryos examined.

dlx5 and sobp expression domains were not sufficiently discrete to measure, so instead the staining intensity on the injected side was visually scored as “fainter”, “darker”, or “same” compared to the uninjected side of the same larva. In uninjected six1-nulls, dlx5 OV expression was fainter compared to wildtypes (Fig. 4I). In comparison, many Six1WT-injected six1-nulls also had fainter dlx5 staining (Fig. 4J), but the proportion was smaller and more larvae had darker or the same staining (Table 2; p>0.05). In uninjected six1-nulls, sobp expression was not detected whereas it was strongly expressed in wildtypes (Fig. 4P). In comparison, the majority of Six1WT-injected six1-nulls had darker sobp staining on the injected side (Fig. 4Q; Table 2; p>0.05). For both dlx5 and sobp, the OV diameters on the Six1WT-injected sides were indistinguishable from the uninjected sides (Fig. 4K, R), indicating that increased staining was not likely due to altered OV diameters. Thus, in six1-nulls Six1WT can partially restore both dlx5 and sobp OV expression.

We next tested whether BOR variants restored OV gene expression as effectively as Six1WT. Notably, neither variant caused a significant change in OV diameter (Fig. 3C, 4M, O, T, V), indicating that the expression differences described below are not likely due to variations in OV diameter. For tbx1, most V17E-injected six1-nulls had visually larger or same size domains (Fig. 4D; Table 2). The L/D indicated that in most embryos the tbx1 domain was the same size on injected versus uninjected sides (Fig. 4E), which was significantly different and thus less effective compared to Six1WT (Fig. 4H). The majority of Y129C-injected six1-nulls had visually smaller tbx1 domains (Fig. 4F; Table 2) and a significantly smaller L/D (Fig. 4G). This also was different from Six1WT, i.e. was less effective, but did not differ from V17E (Fig. 4H). For dlx5, both V17E- and Y129C-injected six1-nulls had proportionally more larvae with darker (e.g., Fig. 4L) or the same (e.g., Fig. 4N) staining compared to Six1WT (Table 2), but these proportions were not significantly different from Six1WT or each other (p>0.05). In most V17E- and all Y129C-injected six1-nulls, sobp staining was undetectable (Fig. 4S, U; Table 2). Both variants were less able to restore sobp compared to Six1WT (V17E, p=0.0346; Y129C, p=0.0002); Y129C was significantly more defective compared to V17E (p=0.0471).

In summary, although BOR is a heterozygous condition, expressing Six1WT and variants in six1-null larvae revealed important differences between their effects, Six1WT partially restored tbx1 and sobp OV expression, whereas both variants were less effective, Y129C significantly so. In contrast, both variants partially restored dlx5 in proportions similar to Six1WT and to each other.

To better model BOR, we next expressed these mRNAs in heterozygotes. On average, control heterozygote OV diameters were similar to wildtypes (Fig. 3B), and for each experimental group there were no significant differences between injected and uninjected OV diameters (Figs. 3D, 5K, M, O, R, T, V), indicating that differences in expression domains reported below are not likely due to changes in OV size. In uninjected heterozygotes, tbx1 is expressed in all OVs, but in a visually smaller or fainter domain in a third (Fig. 5A; Table 2). Most Six1WT-injected heterozygotes also had a visually smaller domain and smaller L/D (Fig. 5B, C; Table 2). dlx5 is expressed in the OV of all uninjected heterozygotes, but is fainter than wildtype staining in ~40% (Fig. 5I; Table 2). In significant contrast, all Six1WT-injected heterozygotes had the same or darker dlx5 OV staining (Fig. 5J; Table 2; p<0.0001). In uninjected heterozygotes, sobp is expressed in all OV’s, and rarely is fainter (Fig. 5P; Table 2). In comparison, although proportionally more Six1WT-injected heterozygotes had fainter sobp OV staining (Fig. 5Q; Table 2), the difference was not significant (p>0.05). Thus, providing additional Six1WT to heterozygotes reduced tbx1 domains, increased dlx5 expression, and had minimal effects on sobp.

Figure 5: In six1-heterozygotes, BOR variants differentially alter OV gene expression.

A. In uninjected six heterozygotes, a third (12/36) had fainter tbx1 OV staining compared to uninjected wildtypes.

B. Most Six1WT-injected six1 heterozygotes had a visually smaller tbx1 domain on the injected (inj) versus uninjected (uninj) side of the same embryo.

C. The lengths of the tbx1 domain/OV diameters (L/D) on each side of Six1WT-injected six1-heterozygotes were significantly smaller. **, p<0.01

D. In most V17E-injected six1-heterozygotes, the tbx1 OV staining appeared the same on both sides.

E. The L/Ds on each side of V17E-injected six1-heterozygotes were not significantly different (ns).

F. In most Y129C-injected six1-heterozygotes, the tbx1 OV staining appeared smaller on the injected side.

G. The L/Ds on each side of Y129C-injected six1-heterozygotes were not significantly different (ns).

H. The LDs of Six1WT-injected versus V17E-injected six1-heterozygotes were significantly different (**, p<0.01). Y129C was not significantly different from Six1WT or V17E (ns).

I. dlx5 OV expression was fainter in 21/52 uninjected six1-heterozygotes compared to wildtypes.

J. In many Six1WT-injected six1-heterozygotes, dlx5 OV staining appeared darker on the injected side.

K. The OV diameter was not significantly different between injected and uninjected sides of Six1WT-injected six1-heterozygotes stained for dlx5.

L. In many V17E-injected six1-heterozygotes, dlx5 OV staining was fainter on the injected side.

M. The OV diameter was not significantly different between injected and uninjected sides of V17E-injected six1-heterozygotes stained for dlx5.

N. In some Y129C-injected six1-heterozygotes, dlx5 OV staining was fainter on the injected side.

O. The OV diameter was not significantly different between injected and uninjected sides of Y129C-injected six1-heterozygotes stained for dlx5.

P. sobp OV staining was rarely fainter (2/18) in uninjected six1-heterozygotes compared to wildtypes.

Q. A few Six1WT-injected six1-heterozygotes had fainter sobp staining on the injected side.

R. The OV diameter was not significantly different between injected and uninjected sides of Six1WT-injected six1-heterozygotes stained for sobp.

S. In most V17E-injected six1-heterozygotes, sobp staining was the same on both sides.

T. The OV diameter was not significantly different between injected and uninjected sides of V17E-injected six1-heterozygotes stained for sobp.

U. An example of fainter sobp staining on the injected side of a Y129C-injected six1-heterozygote.

V. The OV diameter was not significantly different between injected and uninjected sides of Y129C-injected six1-heterozygotes stained for sobp.

We next assessed whether the effects of the BOR variants in heterozygotes were similar to Six1WT. For tbx1, fewer V17E-injected heterozygotes had visually smaller and more had the same staining and L/Ds (Fig. 5D, E; Table 2), and L/Ds were significantly different from Six1WT (Fig. 5H). In contrast, Y129C effects on OV staining intensity and L/D were not different from Six1WT (Fig. 5F-H; Table 2) or V17E (Fig. 5H). For dlx5, a larger proportion of both V17E- and Y129C-injected heterozygotes had fainter staining (Fig. 5L, N; Table 2), but only Y129C was significantly different from Six1WT (p=0.0104). For sobp, most V17E- and Y129C-injected heterozygotes had the same staining intensity on injected and uninjected sides (Fig. 4S, U; Table 2); for both the distribution was indistinguishable from Six1WT or each other (p>0.05).

Expressing Six1WT and variants in six1-heterozygotes revealed important differences compared to uninjected heterozygotes. Six1WT caused more variation in tbx1 OV expression; V17E effects on tbx1 OV expression were similar to uninjected heterozygotes, whereas Y129C effects were similar to Six1WT; Six1WT restored dlx5 OV expression, whereas V17E was slightly less and Y129C was significantly less effective; and neither Six1WT, V17E nor Y129C significantly altered sobp OV expression.

Craniofacial cartilages are differentially perturbed by BOR variants

In mouse and X. tropicalis, loss of Six1 results in hypoplastic otic and jaw cartilages (Laclef et al., 2003; Ozaki et al., 2004; Tavares et al., 2017; Coppenrath et al., 2021), as well as reduced cranial cartilage volume (Naert et al., 2021), tissues that are derived from the cranial NC. To determine whether Six1WT or the BOR-associated variants could rescue these defects, the respective mRNAs were injected into one-cell stage X. tropicalis and cranial cartilages analyzed at tadpole stages (Fig. 1D). First, tadpoles were immunostained for Col2a1 and imaged using mesoSPIM light-sheet microscopy (Vladimirov et al., 2024). A base CranioNet U-Net model (Naert et al., 2021) was then fine-tuned for these recordings to segment and reconstruct the cartilaginous elements in 3-dimensions (Fig. 6A). Qualitatively, injection of Six1WT mRNA into six1-null tadpoles restored cranial cartilage morphology without discernable craniofacial defects. In contrast, V17E- and Y129C-injected six1-null tadpoles displayed gross malformations of cranial cartilages, most notably hypoplasia of the otic capsule and agenesis of the infrarostral cartilage (Fig. 6A, B). Quantitatively, six1-null tadpoles injected with Six1WT mRNA exhibited the largest cranial cartilage volume, which was reduced in V17E-injected (p<0.05) and further decreased in Y129C-injected animals (p<0.05) (Fig. 6C left). To rule out overexpression effects, the same mRNAs were injected into six1-heterozygous and wildtype embryos (described below); in all backgrounds, Six1WT, V17E, and Y129C mRNAs produced similar craniofacial abnormalities indicating that the differences observed in six1-null tadpoles reflect variant-specific alterations in Six1 function rather than nonspecific dosage effects.

Figure 6: Six1 BOR variants differentially rescue craniofacial cartilage formation.

A–B. Representative three-dimensional reconstructions of Col2a1-immunostained six1-null tadpoles injected with Six1WT, V17E, or Y129C mRNA, imaged by mesoSPIM light-sheet microscopy and segmented using a finetuned CranioNet U-Net model. Injection of Six1WT mRNA restored cranial cartilage size without major defects, whereas V17E- and Y129C-injected tadpoles exhibited hypoplasia of multiple cartilaginous elements.

B. Most notable in the variant-injected tadpoles is agenesis of the infrarostral (white arrow) and reduced otic capsule (red arrows).

C. Quantification of total cranial cartilage volume. In six1-null tadpoles (left), Six1WT-injected animals had the largest volumes, which were reduced in V17E-injected and further decreased in Y129C-injected animals. Similar trends were observed in six1^+/− (middle) and six1^+/+ (right) tadpoles. Data are mean ± standard deviation. Statistical tests were applied separately for each genetic background. For six1-nulls, data were not normally distributed (WT: Shapiro p = 0.0091) and Mann–Whitney U tests were used (WT vs V17E, p=0.0238; WT vs Y129C, p=0.0238; V17E vs Y129C, p=0.0476). For heterozygotes and wildtypes, normally distributed data were analyzed by two-sided unpaired t-tests (WT vs V17E, ns; WT vs Y129C, p=0.0053 and p=0.0012, respectively). p-values were corrected for multiple comparisons using the Bonferroni method.

To more specifically ascertain abnormalities in selected cartilages, tadpoles were stained with Alcian Blue, which allows analysis of individual cranial cartilages, and these elements were assessed for changes in staining compared to uninjected tadpoles (Fig. 7A). Elements were scored as abnormal if they were missing, smaller in size, misshaped, faintly stained or unstained (examples shown in Fig. 7B-D). Uninjected six1-null tadpoles frequently had missing infrarostral (IF) and/or misshaped Meckel’s (ME), quadrate (QU), ceratohyal (CH), branchial (BC) and otic capsule (OC) cartilages (n=14; Fig. 7E). In contrast, those injected with Six1WT (n=8) had normal ME, CH and BC (Fig. 7E). In addition, all contained IF, QU and OC, but IF and OC were only faintly stained and half of QU were rounded (Fig. 7E). Thus, providing Six1WT restored many cartilage elements. In comparison, the BOR variants were less effective. All six1-null tadpoles injected with V17E (n=10) had defective IF (30% missing, 50% unstained, 20% misshaped) and many had smaller or faintly stained ME (30% smaller), QU (60% smaller, 10% unstained), CH (60% smaller), BC (80% smaller) and OC (20% unstained, 80% smaller). For each element, V17E was less effective than Six1WT at restoring the cartilage, but the percentages of defects were significantly different only for CH (p<0.05) and BC (p<0.01). Similarly, all six1-null tadpoles injected with Y129C (n=9) had defective IF (33% missing, 45% unstained, 22% faint) and OC (11% unstained, 11% fainter, 78% smaller) and most had defective QU (56% smaller; 11% unstained). However, many more had smaller or misshaped ME (78%), CH (78%), and BC (89%). Like V17E, Y129C was less effective than Six1WT at restoring the cartilages, significantly for ME (p<0.005), CH (p<0.005) and BC (p<0.0001). These data indicate that while Six1WT restored the morphology of many of the cartilage abnormalities observed in uninjected six1-null tadpoles, both variants were similarly defective in the absence of endogenous Six1.

Figure 7: The Y129C variant disrupts morphologies of craniofacial cartilage elements.

A. Normal craniofacial cartilage morphology from a ventral view of Alcian Blue stained head of an uninjected six1^+/+ tadpole. The single infrarostral (IF, white) is located at the midline and articulates with Meckel’s cartilage (ME, green). The quadrate (QU, orange) is triangular in shape, the ceratohyal (CH, light blue) is broad and gently curved and the branchial cartilage (BC, yellow) consists of 4 cartilaginous rows. A’. From a dorsal view of the same tadpole, the oval otic cartilage (OC; dark blue) surrounds the inner ear tissues.

B-D. Examples of dysmorphic cranial cartilages (red arrows) in injected tadpoles.

B. In a V17E-injected six1-null, the IF is barely stained, indicating lack of cartilage differentiation, the MEs are concave and short, one QU is rounded, the CH are narrow, and one BC is small without regular rows.

C. In a V17E-injected heterozygote, the IF and one ME are short, the other ME has a crook, one QU is rounded and one BC is small.

D. In a Y129C-injected heterozygote, both OC are greatly reduced in size (compare to A’).

E. The percentage of six1-nulls showing cranial cartilage abnormalities similar to those illustrated in B-D, in uninjected tadpoles (hatched; n=14) and those injected with Six1WT (green; n=8), V17E (yellow; n=10) or Y129C (blue; n=9). *, p<0.05 compared to uninjected; #, p<0.05 compared to Six1WT-injected.

F. The percentage of six1 heterozygotes showing abnormalities similar to those illustrated in B-D, in uninjected tadpoles (n=26) and those injected with Six1WT (n=52), V17E (n=21) or Y129C (n=22).

G. The percentage of six1 wildtypes showing abnormalities similar to those illustrated in B-D, in uninjected tadpoles (n=8) and those injected with Six1WT (n=19), V17E (n=11) or Y129C (n=12).

Because BOR patients carry only one allele of a Six1 variant, we next analyzed the effects of the various mRNAs on cranial cartilage volume in heterozygous tadpoles (six1^+/−). Compared to Six1WT-injected tadpoles, total cartilage volume was slightly reduced by V17E but significantly decreased by Y129C (p<0.01; Fig. 6C middle). Analysis of Alcian Blue stained tadpoles showed that in uninjected heterozygotes, each cranial cartilage was abnormal in only a low percentage (~20% or lower; Fig. 7F). Injection of Six1WT did not significantly change these percentages in most cartilages, but did restore the morphology of CH (p<0.005) and OC (p<0.01). V17E had similar effects as Six1WT, significantly differing only in its ability to restore OC (p<0.001). In contrast, Y129C caused a significant increase in defects in IF (p<0.05), QU, (p<0.05) and BC (p<0.05) compared to uninjected heterozygotes, as well as a significant increase in defects in ME (p<0.05), CH (p<0.05), BC (p<0.02) and OC (p<0.0001) compared to Six1WT. Thus, defects were more frequent in heterozygotes expressing Y129C.

These results suggest that Y129C may interfere with the activity of residual endogenous Six1. To further assess this possibility, we injected the various mRNAs into six1-wildtype embryos that carry normal levels of Six1. Here, we also found that the total cranial cartilage volume was slightly reduced in V17E-injected but significantly reduced in Y129C-injected tadpoles (p<0.01) compared to Six1WT (Fig. 6C right). Analysis of Alcian blue stained tadpoles showed that Six1WT and V17E caused very few cartilage abnormalities, whereas Y129C caused defects at significantly higher frequencies (Fig. 7G). These data indicate that Y129C likely interferes with the function of endogenous Six1 during cranial cartilage development.

DISCUSSION

Six1 plays a key transcriptional role in the development of the cranial sensory placodes, otic vesicles and branchial arches. Heterozygous inheritance of SIX1 variants in BOR result in highly variable levels of hearing loss and craniofacial dysmorphologies, however it is not clinically obvious whether there are phenotypic differences between patients carrying different variants (Smith and Azaiez, 2025). Because the clinical diagnostic criteria for BOR are the result of a multitude of developmental processes that cannot be interrogated after birth, we posited that exploring the effects of variants during embryogenesis would reveal underlying causes of those phenotypes and their variability. Therefore, we attempted to distinguish between the effects of two functionally distinct variants – V17E and Y129C – on the development of precursor populations that give rise to the craniofacial tissues affected in BOR.

BOR variants differentially affect gene expression in craniofacial progenitor populations

The cranial NC that migrate into the branchial arches give rise to craniofacial skeletal elements, including jaws, middle ear ossicles and outer ear cartilages, whereas the PPE gives rise to the inner ear. Therefore, we first asked whether BOR variants differentially affect NC or PPE gene expression patterns in morphants in which endogenous Six1 was reduced. Previous work in embryos with wildtype Six1 levels reported that V17E caused effects very similar to additional Six1WT, whereas Y129C effects were much weaker (Shah et al., 2020; Mehdizadeh et al., 2021). Herein, we show that the effects of the variants are different in their ability to restore normal gene expression patterns in embryos with reduced Six1. While providing Six1WT to morphants partially restored the normal patterns of foxd3, irx1 and sox11 expression, V17E, which does not bind Eya1, was less effective in restoring foxd3 and irx1 and had a minimal effect on sox11. In comparison, Y129C, which has deficient DNA-binding, only weakly restored foxd3 and irx1, but was more effective than either Six1WT or V17E at restoring sox11. These results demonstrate that SD and HD variants have different effects on genes critical for the early development of NC and PPE progenitor populations. It is tempting to posit that V17E is less effective than Six1WT because it is unable to function as a transcriptional activator due to loss of Eya1 binding (Ikeda et al., 2002; Li et al., 2003; Silver et al., 2003). This is consistent with the previous observation that sox11 PPE expression is expanded by either expressing a Six1-activator construct or co-expressing Six1 + Eya1 (Brugmann et al., 2004). We predict that the opposite effects of Y129C on sox11 expression result from its ability to bind endogenous Eya1, thus making this activating cofactor unavailable to any residual Six1 that might promote sox11 expression. However, the effects of each protein on expression domains were variable, especially for sox11. It remains to be tested whether this variability is due to variable compensation by other genes, such as the highly related Six2, or reflect variable levels of remaining Six1 in the morphants, a possible limitation of this methodology.

We eliminated this potential source of variability in morphants by examining later stages in a mutant line that could be genotyped (see Methods). At these later stages, we assayed three genes known to be required for OV development. Tbx1 is required for the specification of auditory and vestibular precursor cells (Vitelli et al., 2003; Raft et al., 2004; Xu et al., 2007), Dlx5 plays an early role in the vestibular morphogenesis (Merlo et al., 2002), and Sobp regulates Six1 transcriptional activity during OV formation (Tavares et al., 2021). We found that in six1-nulls, Six1WT could partially restore the OV expression of all three genes, whereas V17E was less effective at restoring tbx1 and sobp, and Y129C was less effective than V17E. Thus, clear differences between the BOR variants were revealed in six1-nulls. The BOR variants also had different effects in six1-heterozygotes that best model BOR. For tbx1, Six1WT and Y129C had pleiotropic effects, whereas V17E effects were not different from uninjected six1-heterozygotes. For dlx5, Six1WT restored expression, V17E was slightly less effective and Y129C was significantly less effective. For sobp, all three proteins caused pleiotropic effects that were not significantly different from uninjected six1-heterozygotes or each other.

Thus, in both six1-nulls and heterozygotes, we detected different effects of the BOR variants on OV gene expression, indicating that they act through distinct mechanisms. However, we have yet to elucidate the molecular mechanisms that cause these differences. While their different effects may be related to known differences in cofactor and DNA binding, during OV and branchial arch development there are complex regulatory interactions between Six1, Eya1, Tbx1, Dlx5 and Sobp (Freyer and Morrow 2010; Guo et al., 2011; Li et al., 2020; Lin et al., 2009; Ozaki et al., 2004; Sato et al., 2010; Tavares et al., 2017; Tavares et al., 2021; Zheng et al., 2003). Perhaps when levels of Six1 change or variants are introduced, these other genes or closely related Six family members compensate. Since inner ear development involves hundreds of genes (Chatterjee et al., 2010; Kolla et al., 2020; Baxi et al., 2023) and many additional Six1-binding partners (Neal et al., 2024), much experimentation will be needed to elucidate the precise molecular mechanisms that underlie the different effects of V17E and Y129C that we have uncovered.

BOR variants have different effects on cranial cartilage morphology

BOR patients present with abnormalities in structures derived from the NC that populate the second pharyngeal arch, including the hyoid region, middle ear and external ear. Hypoplasia of the mandible, derived from the NC populating the first pharyngeal arch, also are noted (Smith and Azaiez, 2025). Since Six1 plays a role in pharyngeal arch development (Xu et al., 2002; Guo et al., 2011; Tavares et al., 2017; Coppenrath et al., 2021), we also investigated whether the formation of the cranial cartilages were differentially affected by the BOR variants. Quantitatively, we found that, providing Six1WT restored cranial cartilage volume in six1-nulls whereas it remained significantly reduced in V17E- and Y129C-injected tadpoles. Y129C-injected six1-nulls also had a higher incidence of cranial cartilage malformations. In heterozygotes and wildtypes, only Y129C-injected tadpoles had reduced cartilage volume and significantly more malformations, suggesting that it interferes with the function of endogenous Six1. In comparison, the effects of V17E tended to be similar to those of Six1WT, particularly in heterozygotes and wildtype tadpoles, consistent with a report that another SD variant – R110W – causes milder phenotypes compared to Y129C (Ruf et al., 2004). These results demonstrate that the early alterations in NC gene expression are manifested in later disruptions in morphogenesis. Differential effects on the multiple developmental processes occurring between NC specification and cartilage formation are likely to contribute to the phenotypic variability seen in animal models and patients.

Phenotype variability

Analyses of several BOR families including over 400 individuals clearly show considerable phenotypic variability within and across families harboring the same variants in EYA1 and SIX1 (Smith and Azaiez, 2025). It has long been a goal to explain the underlying mechanisms for this variability. Recent evidence indicates that EYA1 variants are more frequently associated with branchial arch and external ear malformations compared to SIX1 variants (Lee et al., 2023). However, among the reported SIX1 variants there is no clear correlation between the variant and the spectrum of dysmorphologies. For example, both V17E and Y129C patients are characterized by hearing loss, hyoid fistulae and preauricular pits (Ruf et al., 2004; Kochhar et al., 2008). However, since biochemical studies in cell culture clearly show that SD variants have defective interactions with EYA1 and HD variants have reduced ability to promote transcriptional activation (Ruf et al., 2004; Patrick et al., 2009; Shah et al., 2020; Lee et al., 2023), we anticipated that studying these variants in embryos would reveal clear developmental differences. Indeed, we found that the expression levels of several genes in the precursor populations of structures affected in BOR - NC, PPE and OV - as well as cranial cartilage formation were differentially altered by V17E versus Y129C. These results show that the variants cause early developmental changes that impact later morphogenesis.

But there also was notable variability in the effects of each variant even on defined genetic backgrounds. For example, there was variability in expression domain sizes (reduced in some embryos and enhanced in others) in six1-nulls (Fig. 2N, O; Fig. 4 E, G) and heterozygotes (Fig. 5C, E, G). Perhaps this is not surprising since there are differences in the frequency of SD versus HD variants in patients from different ethnic backgrounds (e.g., Lee et al, 2023; Cho et al., 2024), and variable dysmorphologies in six1-null mice carried on different genetic backgrounds (Zheng et al., 2003). It will be important to assess whether the outliers in our datasets are due to modifiers that impact the relative expression levels of the variants and wildtype Six1. Although variants are stable and expressed at similar levels to wildtype Six1 when transfected into cultured cells (Ruf et al., 2004; Patrick et al., 2009; Shah et al., 2020; Lee et al., 2023), this has not yet been established in the embryo environment. Furthermore, there is evidence that Eya1 levels are required for normal Six1 expression (Xu et al., 2002; Lee et al., 2023), which could be a significant factor in the phenotypes observed in heterozygotes. Finally, it is becoming clear that in addition to Eya1, there are many other potential Six1-binding proteins that function during craniofacial development (Neal et al., 2024). It will be important to test these other potentially contributing factors in the presence of BOR variants to better understand the underlying causes of patient phenotypic variation.

MATERIALS AND METHODS

Obtaining embryos and genotyping

Xenopus laevis embryos were obtained by either gonadotropin-induced natural mating or by in vitro fertilization of wildtype, outbred adults, as previously described (Sive et al., 2000; Moody, 2000; Moody, 2018). Embryos were selected at the 2-cell stage if the first cleavage furrow bisected the lightly pigmented region of the animal hemisphere to accurately identify the dorsal-ventral and left-right axes (Klein, 1987).

Xenopus tropicalis embryos were generated by hormone induced in vitro fertilization of adult pairs of −28/+ (Xtr.six1^em2Horb, RRID:NXR_3145) six1 mutants (Coppenrath et al., 2021). Confirmed oocyte positive females were given 20 U of Pregnant Mare Serum Gonadotropin (PMSG) (BioVender #RP17827210000) and 200 U of human chorionic gonadotropin (hCG) (BioVender #RP17825010) (Wlizla et al., 2018), whereas virgin females were given 10 U of PMSG and 100 U of hCG to potentially reduce instances of ovarian hyperstimulation syndrome (Green et al., 2007). Larvae and tadpoles were fixed at appropriate stages and genomic DNA extracted from individual tail clips using DNeasy Blood & Tissue Kit (Qiagen #69506). PCR amplification of the targeted region was done using the following primers: forward primer 5’-CCATGTCTATGCTGCCTTCC-3’ & reverse primer 5’-CCCTCAGTTTCTCTGCTTCC-3’. PCR products were purified using the NucleoSpin PCR Clean-up procedure (Macherey-Nagel #740609.250) and mutations were confirmed by sequencing.

Microinjection

When selected 2-cell stage Xenopus laevis embryos reached the 4-cell stage, both the dorsal and ventral blastomeres on the left side were microinjected in their animal region with an equimolar mixture of two Six1 antisense morpholino oligonucleotides (MOs; 9 ng total; Fig. 1B), according to standard methods (Moody, 2018). These MOs, one of which binds to the ATG start site region and the other to the upstream 5’UTR, were previously validated to be specific and effective (Brugmann et al., 2004; Sullivan et al., 2019). When a subset of morphants reached the 8-cell stage, their dorsal animal and ventral animal daughter cells, which are the major contributors to the cranial neural crest and pre-placodal ectoderm (Moody and Kline, 1990), were microinjected with 150pg of mRNAs encoding either X. laevis Six1WT, V17E or Y129C mixed with 100pg of nßgal mRNA as a lineage tracer, as previously described (Shah et al., 2020). Both injections were done on only the left side of the embryo and the uninjected right side was used as an internal control to compare the sizes of the gene expression domains. Bilateral comparison within the same embryo is an essential control because there is up to a 38% variation in diameter and 2.6-fold difference in volume in Xenopus embryos, even in clutches derived from the same female (Leibovich et al., 2020). These size differences, as well as differences in embryo growth, result in variations in gene expression levels and domain sizes between embryos. Embryos were cultured in a diluted series of Steinberg’s solution until fixation for in situ hybridization (ISH).

Xenopus tropicalis embryos were microinjected at the 1-cell stage (for cartilage analyses) or into the animal region of a single blastomere at the 2-cell stage (for gene expression analyses) with 60 pg of either Six1WT, V17E or Y129C mRNAs mixed with Texas Red-dextran as a lineage marker (ThermoFisher, D1829). Since X. laevis and X. tropicalis Six1 have identical amino acid sequences with the exception of three conservative and one semi-conservative substitutions in the C-terminus (Supplemental Fig. 2), the X. laevis protein is expected to be functionally identical when expressed in X. tropicalis. Microinjections done at the 2-cell stage were targeted to one blastomere at random and screened at early tailbud stages for the fluorescent lineage marker to determine which side was injected. The opposite, uninjected side of the embryo served as an internal control for the ISH analyses. All embryos were reared according to Wlizla et al. (2018) until collected for fixation (Fig. 1C, D).

In Situ Hybridization

Xenopus laevis embryos were cultured to neural plate stages (16–18) and Xenopus tropicalis to larval stages (28–32) (Nieuwkoop and Faber, 1994; Zahn et al., 2022), fixed in 4% paraformaldehyde (in 0.1M MOPS, 2mM EGTA Magnesium, 1mM MgSO₄, pH 7.4) and processed for ISH as described previously (Sive et al., 2000). Dig-labeled antisense RNA probes (dlx5, foxd3, irx1, sobp, sox11, tbx1) were synthesized in vitro (MEGAscript kit; ThermoFisher) from cDNA plasmids as previously described (Sive et al., 2000). The expression patterns were compared between the injected and control sides of the same embryo and visually scored as follows. For neural plate stage embryos, “fainter” means fainter intensity (e.g., Fig. 2C, D); “larger” means same intensity but bigger size (e.g., Fig. 2B, J); and “same” means both same intensity and same size (e.g., Fig. 2M). The proportions of embryos in each category were compared between experimental groups by a 2-sided Fisher’s exact test (GraphPad Prism 11). In a subset of these embryos, the perimeter of the expression domain was measured on both the injected and uninjected sides of the same embryo using the measuring tool in the cellSens Entry software (Olympus/Evident Scientific) (Supplemental Fig. 1). Sizes between injected and uninjected sides were compared by paired, 2-tailed t-tests (GraphPad Prism 11) (e.g., Fig. 2N, O, P). For OV stage larvae, “fainter” means fainter intensity (e.g., Fig. 4J); “smaller” means shorter length (e.g., Fig. 4G); “larger” means same intensity but bigger size (e.g., Fig. 4B); and “same” means both same intensity and same size (e.g., Fig. 4D). The proportions of embryos in each category were compared between experimental groups by a 2-sided Fisher’s exact test (GraphPad Prism 11). In all larvae, the diameter of the OV in the dorsal-ventral axis was measured using the measuring tool in the cellSens Entry software (Olympus/Evident Scientific) (Fig. 3A). For larvae stained for tbx1, the length of the expression domain was measured on both the injected and uninjected sides of the same embryo and expressed as a percentage of the OV on each side (length/diameter) (Fig. 3A). Sizes between injected and uninjected sides of the same embryo were compared by paired, 2-tailed t-tests and those between experimental groups were compared by unpaired, 2-tailed t-tests (GraphPad Prism 11). Whole-mount ISH images were collected with an Olympus SZX16 stereomicroscope coupled to an Olympus UC90 camera.

Tadpole light-sheet microscopy and CranioNet segmentation

Whole-mount Col2a1 immunostaining of Xenopus tropicalis tadpoles was performed as described previously (Naert et al., 2021). Stained specimens were imaged using mesoSPIM light-sheet microscopy at either 5× or 10× magnification (Voigt et al., 2019; Vladimirov et al., 2024). For volumetric normalization across magnifications, reconstructed image stacks were scaled using a cubic correction factor derived from the notochord diameter.

A base 2D CranioNet U-Net model (Naert et al., 2021) was fine-tuned for these recordings to segment and reconstruct cartilaginous elements. For training, every 20th optical section was exported from one representative z-stack per genotype and injection condition using a Fiji macro. The resulting images from all conditions were pooled, and approximately 25% of slices were randomly assigned to the validation dataset using an automated file selection script, ensuring balanced representation across genotypes (six1^+/+, six1^+/−, six1^−/−) and injection conditions (Six1WT, V17E, Y129C). In total, 66 images were used for training and 22 for validation. The CranioNet model was fine-tuned for 20,000 iterations at a learning rate of 1×10⁻⁴, followed by 2,000 iterations at 2×10⁻⁵, reaching an intersection-over-union (IoU) of 0.57 on the validation set.

The resulting segmentation data were further processed using Imaris (Bitplane) for three-dimensional visualization and volumetric quantification. Cranial cartilage and notochord diameters were measured using Imaris measurement tools, and all 3D renderings shown in figures were generated in Imaris. Statistical analysis and data visualization were performed in Python using pandas, numpy, scipy.stats, and altair.

Alcian Blue Staining

Alcian Blue staining of Xenopus tropicalis tadpole cranial cartilage was performed according to Young et al. (2017). Fixed and genotyped tadpoles were incubated in a solution of acid/alcohol containing 0.1% Alcian Blue. When staining was complete, tadpoles were washed in the acid/alcohol solution without Alcian Blue, bleached with a solution containing 1.2% hydrogen peroxide and 5% formamide and cleared in 2% KOH with increasing concentrations of glycerol. The frequency of morphological defects between groups were compared by a 2-sided Fisher’s exact test (GraphPad Prism 11). Whole-mount images were captured using an Olympus SZH16 stereomicroscope coupled with an Olympus UC90 camera and cellSens Entry software.

DATA AND RESOURCE AVAILABILITY:

All relevant data and details of resources can be found within the article.