The Sox10^Dom Mouse: Modeling the Genetic Variation of Waardenburg-Shah (WS4) Syndrome

RESULTS

Sox10^Dom Congenic Lines Segregate Modifiers of Hypopigmentation

The Sox10^Dom mutation originally arose in the C57BL/6J allele of an F₁ hybrid cross and has been maintained on a C57BL/6J × C3HeB/FeJLe-a/a (B6C3F1/J) hybrid background (Lane and Liu 1984). To identify the genetic determinants responsible for the variation in the Sox10^Dom phenotype, we established congenic lines of this mutant on the B6 and C3H parental strain backgrounds. As described previously (Lane and Liu 1984), survival is markedly affected by the strain background. We observed only 30% ofSox10^Dom /+ animals surviving to weaning on the B6 background in contrast to 82% survival on the C3H background. However, in addition to the variation in survival, we demonstrate that the degree of hypopigmentation (spotting) also varies greatly between congenic lines as early as N₂. In the B6.Sox10^Dom /+ pedigree, 3% (1/28) ofSox10^Dom /+ mice demonstrate head spotting, which appears as only a few hairs (Fig. 1). In contrast, 50% of Sox10^Dom /+ mice at N₄ of the C3H pedigree display visible white forelocks (head spots that are significantly larger than those seen in B6.Sox10^Dom /+mice). The absence of head-spotted individuals in the B6. Sox10^Dom /+ congenic line was not a consequence of preweaning mortality of severely affected animals in this pedigree. We were able to assess head spotting in almost all the mice as lethality usually did not occur prior to pigmentation (to 7–10 days postnatum).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

Effect of parental strain on frequency of white forelock inSox10^Dom mice. The frequency of white forelock inSox10^Dom /+ mice on either the B6, B6C3H_F1, or C3H backgrounds is shown. The total number of animals assessed for head spotting in a particular strain background is indicated above each bar on the plot. The increase in head spotting through successive generations of breeding onto the C3H background suggests a recessive C3H modifier.

Modifier loci that influence hypopigmentation have been localized to mouse chromosomes 2, 5, 8, and 10 in crosses segregating theEdnrb^s locus (Pavan et al. 1995). Additional analyses that characterized the influence of the chromosome 10 modifier on dorsal spotting colocalized this locus with two microsatellite markers D10Mit178 and D10Mit96 (H. Rhim and W.J. Pavan, in prep.). To determine whether the same or a closely linked locus could account for the variation in hypopigmentation observed in the Sox10^Dom congenic strains, we performed linkage analysis on a pedigree of Sox10^Dom /+ mice that had been maintained by crosses to B6C3F1/J mice. One hundred and seventy-six Sox10^Dom /+ mice were analyzed from this pedigree and genotypes determined for D10Mit178 andD10Mit96. Of the twenty animals with white forelocks, all were homozygous for the C3H allele at the chromosome 10 locus. Two additional animals had small head spots consisting of only a few hairs similar to that seen infrequently in the B6.Sox10^Dom /+ pedigree. Both animals were homozygous B6 at these markers. There were also nineteen Sox10^Dom /+ animals that were homozygous C3H at the chromosome 10 locus, but did not exhibit white forelocks. This is consistent with the 50% penetrance of the white forelock phenotype observed forSox10^Dom /+ mice in the C3H pedigree at N₄.

Parametric linkage analysis was performed assuming a 50% penetrance with FASTLINK version 4.0P (Cottingham et al. 1993; Schaffer et al. 1994) in which the head-spot trait was modeled as recessive with 50% penetrance and a phenocopy rate was based on the observation that two mice out of 22 were false positives for the trait. This analysis indicated significant linkage for D10Mit178 (lod score of 5.63 with P value of 2.3 × 10⁻⁶) at no recombination (υ = 0.0). Nonparametric linkage analysis with the program SimIBD and 100,000 replicates, 500 simulated null distribution replicates, and 1000 bootstraps gave a P value of 0.009992. The higher P value achieved from the nonparametric analysis is consistent with previous observations that nonparametric linkage methods have lower power than parametric linkage methods when the trait model is correctly specified (Goldin and Weeks 1993). Haplotype inspection revealed only one recombination event betweenD10Mit96 and D10Mit178 in an animal (C3H/C3H and B6/C3H respectively) that lacked a white forelock. In addition all 183Sox10^Dom /+ mice displaying white forelocks from the C3H pedigree were homozygous for the C3H allele on chromosome 10. Homozygosity at these markers would be expected for subsequent generations in this pedigree. We propose that the chromosome 10 C3H locus acts as a recessive modifier of white forelock inSox10^Dom /+ mice with reduced penetrance. The increased penetrance observed in subsequent generations of the C3H pedigree suggests that additional modifier loci are inherited from the C3H background to account for the increasing penetrance of the white forelock trait. Further linkage analyses will be needed to determine the location of the additional loci that influence hypopigmentation.

Comparative Genomic Analysis: Conservation, Expression, Genomic Structure

The neurocristopathies in Sox10^Dom mice, the expression pattern of Sox10, and the altered expression of other neural crest marker genes in these mutant mice indicate a key role for SOX10 in neural crest development (Southard-Smith et al. 1998). To evaluate the conservation of Sox10 in vertebrate evolution and assess the presence of orthologous genes in additional species, a cDNA fragment 3′ to the HMG box of mouse Sox10was hybridized to zoo blots (Fig. 2). Significant hybridizing bands were observed in DNA from all mammals analyzed and from fish. The presence of a Sox10 ortholog in these species is consistent with an essential role for this gene in development (Southard-Smith et al. 1998). Inconsistent hybridization to genomic DNA samples of chicken and yeast between the two blots was observed and thought to be a consequence of either incomplete DNA digestion or strain differences between the samples analyzed.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2.

Cross-species conservation of Sox10. Southern blot autoradiographs from hybridization of EcoRI-digested genomic DNAs with the 3′ end of a mouse Sox10 cDNA probe (left, Quantum Biotechnologies; right, Clontech). Organism names are indicated above each lane. Specific species DNAs applied to the blot at left include Escherichia coli,Saccharomyces cerevisiae, Caenorhabditis elegans,Drosophila melanogaster, Tautoga onitis, Mytilus edulis, Bovis domesticus, Xenopus laevis,Gallus domesticus, Mus musculus, and Homo sapiens. (Left) Positions of molecular size standards in kb. Longer exposures of the filter in B revealed several weakly hybridizing bands within the chicken DNA sample that are not visible in this reproduction. Hybridization of a probe for a second gene to the filter in A produced a similar high molecular weight band and streaking in the chicken lane as seen here with theSox10 3′ cDNA probe.

Northern blot hybridization of mouse and human poly(A)⁺ RNA was performed to compare the expression pattern of Sox10 between mouse and human (Fig. 3). Expression patterns, if comparable, were expected to further validate theSox10^Dom mouse as a human disease model and potentially identify additional tissues that should be evaluated for effects of SOX10 mutations in patients. Northern blot analysis demonstrated that Sox10 expression is not detectable in 7-d.p.c. mouse embryos but is expressed in older embryos through to adulthood. This pattern is consistent with the initiation of expression at day 8.5 d.p.c. observed by in situ hybridization studies in mouse and rat (E.M. Southard-Smith unpubl.; Herbarth et al. 1998; Kuhlbrodt et al. 1998a). In adult mouse tissues, Sox10 mRNA is detected in heart, brain, lung, skeletal muscle, and testes. Expression in human tissue poly(A)⁺ RNA samples was similarly observed in these tissues. Additional sites of Sox10 expression were observed in human pancreas, prostate, ovary, stomach, spinal cord, trachea, and adrenal gland. The high levels of hybridization to human brain and spinal cord RNAs are consistent with Sox10 expression in glial cells and astrocytes (Kuhlbrodt et al. 1998a). SOX10 expression was observed throughout the human digestive tract (stomach, small intestine, and colon) as well, consistent with our previous analyses in the mouse (Southard-Smith et al. 1998). Hybridization signals forSox10 mRNA were consistently absent in kidney and spleen of both mouse and human. The faint signal observed in mouse lung, consistent with expression observed in the developing lung bud of 12.5-d.p.c. embryos (Southard-Smith et al. 1998), was notably absent in hybridizations to human lung poly(A)⁺ RNA.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 3.

Expression profile of Sox10 in mouse and human total tissue RNA samples. Northern blot hybridization of a Sox10 cDNA probe to multiple mouse and human poly(A)⁺ RNA samples (10 μg/lane, Clontech). (Top) Hybridizations with 1.3- and 1.7-kb 3′ end Sox10 cDNA probes that exclude the HMG box for mouse and human samples, respectively; (bottom) hybridizations with the RNA loading control L-32 (Dudov and Perry 1984).

Comparative sequence analysis of Sox10 genomic structure between mouse and human was performed to assess the conservation of exon–intron boundaries and facilitate mutation detection in the human ortholog. Mouse exon–intron boundaries were identified by comparison of the mouse cDNA sequences with genomic sequences obtained from single pass sequencing of BAC43P19 (Southard-Smith et al. 1998; E.M. Southard-Smith, J.E. Collins, J.S. Ellison, K.J. Smith, A.D. Baxevainis, J.W. Touchman, E.D. Green, I. Dunham, and W.J. Pavan, in prep.). Five exons and four introns were identified. Each of the identified splice junctions possesses the consensus for splice donor and acceptor sites. Although the majority of Sox genes exhibit a monoexonic structure, Sox10 is analogous to Sox9,SOX5, Sox17, and SOX20, which possess multiple exons that partition the ORF within the HMG box. By use of the junction sequences from mouse Sox10, the human exon boundaries were identified within cosmid J81I2 and confirmed by sequencing of additional human BAC clones (Table 1). Exon–intron organization and splice-site positions appear conserved between mouse and human; however, some variation in exon size is apparent. Exons 2, 3, and 5 differ by 4, 6, and 170 bases, respectively. Comparison of intron sequence flanking splice junctions demonstrates a considerable lack of conservation outside exonic regions. This information was used to establish a set of primers that amplify human exons for mutation detection analysis.

Our sequence analysis of Sox10 mouse cDNA additionally revealed two potential initiator methionines (Fig. 4) the most 5′, Met_ALT, residing within a representative Kozak consensus sequence (Kozak 1996) and the second residing 309 bp more 3′, Met₁, within a less representative Kozak region. Because no stop codons are apparent in the mouse between these two positions, either methionine could serve as the initiator residue. To investigate which methionine acts as initiator for the Sox10ORF, we extended the human SOX10 cDNA sequence using clones derived from three independent human brain cDNA clones and compared this to cDNA sequences of mouse (Southard-Smith et al. 1998) and rat (Kuhlbrodt et al. 1998a) as well as to mouse genomic sequences (E.M. Southard-Smith and W.J. Pavan, unpubl.). Repeated attempts to generate sequence information for the human SOX10 exon 1 that might contain a Met_ALT analogous to that in the mouse by cDNA isolation, 5′ RACE, and genomic sequencing of human BACs with mouse primers were unsuccessful. However, comparative alignment of the available human sequence with mouse and rat sequences for exons 2 and 3 revealed 90% sequence identity between the murine and human sequences, implying structural relevance. Despite this high degree of conservation, which exceeds the average aligned identity (67%) of human and mouse 5′ UTRs (Makalowski et al. 1996), our sequence analysis of the human SOX10 cDNA clones identified a single base deletion in comparison to the mouse and rat that would frameshift the human ORF. This deletion was observed in three independent human cDNA clones our human genomic clones and has been seen by Pusch and colleagues (Pusch et al. 1998). The single base deletion would preclude the use of an orthologous Met_ALT and supports initiation of the ORF from Met₁ in exon 3.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 4.

Comparison of mouse, human, and rat sequences 5′ of the initiation methionine (Met₁, boxed) proposed by Kuhlbrodt and colleagues (Kuhlbrodt et al. 1998a). Coding sequences derive from cDNA (human; E.M. Southard-Smith, unpubl.; mouse GenBank accession no. AF017182; rat GenBank accession no. AJ001029) or genomic sequences (M. Angrist, unpubl.). An alternative start codon Met_ALT found in the mouse Sox10 5′ UTR is also boxed.

Mutation Detection in Waardenburg-Hirschsprung’s Disease Patients

On the basis of the definition of exon–intron boundaries provided by cross-species comparisons, PCR assays were developed to assess the integrity of SOX10-coding regions in patients that had biopsy-proven HSCR in addition to one or more major stigmata associated with Waardenburg syndrome (sensorineural hearing loss, pigmentary anomalies, and/or bicolored irides). Direct sequencing of the threeSOX10-coding exons in nine patients with HSCR and Waardenburg-associated phenotypes revealed two nonsense mutations. The patient from family 140 (Fig. 5) has short segment HSCR, profound sensorineural hearing loss, hypopigmentation on his abdomen and neck, and is heterozygous for a Y207X mutation. Both parents are phenotypically normal and neither carries the mutation. Y207X occurs in exon 4, 27 residues downstream of the carboxyl end of the HMG box and 14 residues downstream of the corresponding site in the mouse where the insertion responsible for theSox10^Dom phenotype is located (Fig. 6;Southard-Smith et al. 1998).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 5.

Pedigrees of WS4/Sox10 families. (Left) Pedigree of family 140. The Proband (arrow) was found to be heterozygous for theSOX10 mutation Y207X. (−) Mutant allele; (+) wild-type allele. This patient has short segment HSCR, profound deafness, and multiple areas of hypopigmentation on his neck and abdomen. DNA was not available from the proband’s phenotypically normal sister. (Right) Pedigree of family 192. The Proband and his sister were both found to be heterozygous for the SOX10 mutation Q377X. Both siblings have nystagmus and ataxic cerebral palsy and are profoundly deaf. Only the proband has an abnormal enteric neuronal phenotype. Sequence analysis revealed a faint mutant band in the father corresponding to the T nucleotide in the TAG stop codon in addition to the wild-type band of normal intensity, suggesting possible germ-line mosaicism for Q377X.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 6.

Summary of Sox10 mutations detected in mouse and human HSCR. The effect of the Sox10^Dom mutation and multiple independent human mutations [this study (*); Pingault et al. 1998] on SOX10 are diagrammed beneath the protein domains. The category of mutation within the Sox10 coding sequence is indicated below. Note that both 929insG_Dom and 1076delGA produce frameshifts that introduce heterologous amino acid sequence, 99 and 37 residues, respectively, before terminating the protein at the indicated positions on the diagram.

The second mutation was found in the proband of family 192 (Fig. 5), which has sensorineural deafness and variable diagnoses of enteric function ranging from hypoganglionosis to long segment HSCR. The mutation (Q377X) truncates the SOX10 protein within the transcription modulation domain (Fig. 5). In addition to the two mutations described above, Pingault and colleagues (1998) have described four distinct mutations in SOX10. Two of these truncate the protein 3′ of the HMG-binding domain—similar to the mutations we detected—while a third truncates the protein 5′ of the HMG-binding domain. The fourth mutation, a Leu–Arg duplication within the HMG box (482ins6), alters the third helix of the SOX10 DNA-binding domain. To examine further the effect of the 482ins6 mutation on SOX10 function, the predicted structure of the SOX10 HMG domain was evaluated. Model structures were generated for the wild-type SOX10 sequences from human, rat, and mouse, as well as for the human SOX10^482ins6 mutant. Each query sequence was threaded through the NMR coordinates of the second HMG-1 box from rat (rHMG1.2) as described previously in a structural study of the HMG-1 box family of proteins (Baxevanis et al. 1995). For each sequence, all possible placements of the sequence within the structure were considered, with a conformational energy (ΔG _R‖M) being calculated for each placement. Threads with the most favorable conformational energies (i.e., those with the lowest ΔG _R‖M) were selected for further study.

The final placement of each sequence with respect to the NMR structure of rHMG1.2 is shown in the multiple sequence alignment in Figure7. The structural alignment is not vastly different from what would be produced by a traditional, sequence-based alignment method in this particular case. However, because the sequences are being aligned to the HMG-1 box structure, each of the core regions (roughly corresponding to the three α helices) is necessarily constrained to being ungapped. This constraint has a significant effect on the alignment of the 482ins6 mutant, in that the insertion occurs within the third α helix. By virtue of this location, the best thermodynamic solution results in all residues carboxy-terminal to the Leu–Arg insertion being shifted out of position by two amino acids.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 7.

Structural alignment of the HMG-1 box domains of human, mouse, and rat Sox10 proteins. The sequence of HMG-1 box 2 from rat whose NMR structure was used as the basis for the threading experiments (Weir et al. 1993) is shown in the first line of the alignment (rHMG1.2). (Blue) Positions exhibiting absolute identity; (yellow) conserved positions. (▴) The positions of the 2-residue Leu–Arg insertion in human 482ins6. Positions of the secondary structural elements found in the HMG-1 box NMR study are shown below the alignment. Core segments defined for the threading algorithm (Bryant and Lawrence 1993) are boxed. ALSCRIPT V. 2.0 (Barton 1993) was used to format the alignment.

To illustrate the network of pairwise interactions responsible for maintaining the SOX10 structure, as well as to examine the effect of the 482ins6 mutation, a series of energy scaffolds was generated (Fig.8). The view presented focuses on the interactions between core 1 and core 3, as the interactions between core 2 and cores 1 and 3 are, for the most part, identical. Simple visual inspection of the colored bars connecting core 1 and core 3 for both the human SOX10 wild type and 482ins6 mutant immediately shows that there is a definite change in how the inward-facing residues located between the core regions interact with one another. The most significant change is the shift in a large, favorable interaction seen between Val-1 and Asp-64 in wild-type SOX10. In the 482ins6 mutant, this interaction is missing and is instead replaced by a positive interaction between Val-1 and Arg-60, in essence moving the major interaction between Val-1 and core 3 one helical turn amino-terminal on core 3 (from position 64 to 60). The previously favorable interaction between positions 1 and 64 is now replaced by a small, yet negative, one. Two additional unfavorable interhelical interactions are also seen in 482ins6, between Lys-2 and Lys-64, and between Arg-3 and Lys-64. Two significant changes are seen within helix 3: the addition of a favorable interaction between Glu-53 and Arg-60, as well as an unfavorable interaction between Arg-56 and Lys-64. The net effect of these two new intrahelical interactions is the destabilization of helix 3, as their sum produces a positive ΔG.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 8.

Energy scaffolds for wild-type and mutant Sox10 sequences containing the HMG-1 box domain. The α carbon backbone of the protein is depicted as a curving worm. Within the backbone, segments of the HMG-1 box domain comprising the core folding motif are shown in blue, while the intervening loop regions are shown in yellow. Pairwise residue interaction energies between core residues (Bryant and Lawrence 1993) are indicated by the thickness and coloring of the connecting α carbon positions in the protein backbone. Thick, magenta-colored cylinders are the most favorable interactions; thick, cyan-colored cylinders indicate the least favorable interactions. Intermediate colors and cylinder thicknesses represent interactions falling between these extremes. Numbering corresponds to that in the multiple sequence alignment in Fig. 6 and in Baxevanis et al. (1995). Scaffolds were generated by use of the graphics program GRASP (Nicholls et al. 1991). (Top) Human SOX10; (bottom) human 482ins6 mutant, yielding a Leu–Arg insertion at position 59.

Despite the fact that most of the favorable interactions seen in SOX10 still exist in 482ins6, the chance occurrence probability (E _R‖M changes from 0.02 in the wild type to 0.13 in the mutant. Because its E _R‖M > 0.05, the 482ins6 mutant does not produce a statistically significant match of sequence to structure. It appears that the change in the interaction pattern is moving α helices 1 and 3 out of position with respect to one another, and this net change in structure may be responsible for the observed phenotype in 482ins6 mutants.

DISCUSSION

We present data from a series of molecular and genomic studies aimed at understanding the role of SOX10 in HSCR and associated phenotypes (WS4). In WS4, both hypopigmentation and aganglionosis demonstrate variable penetrance and expressivity, as illustrated by hypopigmentation documented in only four of the six SOX10mutations identified to date (Fig. 6; Pingault et al. 1998). This variable penetrance observed in WS4 could arise from effects of modifier loci acting in conjunction with a primary genetic defect inSOX10. Identity-by-descent mapping in HSCR families has already documented the effects of a presumptive chromosome 21 locus that modifies the extra-enteric phenotypes (white forelock, bicolored irides, and hearing loss) associated with HSCR (Puffenberger et al. 1994).

Sox10^Dom /+ mice model the variable penetrance and expressivity of hypopigmentation and aganglionosis observed in HSCR families. In this study, we used congenic lines ofSox10^Dom mice to analyze one of the variable aspects of the WS4 phenotype, hypopigmentation. During construction of our congenic lines, we documented that only 30% of theSox10^Dom /+ animals survived to weaning in the B6 pedigree in contrast with 82% survival in the C3H pedigree. This finding is consistent with the initial report of decreased survival in B6. Sox10^Dom /+ lines (Lane and Liu 1984) and the observation by Kapur et al. (1996) that aganglionosis is increased inSox10^Dom /+ first generation progeny derived from matings to B6. Further analysis is needed to confirm that increased lethality in our B6. Sox10^Dom /+ line stems from increased severity of aganglionosis. The B6.Sox10^Dom /+ and C3H. Sox10^Dom /+ congenic lines will be invaluable in crosses to segregate the modifier locus(i) responsible for variations in neural crest defects. The orthologous gene(s) in humans for this modifier(s) will be an excellent candidate for interactions with HSCR and WS4 disease loci.

Segregation of loci that influence the penetrance of white forelock hypopigmentation of Sox10^Dom /+ mice was also observed with the congenic lines. Linkage analysis demonstrated that penetrance of the white forelock in Sox10^Dom /+ mice segregates with markers D10Mit96 and D10Mit178. This region of mouse chromosome 10 contains a modifier locus that results in variation of white spotting in another mouse model of WS4,Ednrb^s (H. Rhim and W.J. Pavan, in prep.). Although it is not possible to determine whether the same locus accounts for the white forelock phenotype in both Ednrb^s andSox10^Dom mice, the inheritance of the C3H allele is associated with an increased frequency of white forelocks in both mutants. The identity of the chromosome 10 modifier gene has yet to be established; however, the locus cosegregates with mast cell growth factor (MGF). MGF, the ligand for the Kit tyrosine kinase receptor, is essential for melanocyte development and consequently is an excellent candidate for modifying hypopigmentation.

To evaluate further the role of SOX10 in neural crest development and disease, we performed comparative genomic analyses. Zoo blot hybridizations indicate that SOX10 is conserved in all mammals examined. The gene is conserved in fish as well, a model organism relevant to the study of neural crest development. Evidence ofSOX10 gene conservation in fish and the availability of neural crest mutants in Danio rerio are likely to lead to rapid identification of a SOX10 ortholog in this species. In particular, the colorless (cls) mutant, which lacks pigmentation, exhibits ear and otolith defects (Kelsh et al. 1996), and displays an absence of enteric neurons and a reduction in sensory neuron numbers (R.N. Kelsh and J.S. Eisen, unpubl.), is a primary candidate for mutation analysis of the SOX10 gene. However, conservation of the SOX10 gene in other organisms, particularly avians and yeast is questionable. While one zoo blot showed hybridization of high molecular weight bands in chicken genomic DNA, no hybridization was detected on the other blot. These differences could be due to variation in the genomic DNA quality, partial digestion, or strain selected for genomic DNA preparation. Similar inconsistent hybridization to yeast genomic DNA was also observed for the zoo blots. BLASTn analysis of the probe used for hybridization against the yeast genome database did not detect any homologies to ORFs, suggesting that the bands observed in the yeast lane on the one zoo blot are irrelevant. Further evaluation of Sox10 conservation in chickens will be of interest as this is one of the principal model organisms for study of neural crest development.

Extensive sequence conservation is observed on comparison of theSOX10 coding region between mouse, man, and rat (Pingault et al. 1998) although assignment of the initiatior methionine has not been definitively determined. An in-frame stop codon preceeding a single initiator methionine (Met₁) in the rat (Kuhlbrodt et al. 1998a), absence of alternate 5′ methionines in the rat, and conservation of the coding region around Met₁ between rat, human, and mouse suggest this residue is used for initiation of the ORF. We investigated a second methionine upstream of Met₁ in the mouse, Met_ALT, by comparative alignment of sequences in exons 2 and 3 between mouse, human, and rat. Although we cannot exclude the possibility that Met_ALT functions as an initiator codon in the mouse, our analysis of the human SOX10 cDNA sequence identified a single base deletion that rules out usage of an orthologous Met_ALT in the human. These analyses together with the characterization of the rat 5′ UTR supports use of Met₁ as the initiator methionine and, at the same time, suggests a functional role for the highly conserved 5′ UTR sequence, perhaps in message translation or stability. Although analysis of translated products is needed to determine the functional protein(s) encode by the Sox10 locus, these comparisons have identified conserved regions that may be relevant for mutational analysis in WS-4 individuals.

Several SOX10 mutations have been identified in individuals with WS4 phenotypes (Fig. 6). These patients exhibit sensorineural deafness and hypopigmentation in addition to bowel dysfunction (this study; Pingault et al. 1998). Given the highly variable phenotypes observed in theSox10^Dom /+ mouse (Lane and Liu 1984; Kapur et al. 1996) and the identification of hypoganglionic patients possessingSOX10 mutations (this study; Pingault et al. 1998), this gene should be evaluated in patients who have chronic bowel dysfunction, but lack biopsy results pointing to a definitive diagnosis of aganglionosis. One of the families reported here illustrates this point. In family 192 (Fig. 5), the proband (found to carry the Q377X mutation) was biopsied multiple times and given diagnoses ranging from hypoganglionosis to long segment HSCR. In addition to congenital sensorineural hearing loss, the proband also has nystagmus and ataxic cerebral palsy. His sister shares all of his phenotypic features except aganglionosis/hypoganglionosis. Interestingly, although brother and sister are discordant for their enteric neuronal phenotypes, both siblings are heterozygous for Q377X. These patient phenotypes emphasize the variation observed in HSCR disease phenotypes and expand the potential systems affected by SOX10 mutations. The additional effects documented in these patients are consistent with SOX10 being essential for development of multiple aspects of the peripheral nervous system (PNS) as well as enteric neurons and melanocytes (Southard-Smith et al. 1998).

We used structural modeling to analyze one SOX10 defect identified within the HMG DNA-binding domain (Pingault et al. 1998) of a WS4 patient. Our results reveal by both structural alignments and energy scaffolds that the third α helix of the HMG domain is shifted out of its normal conformation by this 482ins6 mutation. Because the HMG-1 box is L shaped, with an angle of ∼80° between the two arms (Weir et al. 1993; Jones et al. 1994), the model presented here predicts that the mutant does not possess the canonical structure believed to facilitate the binding of HMG-1 box proteins to DNA. Our analysis is consistent with recent studies that demonstrate by gel mobility-shift assay the inability of this mutant form of SOX10 to bind the consensus sequence for SOX proteins (Kuhlbrodt et al. 1998b).

In summary, we established congenic lines ofSox10^Dom mice and demonstrated the utility of these lines for identification of modifiers of neural crest defects by analysis of hypopigmentation. We investigated conservation ofSOX10 in multiple species and found similarity of expression for SOX10 in mouse and man. On the basis of the conserved role of SOX10 in neural crest development of mouse and human, we propose that this gene will be relevant to neural crest lineages in other species as well. These studies include the description of two new mutations of the SOX10 gene in WS4 patients. Our identification of WS4 patients with defects in peripheral nervous system components other than the enteric nervous system and variability in aganglionosis indicates the need to evaluate patients with peripheral nervous system defects that may or may not be accompanied by a diagnosis of aganglionosis for mutations in SOX10. Lastly, our structural modeling to evaluate the effects of human mutation on the HMG DNA-binding domain verifies the susceptibility of this region to mutagenesis and is consistent with in vitro DNA-binding assay results.