Polyalanine expansion diseases are proposed to result from unequal crossover of sister chromatids that increases the number of repeats. In this report we suggest an alternative mechanism we put forward while we investigated a new spontaneous mutant that we named “Dyc” for “Digit in Y and Carpe” phenotype. Phenotypic analysis revealed an abnormal limb patterning similar to that of the human inherited congenital disease synpolydactyly (SPD) and to the mouse mutant model Spdh. Both human SPD and mouse Spdh mutations affect the Hoxd13 gene within a 15-residue polyalanine-encoding repeat in the first exon of the gene, leading to a dominant negative HOXD13. Genetic analysis of the Dyc mutant revealed a trinucleotide expansion in the polyalanine-encoding region of the Hoxd13 gene resulting in a 7-alanine expansion. However, unlike the Spdh mutation, this expansion cannot result from a simple duplication of a short segment. Instead, we propose the fork stalling and template switching (FosTeS) described for generation of nonrecurrent genomic rearrangements as a possible mechanism for the Dyc polyalanine extension, as well as for other polyalanine expansions described in the literature and that could not be explained by unequal crossing over.
HIGHLY conserved Hox genes encode transcription factors containing a homeobox and forming multimeric complexes with other specific DNA-binding partners to regulate the transcription of specific genes (Moens and Selleri 2006). They play key roles in the control of positional information during body axis specification and limb patterning. In mammals, Hox genes are organized in four clusters A to D in which they are expressed with a precise spatial and temporal pattern that reflects their genomic organization, with 5′genes expressed late and in more posterior and distal regions (Kmita and Duboule 2003; Deschamps and Van Nes 2005). In the appendices, only the more 5′ located genes (9–13) are expressed, with a predominant role of A- and D-cluster genes. Duplications and deletions analyses (Herault et al. 1998; Kmita et al. 2005) have revealed a complex network of interactions among Hox genes from the different clusters as well as a global regulation of genes within a cluster.
Despite their central role in vertebrate body patterning, human congenital abnormalities attributable to mutations in HOX genes are rare and have only mild effects (for review see Goodman 2002). Mutations in HOX genes have been mostly associated with some human congenital malformation syndromes of the developing limbs. Whereas mutations in HOXA13 cause hand-foot-genitalia syndrome (HFGS, OMIM no. 140000), characterized by short thumbs and halluces, and clinobrachydactyly of the second and fifth fingers (Warot et al. 1997; Goodman et al. 2000), mutations in HOXD13 give rise to a wide spectrum of limb malformations with variable penetrance and expressivity. Missense mutations within the homeodomain of HOXD13 result in a loss of function that leads to brachydactyly type D (BDD, MIM 113200) or to brachydactyly type E (BDE, MIM 113300). Expansion (Akarsu et al. 1996; Muragaki et al. 1996; Goodman et al. 1997) and, in few cases, frameshift deletions (Goodman et al. 1998; Calabrese et al. 2000) of the polyalanine tract in exon 1 of HOXD13 trigger synpolydactyly (SPD, MIM 186000), a semidominant limb malformation syndrome characterized by digit duplication and fusion in heterozygotes and a combination of syndactyly, polydactyly, and shortening of the hands and feet in homozygotes (Caronia et al. 2003; Yucel et al. 2005).
A spontaneous mouse mutant of Hoxd13 was isolated by Johnson and collaborators (1998) that, in its homozygous state, phenotypically and molecularly modeled human SPD. The Spdh (synpolydactyly homolog) mutation is a 21-bp in-frame duplication within the polyalanine stretch of exon 1, expanding the number of alanines from 15 to 22, and it corresponds to the major type of expansion found in human SPD. Spdh heterozygous mice have only a slight shortening of digits 2 and 5, whereas homozygotes present a severe shortening of all fingers from both forelimbs and hindlimbs, associated with a combination of syndactyly and polydactyly due to a substantial delay in the ossification of the limb bony elements. In addition, homozygous mice lack preputial glands and males are not fertile (Bruneau et al. 2001; Albrecht et al. 2002).
We identified a new spontaneous mutant originated in the BALB/c line that presented an alteration of the distal part of the limbs and was hence named Dyc for “Digit in Y shape and Carpe.” This mutation appeared spontaneously in the last backcross of a transgenic line on the BALB/c genetic background, during the final intercross to generate mice homozygous for the transgene. The phenotype was similar to human synpolydactyly and to phenotypes observed in mutations affecting Hox genes. Phenotypic and genetic characterization of Dyc mice demonstrated that Dyc is a new allele of Hoxd13 affecting the polyalanine cluster. Even though the resulting mutant allele contains, like the Spdh mutant, a 7-alanine expansion, sequencing analysis revealed that the nucleotide expansion in the Dyc mutant is different from that of the Spdh. Spdh polyalanine expansion is a straightforward reduplication of a short segment of the imperfect trinucleotide repeat, whereas the Dyc expansion seems to result from two smaller duplications. Unlike classical long unstable trinucleotide repeat expansions that have been described to result from polymerase slippage during DNA replication (see Kovtun et al. 2004 for review), polyalanine expansions are short, imperfect, and stable when transmitted from generation to generation and hence have been proposed to result from an unequal crossing over between two mispaired wild-type alleles (Warren 1997). However, the Dyc expansion, like a few other alanine tract expansions reported in the literature do not all fit with the model of unequal crossing over (Goodman et al. 1997; De Baere et al. 2003; Robinson et al. 2005; Trochet et al. 2005). Until now, no mechanism other than replication slippage, which does not fit with the “nature” of the polyalanine repeats, was proposed (Trochet et al. 2007). In the article, we describe the phenotype and mutation of the Dyc allele of Hoxd13, apply the mechanism of fork stalling and template switching (FosTeS) (Lee et al. 2007), on the basis of the switching forward or backward of the replication fork along the DNA template using microhomology sequences to generate the Dyc expansion, and propose FosTeS as a new mechanism for polyalanine extensions.
MATERIALS AND METHODS
Mouse lines and complementation test:
The Dyc spontaneous mutation was isolated during the intercross of a mutant line backcrossed at N10 on the BALB/c genetic background. C57BL/6J HoxD<Del3> (Zakany and Duboule 1996) mice were obtained from D. Duboule through the European Mouse Mutant Archive (EMMA; www.emmanet.org). The lines were maintained in specific pathogen-free conditions and all the experiments were carried out in accordance with the European and French regulations. Complementation testing was done in a B6C hybrid background.
Preparations of skeletons were realized for adults, newborns, and embryos, according to Johnson et al. (1998). Briefly, the adults and newborns are eviscerated, skinned, and fixed in ethanol. Staining of the bones is obtained with alizarin red and of the cartilage (newborns) with alcian blue. The flesh is clarified in glycerol. The embryos are fixed in Bouin solution, stained with alcian blue, and the flesh is clarified in methyl salicylate.
In situ hybridization:
Whole-mount in situ hybridization was performed using digoxigenin-labeled probes for Hoxd13, Hoxd12, Hoxd11, and Hoxd10 genes as described previously (Herault et al. 1998) and embryos of stages 12.5 days postcoitum (dpc) obtained by time matings from Dyc/+ × Dyc/+. Embryos were collected, fixed in 4% paraformaldehyde at 4°, dehydrated, and stored in methanol at −20° until use. After rehydratation, bleaching with H2O2 and a pretreatment with proteinase K (10 μg/ml), the embryos were incubated over night at 70° with the probes (1 μg/ml). They were then washed several times, incubated with anti-DIG alkaline phosphatase antibody (Roche Biochemicals) at the dilution of 1:5000 overnight at 4°. After washing of the antibody, staining was performed with BM purple (Roche).
Exon sequencing of Hoxd13 to Hoxd10 genes:
To screen for mutation in the coding sequence of Hoxd10-13 genes, specific primers were designed with the DsGene program for amplification of the exons of those genes by PCR (Johnson et al. 1998). Exons were amplified by classical PCR. Then, both strands of the PCR fragments were sequenced using the Big Dye Terminator reaction mix (Applera, France) on an ABI310 sequencer. The sequence was analyzed with the DSGene software (Acceleris, UK).
Dyc mutants show digit malformations and agenesis of preputial glands:
Mice with abnormal digits in fore- and hindlimb feet were discovered in the BALB/c line during a backcross of a transgene to generate homozygous animals. In the progeny of this intercross, 71% of the animals (n = 109 produced progeny) had a digit alteration, with 48% having a slight phenotype affecting digits 2 and 5, and 23% having a stronger phenotype with shortening of all the fingers. This was indicative of the appearance of a semidominant mutation with the group of animals bearing the mild phenotype being heterozygous and the group with the strong phenotype being homozygous and suggested that the parents were heterozygotes. Looking back at the parents, they were indeed found to have the mild digit phenotype.
In heterozygous adult animals, skeletal analysis (Figure 1a) revealed a shortening of digits 2 and 5. Homozygous animals exhibited a complete alteration of the distal part of the five digits and malformation of all the bones from the ankle (metacarpal, carpal, metatarsal, and tarsal bones). These alterations were associated with a syndactyly for digits 4 and 5 and the appearance of an extra sixth postaxial digit (polydactyly). The hindlimbs had an additional strong deformity of the metatarsal bone of digit 1. This new mutant was called Dyc for Digit in Y and Carpe (Y-shaped finger and carpe). No other alteration of the axial skeleton was found (n = 10 wt and 10 Dyc/Dyc; 12 weeks old). Necropsy analyses (n = 3 wt and 4 Dyc/Dyc) did not reveal any other drastic organ malformation and a hematological analysis (n = 3 wt and 4 Dyc/Dyc; 12 weeks old) indicated a normal globular numeration. Homozygous Dyc males are not fertile. Analysis of the urogenital apparatus of these males revealed that preputial glands were absent. Hence, the Dyc mutation produces two major alterations: a severe malformation of the autopod and the lack of preputial glands in males. These two phenotypes were similar to those of the Spdh mutant mouse, a model for human synpolydactyly characterized by an expansion of unstable repeats within the 5′ end of the Hoxd13 gene (Johnson et al. 1998).
Effect of the Dyc mutation on the ossification:
We went further to check whether the origin of the phenotype of the Dyc mutant during bone and cartilage development was similar to the Spdh mutant. Alcian blue/alizarin red staining of the skeleton was performed at several time points during embryonic and postnatal development to compare cartilage and bone development between Dyc/Dyc mutants and wild-type littermates (supporting information,http://www.genetics.org/cgi/data/genetics.109.104695/DC1/1 Figure S1). Alcian blue staining of autopods at E14.5 dpc reveals a developmental delay in individualization of the fingers and cartilage formation in the Dyc/Dyc autopod. Supernumerary fingers are present. At 15.5 dpc, deformation at the level of the metatarsus and metacarpus are visible. At birth (0.5 days postpartum or dpp), ossification centers are detected in all wild-type digits but not in Dyc/Dyc mutant forelimbs. Polydactily is accompanied by syndactily. At 4.5 dpp, while ossification of the phalanx is completed in wild-type autopods, in Dyc/Dyc autopods, the ossification center's location is not clearly defined and there are defects in segmentation of the finger elements, making it hard to distinguish metacarpus and phalanx.
Dyc mutation is an allele of Hoxd genes:
Limb phenotypes associated with the Dyc mutation are more severe than those found in the Hoxd13 mutation obtained by genetic inactivation, but similar to the alterations found in mutants of the posterior Hoxd genes, controlling limb patterning (Zakany et al. 1997; Kmita et al. 2002a,b). Heterozygote HoxDDel3/+ mutant mice (Zakany and Duboule 1996) bearing a deletion that inactivates the expression of Hoxd11, Hoxd12, and Hoxd13 genes simultaneously have a shortening of digits 2 and 5 like Dyc heterozygous animals (Figure 1, b vs. a). HoxDDel3/Del3 homozygous for this deficiency showed reduction in the length of all the digits (ectrodactyly) with a rougher and stiffer aspect of the digits and polydactyly/syndactyly in the hindlimbs (Figure 1b). The Dyc mutation in its homozygous form seems hence to be less severe than the Del3 one. We carried out a complementation test by crossing Dyc animals with HoxDDel3 mutant mice. HoxDDel3/Dyc transheterozygous mice displayed digit alteration similar to the ones present in HoxDDel3/Del3 homozygotes (Figure 1b). Both have a shortening of all fingers with a malformation of toe 1 and polydactyly in the hindlimbs. Interestingly synpolydactily is rarely observed in transheterozygote animals. These results indicate that the Dyc mutation is an allele of one of the genes from the HoxD complex.
Posterior genes from the HoxD do not have their expression perturbed by the Dyc mutation:
We checked whether the Dyc mutation affected the transcriptional control of posterior genes from the HoxD complex in the limb bud. Only the more 5′ located genes (9–13) are expressed in the limbs. They are expressed in overlapping domains in more or less anterior-to-posterior gradients, are involved in the proper patterning of proximodistal elements, and impact the development of the anteroposterior axis of the limbs. Comparing the expression patterns of Hoxd13, Hoxd12, Hoxd11, and Hoxd10 genes in the limb buds of wild-type controls and Dyc/Dyc mutant embryos at 12.5 dpc using whole-mount in situ hybridization revealed identical patterns of expression between mutant and control animals in the distal part of the autopod (Figure S2). Additional hybridization of Hoxd13 at 11.5 dpc and 13.5 dpc and of Hoxa13 at 13.5 dpc did not reveal any abnormal expression of these markers whereas physical alteration is already visible (data not shown).
The Dyc mutation affects the Hoxd13 protein sequence:
After having checked that expression patterns of the posterior genes from the HoxD complex were not perturbed, we looked for the presence of mutations within the coding sequences of those genes. Sequencing analysis of Hoxd10, Hoxd11, and Hoxd12 did not show any mutation within those genes in Dyc mutants. However, PCR amplification of the 5′ part of exon 1 of Hoxd13 indicated an increase in the size of the amplified segment compared to BALB/c and C57BL/6J controls (data not shown). Sequencing of this fragment revealed a duplicated region responsible for the increased size (Figure 2a). This amplification corresponds to 21 bp containing mainly trinucleotide repeats, coding for a series of 15 alanine residues in the wild-type control and for 22 alanine residues in the Dyc mutant. Hence, the Dyc mutation corresponds to polyalanine expansion similar to the Hoxd13Spdh mutation that creates a gain of function of HOXD13 protein. However, alignment between Spdh and Dyc polyalanine stretches (Figure 2b) indicates that the nature of the expansion is different between the two allelic mutants. Whereas the Spdh expansion appears to be a straightforward reduplication of the nucleotide repeats corresponding to alanines 1–7, the Dyc expansion is more complex, with a duplication of nucleotide repeats coding for alanines 1–5 and the two first nucleotides of alanine 7, whereas a small duplication corresponding to the sequence AGCA shifted in front of the first duplication instead of being after as in the wild-type sequence.
The Dyc mutation is a semidominant mutation causing severe limb malformations that are strikingly similar to those of the Spdh mutant, a 21-bp in-frame duplication within the polyalanine-encoding region of the Hoxd13 gene (Johnson et al. 1998). Like Spdh/Spdh mutants, homozygous Dyc mice have delayed distal limb development and altered chondrogenesis resulting in reduced size of all the fingers associated with polydactyly and syndactyly. Heterozygous animals are less affected, with only a slight shortening of fingers 2 and 5, indicating that the Hoxd13 wild-type allele is able to partly compensate for the effect of the Dyc allele. Like in the Spdh mutant, alterations found in Dyc homozygous mice are more severe than those of null mutants of Hoxd13 (Dolle et al. 1993; Davis and Capecchi 1996) but are similar to those found in the loss of function of the three most posterior genes of the HoxD complex (Hoxd13, Hoxd12, and Hoxd11) (Zakany and Duboule 1996). However, normal pattern expression of those genes during limb development of Dyc/Dyc embryos indicates that inactivation does not occur at the level of the transcription as it could have been expected from the transcriptional control loops between Hox genes that have been described in the literature (Gavalas et al. 2003).
Mapping of the Dyc mutation revealed an expansion of the trinucleotide repeat in the upstream exon of Hoxd13 similar to that found in the Spdh spontaneous mutant that was discovered in a mouse colony with the B6C3Fe genetic background. Both Dyc and Spdh alleles give an additional 7 alanines at the 15-alanines stretch, conferring a dominant-negative property to the HOXD13 protein, probably by interfering with the activity of the wild-type HOXD13 and other HOX proteins. This suggests that the polyalanine tract is important either for interaction with other partners or for formation of a functionally important structure of the protein and that whatever the genetic background, there is a threshold length of polyalanine extension above which the extension becomes pathologic and digital anomalies appear. Polyalanine tract expansion is also the common mutation found in the typical form of human SPD (Akarsu et al. 1996; Muragaki et al. 1996; Horsnell et al. 2006), with additional alanine residues varying from 7 to 14 (Goodman et al. 1997). Like in the mouse, a minimum of 7 additional alanine residues is needed in human HOXD13 for a phenotype to appear and the genetic background does not seem to influence the severity of the phenotype that is influenced by the length of the alanine expansion (Goodman et al. 1997). But human and mouse differ in their heterozygous and homozygous states. Whereas digit fusions and duplications are present in human heterozygotes, this phenotype is only visible in mouse homozygotes indicating a difference in molecular interplay between human and mouse.
Polyalanine tract expansions are found in at least nine disorders (Amiel et al. 2004; Brown and Brown 2004), mostly in genes that are coding for transcription factors involved in developmental processes (Karlin et al. 2002; Lavoie et al. 2003). Unlike classical long unstable trinucleotide repeat expansions found in genes implicated in hereditary neurodegenerative diseases, the polyalanine expansions are imperfect, short, and stable through meiosis and mitosis. This suggests a different mechanism of mutation from strand slippage occurring during DNA replication, repair, or recombination in the other diseases of unstable repeat expansion (Sinden et al. 2002; Pearson et al. 2005; Kaplan et al. 2007; Kovtun and McMurray 2008). Indeed, the expanded polyalanine tract in SPD is quite stable (no increase in mutation size) when transmitted from one generation to another. In addition, polyalanine tracts are short, with the longest nonpathogenic tract being of 20 residues (Lavoie et al. 2003), whereas tracts of ∼35 perfect trinucleotide repeats are required for instability and expansion (Eichler et al. 1994; Kunst and Warren 1994). Warren (1997) suggested that polyalanine expansion found in the mouse Spdh allele and human SPD may have resulted from unequal crossing over in the HOXD13 gene during meiosis due to misalignment between triplet repeats. This mechanism could also explain expansions found in other polyalanine disorders (Robinson et al. 2005; Arai et al. 2007). However, unlike the Spdh mutation, the complex duplication found in the Dyc mutant is not a straightforward reduplication and cannot be explained by unequal crossing over. Looking at the 7-alanine expansion tract (Figure 2b), the additional six last alanines can correspond to a reduplication of alanines 1–6, but there is an additional GCA codon in front of those 6 alanines whose origin cannot be explained. We checked for codon polymorphism in the B6C3Fe and BALB/c lines that could explain this atypical insertion, but could not find any. In addition, even if this additional GCA codon is ignored, the 6-alanine expansion cannot result from simple unequal allelic homologous recombination. Other mutations have been reported in the literature (Goodman et al. 1997; De Baere et al. 2003; Robinson et al. 2005; Trochet et al. 2007), which do not fit the unequal allelic recombination model for mutational mechanism described by Warren (1997). Trochet and collaborators (2007) already refuted the unequal crossing-over mechanism as the explanation of polyalanine expansions. However, they only suggested that the replication slippage mechanism proposed for diseases with unstable triplet expansions was also applying to polyalanine extensions.
We therefore tried to find another mechanism that could lead to the Dyc expansion. Among the mechanisms that have kept our attention is the FosTeS described by Lee et al. (2007). FosTeS was proposed to explain some nonrecurrent chromosomal rearrangements (duplications, deletions) causing genomic disorders and that could not be explained by the usual nonallelic homologous recombination (NAHR) and nonhomologous end joining (NHEJ) (Shaw and Lupski 2004; see Gu et al. 2008 for review of the three mechanisms). Performing a breakpoint sequence analysis of nonrecurrent duplications, they uncovered complex arrays of normal, duplicated, and triplicated sequences at the junctions of the duplications and identified few base-pair microhomologies that could act as “bridges” for the DNA replication fork to skip along the chromosome to create these complex arrays. They proposed that the DNA fork could skip forward or backward along the chromosome using these short sequence homologies when encountering a complex genomic architecture or a DNA lesion. Since the proposition of FosTeS as a mechanism to explain the nonrecurrent rearrangements in PMD patients, another genomic disorder has been analyzed for submicroscopic rearrangements, revealing the potential role of FosTeS in an increasing number of complex pathologic rearrangements (Carvalho et al. 2009). We checked if this long-range template switching mechanism could be applied to the Dyc mutation and more generally to polyalanine extensions via short-range template skipping. We made multiple alignment of the wild-type Hoxd13 polyalanine tract sequence to the Dyc sequence at microhomologies to try to reconstitute the Dyc expansion (Figure 2c). We could identify two microhomology regions of 5 bp each that could have made the replication fork skip two times forward along the wild-type polyalanine sequence to create the Dyc extension. We then listed the polyalanine expansions referenced in the literature that could not be explained by unequal crossing over (Table 1) and we tried the same microhomology alignment method between the expanded alleles and their normal control alleles to determine the process of nucleotide amplification. We encounter the same problem as Goodman et al. (1997) to interpret the 9-alanine expansion of pedigree P. This one can be interpreted as either a simple reduplication of alanines 1–9 via the mechanism of unequal crossing over or of FosTeS (Figure 3) only if there is a polymorphism or a mutation in the triplet coding for alanine 8 (GCT–GCG). Both the 15-bp (+5 alanines) and 18-bp (+6 alanines) insertions found in PHOX2B can be interpreted as a skipping “backward” of the replication fork followed by a skipping “forward.” The 21-bp insertion (+7 alanines) could be interpreted as a single skipping “backward.” However, the microhomology found is only of a single nucleotide. The 15-alanine extension in FOXL2 can be recovered by three skipping backward of the replication fork along the normal polyalanine tract. Finally, the +7 alanine extension in the PABPN1 polyalanine can be the result of a first skipping backward at a 2-bp homology followed directly by a second skipping backward of three nucleotides.
Hence, FosTeS was tested on seven atypical polyalanine extensions and allowed to reconstitute the pathologic expansion for at least five of them. We therefore propose FosTeS as an alternative mutational mechanism to unequal crossing over in the generation of polyalanine expansions. Our hypothesis is further supported by the fact that FosTeS is supposed to be engendered, or at least facilitated, by a surrounding high-order genomic architecture that contains cruciform or other non-B structures (Lee et al. 2007). Such alternative DNA structures are facilitated by specific sequence motifs such as low-copy repeats (LCR), symmetrical features, and short, direct, or inverted repeats, and the polyalanine trinucleotide repeat sequence possesses such sequence motifs and symmetry elements. Looking for specific features flanking the breakpoints of complex rearrangements occurring in the MECP2 region and proposed to result from FosTeS, researchers found increased frequency of the sequences 5′-CTG-3′/5′-CAG-3′ (Carvalho et al. 2009). They suggest that these motifs might represent a cis-acting sequence that may be a recognition site for proteins involved in priming DNA replication in eukaryotes. Interestingly, 5′-CTG-3′/5′-CAG-3′ are motifs that are also found around the breakpoints in the polyalanine expansion tracts (see Figures 2c and 3).
Hence, after having been described as a mechanism for large genomic rearrangements, our findings that FosTeS could also be implicated in monogenic traits via very short stretches of sequence duplications reveals that a variety of genomic rearrangements ranging from gross DNA changes of several kilobases or megabases to rearrangements involving only few nucleotide expansions within single genes could originate from the same molecular mechanism.
Supporting information is available online at http://www.genetics.org/cgi/content/full/genetics.109.104695/DC1.
↵1 Present address: Université de Limoges, 87060 Limoges, France.
Communicating editor: N. Arnheim
- Received May 5, 2009.
- Accepted June 15, 2009.
- Copyright © 2009 by the Genetics Society of America