Introduction

Mutations in the Aristaless-related homeobox (ARX) gene are frequent and pleiotrophic, giving rise to non-syndromic intellectual disability, infantile spasms or serious brain malformations.1 More than half of the ARX cases reported are due to polyalanine expansion mutations, in two different polyalanine tracts. The c.429–452dup mutation affecting the second tract is the most frequent ARX mutation. Follow-up work on these c.429–452dup families revealed a greater than expected number of affected males among the offspring of the obligate carrier females.

According to the fundamental principle of Mendelian genetics, two alleles of a diploid locus have an equal probability (50:50) to be transmitted to an offspring. The deviation from this rule is known as transmission ratio distortion (TRD).2 There are several examples of TRD suggested in human genetics, however this phenomenon remains inconclusive in man, highlighting the importance of determining this phenomenon robustly. In the case of autosomal dominant myotonic dystrophy, TRD has been reported in live-born offspring from affected female parents but not affected male parents.3, 4, 5 When looking at meioses from unaffected parents heterozygote for different repeat lengths (non-disease causing) at the DM locus, there was a significant distortion in favour of transmission of the larger allele in males.6 In contrast, no TRD was determined for any allele length when sperm from heterozygous unaffected donors were analysed7or when pre-natal diagnoses were analysed from either affected mothers or fathers.8

Other autosomal dominant diseases also have mixed reports of TRD. With respect to Long QT syndrome, a large study investigating over 750 nuclear families, a higher than expected transmission of the mutant allele from female carriers of type I disease and from either male or female carriers with type II disease was reported.9 Moreover, this study indicated there was an increased transmission to female offspring. A later study of 182 families from two particular Long QT syndromes did not find the same distortion to the expected 50:50 Mendelian transmission.10 Diseases caused by expansion of CAG repeats, such as Machado–Joseph disease11, 12 and spinocerebellar ataxia 713 are suggested to show a prevalence of transmission of the mutant allele, but again have conflicting reports. Investigating transmission at two CAG disease repeat loci, TRD was noted in male meioses in live-born offspring11 and in sperm typing from MJD patients14 and preferentially transmitted to live-born offspring by female carriers.12 A study of heterozygotes of non-disease CAG repeats in unaffected families15 and unaffected twins and parent groups16 did not find TRD in male meioses. This may be, in part, due to the difficulties in comparing the findings from normal alleles with alleles in the disease range. Interestingly, the study of Rubinsztein et al15 found in 57% of female meiosis smaller non-disease causing CAG repeats, suggesting meiotic drive may occur among some non-disease causing CAG repeats sizes.

Among recessive disorders, TRD has been reported in the congenital disorder of glycosylation type Ia (CGD-Ia).17 With respect to tri-nucleotide repeats in X-chromosome genes, a TRD in favour of the mutant alleles of intermediate length of FMR1 has been shown in the X-linked Fragile X syndrome,18 but for neither the FRAXA nor the FRAXE loci when common, intermediate pre- and full-mutation alleles were analysed separately.19

Similar to FMR1, ARX is located on the X chromosome and as such, heterozygote carrier females can transmit either the normal or mutant allele to their daughters and sons. If a daughter inherits the mutant allele from the carrier mother and a normal allele from her unaffected father, she too will be a heterozygote carrier.1 As the ARX gene is randomly X-inactivated, these women will have half of the ARX-expressing cells with normal ARX and half with the mutant ARX. However, if a son inherits the mutant allele he will be affected as there is no ARX paralog on the Y chromosome and all ARX-expressing cells will produce the mutant version.

In this study, we have investigated the transmission ratio of the c.429–452dup mutation in a large cohort of published and unpublished ARX families. We have found that contrary to the expectation of Mendelian genetics, carrier females preferentially transmit the c.429–452dup mutation to the next generation as judged by significantly skewed numbers of affected males versus normal males.

Materials and methods

For the purpose of assessing the transmission ratio of the c.429–452dup mutation, we collected and assessed all published families (n=33) as well as some unpublished pedigrees (n=6) segregating the c.429–452dup mutation (see Table 1 and references therein). We counted the number of obligate carrier females and the number of normal and affected males they had. All families analysed had at least two affected males to correct for the ascertainment bias (see below). We did not score any clinically affected females, albeit rare, as the contribution of the c.429–452dup mutation to their phenotype has not yet been established.

Table 1 Numbers of carrier females and offspring with the c.429–452dup mutation in ARX

Statistical analaysis

The null hypothesis we tested was the Mendelian X-chromosome allele segregation ratio of 50:50. To prevent ascertainment bias, we excluded all probands from our transmission ratio analysis, automatically removing any families with a single affected male. We used the Chi-square test to establish the significance of our findings. A Chi-square table with one degree of freedom assesses the number of possible outcomes minus one (either affected or unaffected). A P-value of less than 0.05 was used to indicate statistical significance.

Results

A total of 39 families segregating the ARX c.429–452dup mutation were analysed (Table 1). All together, we counted 281 females, of which 137 were presumed non-carriers or not tested and 144 were carriers (Table 1). The 144 carrier females had 301 sons with 188 affected males (both alive and deceased) and 98 unaffected males (χ2=28.32; P<0.0001; Table 2). Removal of the 39 probands from the analysis to correct for the ascertainment left 149 affected males and 98 normal males, a total of 247 male offspring. The expected Mendelian segregation ratio from this cohort would be 123.5 males in each clinical category. The distortion of the transmission ratio to 149:98 was 60% in favour of the males with the c.429–452dup mutation. When tested using Chi-square (χ2=10.53) it was significant at P<0.002 (Table 2). To address a possible ascertainment bias, we also used a more stringent approach by removing two probands from the nine MRX families and one proband from all other families, with a remaining total of 140 affected males and 98 normal males. This conservative test still gave significant results (χ2=7.412; P<0.0065).

Table 2 Chi-square goodness of fit analysis of transmission of the mutant c.429–452dup ARX mutant allele to male offspring

Across the 39 families evaluated, there were also an additional 15 males born to carrier females that were indicated as deceased on respective pedigrees. In the majority of these cases, the genotype, cause of death and the age at which they died was not known. If we conservatively assume that these males were not affected, that is, did not carry the c.429–452dup mutation and include in the analysis, the expected transmission of the mutant allele was still distorted (χ2=4.95; P<0.05).

Discussion

Our data suggest that there is a distortion of the expected Mendelian transmission ratio of the ARX alleles in favour of the deleterious c.429–452dup mutation. This finding may have implications for genetic counselling of families segregating this particular mutation. To assess whether the TRD is seen also for other ARX mutations, we attempted to extend our analysis to families with the c.304ins(GCG)7 mutation. This particular mutation leads to expansion of the first polyalanine tract and is associated with more severe clinical outcomes of X-linked infantile spasms,20, 21, 22, 23, 24 infantile epileptic-dyskinetic encephalopathy22 or Ohtahara syndrome.25 However, of the 11 published families with this mutation, many include only a single affected individual, thereby excluding them from this type of analysis. Mouse models of both of these mutations have recently been generated.26, 27 However, there is no data as yet available regarding the transmission of the mutant alleles. Provided the TRD holds true in these knock-in mice, it will provide an opportunity to investigate the molecular mechanisms of this phenomenon in greater detail.

The causes underlying TRD include segregation distortion, which might be due to selection, and meiotic drive. Segregation distortion typically occurs after meiosis, but prior to fertilisation. We cannot formally rule out that a specific cis-allele at another locus, in a close proximity to the ARX gene, is responsible for this transmission distortion rather than the ARX mutation itself. However, this is unlikely given the recurrent (in contrary to the identity by descent) origin of the c.429–452dup mutations.20 The deviation from the Mendelian inheritance may occur as a result of post-fertilisation lethality of embryos or neonates of a particular genotype. TRD was reported for myotonic dystrophy, an autosomal dominant disorder with anticipation, where preferential transmission of the longer alleles (from heterozygous carrier females) had been observed.4 Although subsequent studies have narrowed down the window of TRD to the time prior to pre-implantation, the mechanism still remains elusive.28 In the case of the c.429–452dup mutation in ARX, we would not expect an increase in lethality of the embryo carrying normal, maternally derived ARX alleles, effectively ruling out this type of deviation from Mendelian inheritance as an explanation.

Meiotic drive, on the other hand, requires TRD to occur during female meiosis. This means that the resulting gametes are not preferentially lost nor is fertility itself affected, only the inheritance of the neutral polymorphism is reduced.29 This appears to fit with the preferential inheritance of the c.429–452dup mutation. The mechanism driving the TRD remains unknown. An elegant study directly testing meiotic drive, in particular during the second meiosis, was reported for TRD of the DDK allele at the mouse Om locus.30 Originally, the embryo lethality of DDK female cross with a male from any other inbred mouse strain was attributed to single-locus lethality model.31 However, in the study by Wu et al,30 the male pronuclei, female pronuclei and second polar bodies were recovered from single cell embryos, and the genotypes of single chromatids at the Om locus and the associated centromeres were established. This study demonstrated preferential segregation of chromatids carrying the DDK allele at the Om locus to the maternal pronucleus and the reciprocal preferential segregation of chromatid carrying the wild-type allele to second polar body. This only occurred when the dyad obtained from the first meiosis was heteromorphic, that is, recombined between the centromere and the Om locus. Essentially, this means that TRD only occurs when it is possible for the ova to make a segregational choice between the DDK and Wt alleles during meiosis MII. This is consistent with non-random segregation. Of interest, the authors note that the levels of distortion to transmission of the DDK allele show striking uniformity with TRD in univalent X chromosomes in MI, in mouse and human Robertsonian translocations.30 Similar to these observations, we identified comparable levels of distortion of the transmission of the c.429–452dup mutation in ARX to male offspring in our study. This similarity may indicate the asymmetry of the oocyte meiotic spindle, occurring at either MI or MII, as the common underlying mechanism of TRD. The mechanism of asymmetry of meiosis in the oocyte having an important role in TRD is further supported by cases of de novo mutation of the c.429–452dup32, 33, 34 and also c.430–456dup,35 both identified through maternal gonadal mosaicism. There are no reports to date of gonadal mosaicism in the (grand)father for any expanded polyalanine tracts in ARX. Taken together, a better understanding of the role of ARX in oocyte maturation may shed light on the forces underlying the phenomenon of TRD.