|Home | About | Journals | Submit | Contact Us | Français|
Orofacial clefts are common birth defects with a complex etiology. While underlying mechanisms are still largely unknown, altered gender ratios for clefting phenotypes, evidence for linkage to the X chromosome and the occurrence of several X-linked clefting syndromes suggest that skewed X chromosome inactivation (XCI) may contribute to the etiology of orofacial clefting. We tested this hypothesis in a sample set of female monozygotic (MZ) twins and sister pairs discordant for clefting.
We determined XCI in peripheral blood lymphocyte DNA using a methylation based androgen receptor gene assay. We measured skewing of XCI as the deviation in XCI pattern from a 50:50 ratio and used a paired t-test to compare the degree of skewing in cases and their unaffected sisters.
Our analysis revealed no significant difference in the degree of skewing between twin pairs (P=0.3). However, significant differences were observed in the sister pairs (P=0.02), with the cleft lip with cleft palate (CL+P) group showing the most significant result (P=0.01). Results from the cleft lip only (P=0.79) and cleft palate only (P=0.75) groups were not significant.
We did not find evidence for involvement of skewed XCI in the discordance for clefting in our sample of female MZ twins. However, results from the paired sister study suggest that skewed XCI may be important in orofacial clefting, particularly CL+P.
Orofacial clefts are common congenital defects with a complex etiology that involves both genetic and environmental factors. They can be classified as either syndromic or nonsyndromic (isolated) cases, based on the presence of other developmental defects. The latter can be further classified into cleft lip only (CLO), cleft lip with cleft palate (CL+P) or cleft palate only (CPO).1 CPO is considered etiologically distinct from cleft lip with or without cleft palate (CL/P) and is less common, occurring in approximately 1/1500-2000 births, compared to 1-2/1000 births for CL/P.2 They also differ in their sex ratios with CL/P occurring more frequently in males than females, but the reverse is observed for CPO.3 While the etiology of clefting is still poorly understood, genetic mapping approaches and evaluation of candidate genes has found important roles for IRF6,4 MSX1,5 and some other genes,6 but disease-causing mutations are yet to be identified. The gene identification process is hindered by the complex nature of clefting, which includes genetic heterogeneity. In this study, we examined the role of skewed X chromosome inactivation (XCI) in females with isolated clefting.
XCI occurs early in female embryogenesis by the random inactivation of one X chromosome, to achieve equal X-linked gene dosage between females and males. This process is generally random, but a skewed XCI pattern in which the same X chromosome is inactivated in the majority of the cells in a tissue can occur due to selection during development, mutations in genes involved in the X-inactivation process, or stochastic processes.7, 8 In a large study, Amos-Landgraf et al. examined the distribution of XCI patterns in blood samples from 1,005 phenotypically unaffected females, and found that only 8% of the samples had >80:20 skewing. Since highly skewed XCI is rare, a finding of significantly skewed XCI is unlikely to be a purely stochastic event.9 Female carriers of X-linked recessive mutations do not usually express the phenotype, but highly skewed XCI favoring gene expression from the chromosome with the mutant allele can result in manifestation the disease phenotype, and has been reported for various X-linked diseases. Similarly, XCI can modulate expression of X-linked dominant traits.
Differences in XCI have been shown to underlie phenotypic discordance for X-linked diseases in monozygotic (MZ) female twins such as Duchenne muscular dystrophy 10 and red-green color blindness.11 This results from the preferential inactivation of the normal chromosome in the affected twin, while the unaffected twin has predominant inactivation of the mutant chromosome or random XCI.12 Twin studies have previously confirmed a genetic component of clefting, with a higher concordance rate in monozygotic (25–50%) compared to dizygotic (3–6%) twins.13 Female MZ discordant twins provide a valuable resource to examine the role of XCI in clefting. In addition, non-random XCI could result in expression of X-linked diseases in carrier females due to predominant inactivation of the normal chromosome.14 It has also been suggested that XCI may contribute to disease variability between males and females,7 such as the observed altered gender ratios for clefting phenotypes3. Evidence for linkage of clefting to the X chromosome,15 and the possible contribution of genes for X-linked clefting syndromes further support this proposition.
In this study, we have compared XCI pattern in a sample set of MZ female twins and sister pairs discordant for clefting. Our results show that differences in XCI are not likely etiologic in the discordance for clefting in the MZ female twins. However, significantly greater XCI skewing in unaffected females compared to their affected sisters suggests a role for XCI in orofacial clefting, particularly the CL+P phenotype.
For this study, we used DNA previously extracted from whole blood. We had samples from various countries: 13 pairs of MZ discordant twins that are part of an ongoing twin study 16 (Table 1A) and 171 sister pairs (Table 1B). We also obtained both blood and palatal tissue DNA from 11 patients who were undergoing cleft surgical repair at the University of Iowa Hospitals and Clinics. Genotyping of DNA markers confirmed zygosity in the twin samples. All samples were obtained following informed consent and Institutional Review Board approval.
The XCI pattern refers to the relative proportion of the cells with either the maternal or the paternal X chromosome active, and was determined using the androgen receptor (AR) polymerase chain reaction (PCR) assay. The assay is based on the differential methylation of CpG sites in first exon of the AR gene, and the two copies of the X chromosome can be distinguished from each other through a highly polymorphic trinucleotide repeat polymorphism.17 There is a high correlation between XCI patterns determined by the AR-PCR assay with those measured by allele-specific expression at the X-inactive specific transcript (XIST) locus, thus providing support for its accuracy.9 For each sample, one reaction was set up with 10 U HpaII (New England Biolabs), while a control reaction was set up with the enzyme digestion buffer, but with no restriction enzyme. Incubations were 16h at 37°C and the reaction was terminated by incubation at 65°C for 20 min. From this reaction, 1–2ul was used for PCR reaction. All reactions were performed in triplicate.
PCR was carried out in Applied Biosystems Gene Amp PCR System 9700 with an initial denaturation at 94°C for 5 min, 35 cycles at 94°C for 30 s, 59°C for 30 s, and 72°C for 30 s, and a final extension at 72°C for 5 min. The sequence of the forward primer with fluorescent dye 56-FAM (Integrated DNA Technologies) was 5’-TCC AGA ATC TGT TCC AGA GCG TGC –3’. The sequence of the reverse primer was 5’-CTC TAC GAT GGG CTT GGG GAG AAC –3’ 17. 2µL of PCR products were mixed with 9.5µL HD formamide and 0.5 µL GeneScan-500 ROX size standards (Applied Biosystems). The mixture was denatured at 96°C for 1 min and then cooled at 4°C for >15 minutes, before electrophoretic separation on an ABI 3100 Genetic Analyzer (Applied Biosystems).
The XCI pattern was recorded as the proportion of PCR product from one allele relative to the total amount of PCR product from both alleles, and it was calculated using the peak areas generated by the GeneMapper software as previously described.18 The proportion of inactivation of the lower molecular weight allele (allele 1) was calculated using the following formula, which normalizes occasional biases in allele amplification: Proportion (allele 1) = [(d1/u1)/(d1/u1+d2/u2)] x100, where d1 and d2 represent the two peak areas from the digested samples and u1 and u2 are the corresponding peaks from the undigested samples (Figure 1). The proportion of the higher molecular weight allele (allele 2) was obtained by subtracting the proportion of allele 1 from 100. Results were averaged from three replicates of the experiment. When the two alleles differed in length by one (3 base pairs) or two repeats (6 base pairs), the peak area was adjusted by 25% and 4% respectively due to the overlap of stutter bands. XCI skewing was expressed as the deviation, in absolute value, from a 50:50 ratio of the two X chromosomes. It varies between 0 and 50 with 0 designating random X-inactivation and 50 indicating completely skewed X-inactivation.19
We used a paired t-test to compare X-inactivation skewing between the affected females and their unaffected sisters. We also carried out McNemar’s test for the MZ twins due to its small sample size (n=8). The Pearson’s correlation coefficient (r) was calculated to test for the association between X-inactivation skewing in blood and palatal tissue DNA.
Results of the twin samples are shown in Figure 1. Of the 13 twin pairs studied, 8 were heterozygotes for the trinucleotide repeat polymorphism in the AR gene and therefore informative for X-inactivation analysis. The P value was 0.3 for the t-test and 1 for the McNemar’s test, indicating no significant differences in X-inactivation skewing between the twin pairs. The same X chromosome was always preferentially inactivated within each twin pair.
Differences in XCI skewing between affected females and their unaffected sisters were randomly distributed, with a minor skew towards values < 0 (Figure 2). Results of the paired t-test are presented in Table 2. Overall, XCI skewing differed significantly between the affected and unaffected individuals (P=0.02), with greater skewing observed in the unaffected females (t-value < 0). However, when classified by phenotype, the results for the CLO and CPO phenotypic groups were not significant, indicating equivalent XCI skewing in both affected and unaffected sisters. For the CPO group, the small number of samples limits the power to detect a significant difference if it indeed exists. Our data also provided an opportunity to examine skewing differences in sister pairs who inherited the same maternal AR gene allele versus those who didn’t. Interestingly, significantly different skewing was only observed in the stratum where the sisters shared the maternal allele and with similar P values as observed in the aggregated analysis. However, the t statistics remained < 0, validating the overall observation of greater skewing in the unaffected sisters.
To measure how well the XCI pattern in blood lymphocyte DNA may reflect that in the palatal tissue, we studied 11 palatal tissue samples from the cleft patients, 8 of which were informative for X-inactivation analysis. A Pearson’s correlation coefficient (r) of 0.92 was observed between X-inactivation skewing in blood and palatal tissue DNA.
Differential XCI patterns in female MZ twins may account for discordant phenotypes. In this study, we did not observe significant differences in XCI skewing between the discordant twin pairs. Furthermore, each twin pair exhibited skewing in the same direction, so there was no bias in the initial choice of the X chromosome to inactivate. Our results do not support skewed XCI as a mechanism for discordance in these samples. Previous studies have identified differences in XCI as the basis for discordance in MZ twins for various diseases.10, 11 Other factors that could result in discordance include somatic mutations,20, 21 differential methylation,22 variation in gene expression,23 and stochastic factors. Sequencing of a panel of CL/P candidate genes in these samples has not identified any DNA sequence differences between pairs.16 In addition, prenatal non-genetic factors may also be involved in MZ twin discordance.
Reciprocal skewed XCI has been proposed to trigger the twinning process in females, through the segregation of two clones of cells with opposite XCI patterns.12 Our result showed similar XCI patterns within all twin pairs and did not provide support for this hypothesis. While chorion and placenta records were unavailable for this study, identical XCI patterns between our twin pairs suggested monochorionic (MC) placentation. MC twins exhibit similar XCI patterns, compared to dichorinoic twins due to the timing of twinning, relative to onset of XCI.24 Consequently, MC twins are less likely to differ due to XCI patterns.24, 25
One should note tthat we only have 8 informative twins in our analysis, which provides us with minimal power in detecting an effect. Assuming that the effect size in the twins is the same as what observed in the CL+P group, a samples size of 87 would be needed to achieve 70% power at an α level of 0.05. Studies with larger sample size are needed to further investigate this hypothesis.
This study also examined the role of XCI in clefting by comparing XCI skewing in affected females and their unaffected sisters. The matching between sib pairs in this study minimizes confounding factors such as environmental factors and population XCI pattern variations, thus providing a more robust test than population based case-control studies. Overall, we observed significantly greater XCI skewing in unaffected females than in their affected sisters, a finding that was more evident in sister pairs that inherited the same maternal allele at the AR gene locus. The implication of this observation was not clear, but it raises questions about the control and heritability of XCI. Additionally, classification by phenotype revealed that only the CL+P group showed a significant difference in XCI skewing, perhaps due to the severity of the phenotype or the large sample numbers, which increases the power to detect an effect.
Significantly greater XCI skewing in unaffected than in affected females is not surprising since predominant inactivation of a chromosome bearing a hypothetical X-linked gene predisposing to clefting would have a protective effect on the unaffected individual, resulting in a normal phenotype. Conversely, less skewing as observed the in affected females could result in sufficient gene expression from a variant allele, which would trigger the clefting phenotype. Such variation in XCI may underlie differences in disease expression between females with an otherwise similar genetic background. However, for a complex trait such as CL/P with complex inheritance, variable XCI might only have a minor contribution to disease etiology. Genes for X-inked syndromic forms of clefting may play a role in isolated clefting, and thus skewing of XCI could potentially influence their expression levels. Online Mendelian Inheritance in Man (OMIM) lists more than 20 X-linked diseases with clefting as part of the phenotype. An example is X-linked cleft palate and ankyloglossia (CPX; OMIM 303400) caused by mutations in TBX22.26 While most female carriers are asymptomatic, 16% have a cleft palate that may be attributed to skewed XCI. Interestingly, this gene lies in a region that has been linked to isolated clefting.15
Differences in XCI are consistent with an altered gender ratio in disease prevalence, with CL/P occurring more frequently in males, and CPO more frequently in females.3 These differences have previously been attributed to differences in the timing of palate development,27, 28 and embryonic face shape.29 XCI skewing may also afford females a higher threshold than males, who carry one X-chromosome. It is conceivable that a male who inherits an X-chromosome predisposing to clefting will be at a greater risk than a female, for whom gene expression is subject to modification through XCI. Apart from the sex differences in disease prevalence, there is one report of a sex dependent association of clefting with MSX1.30
An important consideration for XCI is that skewing in peripheral blood increases as women age, and is more pronounced after the age of 50–60 years.19 In this study, subjects’ ages were available for ~75% of the affected individuals, ranging from 1–42 years (mean=11 years). The ages of the unaffecteds were mostly unavailable, but would be expected to fall within the same range as that of the affecteds. According to Kristiansen et al’s definition,19 our study subjects fall in the ‘young’ age group and therefore we do not expect that age-related skewing would significantly alter our observations.
Another concern is that XCI patterns may vary from tissue to tissue, especially those originating from different germ layers.8 Peripheral blood lymphocyte DNA is of mesodermal origin, while the tissues that form the lip and palate are of ectodermal and neural crest origin. Comparison of XCI skewing in blood lymphocyte and palatal tissue DNA in a modest number of cases showed a high Pearson correlation coefficient (r) of 0.92, suggesting that XCI skewing patterns in palatal tissue DNA may be reflected in blood lymphocyte DNA.
In summary, we report greater XCI skewing in unaffected females compared to their affected sisters. Our results suggest a role for this mechanism in clefting, particularly CL+P. We are unaware of similar XCI studies in females with CL/P; therefore, additional confirmatory studies are guaranteed.
We would like to thank all the families who willingly contributed samples and information for this study. We particularly thank Laura Mitchell and Georgia Chevenix-Trench for providing samples for the twin study, Lina Moreno for providing samples from Colombia, Andrew Lidral for many helpful discussions, and Maria Adela Mansilla for managing the samples and performing zygosity tests. Many thanks to Dr. Frederic Deleyiannis, Dr. Javier De Salamanca, Dr. Beatriz Gonzáles Meli, Dr. Katherine Neiswanger, Seth Weinberg, Carla Brandon, Kathleen M. Bardi, Stephanie Petriprin, Negin Noorchashm and Nicole Scott for their assistance with subject recruitment. This work was supported in part by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences. This work was supported by NIH grants R37-DE-08559, R01-DE-09886, R01-DE016148, R01-DE011948, R01-DE014667, R21-DE016930, P50-DE016215 and R01-DE-015197.