Skewed X-inactivation is common in the general female population

Abstract

X-inactivation is a well-established dosage compensation mechanism ensuring that X-chromosomal genes are expressed at comparable levels in males and females. Skewed X-inactivation is often explained by negative selection of one of the alleles. We demonstrate that imbalanced expression of the paternal and maternal X-chromosomes is common in the general population and that the random nature of the X-inactivation mechanism can be sufficient to explain the imbalance. To this end, we analyzed blood-derived RNA and whole-genome sequencing data from 79 female children and their parents from the Genome of the Netherlands project. We calculated the median ratio of the paternal over total counts at all X-chromosomal heterozygous single-nucleotide variants with coverage ≥10. We identified two individuals where the same X-chromosome was inactivated in all cells. Imbalanced expression of the two X-chromosomes (ratios ≤0.35 or ≥0.65) was observed in nearly 50% of the population. The empirically observed skewing is explained by a theoretical model where X-inactivation takes place in an embryonic stage in which eight cells give rise to the hematopoietic compartment. Genes escaping X-inactivation are expressed from both alleles and therefore demonstrate less skewing than inactivated genes. Using this characteristic, we identified three novel escapee genes (SSR4, REPS2, and SEPT6), but did not find support for many previously reported escapee genes in blood. Our collective data suggest that skewed X-inactivation is common in the general population. This may contribute to manifestation of symptoms in carriers of recessive X-linked disorders. We recommend that X-inactivation results should not be used lightly in the interpretation of X-linked variants.

Fig. 1

Distribution of mean allelic (a) and paternal (b) ratios for each individual. Black lines are the smoothed density curves corresponding with the obtained distributions

Fig. 2

Lack of association between X-inactivation status in mothers and daughters. Scatter plot of median measure of balance for mothers (x-axis) and daughters (y-axis) at least ten reads coverage on heterozygous SNVs. There is no significant correlation (Pearson’s ρ = 0.038, p value = 0.7934)

Fig. 3

Theoretical assessment of cell numbers at the time of X-inactivation. Comparison of the empirical median paternal ratio distribution for heterozygous SNVs with more than ten reads per individual (orange line) with theoretical distributions under the hypothesis that X-inactivation takes place at the 4 (dotted black line), 8 (dashed black line), 16 (long dashed black line), and 32 (solid black line) precursor stage. Theoretical distribution at eight initial lineage-restricted precursor cells is most comparable with empirical distribution (highest p value = 0.011, two-sample Kolmogorov–Smirnov test)

Fig. 4

Assessment of escape from X-inactivation. Histogram of the skew factor for the entire X-chromosome (black bars) and for specific example genes (gray bars) in all skewed individuals (one bar for each individual) with coverage ≥10 on heterozygous SNVs in those genes. A one-sided test was used to test whether the ratios for a given gene were significantly different from the median ratio for the entire X-chromosome. In a, b, two “known” escapee genes: [, –23] PUDP (ENSG00000130021) appears to escape X-inactivation, whereas TRAPPC2 (ENSG00000196459) does not. In c–g, several genes not known to escape X-inactivation: c TFE3 (ENSG00000068323) does not escape X-inactivation (in line with the literature), whereas SSR4 (ENSG00000180879), REPS2 (ENSG00000169891), and SEPT6 (ENSG00000125354) were identified to escape X-inactivation for the first time in our study. g GAPDH65 (ENSG00000235587) was found to be significant, but is a pseudogene and a likely false-positive gene due to inaccurate read mapping

Fig. 5

Overview of X-chromosomal genes that do (significant, p < 0.05) or do not (non-significant, p ≥ 0.05) escape X-inactivation in blood in our study in comparison to previous studies. Note: The first number in each cell corresponds to the number of escapee (significant, p < 0.05) or non-escapee (non-significant, p > =0.05) genes in our study, status of which matches the corresponding status in the literature [,–23]. The second number in each cell shows how many overlapping genes between our study and each of the referenced studies have the corresponding status according to the literature. Shading of the cells reflects degree of overlap (white: 0%, light grey: 1-50 %, grey: 51-99%, dark grey: 100%) Tissues analyzed: Carrel - primary human fibroblast cell lines, rodent/human somatic cell hybrids [14] Park - primary human fibroblast cell lines, rodent/human somatic cell hybrids [14] Zhang - immortalized human B-cells [22] Cotton - human fibroblast cell lines [20] Tukiainen - diverse human tissues [23]