Evidence for compensatory upregulation of expressed X-linked genes in mammals, Caenorhabditis elegans and Drosophila melanogaster

Abstract

Many animal species use a chromosome-based mechanism of sex determination, which has led to the coordinate evolution of dosage-compensation systems. Dosage compensation not only corrects the imbalance in the number of X chromosomes between the sexes but also is hypothesized to correct dosage imbalance within cells that is due to monoallelic X-linked expression and biallelic autosomal expression, by upregulating X-linked genes twofold (termed 'Ohno's hypothesis'). Although this hypothesis is well supported by expression analyses of individual X-linked genes and by microarray-based transcriptome analyses, it was challenged by a recent study using RNA sequencing and proteomics. We obtained new, independent RNA-seq data, measured RNA polymerase distribution and reanalyzed published expression data in mammals, C. elegans and Drosophila. Our analyses, which take into account the skewed gene content of the X chromosome, support the hypothesis of upregulation of expressed X-linked genes to balance expression of the genome.

Figure 1

Distributions of gene expression are similar between the X chromosome and autosomes in human, except for reproduction-related X-linked genes not expressed in somatic tissues. (a) Expression distributions are similar for X-linked (red) and autosomal genes (blue) in human brain based on RNA-seq data (P = 0.71, by Kolmogorov-Smirnov test). Left, frequency of genes with 0 FPKM; center, histograms of X-linked and autosomal expression distributions; right, cumulative frequencies for genes with >0 FPKM. A theoretical curve (dotted red line) generated by doubling X-linked expression does not result in equal X-linked and autosomal distributions (P = 1 × 10⁻¹², by Kolmogorov-Smirnov test). (b) Box plots of expression of genes with >0 FPKM are similar from each human chromosome in brain. (c) Left, the frequency of genes with 0 FPKM is significantly higher on the X chromosome than each autosome except for chr. 21 in 41 lymphoblastoid cell lines (P < 0.05, by Fisher’s exact test). Right, histograms of expression distribution for genes with >0 FPKM on each human chromosome. (d) X:A median expression ratios increase depending on FPKM cutoffs (>0, ≥0.1, ≥0.5, ≥1 and ≥2). Ratios calculated for individual lymphoblastoid cell lines (17 female, orange and 24 male, green) reveal variability between lines but no differences between males and females (P > 0.05, by Student’s t-test). (e) X:A median expression ratios calculated after separately rank-ordering X-linked and autosomal genes with >0 FPKM by dividing them into 16 bins based on expression. Bins 1–5 contain genes with <1 FPKM (shadowed). (f) Pairwise comparison of the distribution of gene expression in the testis versus liver. Histograms of expression distribution for the subsets of X-linked (X) and autosomal (A) genes expressed in testis (>0 FPKM) but not expressed (0 FPKM) in liver.

Figure 2

Distributions of gene expression are similar between the X chromosome and autosomes in mouse tissues. (a) Expression distributions are similar for X-linked (red) and autosomal genes (blue) in mouse brain (embryonic day 15 (E15) brain) based on RNA-seq data (P = 0.04, by Kolmogorov-Smirnov test). Left, frequency of genes with 0 FPKM; center, histograms of X-linked and autosomal expression distributions; right, cumulative frequencies for genes with >0 FPKM. A theoretical curve (dotted red line) generated by doubling X-linked expression does not result in equal X-linked and autosomal distributions (P = 1 × 10⁻¹³ for E15 brain, by Kolmogorov-Smirnov test). (b) Box plots of expression of genes with >0 FPKM are similar from each mouse chromosome in E15 brain (P > 0.05 for expression comparison of the X chromosome versus 14 of 19 autosomes, one-way ANOVA test). (c) Median X:A expression ratios increase depending on FPKM cutoffs (>0, ≥0.1, ≥0.5, ≥1 and ≥2). Reanalysis of RNA-seq data for mouse brain, liver and muscle. Error bars show 95% bootstrap confidence intervals.

Figure 3

Expressed X-linked genes are enriched in RNA PolII-S5p. (a) Average PolII-S5p occupancy at the 5′ end of 647 X-linked genes (red) is higher in a 1-kb window downstream of the transcription start site (TSS) compared to 16,141 autosomal genes (blue) in undifferentiated female ES cells. Average occupancy (log₂ ChIP: input ratio) is plotted using a 500-bp sliding window (100-bp interval) 3 kb up- and downstream of the TSS. (b) Snapshot of PolII-S5p occupancy in a 2-kb window downstream of the TSS relative to gene expression. Occupancy is higher on X-linked genes (red) compared to autosomal genes (blue) for expressed genes (bins 6–9) but not for weakly- or non-expressed genes (bins 1–5). X-linked and autosomal genes sorted in nine bins based on expression as determined by RNA-seq. Bin 1 contains 124 X-linked and 2,905 autosomal genes with 0 FPKM; bins 2–9 each contain 63 X-linked and 1,596 autosomal genes. X:A median expression ratios are shown in black for bins 2–9. Same ChIP-chip analysis as in a. Error bars show 95% bootstrap confidence intervals. (c) Gene expression is correlated to PolII-S5p occupancy at the 5′ end of genes. Scatter plots of average PolII-S5p occupancy in a 1-kb region downstream of the TSS against X-linked (red) and autosomal (blue) gene expression in log₂ FPKM. Same ChIP-chip analysis as in a.

Figure 4

X:A expression ratios in adult C. elegans result from the presence of germ cells in which the X chromosomes are silenced. We rank-ordered X-linked genes according to their expression values and divided them into 55 bins of 50 genes each. Bin 1 contained the lowest-expressed genes, and bin 55 the highest. We likewise ranked the autosomal genes and divided them into 55 bins. The X:A ratio of expression levels were calculated for each bin. Genes with zero reads were included in the binning. Bins 22–48, which represent the middle of the distribution of expression values, are plotted. This ordering removes the effect of very highly or very weakly expressed genes (for example, in early embryo, bins 1–21 account for 0.3% of X-linked expression and 0.06% of autosomal expression). Weakly expressed genes (for example, bins 22–31) may have higher X:A expression ratios if dosage-compensation mechanisms work most efficiently on appreciably expressed genes.

Figure 5

X chromosome dosage compensation in early mitotic cells in the Drosophila germline. Distributions of gene-level expression values in log₂ FPKM are shown by chromosome arm (≥1 FPKM). (a–d) Red dashed lines indicate the median value for genes on the X chromosome for ovary (a) and testis (b) from wild-type Drosophila and for ovary (c) and testis (d) from bam mutant Drosophila.