Heterogeneous gene expression from the inactive X chromosome: an X-linked gene that escapes X inactivation in some human cell lines but is inactivated in others

Abstract

In mammalian females, most genes on one X chromosome are transcriptionally silenced as a result of X chromosome inactivation. Whereas it is well established that some X-linked genes "escape" X inactivation and are expressed from both active (Xa) and inactive (Xi) X chromosomes, most models for the chromosomal control of X-linked gene expression assume that the X inactivation status of a given gene is constant among different females within a population. In this report, we test the expression of human X-linked genes in primary cell lines from females with complete nonrandom X inactivation, by using transcribed polymorphisms to distinguish Xa and Xi expression. Six X-linked genes used to document this assay system showed monoallelic expression in all informative cell lines, consistent with X inactivation. However, a novel pattern of expression was observed for another gene, REP1; monoallelic expression, indicating inactivation, was detected in some lines, whereas biallelic expression, indicating escape from inactivation, was detected in others. Furthermore, levels of Xi expression varied among cell lines that expressed REP1. The cellular basis of Xi expression was examined by expression assays in single cells. These data indicate that REP1 is expressed from the Xi in all cells, but that the level of expression relative to Xa levels is reduced. These findings suggest that Xi gene expression is under a previously unsuspected level of genetic or epigenetic control, likely involving local or regional changes in chromatin organization that determine whether a gene escapes or is subject to X inactivation.

Figure 1

Nonrandom inactivation in females with structurally abnormal Xs. (A) Fibroblast cell lines were studied from individuals carrying balanced [t(X;A)] or unbalanced [der(X;A)] X; autosome translocations, iso- or isodicentric chromosomes [i(X), idic(X)], or deleted X chromosomes [del(X)]. X chromosomes (normal or with structural rearrangements) that contain the XIST gene can be inactivated (shaded). In carriers of balanced X;A translocations, inactivation of the normal X ensures proper dosage of X-linked gene products (41). In other cases, structurally abnormal Xs are inactivated in females, establishing proper dosage from the active normal X (42). (B) Methylation assay at the androgen receptor gene (AR) to establish nonrandom inactivation. An HpaII site is adjacent to the polymorphic trinucleotide repeat. Digestion of genomic DNA with HpaII prior to amplification digests the unmethylated allele on the Xa chromosome. Subsequent PCR amplifies only the methylated allele(s) on Xi chromosomes. The relative representation of each allele in the HpaII-digested samples (+HpaII), as compared with the amplified undigested samples (−HpaII), demonstrates the randomness (as seen in the sample from a normal XX female) or nonrandomness of X inactivation (cases 48 and 53).

Figure 2

Gene expression of X-linked genes. (A) XIST, G6PD, and DXS6673E show monoallelic expression in all nonrandomly inactivated fibroblast lines tested. D indicates amplification of DNA, and a + or − refers to RNA that has been amplified with or without prior reverse transcription. For XIST and G6PD, samples are digested with HinfI or RsaI (respectively) to differentiate alleles. The DXS6673E PCR products were separated on an ABI 373, and the electropherogram traces are shown. Different normal control female samples (XX) showing various degrees of random and skewed inactivation are shown, in contrast to the nonrandomly inactivated cell lines from the females with structural abnormalities (case numbers indicated). (B) Quantitative amplification of REP1 alleles in fibroblast cell lines from cases 51, 53, and 128. PCR products at the top are digested with HhaI and on the bottom with ApaLI. D, − and +, as in A.

Figure 3

Expression from Xa and Xi chromosomes in mouse/human somatic cell hybrids for DXS6673E and REP1. Negative images of ethidium bromide-stained PCR products are shown. As controls, PGK1 is subject to X inactivation and verifies the inactivation status of the hybrids, and MIC2, a gene that escapes inactivation, demonstrates the presence of amplifiable human X chromosome cDNA from each hybrid sample (23). Amplification products in lanes 1 and 2 are from human (H) and mouse (M) controls. Lanes 3–5, hybrids carrying Xa chromosomes; lanes 6–15, hybrids carrying Xi chromosomes.