On human Xq24, highly homologous copies of CT47 are organized into an MSR of tandemly arranged units of each 4.8
kb in size. This polymorphic MSR varies from 4 to 17 copies in length (manuscript in preparation and Chen et al19
). As CT47
has a very specific expression profile, it is a good model to study the tissue-specific epigenetic regulation of MSRs. CTAs form a large family of protein coding sequences with specific expression pattern: they are expressed in the testis and certain types of cancers. Our study presents the first detailed analysis of the chromatin structure of a CTA in non-malignant and malignant cells and during early stages of development. Although biological processes leading to cancer should preferably studied on human tumor material, the established SCLCs represent a valuable and readily available tool for studying genetic29
and epigenetic changes30
Our expression data are in a good agreement with earlier studies showing that CT47
expression is restricted to the testis and malignant cells,19
albeit that the expression in SCLCs is only a fraction of that observed in the testis. We cannot exclude the possibility that only a small proportion of SCLC cells in culture express CT47
, analogous to the presence of small subpopulation of solid tumors.2
We did not observe the expression of CT47
in hESC cells and hEBs. The CTAs MAGE-A, GAGE and NY-ESO-1 were previously shown not to be expressed in hESCs,31
whereas MAGE-A2 and MAGE-A6 expression could be detected in 20-day-old hEBs. Although we cannot exclude the possibility that the absence of CT47
transcript in our hESC and hEB samples is the result of the epigenetic fluidity of human hESCs,23, 32, 33
our results suggest that CT47
transcription is induced at later stages of development, possibly between the pluripotent stem cell and primordial germ cell stages, from which mature germ cells are developing.
Chromatin analysis of non-expressing cells showed high levels of heterochromatic histone modifications H3K27me3 and H3K9me3, and relatively low levels of the euchromatic histone modification H3K4me2 in the promoter region of CT47
. We could not detect a correlation between CT47 array length and its chromatin structure in analogy with FSHD, where contractions only below a threshold of 11 units result in D4Z4 chromatin relaxation.11
Rather, we observed a high variation in the abundance of different histone modifications in samples with the same array lengths ( and ). As D4Z4 shows similar variability in levels of histone modifications with repeat lengths >11 units, it is likely that 4 units of CT47 is sufficient to ensure a heterochromatic chromatin environment in non-germline cells and opens up the possibility that each MSR needs a minimal, but for each MSR unique, number of repeat units to establish its appropriate chromatin structure.
We observed a relative higher abundance of H3K27me3 to H3K9me3 in female samples. The observation that females have relatively higher levels of H3K27me3 than males suggests that repression of CT47 is differently established on Xa and Xi. Indeed, Xi in female samples can be more densely populated with H3K27me3 than Xa.34
Nevertheless, this is different from the X-chromosomal MSR DXZ4, showing that two MSR on Xq can behave very differently in chromatin structure on the Xa and Xi.
A unique feature of the pluripotent state is the so-called bivalent chromatin state defined by the presence of permissive (H3K4me3) and repressive (H3K27me3) histone modifications.35
Bivalent states are found at promoter regions of genes expressed at low levels in hESCs that are associated with differentiation. We detected identical low levels of H3K4me2, H3K4me3 and H3K27me3 in hESCs and hEBs at the CT47
promoter region, suggesting that the product of this gene is not necessary for early development.
Repetitive elements were identified as targets of PRC1 and PRC2 complexes in the pluripotent state.36
Our results show that the MSR CT47 has, like D4Z4,11
a heterochromatic structure already at early stages of development, marked by abundant H3K9me3 levels, and that this chromatin structure does not change during differentiation. The high levels of H3K9me3 suggest that repression of MSR is mainly governed by SUV39H1 and SUV39H2 and not so much by PRC complexes.
SCLC cell lines, which do express CT47, show a disturbed chromatin structure of this MSR. In addition, male cell lines NCI-H69 and NCI-H187 harbor three arrays and NCI-H82 two arrays of CT47 MSR, suggesting that they have become aneuploid for the X chromosome. We detected significantly lower levels of H3K9me3 in these cancer cell lines without a concomitant relative increase in the abundance of the transcriptionally permissive histone modification H3K4me2. This might explain the relatively low levels of CT47 expression in these cells. Rather than transcriptional reactivation, SCLCs seem to show a relative loss of transcriptional repression, resulting in leaky expression of CT47. Owing to technical limitations, we can currently not exclude the possibility that individual units within an array are differently organized.
In contrast to H3K9me3, H3K27me3 levels are not significantly lower in SCLCs, whereas EZH2 levels are significantly decreased in these samples. This shows that SCLC cells have serious defects in coordinated epigenetic regulation of this MSR, consistent with reports of genome-wide epigenetic changes in other malignancies.16
CTA gene expression is often correlated with DNA promoter hypomethylation. We observed that all of the CpG dinucleotides examined in the promoter region of CT47 show a significant decrease in DNA methylation in SCLCs compared with the otherwise high methylation levels in non-malignant cells. Epigenetic factors like DNA methylation and histone modifications are acting in a coordinated way to establish normal development. Our study reveals that transcriptional regulation/derepression of CT47 is dominated by the loss of repressive markers such as DNA methylation and H3K9me3: SCLCs with significantly lower levels of DNA methylation and H3K9me3 at the promoter region are expressing CT47, despite the absence of open chromatin markers.
We observed striking commonalities between the autosomal MSR D4Z4 and the X-chromosomal MSR CT47. In normal somatic cells, hESCs and hEBs, both MSRs are highly methylated and have high levels of repressive chromatin markers. In the adult body, DUX4
encoded by D4Z4, and CT47
transcripts are highly abundant in the testis and ovary. In disease conditions, like FSHD or cancer, there is a loss of DNA methylation and repressive chromatin markers, but no increase in permissive chromatin modifications, leading to low or leaky expression levels of the MSR-encoded genes. Like CT47
expression in SCLCs, evidence was presented that DUX4
is expressed in malignant adult and fetal human rhabdomyosarcoma cell lines,38
and therefore DUX4
also fits the definition of a CTA. The shared features of epigenetic regulation and expression patterns of CT47 and D4Z4 suggest that other MSRs may also be regulated similarly. A recent publication shows that indeed the family of MSRs exclusively expressed in the germ line and testis is expanding and includes new members like TAF11-Like, PRR20 and ZAV.9
Understanding the epigenetic organization of MSRs in health and disease, which cover a high portion of the human genome, may therefore significantly advance our understanding of the epigenetic map of the human genome and of germline biology.