Allele-specific analysis of cell fusion-mediated pluripotent reprograming reveals distinct and predictive susceptibilities of human X-linked genes to reactivation
© The Author(s). 2017
Received: 4 August 2016
Accepted: 14 December 2016
Published: 25 January 2017
Inactivation of one X chromosome is established early in female mammalian development and can be reversed in vivo and in vitro when pluripotency factors are re-expressed. The extent of reactivation along the inactive X chromosome (Xi) and the determinants of locus susceptibility are, however, poorly understood. Here we use cell fusion-mediated pluripotent reprograming to study human Xi reactivation and allele-specific single nucleotide polymorphisms (SNPs) to identify reactivated loci.
We show that a subset of human Xi genes is rapidly reactivated upon re-expression of the pluripotency network. These genes lie within the most evolutionary recent segments of the human X chromosome that are depleted of LINE1 and enriched for SINE elements, predicted to impair XIST spreading. Interestingly, this cadre of genes displays stochastic Xi expression in human fibroblasts ahead of reprograming. This stochastic variability is evident between clones, by RNA-sequencing, and at the single-cell level, by RNA-FISH, and is not attributable to differences in repressive histone H3K9me3 or H3K27me3 levels. Treatment with the DNA demethylating agent 5-deoxy-azacytidine does not increase Xi expression ahead of reprograming, but instead reveals a second cadre of genes that only become susceptible to reactivation upon induction of pluripotency.
Collectively, these data not only underscore the multiple pathways that contribute to maintaining silencing along the human Xi chromosome but also suggest that transcriptional stochasticity among human cells could be useful for predicting and engineering epigenetic strategies to achieve locus-specific or domain-specific human Xi gene reactivation.
KeywordsReprograming X chromosome reactivation Stochasticity
In mammals, one of the two female X chromosomes is randomly inactivated to compensate gene expression between males (XY) and females (XX) . This process takes place in the pluripotent cells of the early embryo and is inherited thereafter through cell division. As a consequence, female tissues comprise a mosaic of cells with reciprocal X chromosome inactivation (XCI) patterns that have been clonally inherited through development .
Silencing along the inactive X chromosome (Xi) leads to the formation of a condensed nuclear compartment, which excludes RNA polymerase and is enriched of repressive chromatin modifications (reviewed in ). Xist long non-coding RNA triggers the formation of this compartment by binding along the Xi  and then recruiting chromatin modifiers such as polycomb repressive complexes [5, 6]. Following these early events, gene silencing is established and spreads along the Xi with different kinetics [7–9]. Not all the genes on the Xi are effectively silenced and some genes escape inactivation and instead remain transcriptionally active [10, 11]. The heterogeneity in kinetics and efficiency of silencing suggests that local genetic and epigenetic features might determine the susceptibility of X-linked loci to inactivation. This hypothesis is supported by recent studies showing that the pre-existing chromatin context favours the recruitment of different sets of silencing factors and influences the efficiency with which XCI is established and propagated [12, 13]. Whether the local gene context also affects the maintenance of silencing and, therefore, the susceptibility of gene loci to be reactivated has been more difficult to interrogate.
The cell type, age and developmental context might also influence locus susceptibility to silencing and reactivation. Supporting this idea, recent studies in different mouse cell-types and tissues showed that some Xi genes escape silencing in a tissue-specific manner . Furthermore, in mouse it has been shown that some genes escape silencing at the onset of XCI , whereas others are initially silenced in the embryo and are then reactivated later in development . As genes that escape XCI lack Xist coating  and repressive H3K27me3 and H3K9me3 histone modifications [17, 18], one explanation could be that they retain an accessible chromatin during XCI and therefore are susceptible to subsequent reactivation. Supporting this hypothesis, RNA polymerase II has been shown to bind a subset of Xi genes that are silenced in some cell lines but active in others, possibly marking genes permissive to transcription .
Dynamic studies of the global Xi status during cell fate reprograming offer an opportunity to dissect the influence of pre-existing chromatin and cellular environments on Xi gene reactivation. Towards this aim, we examined Xi-specific expression during pluripotent reprograming of human female fibroblasts. XCI in humans has some intrinsic advantages for studying the impact of gene context and cellular environment on silencing or transcription. First, the percentage of genes escaping silencing in human is higher than in mouse (15% versus 3–7% of mouse), with 10% of genes showing a variable XCI status (i.e. escape or are subject to silencing) in different females or tissues . This suggests that spreading of XCI in human might be less efficient and has evolved independently from different mammalian species. Second, compartmentalisation of the human X chromosome based on evolutionary, genomic and chromatin features has revealed some important associations between locus context and probability of silencing or escape. For example, many genes escaping XCI are contained within the evolutionary recent X Added Region (XAR) enriched for Alu repeats and depleted of Long Interspersed Elements (LINE) L1 and H3K27me3/H3K9me3 repressive histone modifications [17, 18, 21–25]. These findings may implicate a role for DNA elements, chromatin features and chromosomal domains in the spreading of human XCI.
Pluripotent reprograming approaches allow reproducible studies of Xi reactivation  and offer the possibility of investigating the predisposition of different loci. Interestingly, several studies in human induced pluripotent stem cells (iPSCs) have shown that the Xi can be either fully or partially reactivated in different clones [27–29]. A comparative analysis of the X:autosome expression ratio in several iPSC lines suggested that telomeric chromosome regions distal from XIST locus are preferentially reactivated . However, a general consensus about the susceptibility of genes to X chromosome reactivation (XCR) and their location has not been achieved, mainly due to the variable outcomes of human iPSC reprograming and a paucity of studies that directly assess the expression of Xi-specific alleles across the entire X chromosome. Here, we used RNA-sequencing (RNA-seq) and allele-specific analyses to investigate human XCR during pluripotent reprograming of single-cell derived female fibroblast clones. Because XCR is a late event in iPSC-reprograming, is unstable and varies depending on the combination of inducing pluripotency factors and culture conditions (reviewed in ), we have used an alternative model system in which human fibroblasts (hFs) are reprogrammed by cell-fusion with mouse embryonic stem cells (mESCs). We have previously shown that interspecies cell fusions rapidly induce human pluripotency genes, giving us an opportunity to access the earliest events and the kinetics of the transition from the somatic to the pluripotent state [31, 32]. In this study, we determined the expression of human Xi genes in isogenic female fibroblast clones with different active/inactive X chromosomes ahead of and after cell fusion-mediated reprograming. We showed that a subset of genes (10% of those identified) is rapidly and consistently reactivated upon the induction of the human pluripotency gene network. Importantly, this group of genes were also shown to display variable Xi-allelic expression among individual hFs and between different single-cell derived clones, highlighting an intrinsic susceptibility to be expressed from the Xi.
SNP detection by RNA-seq of female fibroblast clones discriminate the Xi and Xa chromosomes
Genes with significant Xi expression in hF clone 12 or 34
For 76 genes containing more than one SNP, we also calculated the probability of Xi expression for each single SNP position (Additional file 2: Table S1). This analysis showed that more than 98% of SNPs had concordant Xi-allelic expression within the same gene. The few SNPs that showed discordant Xi expression probability with others along the same gene (7 and 5 SNPs out of 328 total in clones 12 and 34, respectively) had low Xi reads and did not change the probability of Xi expression when summed up. Overall, these data confirm that RNA-seq allowed robust identification of heterozygous SNPs along the Xa and Xi and thereby to discriminate Xi genes that expressed or silenced.
Cell-fusion between female hF clones and mESCs induces expression of the human pluripotency network
X chromosome-wide analysis of allelic expression reveals rapid reactivation of a subset of Xi genes during human pluripotent reprograming
Local chromatin context and reprograming-mediated human Xi gene reactivation
An alignment along the linear DNA sequence of the human X chromosome revealed that most of the reprograming-reactivated gene loci clustered on the short Xp arm within topologically associated domains (TADs) that are enriched for genes escaping XCI in somatic cells (Additional file 1: Figure S3A) [45–47]. Genes neighbouring other regions prone to escape, for example in the distal Xq arm, were not particularly sensitive to reactivation upon reprograming. Furthermore, regions adjacent to the XIST locus or close to the centromere were not immune to reactivation, as exemplified by PIN4. Instead, reactivated genes appeared to be preferentially located in evolutionarily recent regions (strata S3–S5 illustrated in Additional file 1: Figure S3A) that were enriched for Alu repeats and depleted of LINE L1 elements [48, 49].
Since reprograming induced upregulation of human pluripotency-associated factors including OCT4 and NANOG, we asked whether the sensitivity of target genes to reactivation might correlate with either the extent of gene induction or the binding of such factors. As anticipated, reactivated genes were upregulated upon reprograming (Additional file 1: Figure S3B, green), while genes that remained active (purple) or inactive (red) throughout reprogramming showed no change in expression. Further analysis of Xa-specific and Xi-specific reads showed that, although induction reflected both re-expression of silent Xi alleles and increased Xa transcription (Additional file 1: Figure S3C), these changes were unlikely to be directly mediated by OCT4 or NANOG as these factors bound similarly to reactivated, active and inactive genes in hESCs (Additional file 1: Figure S3D).
To investigate whether the underlying chromatin state influence locus susceptibility to reactivation, we analysed H3K27me3 and H3K9me3 chromatin immunoprecipitation sequencing (ChIP-seq) from published datasets of immortalised human fibroblasts  (Additional file 1: Figure S3A and E). These repressive histone modifications are preferentially enriched along distinct regions of the Xi (Additional file 1: Figure S3A), which were previously shown to form two spatially distinct compartments [50, 51]. Xi genes that were reactivated upon cell fusion-mediated reprograming were preferentially localised to H3K27me3-enriched macro-domains (Additional file 1: Figure S3A, dashed rectangles). At the level of individual genes, however, the average enrichment of H3K27me3 or H3K9me3 along the gene body did not show significant differences between candidate reactivated and inactive genes (Additional file 1: Figure S3E). As the correlation between histone modifications and gene reactivation might be masked where values for Xa and Xi were combined, rather than separately assayed, we also performed ChIP for H3K27me3 and H3K9me3 in reciprocal hF clones 12 and 34 and analysed the ChIP by allele-specific probes (Additional file 1: Figure S4A and B). This analysis of candidates occupying three distinct TADs showed that the Xi-specific alleles of reactivated genes generally had low H3K9me3 along the gene body, but variable levels of H3K27me3, a profile that resembled genes that were already significantly expressed from the Xi before reprogramming, such as XG in clone12 and 34. Detailed comparisons of reactivation-sensitive (green) and inactive loci (red) indicated that H3K27me3 and H3K9me3 levels did not however determine the sensitivity of loci to reactivation-mediated reprograming (Additional file 1: Figure S4C).
Human Xi genes that are susceptible to reprograming-mediated reactivation show stochastic expression in human fibroblasts
Xi genes that are sensitive to DNA demethylation are revealed through reprograming
Previous studies have shown that DNA demethylation triggers the sporadic reactivation of Xi alleles in somatic cells [52, 53] and is required for reprograming-induced Xi reactivation [29, 54]. We therefore checked whether global DNA demethylation induced by 5-deoxy-azacytidine (5azaC; Additional file 1: Figure S5A) would increase Xi transcription at active, reactivated or stably inactive genes. Although 5azaC treatment was able to significantly increase Xi expression ahead of reprograming (from 18% to 28%) at XG locus in clone 34 (Additional file 1: Figure S5B), we did not observe significant changes in Xi expression at any of the candidate reactivation-sensitive (GYG2, PRKX, CTPS2, FUNDC1, PIN4 and SRPX2) or reactivation-resistant genes (TBL1X, PDHA1, APOO, MAGED2, HDAC8 and IDS) tested. This suggested that DNA methylation was unlikely to be the cause of the stochastic Xi re-expression in hF clones.
Our results show that loci along the human X chromosome have distinct susceptibilities to reprograming-mediated reactivation that reflects both their genomic context and the local chromatin environment that is resident (or induced) within somatic cells. Among the 87 human X genes sampled here, we have identified nine that are reactivated upon reprograming (10%) and 62 that remain inactive (71%). Seven genes within the inactive cohort (11%) were shown to be DNA methylation-sensitive. In addition, genes identified as being particularly prone to reprograming-mediated XCR also showed evidence of variable Xi expression within individual hFs and hF clones, suggesting that these two processes are mechanistically related.
Our study set out to examine human X chromosome reactivation directly during pluripotent reprograming of somatic cells using allele-specific RNA-seq approaches. We have examined 87 X-linked genes and provide the first direct and quantitative evidence of Xi transcription in several hF clones before, and immediately after, the induction of pluripotency. Previous iPS studies have generally used indirect measures of XCR, such as X:autosome expression ratios [28, 56] or loss of DNA methylation [29, 57], to gauge Xi gene expression. Here, we have instead used allele-specific RNA-seq to discriminate transcripts originating from the Xa and Xi and to directly quantify their expression during reprograming. We have applied statistical modelling to directly assess significant Xi expression rather than relying on empirical definitions of XCI escape as being greater than 10% of Xa expression. Recent studies in mouse cells and tissues have shown that allele-specific RNA-seq detects a wider range of Xi expression and allows genes that escape XCI to be estimated with a higher level of sensitivity than previous techniques [14, 36]. Using a similar approach, we show selective reactivation of genes along the human Xi upon induction of the human pluripotency network. Due to the epigenetic instability of the Xi in human pluripotent cells [27, 58], and differences in the reprograming routes of iPSC versus cell–cell fusions, direct comparisons between different approaches is problematic. However, one of the genes described in our analysis, GYG2, has been shown to reactivate during early iPSC reprograming . Recent reports also show that Xi genes that are reactivated at early iPSC stages can undergo subsequent de novo inactivation, probably due to inappropriate culture conditions [59, 60]. Interestingly, some genes that have been shown to be reactivated at late iPSC stages, including PDHA1 and IDS, are reactivated by cell fusion-mediated reprograming after 5azaC pre-treatment of somatic clones. This suggests that fusion with mESCs promotes a different (and potentially faster) reprograming route than iPSCs and supports claims that DNA demethylation is required for the reactivation of a subset of human Xi genes during pluripotent reprograming .
From our RNA-seq analysis, we estimate that Xi reactivation occurs in about 10% of human X-linked genes. This selectivity could reflect the mechanism by which silencing is propagated along the X chromosome and, therefore, underlying genetic and epigenetic differences. It was originally proposed that silencing spreads linearly from the XIST locus into adjacent and more distal DNA, aided by specific repetitive sequences . However, recent studies in mouse showed that Xist RNA first binds to loci that are in close spatial proximity to its locus [8, 9] and then propagates to inactivated genes, suggesting that ‘spatial’ spreading may underpin XCI. Here, we show that reactivation of genes along the human Xi is not limited to telomeric regions distal to the XIST locus, as previously suggested . We observe instead that several reactivated genes localise within TADs enriched of genes constitutively escaping XCI (i.e. GYG2, PRKX, RP11-706O15.1 and CTPS2). Recent studies showed that chromatin domains along the Xa and Xi have a distinct spatial organisation  with TAD-like structures that retain DNA accessibility being present along the Xi only around genes escaping XCI . As loci of escaping Xi genes have been shown to interact within each other  spatially segregating from inactive loci [64, 65], our data support the hypothesis that three-dimensional chromatin organisation might influence gene susceptibility to expression. Interestingly, reprograming-reactivated genes within the same TAD of loci that remained inactive were enriched in SINEs and depleted in L1 LINEs, confirming that repeat sequences also contribute to the spread of silencing along the X chromosome .
Since two spatially distinct heterochromatin compartments can be distinguished along the human Xi based on the preferential enrichment of H3K9me3 or H3K27me3 [50, 51], we also investigated whether these chromatin features influence the predisposition of genes to reactivation. Analysis of published ChIP-seq data showed that reactivated genes preferentially localised within macro-domains enriched in H3K27me3 and depleted in H3K9me3 (dashed squares in Additional file 1: Figure S3A). Allele-specific ChIP analysis in our hF clones also confirmed that Xi loci of reactivated genes have low levels of H3K9me3 ahead of reprograming, consistent with previous reports that H3K9me3 can act as a barrier for reprograming [66, 67]. Interestingly, a recent study has shown that XIST RNA mainly associates with H3K27me3-enriched domains in hFs, whereas the heterochromatin protein HP1α co-localises with H3K9me3 domains . Furthermore, loss of XIST RNA in hESC lines that have already undergone XCI leads to the reactivation of genes within H3K27me3 domains . Collectively, these data suggest that silencing of distinct loci along the human Xi might be maintained by partially overlapping XIST-dependent and XIST-independent mechanisms, and that genes within H3K27me3 domains (and depleted in H3K9me3) might be more susceptible to reactivation in a pluripotent cellular environment. Consistent with this hypothesis, in a recent study we have shown that the delocalisation of XIST RNA and loss of H3K27me3 from the Xi are early events that precede partial Xi reactivation during cell fusion-mediated reprograming . Interestingly, we show here that a second set of genes become susceptible to reprograming-mediated reactivation following DNA demethylation. These genes show high H3K9me3 and low H3K27me3 levels, suggesting that DNA methylation might be differentially regulated across the human Xi through proteins such as SMCHD1 that can bind H3K9me3 domains and mediate DNA methylation [69, 70].
Perhaps the most surprising result from our study is the finding that Xi gene reactivation can be predicted from detailed transcript analysis in single cell-derived clones ahead of reprograming. Evidence that human Xi genes that are sensitive to reprograming-mediated reactivation show stochastic Xi expression in hFs was derived both from RNA-seq analysis of hF clones and from single-cell RNA-FISH studies. It seems unlikely that stochastic expression reflects escape from either DNA methylation or facultative heterochromatin because these genes had similar levels of H3K27me3 and H3K9me3 in expressing and non-expressing clones, and 5azaC treatment was not able to induce Xi expression. Treatment with 5azaC did, however, reveal the reactivation of a different subset of Xi genes upon reprograming.
Overall, these results have two important implications. First, they suggest that multiple different modes of silencing exist along the human Xi. Silencing of reprograming-reactivated genes might depend on the intrinsic expression probability of the gene and rely on stochastic transcriptional events that are stabilised upon pluripotent conversion. Consistent with this hypothesis, we noted that increases in the expression of X-reactivated genes upon cell fusion-mediated reprograming originated from both the Xa and Xi. Interestingly, stochasticity has been proposed to regulate Xi expression across the regions of the human X chromosome that have been most recently added during evolution (i.e. XAR)  and where most of the reprograming-reactivated candidate genes lie. Distinct Xi regions might instead be silenced by multiple epigenetic modifications that need to be erased to permit reactivation. This was highlighted by the discovery of a second cohort of Xi genes that were reactivated only after DNA demethylation. Second, our data suggest that Xi genes that will be successfully reactivated following pluripotent reprograming can be reliably predicted ahead of fusion by comparing allelic expression patterns in somatic cell clones. This result underscores the importance of understanding the underlying causes of intrinsic transcriptional variability and mosaicism that occur in normal female tissues. It also raises the interesting possibility that cell types showing maximal Xi-allelic variation might be the most suitable targets for human iPSC reprograming where reactivation of the X chromosome is desired. Given the high incidence of disease-associated genes residing on the X chromosome [71, 72], investigating XCI in single cells and their clonal derivatives may help to better design rational strategies to achieve locus-specific or domain-specific Xi reactivation.
Cell culture and fusion
Human fibroblasts NHDF17914 (Lonza) were immortalised with pBABE-hTERT-blast and single-cell clones were derived in 3% O2. In order to confirm the expression of a single X chromosome, clones were screened by PDHA1 and ATRX RFLP analysis as previously described . Mouse ESCs E14Tg2a HPRT-/ were transfected with pCAG-puro  to derive a stable clonal line.
For fusion experiments, cells were mixed in 1:1 ratio and treated in suspension with poly-ethylene-glycol (PEG 1500, Roche), as previously described . Briefly, for RNA-seq experiments, 7 × 107 hFs were harvested and fused to an equal number of mESCs by adding 2 mL of 50% PEG and stirring at 37 °C for 90 s. PEG was subsequently diluted with 4 volumes of KO-DMEM, resuspended in mESC medium (KO-DMEM supplemented with 10% FCS, 1% non-essential amino acids, 2 mM L-glutamine, 50 μM β-mercaptoethanol and 10 μg/mL penicillin/streptomycin) and plated onto gelatinised dishes at a density of 106 cells per 140 mm dish. Heterokaryons and hybrids were cultured at 3% O2 in mESC media supplemented with LIF and selected after 12 h by adding puromycin (1.5 μg/mL) and HAT (20 μM hypoxantine, 0.08 mM aminopterin and 3.2 mM thymidine; Sigma). At each timepoint, 105 cells were collected for RNA extraction. All experiments have been performed with hF clones between passages 8 and 14.
Allele-specific Taqman PCR
Taqman PCR was performed with Taqman Universal Master Mix No AmpEraseUNG (Applied Biosystems) on RNAse-H treated single-strand cDNA (First Strand Synthesis kit, Invitrogen) or genomic DNA. Allele-specific Taqman probes (Applied Biosystem) were used for the analysis.
Total RNA was extracted using RNABee (Amsbio) and treated with 0.5 units/μg of DNAseI (Turbo DNAse kit; Ambion) to degrade traces of gDNA. Complete removal of gDNA contamination was tested by PCR amplification of total RNA and only samples with no amplicon upon 40 PCR cycles were processed. Libraries were prepared from 0.6 μg of total RNA (RIN >7.5) by True-seq RNA sample prep v2 kit (Illumina) and were amplified by 13 PCR cycles. Paired-end 100 bp reads were generated using HiSeq2500 sequencer (Illumina). GEO access number: GSE60308.
Heterozygous SNP identification from RNA-seq
To identify heterozygous SNPs along the X chromosomes of NHDF17914 cells, we sequenced RNA libraries obtained from two clones with reciprocal XCI patterns (Strategy 1, Additional file 1: Figure S1). We mapped RNA-seq reads obtained from two biological replicates of each clone to the Human genome (hg19) using Tophat v2.0.8/Bowtie 2.1.0 [74, 75]. Biological replicates were merged following alignment. For each clone, we used SAMtools mpileup and BCFtools to separately identify the observed base (i.e. the most abundant base) and its phred scaled quality score at every genomic position along the X chromosome. We considered only positions with a minimum of eight overlapping reads and phred scaled quality score ≥20 in both clones. The positions at which a different base was identified as the most abundant in reciprocal clones were selected as putative heterozygous SNPs. We annotated the SNPs using Annovar , discarded SNPs overlapping more than one gene and retained only the SNPs within mature transcript regions (i.e. exons and UTR). The final dataset is composed of 379 heterozygous SNPs across 183 genes. Full list of the SNPs is reported in Additional file 2: Table S1.
An alternative/independent approach (Strategy2 in Additional file 1: Figure S1) was used to validate the identified SNPs. This second approach is based on the different assumption that gene expression levels will be almost equivalent in the two clones thus mirroring genomic ratios. Briefly, SAMtools was used for directly identifying heterozygosity (Hardy-Weinberg equilibrium hypothesis) from merged clone 12 and clone 34 alignments. This second strategy found 355 heterozygous SNPs out of which 354 were in common with previous analysis, thus showing the robustness of our approach.
Human-specific alignments and allelic expression analysis of X-linked genes
RNA-seq was performed on hF clone 12 and 34 before and at 0, 4 and 6 days after mESC fusion in two independent fusion experiments per clone. To analyse the human-specific transcripts, RNA-seq reads were aligned independently to human (hg19, Ensembl gene version 72) and mouse reference genome (mm9, Ensembl gene version 67) using Tophat v2.0.8 and the reads aligning to both human and mouse genomes were excluded from further analysis. The number of SNP-overlapping reads for each allele was obtained using SAMtools mpileup.
In order to reconstruct the haplotypes of the two X chromosomes, at each SNP position the base with the highest read depth in hF clone 12 was considered as the Xa allele in this clone and the Xi one in clone 34. The Xa and Xi alleles were considered the same for all the samples in a time series.
To calculate allelic expression, we summed up the reads overlapping the Xa or the Xi alleles at the different SNPs along the same gene and excluded SNP positions overlapping two or more gene transcripts.
We used a beta-binomial model to assess the XCI status quantitatively and estimate the probability that a gene is inactivated. The beta-binomial distribution has been previously used to model differences in allelic expression  and estimate XCI [36, 78] by taking into account read count information (e.g. nucleotide variation and base sequencing quality) and overcoming the over dispersion of RNA-seq data. For each gene, we modelled the probability of inactivation as previously described . Briefly, we estimated the total number of Xi-specific reads, the sum of base quality scores within these reads and the Xi ratio versus total; we fitted these data to a mixture of two beta-binomial distributions, one accounting for genes subject to inactivation and one for genes expressed from the Xi. When observing a probability of inactivation (τi0) <0.05 we considered a gene significantly expressed from the Xi with 95% confidence.
In order to distinguish between genes that are reactivated (green) and the ones that are already active ahead of reprogramming (purple) or remain subject to XCI (red) across reprograming, we separately estimated the probability of inactivation (τi0, see Additional file 4: Table S3) in hF and at 0, 4 and 6 days after mESC-fusion. To increase confidence in annotation, only genes with minimum 20 SNP-overlapping reads (in hF or day0 and in day4 and day6) were considered (87 genes) and classified as follows: ‘reactivated’ genes, which are not expressed from the Xi (τi0 > 0.05) in hF/day0 but acquire significant Xi expression (τi0 ≤ 0.05) at one data-point of the time series in at least one clone after reprograming (green); ‘active’ genes, which have significant Xi expression (τi0 ≤ 0.05) in hF/day0 before reprograming and at one data-point of the timeseries after reprograming in at least one clone (purple); ‘inactive’ genes, which are not significantly expressed from the Xi (τi0 > 0.05) in hF and all data-points of the time series in both clones (red). Genes with less than 20 SNP-overlapping reads across the time series were not evaluated.
Data obtained from two independent biological replicates for each clone were either merged by summing up Xi and Xa SNP-specific reads and their base quality scores (Additional file 4: Table S3) or analysed separately. Single replicates showed a good concordance both for Xi expression ratios (mean ± standard error is shown Fig. 3) and for beta-binomial statistics (concordance between replicates for SNPs with more than 10 overlapping reads was of 98%, 99%, 95% and 97% in clone 12 for hF, days 0, 4 and 6 respectively; and 94%, 97% and 89% in clone 34 for hF, days 0 and 4, respectively) with variations mainly due to different sequencing depth among replicates and not affecting the final classification compared to merged data.
Gene expression analysis
We counted the reads overlapping with genes using HTseq and identified the differentially expressed genes using edgeR . FPKM values were estimated using R script.
Reproducibility between independent biological replicates (i.e. separate fusion experiments), or between distinct clones, was assessed by Spearman correlation analysis using FPKMs and R script. Correlation coefficients (r) comparing gene expression between independent replicates were higher than 0.90 and 0.87 in clones 12 and 34, respectively. The same comparison between clones 12 and 34 gave r values ≥ 0.88 for all timepoints.
RNA-FISH analysis and probes
Probes used for RNA-FISH analysis were obtained by nick translation of appropriate phosmids/BACs containing HDAC8, TBLX1, CTPS2 and RP11-706O15.1 (i.e. RP11-1021B19, RP11-451G24, CTD-2277I2 and WI2-1543 K8, respectively), in the presence of fluorophore-coupled dUTPs as previously described . Locus-specific transcriptional signals were detected in hF nuclei and scored according to the detection of a single focus, of two separate foci (>1 microns apart) or >2 signals per nucleus . More than 200 cells were scored in each sample.
This work was supported by ERC-2011-ADG grant (REPLENICHE project) to AGF, HFSP and EMBO long-term fellowships to IC.
Availability of data and materials
The RNA-seq datasets supporting the conclusions of this article are available from the gene expression omnibus (GEO) with accession number: GSE60308. http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE60308. Taqman data and RNA-FISH images are available from Figshare https://doi.org/10.6084/m9.figshare.4275617  and https://doi.org/10.6084/m9.figshare.4276373 .
IC conceived the study, designed and performed all the experiments and analysed data. IC and GD established the workflow for identification of heterozygous SNPs and allelic expression analysis. GD and YWC performed bioinformatics analysis and BL provided input for graphical representation of data. ACK helped perform qPCR. MM and AGF provided guidance and helped in interpretation of the results. IC and AGF drafted the manuscript and all authors approved it.
The authors declare that they have no competing interests.
Ethics approval and consent to participate
No ethics approval was needed for this study.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Lyon MF. Gene action in the X-chromosome of the mouse (Mus musculus L.). Nature. 1961;190:372–3.View ArticlePubMedGoogle Scholar
- Migeon B. Females are mosaics: X inactivation and sex differences in disease. Oxford: Oxford University Press; 2014.Google Scholar
- Nora EP, Heard E. Chromatin structure and nuclear organization dynamics during X-chromosome inactivation. Cold Spring Harb Symp Quant Biol. 2010;75:333–44.View ArticlePubMedGoogle Scholar
- Clemson CM, McNeil JA, Willard HF, Lawrence JB. XIST RNA paints the inactive X chromosome at interphase: evidence for a novel RNA involved in nuclear/chromosome structure. J Cell Biol. 1996;132:259–75.View ArticlePubMedGoogle Scholar
- Kohlmaier A, Savarese F, Lachner M, Martens J, Jenuwein T, Wutz A. A chromosomal memory triggered by Xist regulates histone methylation in X inactivation. PLoS Biol. 2004;2:E171.View ArticlePubMedPubMed CentralGoogle Scholar
- Wutz A, Jaenisch R. A shift from reversible to irreversible X inactivation is triggered during ES cell differentiation. Mol Cell. 2000;5:695–705.View ArticlePubMedGoogle Scholar
- Chow JC, Ciaudo C, Fazzari MJ, Mise N, Servant N, Glass JL, et al. LINE-1 activity in facultative heterochromatin formation during X chromosome inactivation. Cell. 2010;141:956–69.View ArticlePubMedGoogle Scholar
- Engreitz JM, Pandya-Jones A, McDonel P, Shishkin A, Sirokman K, Surka C, et al. The Xist lncRNA exploits three-dimensional genome architecture to spread across the X chromosome. Science. 2013;341:1237973.View ArticlePubMedPubMed CentralGoogle Scholar
- Simon MD, Pinter SF, Fang R, Sarma K, Rutenberg-Schoenberg M, Bowman SK, et al. High-resolution Xist binding maps reveal two-step spreading during X-chromosome inactivation. Nature. 2013;504:465–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Carrel L, Willard HF. X-inactivation profile reveals extensive variability in X-linked gene expression in females. Nature. 2005;434:400–4.View ArticlePubMedGoogle Scholar
- Yang F, Babak T, Shendure J, Disteche CM. Global survey of escape from X inactivation by RNA-sequencing in mouse. Genome Res. 2010;20:614–22.View ArticlePubMedPubMed CentralGoogle Scholar
- Kelsey AD, Yang C, Leung D, Minks J, Dixon-McDougall T, Baldry SE, et al. Impact of flanking chromosomal sequences on localization and silencing by the human non-coding RNA XIST. Genome Biol. 2015;16:208.View ArticlePubMedPubMed CentralGoogle Scholar
- Cotton AM, Chen CY, Lam LL, Wasserman WW, Kobor MS, Brown CJ. Spread of X-chromosome inactivation into autosomal sequences: role for DNA elements, chromatin features and chromosomal domains. Hum Mol Genet. 2014;23:1211–23.View ArticlePubMedGoogle Scholar
- Berletch JB, Ma W, Yang F, Shendure J, Noble WS, Disteche CM, et al. Escape from X inactivation varies in mouse tissues. PLoS Genet. 2015;11:e1005079.View ArticlePubMedPubMed CentralGoogle Scholar
- Patrat C, Okamoto I, Diabangouaya P, Vialon V, Le Baccon P, Chow J, et al. Dynamic changes in paternal X-chromosome activity during imprinted X-chromosome inactivation in mice. Proc Natl Acad Sci U S A. 2009;106:5198–203.View ArticlePubMedPubMed CentralGoogle Scholar
- Lingenfelter PA, Adler DA, Poslinski D, Thomas S, Elliott RW, Chapman VM, et al. Escape from X inactivation of Smcx is preceded by silencing during mouse development. Nat Genet. 1998;18:212–3.View ArticlePubMedGoogle Scholar
- Brinkman AB, Roelofsen T, Pennings SW, Martens JH, Jenuwein T, Stunnenberg HG. Histone modification patterns associated with the human X chromosome. EMBO Rep. 2006;7:628–34.PubMedPubMed CentralGoogle Scholar
- Valley CM, Pertz LM, Balakumaran BS, Willard HF. Chromosome-wide, allele-specific analysis of the histone code on the human X chromosome. Hum Mol Genet. 2006;15:2335–47.View ArticlePubMedGoogle Scholar
- Kucera KS, Reddy TE, Pauli F, Gertz J, Logan JE, Myers RM, et al. Allele-specific distribution of RNA polymerase II on female X chromosomes. Hum Mol Genet. 2011;20:3964–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Peeters SB, Cotton AM, Brown CJ. Variable escape from X-chromosome inactivation: Identifying factors that tip the scales towards expression. Bioessays. 2014;36:746–56.View ArticlePubMedPubMed CentralGoogle Scholar
- Lahn BT, Page DC. Four evolutionary strata on the human X chromosome. Science. 1999;286:964–7.View ArticlePubMedGoogle Scholar
- Al Nadaf S, Deakin JE, Gilbert C, Robinson TJ, Graves JA, Waters PD. A cross-species comparison of escape from X inactivation in Eutheria: implications for evolution of X chromosome inactivation. Chromosoma. 2012;121:71–8.View ArticlePubMedGoogle Scholar
- Bailey JA, Carrel L, Chakravarti A, Eichler EE. Molecular evidence for a relationship between LINE-1 elements and X chromosome inactivation: the Lyon repeat hypothesis. Proc Natl Acad Sci U S A. 2000;97:6634–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang Z, Willard HF, Mukherjee S, Furey TS. Evidence of influence of genomic DNA sequence on human X chromosome inactivation. PLoS Comput Biol. 2006;2:e113.View ArticlePubMedPubMed CentralGoogle Scholar
- Cotton AM, Ge B, Light N, Adoue V, Pastinen T, Brown CJ. Analysis of expressed SNPs identifies variable extents of expression from the human inactive X chromosome. Genome Biol. 2013;14:R122.View ArticlePubMedPubMed CentralGoogle Scholar
- Ohhata T, Wutz A. Reactivation of the inactive X chromosome in development and reprogramming. Cell Mol Life Sci. 2013;70:2443–61.View ArticlePubMedGoogle Scholar
- Anguera MC, Sadreyev R, Zhang Z, Szanto A, Payer B, Sheridan SD, et al. Molecular signatures of human induced pluripotent stem cells highlight sex differences and cancer genes. Cell Stem Cell. 2012;11:75–90.View ArticlePubMedPubMed CentralGoogle Scholar
- Bruck T, Benvenisty N. Meta-analysis of the heterogeneity of X chromosome inactivation in human pluripotent stem cells. Stem Cell Res. 2011;6:187–93.View ArticlePubMedGoogle Scholar
- Nazor KL, Altun G, Lynch C, Tran H, Harness JV, Slavin I, et al. Recurrent variations in DNA methylation in human pluripotent stem cells and their differentiated derivatives. Cell Stem Cell. 2012;10:620–34.View ArticlePubMedPubMed CentralGoogle Scholar
- Wutz A. Epigenetic alterations in human pluripotent stem cells: a tale of two cultures. Cell Stem Cell. 2012;11:9–15.View ArticlePubMedGoogle Scholar
- Pereira CF, Piccolo FM, Tsubouchi T, Sauer S, Ryan NK, Bruno L, et al. ESCs require PRC2 to direct the successful reprogramming of differentiated cells toward pluripotency. Cell Stem Cell. 2010;6:547–56.View ArticlePubMedGoogle Scholar
- Pereira CF, Terranova R, Ryan NK, Santos J, Morris KJ, Cui W, et al. Heterokaryon-based reprogramming of human B lymphocytes for pluripotency requires Oct4 but not Sox2. PLoS Genet. 2008;4:e1000170.View ArticlePubMedPubMed CentralGoogle Scholar
- Tchieu J, Kuoy E, Chin MH, Trinh H, Patterson M, Sherman SP, et al. Female human iPSCs retain an inactive X chromosome. Cell Stem Cell. 2010;7:329–42.View ArticlePubMedPubMed CentralGoogle Scholar
- Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, et al. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 2007;35:D5–D12.View ArticlePubMedGoogle Scholar
- Harvey CT, Moyerbrailean GA, Davis GO, Wen X, Luca F, Pique-Regi R. QuASAR: quantitative allele-specific analysis of reads. Bioinformatics. 2015;31:1235–42.View ArticlePubMedGoogle Scholar
- Calabrese JM, Sun W, Song L, Mugford JW, Williams L, Yee D, et al. Site-specific silencing of regulatory elements as a mechanism of X inactivation. Cell. 2012;151:951–63.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhang Y, Castillo-Morales A, Jiang M, Zhu Y, Hu L, Urrutia AO, et al. Genes that escape X-inactivation in humans have high intraspecific variability in expression, are associated with mental impairment but are not slow evolving. Mol Biol Evol. 2013;30:2588–601.View ArticlePubMedPubMed CentralGoogle Scholar
- James D, Levine AJ, Besser D, Hemmati-Brivanlou A. TGFbeta/activin/nodal signaling is necessary for the maintenance of pluripotency in human embryonic stem cells. Development. 2005;132:1273–82.View ArticlePubMedGoogle Scholar
- Wang L, Schulz TC, Sherrer ES, Dauphin DS, Shin S, Nelson AM, et al. Self-renewal of human embryonic stem cells requires insulin-like growth factor-1 receptor and ERBB2 receptor signaling. Blood. 2007;110:4111–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Pauklin S, Vallier L. The cell-cycle state of stem cells determines cell fate propensity. Cell. 2013;155:135–47.View ArticlePubMedPubMed CentralGoogle Scholar
- Gerstein MB, Kundaje A, Hariharan M, Landt SG, Yan KK, Cheng C, et al. Architecture of the human regulatory network derived from ENCODE data. Nature. 2012;489:91–100.View ArticlePubMedPubMed CentralGoogle Scholar
- Tsubouchi T, Soza-Ried J, Brown K, Piccolo FM, Cantone I, Landeira D, et al. DNA synthesis is required for reprogramming mediated by stem cell fusion. Cell. 2013;152:873–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Cahan P, Li H, Morris SA, Lummertz da Rocha E, Daley GQ, Collins JJ. Cell Net: network biology applied to stem cell engineering. Cell. 2014;158:903–15.View ArticlePubMedPubMed CentralGoogle Scholar
- Takashima Y, Guo G, Loos R, Nichols J, Ficz G, Krueger F, et al. Resetting transcription factor control circuitry toward ground-state pluripotency in human. Cell. 2014;158:1254–69.View ArticlePubMedPubMed CentralGoogle Scholar
- Dixon JR, Selvaraj S, Yue F, Kim A, Li Y, Shen Y, et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature. 2012;485:376–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Marks H, Kerstens HH, Barakat TS, Splinter E, Dirks RA, van Mierlo G, et al. Dynamics of gene silencing during X inactivation using allele-specific RNA-seq. Genome Biol. 2015;16:149.View ArticlePubMedPubMed CentralGoogle Scholar
- Giorgetti L, Lajoie BR, Carter AC, Attia M, Zhan Y, Xu J, et al. Structural organization of the inactive X chromosome in the mouse. Nature. 2016;535:575–9.View ArticlePubMedGoogle Scholar
- Ross MT, Grafham DV, Coffey AJ, Scherer S, McLay K, Muzny D, et al. The DNA sequence of the human X chromosome. Nature. 2005;434:325–37.View ArticlePubMedPubMed CentralGoogle Scholar
- Kelkar A, Thakur V, Ramaswamy R, Deobagkar D. Characterisation of inactivation domains and evolutionary strata in human X chromosome through Markov segmentation. PLoS One. 2009;4:e7885.View ArticlePubMedPubMed CentralGoogle Scholar
- Nozawa RS, Nagao K, Igami KT, Shibata S, Shirai N, Nozaki N, et al. Human inactive X chromosome is compacted through a PRC2-independent SMCHD1-HBiX1 pathway. Nat Struct Mol Biol. 2013;20:566–73.View ArticlePubMedGoogle Scholar
- Chadwick BP, Willard HF. Multiple spatially distinct types of facultative heterochromatin on the human inactive X chromosome. Proc Natl Acad Sci U S A. 2004;101:17450–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Csankovszki G, Nagy A, Jaenisch R. Synergism of Xist RNA, DNA methylation, and histone hypoacetylation in maintaining X chromosome inactivation. J Cell Biol. 2001;153:773–84.View ArticlePubMedPubMed CentralGoogle Scholar
- Mohandas T, Sparkes RS, Shapiro LJ. Reactivation of an inactive human X chromosome: evidence for X inactivation by DNA methylation. Science. 1981;211:393–6.View ArticlePubMedGoogle Scholar
- Pasque V, Tchieu J, Karnik R, Uyeda M, Sadhu Dimashkie A, Case D, et al. X chromosome reactivation dynamics reveal stages of reprogramming to pluripotency. Cell. 2014;159:1681–97.View ArticlePubMedPubMed CentralGoogle Scholar
- Csankovszki G, Panning B, Bates B, Pehrson JR, Jaenisch R. Conditional deletion of Xist disrupts histone macroH2A localization but not maintenance of X inactivation. Nat Genet. 1999;22:323–4.View ArticlePubMedGoogle Scholar
- Tomoda K, Takahashi K, Leung K, Okada A, Narita M, Yamada NA, et al. Derivation conditions impact X-inactivation status in female human induced pluripotent stem cells. Cell Stem Cell. 2012;11:91–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Mekhoubad S, Bock C, de Boer AS, Kiskinis E, Meissner A, Eggan K. Erosion of dosage compensation impacts human iPSC disease modeling. Cell Stem Cell. 2012;10:595–609.View ArticlePubMedPubMed CentralGoogle Scholar
- Vallot C, Ouimette JF, Makhlouf M, Feraud O, Pontis J, Come J, et al. Erosion of X chromosome inactivation in human pluripotent cells initiates with XACT coating and depends on a specific heterochromatin landscape. Cell Stem Cell. 2015;16:533–46.View ArticlePubMedGoogle Scholar
- Kim KY, Hysolli E, Tanaka Y, Wang B, Jung YW, Pan X, et al. X chromosome of female cells shows dynamic changes in status during human somatic cell reprogramming. Stem Cell Rep. 2014;2:896–909.View ArticleGoogle Scholar
- Barakat TS, Ghazvini M, de Hoon B, Li T, Eussen B, Douben H, et al. Stable X chromosome reactivation in female human induced pluripotent stem cells. Stem Cell Rep. 2015;4:199–208.View ArticleGoogle Scholar
- Gartler SM, Riggs AD. Mammalian X-chromosome inactivation. Annu Rev Genet. 1983;17:155–90.View ArticlePubMedGoogle Scholar
- Wang S, Su JH, Beliveau BJ, Bintu B, Moffitt JR, Wu CT, Zhuang X. Spatial organization of chromatin domains and compartments in single chromosomes. Science. 2016;353:598–602.View ArticlePubMedGoogle Scholar
- Splinter E, de Wit E, Nora EP, Klous P, van de Werken HJ, Zhu Y, et al. The inactive X chromosome adopts a unique three-dimensional conformation that is dependent on Xist RNA. Genes Dev. 2011;25:1371–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Clemson CM, Hall LL, Byron M, McNeil J, Lawrence JB. The X chromosome is organized into a gene-rich outer rim and an internal core containing silenced nongenic sequences. Proc Natl Acad Sci U S A. 2006;103:7688–93.View ArticlePubMedPubMed CentralGoogle Scholar
- Deng X, Ma W, Ramani V, Hill A, Yang F, Ay F, et al. Bipartite structure of the inactive mouse X chromosome. Genome Biol. 2015;16:152.View ArticlePubMedPubMed CentralGoogle Scholar
- Chen J, Liu H, Liu J, Qi J, Wei B, Yang J, et al. H3K9 methylation is a barrier during somatic cell reprogramming into iPSCs. Nat Genet. 2013;45:34–42.View ArticlePubMedGoogle Scholar
- Soufi A, Donahue G, Zaret KS. Facilitators and impediments of the pluripotency reprogramming factors’ initial engagement with the genome. Cell. 2012;151:994–1004.View ArticlePubMedPubMed CentralGoogle Scholar
- Cantone I, Bagci H, Dormann D, Dharmalingam G, Nesterova T, Brockdorff N, et al. Ordered chromatin changes and human X chromosome reactivation by cell fusion-mediated pluripotent reprogramming. Nat Commun. 2016;7:12354.View ArticlePubMedPubMed CentralGoogle Scholar
- Brideau NJ, Coker H, Gendrel AV, Siebert CA, Bezstarosti K, Demmers J, et al. Independent mechanisms target SMCHD1 to H3K9me3-modified chromatin and the inactive X chromosome. Mol Cell Biol. 2015;35:4053–68.View ArticlePubMedPubMed CentralGoogle Scholar
- Gendrel AV, Apedaile A, Coker H, Termanis A, Zvetkova I, Godwin J, et al. Smchd1-dependent and -independent pathways determine developmental dynamics of CpG island methylation on the inactive X chromosome. Dev Cell. 2012;23:265–79.View ArticlePubMedPubMed CentralGoogle Scholar
- Ballabio A, Nelson D, Rozen S. Genetics of disease The sex chromosomes and human disease. Curr Opin Genet Dev. 2006;16:209–12.View ArticlePubMedGoogle Scholar
- Gecz J, Shoubridge C, Corbett M. The genetic landscape of intellectual disability arising from chromosome X. Trends Genet. 2009;25:308–16.View ArticlePubMedGoogle Scholar
- Niwa H, Masui S, Chambers I, Smith AG, Miyazaki J. Phenotypic complementation establishes requirements for specific POU domain and generic transactivation function of Oct-3/4 in embryonic stem cells. Mol Cell Biol. 2002;22:1526–36.View ArticlePubMedPubMed CentralGoogle Scholar
- Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. 2013;14:R36.View ArticlePubMedPubMed CentralGoogle Scholar
- Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods. 2012;9:357–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38:e164.View ArticlePubMedPubMed CentralGoogle Scholar
- Pickrell JK, Marioni JC, Pai AA, Degner JF, Engelhardt BE, Nkadori E, et al. Understanding mechanisms underlying human gene expression variation with RNA sequencing. Nature. 2010;464:768–72.View ArticlePubMedPubMed CentralGoogle Scholar
- Zou F, Sun W, Crowley JJ, Zhabotynsky V, Sullivan PF, Pardo-Manuel de Villena F. A novel statistical approach for jointly analyzing RNA-Seq data from F1 reciprocal crosses and inbred lines. Genetics. 2014;197:389–99.View ArticlePubMedPubMed CentralGoogle Scholar
- Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40.View ArticlePubMedGoogle Scholar
- Chaumeil J, Augui S, Chow JC, Heard E. Combined immunofluorescence, RNA fluorescent in situ hybridization, and DNA fluorescent in situ hybridization to study chromatin changes, transcriptional activity, nuclear organization, and X-chromosome inactivation. Methods Mol Biol. 2008;463:297–308.View ArticlePubMedGoogle Scholar
- Cantone I. Allele-specific Taqman analysis of human X linked genes during cell fusion-mediated pluripotent reprogramming.xlsx. 2016. figshare.https://dx.doi.org/10.6084/m9.figshare.4275617.v1.
- Cantone I. Nascent RNA-FISH of X-linked genes in human fibroblast clones. 2016. figshare.https://dx.doi.org/10.6084/m9.figshare.4276373.v1.