Xist-mediated chromatin changes that establish silencing of an entire X chromosome in mammals

X chromosome inactivation (XCI) ensures an equal gene dosage between the sexes in placental mammals. Xist, a modular multi-domain X-encoded long non-coding RNA coats the X chromosome in cis during XCI. Xist recruits chromatin remodelers and repressor complexes ensuring silencing of the inactive X (Xi). Here, we review the recent work focused on the role of Xist functional repeats and interacting RNA-binding factors in the establishment of the silent state. Xist orches-trates recruitment of remodelers and repressors that first facilitate removal of the active chromatin landscape and sub-sequently direct the transition into a repressive heterochro-matic environment. Some of these factors affect silencing on a chromosome-wide scale, while others display gene-specific silencing defects. The temporal order of recruitment shows each silencing step is party dependent on one another. After the Xi is established, many of the factors are dispensable, and a different repertoire of proteins ensure the silenced Xi is maintained and propagated. deposition at Xist Entry Sites (1A). PcG-mark accumulation largely follows Xist-spreading, showing enrichment at moderate sites first, spreading mostly to intergenic regions (2). PcG-marks spread into promoters and gene bodies in response to silencing (3). forward recruitment of PRC2 and PRC1 ensures PcG-marks accumulate and facilitate Polycomb-domain formation of the Xi. Alternatively, pre-marked CpG islands act as nucleation sites independent of Xist (1B), leading to accumulation (2). establishment requires Repeat B and Xist-mediated PcG recruitment, as silencing is relaxed in prolonged differentiation models if Repeat B is lost. Systematic allelic analysis defines the interplay of key path- ways in X chromosome inactivation 15. This study systematically investigated the importance of Xist interactors in XCI-inducible cell models and shows that SPEN and Repeat A are crucial in silencing of the Xi. Loss of other silencing pathways does not lead to global defects in silencing of the X. PcG-proteins are essential for silencing of lowly expressed genes pre-marked with PcG-modifications. assesses the temporal order of chromatin changes that occur in the early establishment of XCI. Deacetylation and demethylation of chromatin leads to transcriptional silencing and decom- missioning of enhancers and promoters. PcG-mark accumulation requires loss of active chromatin landscape for efficient spreading. H2AK119ub1 accumulation comes first and directs depositon of H3K27me3, further substantiating the order of Polycomb recruitment. Analysis of the role of SPEN in initiation of XCI shows that SPEN is required for the early phase of establishing the silent state. CUT&RUN analysis shows that SPEN binding overlaps at enhancers and pro- moters with HDAC3 and NuRD-complex members, suggesting a po-tential role for SPEN in distributing and/or activating repressor complexes. This study also shows SPEN is not required for maintenance of the Xi in NPC models. hnRNPK a Xist-mediated vPRC1 recruitment, providing insight into the order of of PcG Xi coupled to Xist This study shows PcG proteins can be recruited in multiple ways to the Xi. Disrupting Xist Repeat B or hnRNPK shows recruitment of PcG-proteins persists, pointing toward a secondary pathway that occurs independent of Xist in the maintenance phase. The data presented also suggest that Xist relies on PcG-marks for efficient spreading on the Xi.


Introduction
Sex chromosomes in Therian mammals evolved from a pair of autosomes (Reviewed in Ref. [1]). During evolution, the Y chromosome lost most of its genes resulting in a gene imbalance between male and female placental mammals. X chromosome inactivation (XCI) is the process via which one of the two female X chromosomes is epigenetically silenced to ensure equal dosage of Xlinked genes between sexes [2]. The X inactive-specific transcript (Xist) is a long non-coding RNA encoded on the X-chromosome with a pivotal role in orchestrating this process [3,4]. Xist has a modular structure harboring six conserved repetitive domains with distinct characteristics that expanded its function in evolution (Repeat A till Repeat F) [5e7]. The sequence and conformation of the domains allow for recruitment of a large repertoire of proteins that aid Xist in separate phases of chromosome silencing which have been extensively researched.
Xist-mediated silencing can be divided into three functionally distinct phases: initiation, establishment, and maintenance. The initiation phase starts with determining the X to autosome ratio, leading to female exclusive mono-allelic upregulation of Xist and spreading in cis [8,9]. In the establishment phase, active chromatin marks are removed and replaced by repressive marks in a stepwise manner leading to heterochromatinization and CpG island methylation of the Xi [10,11]. In the maintenance phase, Xist remains associated with the inactive X while the silent state is propagated through a near-infinite number of cell divisions. Here, we review new insights in the temporal order of Xist-mediated chromatin changes and factors involved in establishment of the silent state. between repressor complexes and Xist ( Figure 1)  Repressor complexes mediate deacetylation as a crucial early step in establishing silencing [26]. Enhancer deacetylation in XCI mostly requires HDAC3, as part of the NCoR/SMRT complex. Multiple HDACs are present in mammals, but exclusively loss of HDAC3 affects Xi silencing [17,22]. HDAC3 was suggested to be prebound at enhancers in a catalytically inactive state [17]. SPEN potentially activates part of the pre-bound HDAC3 population by delivering NCoR/SMRT complex members, leading to histone deacetylation and facilitating H3K4me1 demethylation [17,21,22,27]. SPEN localization shows overlap with NuRD complex members predominantly at promoters, which are deacetylated and later demethylated in a stepwise manner [17,21]. Ultimately, removal of active marks leads to loss of transcription and decommissioning of enhancers and promoters. Even though these results hint at an exclusive role for SPEN in enhancer and promoter deacetylation through NCoR/SMRT and NuRD, both complexes likely function partially redundantly, which is evident from the HDAC3 knock-out phenotype that to a lesser extent also delays promoter deacetylation [17] and localization of NuRD complex members to enhancer regions [21,28].
SPEN pioneers Xist-dependent recruitment and/or activation of pre-bound repressor complexes required for removing the active chromatin landscape as the initial first step in XCI [17,21]. These changes are essential for recruitment of downstream chromatin writers and erasers that further lock in the silencing process. Interestingly, loss of SPEN after the Xi is established, does not result in Xi reactivation [17]. It is tempting to speculate that Repeat A and SPEN are

Current Opinion in Cell Biology
Xist Repeat A facilitates the removal of the active chromatin landscape through the recruitment of SPEN and repressor complexes. SPEN interacts with Xist Repeat A leading to activation and/or distribution of repressor complexes NCoR/SMRT-HDAC3 to enhancers as well as NuRD to promoters. Histone deacetylation leads to demethylation and facilitates enhancer-promoter decommissioning, allowing for deposition of additional repressive marks. RBM15/RBM15b recruit m6A RNA methylation complex members, leading to m6A methylation of Xist and recruitment of YTHDC1. exclusively required in early establishment of silencing, after which downstream chromatin remodelers and writers stabilize and propagate the silenced state.

Specialized role in silencing for other repressing complexes
The role of additional Repeat A interacting proteins in establishment of the Xi have been up for debate. RBM15 and RBM15b, two SPEN homologues, direct RNA m6A methylation complexes toward Xist [22,29,30]. Many eukaryotic RNAs are regulated by this RNA modification, which is involved in folding, stability, splicing, and recruitment of binding partners (Reviewed in Ref. [31]). Knock-down of RBM15 and m6A methylation complex members WTAP and METTL3 was found to reduce efficiency of Xist-mediated silencing, suggesting that m6A methylation is required for recruitment of proteins involved in silencing [20,30]. Indeed, binding of the m6A-reader YTHDC1 to Xist was lost in the absence of m6A and tethering YTHDC1 to Xist partially rescued the silencing defects observed in Repeat-A mutants as well as RBM15/RBM15b and METTL3 knock-down conditions [26,30]. Other studies report m6A methylation of Xist only plays a minor specialized role, as limited effects on silencing were observed in RBM15-deficient or m6A methylation complex members WTAP-and METTL3-deficient cells [15,22]. The exact role of m6A methylation, YTHDC1 and how they affect chromatin changes in XCI, therefore, needs further investigation.
During Xist accumulation histone modifications associated with gene activity are removed (H3K9ac, H3K27ac, H3K36me3, H4ac) and repressive histone modifications are deposited (H2A119ub1, H3K9me2/3, H3K27me3, H4K20me1) ( Figure 2) [17,32e36]. SETDB1, important in catalyzing H3K9me3, is required for silencing a subset of genes exclusively in the establishment phase, while cross-talk with the DNA methylation machinery stabilizes repression in the maintenance phase [37,38]. SET8 (also known as PR-Set7) catalyzes H4K20me1, regulating chromosome compaction, but the role of SET8 in XCI is unclear [34,39,40]. In Caenorhabditis elegans SET8 orthologues catalyze H4K20me1 to affect the chromosome architecture, contributing to dosage compensation where gene expression is reduced by half [41,42]. Also, the nucleosome composition changes to stabilize the Xi as H2A is exchanged with MacroH2A in a cell cyclee dependent manner [43e45]. MacroH2A enrichment is strongest at stably silenced regions mostly in the maintenance phase of XCI. It has been suggested that the structural differences of the MacroH2A linkers inhibits chromatin remodelers, preventing transcription factor binding and promoting heterochromatin formation [46]. Many of the other changes in histone modifications on the Xi are less well understood, and factors involved are not yet identified.

Establishment and maintenance of gene silencing by Polycomb complexes
Polycomb Group (PcG) protein complexes have been implicated to play a crucial role in maintaining silencing of the Xi and altering the organization of the chromosome [35,36]. PcG proteins are conserved in evolution and their expansion during the emergence of vertebrates is thought to accompany the increased complexity of regulating gene expression in a broad range of cellular processes [47,48]. PcG proteins can be divided into two main complexes with several variants. Polycomb Repressive Complex (PRC) 1 catalyzes monoubiquitination of H2AK119 while PRC2 catalyzes mono-, di-, and trimethylation of H3K27.
Heterogeneous nuclear ribonucleoprotein K (hnRNPK) acts as a molecular bridge between variant noncanonical PRC1 (vPRC1) and Xist Repeat B and to a lesser extend Repeat C through poly-cytosine-rich repeats [49e51]. The vPRC1 complex containing either PCGF3 or PCGF5 deposits H2AK119ub1, the accumulation of which is partly dictated by Xist spreading [17, 49,51]. Xist entry sites, which are gene-dense regions in topological close proximity with the Xist locus, show early accumulation of Xist RNA and overlap with   . In a feedforward loop, deposition and spreading in cis of both PcG-marks is stimulated and reinforced, facilitating the transition to a repressive chromatin environment of the Xi (Figure 3). In line with this, recent reports show that vPRC1-mediated deposition of H2AK119ub1 is poorly propagated through multiple cell divisions, which is an absolute requirement for XCI [62]. Therefore, it is likely that silencing in XCI is reinforced by a mutual requirement of cPRC1 as well as PRC2 in facilitating maintenance of Xi silencing.

The interplay of SPEN and PcGs in establishing silencing
Efficient spreading of PcG-marks requires removal of the active chromatin landscape [17]. In the absence of transcriptional silencing in Repeat A mutants, PcG proteins can be recruited, but PcG-associated histone modifications remain mostly restricted to intergenic regions, while spreading into active gene bodies is limited. This observation highlights the notion that PcGs largely act in a secondary step, downstream of Repeat A and SPEN-mediated silencing and PcGmodifications mostly follow rather than direct silencing of target genes as postulated previously [63,64]. This is in line with findings indicating that deacetylation by NuRD and HDAC3 are required for H3K27me3 deposition by PRC2 in gene silencing [17,22,65]. PcG proteins can also be recruited independent of Xist to specific CpG islands of genes containing PcG-associated marks [14,53,66,67]. These CpG islands act as nucleation sites and could allow for limited local accumulation of PcG independent of Xist Repeat B/C but likely require neighbouring genes to be silenced for spreading. Therefore, predominantly Xistdependent and to a lesser extent independent recruitment of PcGs is instrumental in the polycomb domain formation and stabilizing the Xi.
The importance of Repeat B/C and PcGs in establishing early gene silencing in XCI is still under debate. Loss of

Current Opinion in Cell Biology
Xist-mediated recruitment of PcG proteins and PcG-mark accumulation. Repeat B/C of Xist recruit vPRC1 through interacting with hnRNPK, resulting in H2AK119ub1 deposition at Xist Entry Sites (1A). PcG-mark accumulation largely follows Xist-spreading, showing enrichment at moderate sites first, spreading mostly to intergenic regions (2). PcG-marks spread into promoters and gene bodies in response to silencing (3). Feed forward recruitment of PRC2 and PRC1 ensures PcG-marks accumulate and facilitate Polycomb-domain formation of the Xi. Alternatively, pre-marked CpG islands act as nucleation sites independent of Xist (1B), leading to accumulation (2).
Xist Repeat B/C and PcG-recruitment has been implicated in loss of silencing efficiency [15, 49,66] and reduced Xist accumulation [68]. Other studies indicate that loss of Repeat B is dispensable for the initiation of silencing, exemplified by initial silencing of most Xlinked genes, but silencing erodes later in differentiation [14]. A subset of mostly lowly expressed genes, however, are unable to be silenced in absence of Repeat B/C and rely on PcG activity [15]. Moreover, PRC2 knock-out embryos do initiate XCI in the developing epiblast, although loss of PRC2 is lethal at a later stage [69]. This suggests that initiation of silencing of most genes can occur independent of PcG proteins; however, this inactive state is likely transient, poorly maintained, and inefficiently propagated. PcG proteins, therefore, reinforce the silenced X and in turn direct recruitment of downstream proteins that further maintain silencing. This affects the nuclear organization and facilitates DNA methylation (Reviewed in Ref. [70]). PcG proteins, PcG-marks, and Polycomb domain formation, therefore, appear to form the bridge between the transient state of the silenced X and the propagated stably silenced Xi with methylated CpG islands present in somatic cells.

Concluding remarks
XCI serves as an excellent model to unravel gene silencing on a chromosome-wide scale. The different silencing defects observed in Xist Repeat and Xistinteracting protein knock-out studies at various stages of XCI highlight an Xist-directed multi-phasic process requiring a separate orchestra of chromatin-modifying complexes, each dependent on the activity of one another. The temporal order of events show Xistmediated deacetylation and gene silencing as the crucial first steps. This ultimately allows for Polycomb domain formation and the transition to the heterochromatic Xi. Still, many additional layers of chromatin regulation are unexplored, and histone changes less well understood, providing excellent starting points for future work. More chromatin remodelers and repressor complexes might be recruited in a Xist-dependent matter, while others mostly respond to chromatin changes that occur. Global defects show the generality of factors involved in silencing, as shown by the loss of SPEN and Repeat A [5,15,21], while gene-specific defects highlight requirements for specific repressor complexes, evident from SETDB1-loss [37]. Further insights in XCI will aid in elucidating the missing puzzle pieces of how chromatin remodelers and repressor complexes execute gene silencing chromosome-wide and gene-specifically, as well as how these silencing pathways depend and communicate to one another.

Credit author statement
JB and JG were both involved conceptualization and writing the manuscript.

Conflict of interest statement
Nothing declared.  Analysis of the role of SPEN in initiation of XCI shows that SPEN is required for the early phase of establishing the silent state. CUT&RUN analysis shows that SPEN binding overlaps at enhancers and promoters with HDAC3 and NuRD-complex members, suggesting a potential role for SPEN in distributing and/or activating repressor complexes. This study also shows SPEN is not required for maintenance of the Xi in NPC models.