The nuclear matrix protein SAF-A (hnRNP U) has been shown to play an essential role in XCI and Xist RNA attachment. Initial evidence came from the observation that SAF-A:GFP fusion proteins are enriched at the Xi in somatic cells and that knock-down of SAF-A led to release of Xist RNA from the Xi (Hasegawa et al., 2010; Helbig and Fackelmayer, 2003). For long time, SAF-A has been known to
possess a DNA- and a RNA-binding motif and a recent in silico study predicted a high interaction probability between SAF-A and Xist RNA (Agostini et al., 2013; Fackelmayer et al., 1994). Yet, final evidence remained inconclusive as experiments with conventional microscopy showed no colocalization with Xist RNA. SAF-A even appeared to be excluded from the Xi when using SAF-A specific antibodies (for reviews see Hasegawa and Nakagawa, 2011; Tattermusch and Brockdorff, 2011).
The strong spatial correlation between SAF-A and Xist RNA observed with 3D-SIM provide a plausible explanation for the previously reported discrepancy between microscopic data and biochemical analyses. The results further indicate a Xi-specific post-translational modification of SAF-A, which can be concluded from the observed epitope masking effect, and which may be a prerequisite or the result of its interaction with Xist RNA. It can also be speculated that binding of Xist RNA is causing multimerization of SAF-A molecules, which in turn may act as a scaffold for Xist RNA spreading along the Xi, after its initial tethering by YY1 (Jeon and Lee, 2011). This view is very intriguing as it offers another explanation how Xist RNA is able to spread along the Xi in a sequence-independent manner within the ANC of the Barr body and subsequently accumulating at certain (gene-rich) genomic sites, where it may cause its inhibiting function.
This is further substantiated by a very low spatial interaction observed for Xist RNA and markers of repressed chromatin, like H3K27me3 and macroH2A1. Using 3D-SIM, the data provided support for the CT-IC model, as Xist RNA and SAF-A have been found to be localized within the decondensed ANC of the Barr body as opposed by repressive chromatin marks, which have been shown to be localized within the compacted PNC. Together, these two observations argue for a functional relevance of the spatial separation of these two compartments and moreover provide a novel model of Xist RNA spreading and function.
5.2.4 Localization of Xist RNA within the ANC is an intrinsic property already at early differentiating XX ESCs
Analysis of early differentiation at the onset of XCI provided evidence that the localization of Xist RNA into decondensed chromatin sites – reflecting the ANC – is an intrinsic property, which is not just established in fully differentiated cells. Initial Xist RNA spreading is followed by an exclusion of RNAP II, which occurred concomitantly with chromatin compaction (Chaumeil et al., 2006). Of note, chromatin compaction and full RNAP II exclusion was observed on Day 5 of differentiation, which is one day after the initial spreading of Xist RNA, under the experimental conditions used in this work. It could have been shown that chromatin compaction on the Xi can be induced through a novel PRC2- and H3K27me3-independent pathway involving SMCHD1 and the HP1-binding HBiX1 (Nozawa et al., 2013). This is in line with another recent study showing the need for chromatin compaction as a
prerequisite of PRC2-induced H3K27me3-enrichment during early XCI (Yuan et al., 2012). Thus, one may speculate that spreading of Xist RNA leads to inactivation of X-linked gene promoters as a first step through the loss of active histone modifications (Marks et al., 2009). This in turn may lead to RNAP II exclusion and concomitant local chromatin compaction. These early events, which may be partially facilitated by a structural role of Xist RNA as suggested by the results of the present work, are then followed by the accumulation of silencing chromatin modifications and subsequent long- term silencing. The observed increased chromatin compaction after 9 days of differentiation, which is not distinguishable from somatic C2C12 cells, as visualized by 3D-SIM, is likely to be the result of these progressively acquired long-term silencing modifications, like DNA-methylation, histone de- acetylation or incorporation of repressive histone variants.
5.2.5 3D-SIM reveals incomplete silencing within an autosomal Barr body
Xist RNA was predominantly found within the ANC of the tr36 inducible autosomal ES cell line indicating that spreading occurs normally in respect to functional chromatin compartments and providing further evidence that the localization of Xist RNA within this active compartment is indeed an intrinsic property. Nevertheless, Xist RNA obviously fails to induce the generation of an autosomal Barr body with the same features as its Xi counterpart, as seen with 3D-SIM, which argues for an impaired silencing capacity within this autosomal context. It has been reported by one group that Xist RNA is tethered to the Xi via YY1 within the so called nucleation center at the XIC (Jeon and Lee, 2011). One possibility, which may account for the widely recognized incomplete Xist RNA-induced gene silencing of autosomes is that autosomal chromatin around the transgene insertion site lack this YY1 nucleation center. As a result, Xist RNA may only be inadequately tethered to the autosome, which would also explain the observed vast expansion of the Xist RNA volume that has an about 2- fold increased diameter than its X-chromosomal counterpart. Still, this option does not explain why the present Xist RNA fails to induce proper gene silencing as it does on the X chromosome.
Undisputedly, the X chromosome has acquired some evolutionary adaptation, which facilitates its susceptibility for chromosome-wide silencing. One of the oldest hints for this idea was the discovery of the increased fraction of repetitive LINE-1 sequences on the X chromosome, which has led to the formulation of the LINE-1 hypothesis, claiming that these repetitive elements act as waystations for Xist RNA spreading (Bailey et al., 2000; Lyon, 1998). Although the data presented in this thesis argue against a direct interaction between Xist RNA and LINE-1 sequences, those repetitive elements might still facilitate the formation of a repressive chromatin compartment (Tang et al., 2010). The obtained 3D-SIM data presented in this thesis suggests that LINE-1 sequences, instead of being direct waystations for Xist RNA spreading, facilitate the formation of a repressive chromatin compartment,
which is hampered in the autosomal context and thus leads to the observable incomplete silencing effects.