Female mammals have two X chromosomes, whereas males have only one. A remarkable solution has therefore evolved to prevent a gross imbalance in gene expression occurring between the sexes: in every cell that has two X chromosomes, one entire X chromosome is ‘silenced’ to prevent RNA from being transcribed from it. This process is called X-chromosome inactivation (XCI) and initiates early in the development of female embryos. Once complete, XCI is essentially stable for life1 — thus, by extension, a human X chromosome can be propagated in the silenced state for more than 100 years.
XCI has become a paradigm for epigenetic processes — those in which DNA and associated proteins are modified to alter gene expression — and has been intensively studied for decades. For the past 25 years, much of this research has centred on a long non-coding RNA (lncRNA) called Xist, which is needed to orchestrate XCI. However, the details of Xist’s silencing mechanism have been elusive. Writing in Nature, Dossin et al.2 report a stunning series of experiments that reveal how Xist silences genes by partnering with a protein called SPEN.
Xist is expressed exclusively from the X chromosome that will be inactivated, where it spreads locally and silences nearly every gene on the chromosome by associating with an array of proteins. For example, Xist engages the Polycomb protein complexes (which modify the histone proteins that package DNA into a condensed form called chromatin) to maintain gene silencing on the inactivated X chromosome3,4. Although this maintenance function is well documented, how Xist silences active genes in the first place has remained a mystery — in part because the majority of Xist’s protein partners were unknown. But in 2015, a series of studies5–9 revealed a comprehensive set of proteins involved in XCI. These screens all identified SPEN as a Xist-binding protein that is essential for XCI.
SPEN belongs to an evolutionarily conserved family of RNA-binding proteins that have been implicated in transcriptional silencing and, curiously, RNA processing in both animals and plants10. To interrogate SPEN’s role in XCI, Dossin et al. first used a biological system known as an auxin-inducible degron to rapidly degrade SPEN in mouse embryonic stem cells. Consistent with a 2019 report11, the authors observed that Xist is almost completely unable to silence genes along the X chromosome in the absence of SPEN. In an important first, the authors demonstrated that SPEN is required for successful XCI in vivo in mice. They also found that SPEN was needed to dampen expression of ‘escapees’ — genes on the silenced X chromosome that partially evade XCI.
By observing fluorescently labelled molecules in living cells, Dossin et al. found that SPEN is recruited to the X chromosome as soon as Xist expression begins at the onset of XCI. SPEN contains four RNA-binding domains (called RRMs) at its amino-terminal end and an evolutionarily conserved SPOC domain at its carboxy-terminal end. The authors found that, although RRMs 2–4 are required to bind Xist, the SPOC domain is the essential mediator of gene silencing. As suggested by previously reported experiments12, forcing an interaction between Xist and the SPOC domain alone was enough to restore XCI in cells that lack SPEN.
It has been proposed7,13 that SPEN confers gene-silencing capabilities on Xist by recruiting and/or locally activating the enzyme HDAC3, which removes gene-activating acetyl groups from histones. However, HDAC3 accounts for only part of the gene silencing that occurs during the early stages of XCI13. To find other mechanisms by which SPEN might bring about silencing, Dossin et al. used a mass spectrometry technique to identify proteins that interact with the SPOC domain.
Confirming earlier work14, the authors found that SPEN’s SPOC domain interacts not only with HDAC3, but also with the associated co-repressor proteins NCOR1 and NCOR2 (also called SMRT), and with components of the nucleosome remodelling and deacetylase (NuRD) complex, all of which are epigenetic silencers. Moreover, the authors observed that the SPOC domain interacts with parts of the machinery used for transcription and splicing (the process by which newly made RNA transcripts are turned into messenger RNA), including RNA polymerase II, the enzyme that catalyses transcription. Dossin and colleagues identified interactions with components of the N6-methyladenosine (m6A) methyltransferase complex, several of which have been linked to XCI6,11,15. Accordingly, SPEN and its array of associated proteins might function like a molecular multi-tool to silence genes in various genomic contexts. Although much of SPEN’s silencing function might derive from its interactions with known epigenetic silencers, its association with transcription and RNA-processing machineries leaves open the possibility that SPEN can also silence genes through another, as-yet-undefined mechanism.
Perhaps most strikingly, Dossin et al. adapted a technique called CUT&RUN to map the location of SPEN on an X chromosome that was being inactivated. This revealed that, shortly after Xist starts to be expressed, SPEN associates with active gene promoters and enhancers (DNA regions that initiate and increase the likelihood of transcription, respectively), but then disengages from these sites after it has silenced transcription. These discoveries imply that SPEN is part of a system that recruits silencing machinery specifically to transcriptionally active regulatory elements at the onset of XCI (Fig. 1). Whether this mechanism also requires chromatin modifications, RNA polymerase II, actively transcribed RNA or other factors should be addressed in the future. Another issue that should be investigated is why Xist isn’t silenced by SPEN, given that a large amount of SPEN accumulates over the Xist gene.
SPEN binds to a region of Xist RNA called Repeat A, which is required to initiate gene silencing5,8,16. Because deleting the Spen gene largely mirrors the effects of deleting Repeat A11, SPEN seems to be responsible for most of Repeat A’s silencing ability. However, Repeat A also binds to other proteins, including those that normally promote splicing, as well as to RBM15 and RBM15B, SPEN’s SPOC-domain-containing cousins5,15,17. Therefore, it is now crucial to determine how these proteins might compete or cooperate with SPEN to initiate gene silencing. Moreover, deletion of Repeat A drastically reduces levels of the Xist RNA itself18, and, in certain contexts, deletion of SPEN similarly reduces levels of Xist11. How Repeat A is required for the production of Xist, and how its role in Xist production relates to its ability to initiate silencing, are key questions for the future.
For decades, Xist has served as a leading example of RNA’s role in regulating gene expression. Most notably, Xist was one of the first mammalian RNAs shown to be involved in Polycomb-mediated silencing3,4. It therefore seems appropriate that, by studying this RNA, Dossin et al. might have uncovered a new and fundamental aspect of gene regulation — the transient recruitment of SPEN to regulatory elements by RNAs, or even by proteins, which could be a general mechanism for silencing transcription throughout the mammalian genome.
Nature 578, 365-366 (2020)
Vallot, C., Ouimette, J.-F. & Rougeulle, C. BioEssays 38, 869–880 (2016).
Dossin, F. et al. Nature 578, 455–460 (2020).
Silva, J. et al. Dev. Cell 4, 481–495 (2003).
Plath, K. et al. Science 300, 131–135 (2003).
Chu, C. et al. Cell 161, 404–416 (2015).
Moindrot, B. et al. Cell Rep. 12, 562–572 (2015).
McHugh, C. A. et al. Nature 521, 232–236 (2015).
Monfort, A. et al. Cell Rep. 12, 554–561 (2015).
Minajigi, A. et al. Science 349, aab2276 (2015).
Su, H., Liu, Y. & Zhao, X. Cancer Transl. Med. 1, 21–25 (2015).
Nesterova, T. B. et al. Nature Commun. 10, 3129 (2019).
Ha, N. et al. iScience 8, 1–14 (2018).
ylicz, J. J. et al. Cell 176, 182–197 (2019).
Shi, Y. et al. Genes Dev. 15, 1140–1151 (2001).
Patil, D. P. et al. Nature 537, 369–373 (2016).
Wutz, A., Rasmussen, T. P. & Jaenisch, R. Nature Genet. 30, 167–174 (2002).
Pintacuda, G. et al. Mol. Cell 68, 955–969 (2017).
Hoki, Y. et al. Development 136, 139–146 (2009).