You have full access to this article via your institution.

In a recent study, Ding et al. investigated the role of hnRNPK phase separation in mammalian dosage compensation. The study stands out by linking biophysical and biochemical measurements with genetic and cell biological experimentation, providing wide-ranging evidence for specific mechanistic aspects of X chromosome inactivation, while expanding the potential repertoire for phase separation in biology.

In mammals, X chromosome inactivation (XCI) compensates for the dosage difference between XX and XY sex chromosomes in females and males, respectively, by silencing one of the two X chromosomes in females. This process is initiated by the X-inactive specific transcript (Xist), a non-coding RNA which is expressed from and spreads over the inactive X (Xi) chromosome. As Xist spreads, it facilitates the recruitment of chromatin-modifying complexes of the Polycomb group and initiates gene silencing. However, how Xist spreads from its site of transcription over the Xi remains a topic of investigation.

Ding et al.1 propose that liquid-liquid phase separation (LLPS) of the RNA-binding protein hnRNPK plays a role in facilitating Xist spreading and recruiting silencing factors to the Xi. Previously, Xist spreading was known to depend on the interplay between the RepB motif on Xist and Polycomb complexes, with hnRNPK acting as a mediator indirectly linking Xist to Polycomb-associated silencing factors.2 hnRNPK contains multiple RNA-binding domains and was known to undergo LLPS mediated by its RGG domain, an intrinsically disordered region (IDR).3,4 LLPS is increasingly recognized as a fundamental mechanism in forming biomolecular condensates, dynamic compartments that compartmentalize biochemical processes implicated in diverse cellular mechanisms.5 Such condensates emerge from multivalent, weak interactions, including those mediated by IDRs and protein–nucleic acid interactions.6 hnRNPK association with the Xi does not require Xist and occurs even before Xist expression. This raises the question: could hnRNPK and RepB on Xist cooperatively drive an LLPS mechanism that facilitates Xist spreading and silencing factor recruitment?

Condensates typically consist of different scaffold components, including proteins and nucleic acids, and can recruit specific client molecules. The study shows that hnRNPK condensates can internalize Xist RNA, promoting LLPS through interactions between hnRNPK’s RGG domain and the repeat B (RepB) motif of Xist. These hnRNPK–RepB condensates further localize essential XCI factors, including YY1, RING1B, SPEN, and SMCHD1, all of which possess IDRs. Recent research suggests that condensates influence biochemical activity beyond merely concentrating molecules by modulating their local physicochemical environment.7 This suggests that hnRNPK–RepB condensates may provide specialized microenvironments that enhance interactions between Xist and silencing factors.

Condensate composition can tune their biophysical properties. The study demonstrates that Xist alters the deformability, adhesiveness, and wetting behavior of hnRNPK droplets in vitro. Wetting phenomena, ubiquitous in nature and technology — including for instance painting — describe how liquids spread over surfaces due to surface tension forces.8 This spreading depends on a balance of adhesive forces and cohesive forces. If the interaction between the liquid and surface is strong, the liquid spreads efficiently, coating the surface. To probe this wetting behavior, the authors studied how hnRNPK condensates wet glass surfaces with and without RepB coating, finding increased contact area on RepB-coated slides. These adhesive forces between hnRNPK droplets and RepB would offer a potential biophysical mechanism for Xist spreading. How far such in vitro observations translate to cells — where condensates may contain fewer molecules — remains an important question.

At first glance, under the microscope, the Xi territory in cells does not appear to resemble droplets, as no enrichment of hnRNPK has been observed. hnRNPK is not enriched over Xi and associates also with autosomal chromatin. This might not simply be a detection issue, as Xi enrichment has been observed for several other factors, including Polycomb proteins, SPEN, hnRNPU/SAF-A, and CIZ1.9 Yet, smaller hnRNPK clusters in cells might still share properties with in vitro droplets. If so, conformational changes underlying LLPS might also be relevant when Xist establishes its compartment on the Xi. Observing LLPS could provide an opportunity for detecting such conformational properties, which would be difficult to infer from genetics or cell biology. The study introduces an important control: a 6A hnRNPK mutant, in which six prolines in the RGG domain are replaced with alanines. This mutant lacks LLPS properties but retains Xist RNA binding. Its failure to support proper Xist spreading strongly links LLPS to properties within cells. The ability of hnRNPK to undergo LLPS is critical for the wetting behavior of Xist.

So what does LLPS do for spreading Xist? The authors approach this through the interaction of Xist repeat B with hnRNPK. In cells, Xist mutants lacking repeats B and C still form clusters.10 The X-linked genes are repressed by these mutants further suggesting that Xist is localized on the Xi. However, evidence that mutations in Xist repeat B10 and hnRNPK do not abrogate Xist clusters suggests a more subtle spreading defect. The dynamics or pattern of Xist on the chromosome needs to be considered here. Bousard et al.10 observed that mutations of Xist repeats B and C might not impair X-linked gene silencing but causes a loss of Polycomb-associated histone modifications. Although the relationship between Polycomb complexes and spreading remains to be fully defined, Xist is still visible on Xi, which does not mean that spreading is normal. Xist clustering does not equal effective spreading, and hnRNPKʼs role may involve shaping Xistʼs distribution, not just attaching it to Xi.2 Ding et al.1 show that only wild-type repeat B, not a mutant version supports hnRNPK LLPS wetting properties. This control provides convincing evidence of the specificity of this connection between hnRNPK LLPS and Xist.

While hnRNPK LLPS and some aggregation mechanisms in XCI have been proposed before, the study advances the field by demonstrating the specificity of LLPS in Xist spreading. Recently, CIZ1 has also been shown to form in vitro aggregates, with a prion-like mechanism proposed.9 It should hardly come as a surprise if properties facilitating compartmentalization are involved in organizing a domain at the Xi within the cell nucleus. CIZ1 and hnRNPU accumulate on Xi and are required for chromosomal attachment of Xist in specific cell types. The cell type specificity of their requirement for Xist cluster formation suggests potentially redundant mechanisms. It is conceivable that hnRNPK might also be essential for Xist cluster formation in particular contexts, a possibility that the 6A hnRNP allows to explore further.

The paper by Ding et al.1 is a landmark study integrating cell biology and genetics evidence with biochemistry and biophysical measurements into a coherent mechanistic model. It also suggests chromatin regulation as a promising area for biophysics to expand into. General principles start to emerge for bridging results of these traditionally more distant scientific disciplines that can be expected to shape further initiatives.