Directed stochasticity: Building biomolecular condensates in the right place

Many biomolecular condensates have been documented within eukaryotic cells (1). These membraneless organelles vary in size, composition, and location and are associated with a variety of cellular processes, including transcription and RNA processing. Determining how biomolecular condensates form and dissolve is important for understanding fundamental cell biology as well as how condensate misregulation contributes to human disease. Many biomolecular condensates form “bodies” via liquid–liquid phase separation, an inherently stochastic process, in either the nucleus or the cytoplasm. These bodies form in their respective compartments but not the other, but how cells prevent formation in improper locations within the cell is unclear. This issue is especially relevant for nuclear proteins that are produced in the cytoplasm but only form condensates in the nucleus. By studying the histone locus body (HLB), Xu et al. (2) address whether mechanisms exist to prevent nuclear body proteins from inappropriately forming condensates in the cytoplasm.

The HLB forms at replication-dependent (RD) histone genes in metazoans (3). The expression of these genes is tightly coupled to the S phase of the cell cycle when large amounts of histone proteins are synthesized to package the newly replicated DNA. RD histone genes do not contain introns and encode the only mRNAs that terminate in a stem-loop rather than a polytail (4). HLBs concentrate factors needed for proper transcription and 3′-end processing of these unique mRNAs. The primary factor in human cells necessary for HLB formation is the large (220 kD), mostly intrinsically disordered protein, nuclear protein at the ataxia-telangiectasia locus (NPAT). RD histone mRNA synthesis is activated as cells enter the S phase via phosphorylation of NPAT by the kinase cyclin E/Cdk2 kinase (5). Xu et al. uncovered mechanisms preventing NPAT from inappropriately forming condensates in the cytoplasm by determining how NPAT enters the nucleus.

Large proteins are imported into the nucleus by karyopherins, which function as molecular chaperones by binding to a nuclear localization sequence (NLS) within their cargo protein in the cytoplasm and unloading the cargo in the nucleus after transportation through the nuclear pore complex. Xu et al. used the karyopherin α-specific inhibitor, ivermectin, to show that loss of KPNA function led to reduced numbers of NPAT foci in the nucleus and decreased expression of RD histone genes. They found that KPNA3 is the only KPNA protein to interact with NPAT. A region of KPNA3 containing three armadillo (Arm) repeats binds directly to NPAT’s C-terminal region, which contains an NLS. Knockdown of KPNA3 demonstrated its requirement for nuclear import of NPAT and expression of the RD histone genes.

The authors observed that strongly suppressing NPAT nuclear transport by combining knockdown of KPNA3 with ivermectin treatment results in the appearance of cytoplasmic NPAT foci. A similar result occurs when the NPAT NLS is deleted. Whether these cytoplasmic NPAT foci also contain other HLB factors and thus represent ectopic “HLB formation” was not assessed and remains an interesting outstanding question. Nevertheless, this result raises the intriguing possibility that rapid transport of NPAT into the nucleus after its translation in the cytoplasm is important for suppressing inappropriate condensation in the cytoplasm, an idea that could be applied to factors that drive the assembly of other nuclear bodies.

Interestingly, the authors found that the transport of NPAT into the nucleus is not the only way that KPNA3 inhibits inappropriate NPAT condensate formation in the cytoplasm. This observation first required an understanding of the mechanisms of NPAT condensation. A C-terminal fragment of NPAT containing the NLS accumulates in the nucleus but cannot form distinct puncta, suggesting that other regions of NPAT are needed to form HLBs. Work on the Drosophila ortholog of NPAT, Mxc, provided a clue. A LisH domain and a self-interaction facilitator (SIF) domain within the N terminus of Mxc are each required for HLB formation and Mxc function in vivo (6). Similarly, Xu et al. identified a C-terminal SIF (C-SIF) of NPAT proximal to the NLS that interacts directly with the central portion of NPAT. C-SIF deletion reduces NPAT’s self-association in vitro and in vivo but does not affect KPNA3’s binding and nuclear import of NPAT. Thus, self-interaction that promotes biomolecular condensation is conserved between NPAT and Mxc and is a feature of proteins that scaffold other nuclear bodies such as the Cajal body (7, 8).

In a clever set of experiments, Xu et al. demonstrated that KPNA3 binding to the NPAT NLS prevents NPAT self-interaction by sterically hindering the C-SIF domain. Purified KPNA3-reduced NPAT condensation and liquid–liquid phase separation in vitro and overexpression of KPNA3 in cells promoted the nuclear import of NPAT while suppressing the formation of HLBs. However, the Arm 3–5 repeats of KPNA3 that bind NPAT were not enough on their own to inhibit NPAT self-interaction. Only by fusing the KPNA3 Arm 3–5 repeats to another KPNA or to GFP did the authors recapitulate the inhibitory effect of KPNA3 on NPAT condensation. The simplest interpretation of these data is that once KPNA3 binds the NPAT NLS, it sterically hinders C-SIF from interacting in trans with the central region of another NPAT molecule, thereby suppressing condensation. Thus, KPNA3 prevents ectopic HLB assembly in the cytoplasm in two ways: by binding to and blocking the ability of NPAT to self-associate and by transporting NPAT to the nucleus.

The Xu et al. study raises several interesting questions (Fig. 1). For instance, after KPNA3 releases NPAT in the nucleus, and thus from inhibition of self-association, what mechanisms ensure that NPAT condensation occurs only at RD histone loci? In Drosophila, the RD histone locus seeds formation of the HLB (9), but how Mxc initially recognizes the histone locus remains unclear. Although sequence-specific DNA-binding proteins interact with NPAT (10), a factor that targets NPAT to all RD histone genes has not been identified. The dynamics of NPAT release from KPNA3 are unknown and could provide another point of regulation for HLB assembly. Does KPNA3 release NPAT only in the vicinity of RD histone genes to prevent ectopic NPAT condensation in the nucleoplasm or at other loci? The C terminus of NPAT binds directly to HLB proteins FLICE-associated huge protein (FLASH) and YARP/Gon4L (11), raising the possibility that NPAT-binding proteins other than KPNA3 modulate NPAT condensation. FLASH contributes to HLB assembly in both flies and human cells (5, 12), and thus interactions between NPAT and its other binding partners are likely important for HLB assembly. Future studies of the HLB should continue to provide insights into the functional relationship between nuclear bodies and genome function.