Balanced between order and disorder: a new phase in transcription elongation control and beyond

ABSTRACT

We recently reported that the cyclin T1 histidine-rich domain creates a phase-separated environment to promote hyperphosphorylation of RNA polymerase II C-terminal domain and robust transcriptional elongation by P-TEFb. Here, we discuss this and several other recent discoveries to demonstrate that phase separation is important for controlling various aspects of transcription.

Introduction

The accurate and yet efficient spatiotemporal regulation of numerous biochemical reactions inside a cell is crucial for all biological functions. Most in vivo enzymatic reaction components must be appropriately coordinated and concentrated within specialized subcellular compartments to ensure their highly efficient reactions. In many cases, the intracellular compartmentalization is achieved through encapsidation within membrane-bound organelles such as the endoplasmic reticulum, Golgi apparatus and mitochondria. For these organelles, the surrounding lipid bilayer membrane enables the proper control of composition and biomolecule concentration within the organelles. However, cells also contain many non-membrane-bound compartments. How these structures organize and concentrate specific sets of molecules without the surrounding membranes has remained a mystery for a long time. Recent studies have shown that these membraneless compartments, such as P granules in germ cells of Caenorhabditis elegans, exhibit remarkable liquid-like properties and can continuously exchange their components with the surrounding environment, which suggest that these structures represent liquid-phase condensates that are formed via the biologically regulated liquid-liquid phase separation process [1]. It is now generally believed that phase separation could be a fundamental mechanism for regulating diverse physiological as well as pathological processes including cell division [2], signal transduction [3,4], cancer and neurodegenerative diseases [5–7].

The main force that drives phase separation is believed to be weak, multivalent protein-protein interactions [8]. For example, proteins composed of repeated modular domains have been shown to drive the intracellular phase separation, primarily due to the assembly of high-order complexes by intra- or inter-molecular multivalent interactions. In addition, proteins containing low-complexity and intrinsically disordered regions (IDRs) represent another large group of macromolecules to exhibit this unique molecular signature that promotes phase separation. Although the IDRs lack well-defined 3D structures, they often contain repetitive sequence elements that can serve as the basis for charge-dependent multivalent interactions. Interestingly, IDRs exist widely in transcription factors/co-factors, but their exact roles in regulating transcription have remained understudied [9].

Like all highly regulated biological functions, the transcription of protein-coding genes by RNA polymerase (Pol) II is a complex biochemical process that is accomplished by the concerted actions of numerous transcription factors/co-factors. Multiple lines of evidence have suggested that some of these factors can assemble local foci that are enriched in the transcription machinery and Pol II to efficiently promote gene transcription, which have led to the model of the so-called “transcription factories” for the spatial control of transcription inside the nucleus [10]. In addition to the transcription factories, the nucleus also contains many other compartments that can gather distinct types of machinery for controlling various nuclear activities. For example, nuclear speckles are self-organized nuclear domains located in the interchromatin regions of nucleoplasm [11]. Enriched in pre-mRNA splicing factors and additional factors involved in transcription, chromatin modification, and 3′-end RNA processing, nuclear speckles are often found in the vicinity of active transcription sites, suggesting their close relationship with efficient transcription/RNA processing.

In light of recent advances in understanding the role of phase separation in modulating a number of biological processes and contributing to intracellular membrane-less compartmentalization, this phenomenon has also been proposed to regulate transcription [12]. Indeed, several recent reports, including one from our group, have provided experimental evidence that certain transcription factors and co-factors harboring low complexity domains or IDRs can form biomolecular condensates via phase separation to control various aspects of transcription (Figure 1) [13,14]. Here, we discuss the transcription elongation control by the ordered and disordered regions of P-TEFb, with an emphasis on the role of the CycT1 histidine-rich domain (HRD) in creating a phase-separated environment for hyperphosphorylation of the Pol II CTD and robust transcriptional elongation. We will connect our observations with other recent findings, which have revealed an important role of phase separation in controlling other key aspects of gene transcription process.

Figure 1.

Phase separation controls at least three main steps during gene transcription. Within the transcription factories, various transcription factors/co-factors are organized into different compartments to ensure highly efficient and accurate regulation of gene expression. At the super-enhancers, transcriptional coactivators such as Brd4 and MED1 can form phase-separated condensates through their intrinsically disordered regions (IDRs), which allow the compartmentalization and concentration of transcription apparatus for the super-enhancer-driven transcriptional activation. E: enhancer. Many sequence-specific transcription factors such as the FET (FUS/EWS/TAF15) family proteins that contain low complexity sequences can self-assemble into transactivation hubs and transiently recruit Pol II with unphosphorylated CTD to increase transcriptional initiation from gene promoters. GGAA-microsatellites: Highly repetitive GGAA-containing microsatellite DNA elements. Upon phosphorylation by CDK7 present in the general transcription factor TFIIH, the CTD incorporates into phase-separated droplets formed by the CycT1 histidine-rich domain (HRD). This phase-separated environment enhances the functional interactions between P-TEFb and Pol II to promote efficient biochemical reactions that lead to CTD hyperphosphorylation and robust transcriptional elongation.

CDK9 hyperphosphorylates the CTD

Consisting of CDK9 and CycT1, P-TEFb is a master regulator of Pol II transcriptional elongation. One of the critical substrates phosphorylated by CDK9 during this process is the C-terminal domain (CTD) of Pol II. As a long unstructured protein domain, the CTD is composed of multiple heptapeptide repeats with the consensus sequence YSPTSPS [15]. Although the CTD is not required for the catalytic activity of Pol II, it is essential for optimal transcription and co-transcriptional RNA processing. This essential function stems from the fact that the CTD is subjected to dynamic post-translational modifications (PTMs) during the transcription cycle, which creates the “CTD code” for recruiting and modulating the activities of the various transcription and RNA processing factors [16,17].

To date, the best-characterized CTD modification is phosphorylation. Depending on the level of CTD phosphorylation, Pol II exists in either the hypophosphorylated (IIa) or hyperphosphorylated (IIo) form. Recent analyzes by mass spectrometry have revealed that the phosphorylation of serines at the second (Ser2) and fifth positon (Ser5) in a heptapeptide repeat is the most abundant CTD modification in both yeast and humans [18,19], suggesting that the hyperphosphorylated Pol II is mostly modified on these two residues. In addition, the biological significance of CTD hyperphosphorylation has also been well documented. Early studies using in vitro reconstituted transcription systems demonstrated that the transition of Pol II from initiation to productive elongation is accompanied by the hyperphosphorylation of the Pol II CTD [20]. Consistent with the demonstrations that the general transcription elongation activator P-TEFb is required for the release of Pol II from promoter-proximal pausing [21], the CDK9 kinase activity has been shown to be responsible for the CTD hyperphosphorylation on Ser2 in vivo during transcriptional elongation [22,23].

HRD-dependent phase separation by CycT1-IDR promotes Pol II CTD hyperphosphorylation and transcriptional elongation

In contrast to the N-terminal region of CycT1 that contains well-folded structural domains such as the cyclin-box and the Tat/TAR-recognition motif (TRM) for specific interactions with CDK9 and other proteins and RNA, the long and extended C-terminal region of CycT1 is mostly unstructured and only contains a few interspersed short motifs. In an effort to determine the exact role of this region in regulating P-TEFb activity, we have recently identified an unanticipated role of the CycT1 histidine-rich domain (HRD) for CTD hyperphosphorylation and transcriptional activation.

Our study began with a series of in vitro kinase reactions, in which we discovered that the deletion of the HRD from CycT1 severely diminished the ability of CDK9 to hyperphosphorylate the full-length Pol II CTD containing all 52 repeats, but not its ability to autophosphorylate or phosphorylate a short CTD substrate containing only nine repeats. Taking advantage of the fact that the HIV-1 transcription is highly sensitive to changes in P-TEFb activity [24], we used the Gal4-tethering system to directly deliver wild-type or the various CycT1 mutants fused to the Gal4 DNA-binding domain to the HIV-1 promoter that contains upstream Gal4-binding sites (UAS). This strategy allowed us to measure the CycT1 activity independent of other factors that may recruit P-TEFb to the DNA template. Correlating with the dependence on the HRD for CTD hyperphosphorylation, we found that the deletion of HRD decreased the Gal4-CycT1-mediated HIV transcription. Notably, the transcription of several well-known P-TEFb target genes also decreased when endogenous CycT1 was replaced with the ∆HRD mutant. These in vitro and in vivo results implicated the HRD as essential for the CTD hyperphosphorylation and gene transcription.

To explore the molecular mechanism underlying the HRD dependence, we applied the state-of-the-art imaging techniques to monitor the CycT1 behavior inside the cell nucleus. Our single particle tracking and Fluorescence Recovery After Photobleaching (FRAP) analyzes demonstrate that the HRD was required for longer residence time of CycT1 on the chromatin templates, presumably due to the stronger association of P-TEFb with the actively transcribed genes.

The HRD is located within a broader Intrinsically Disordered Region of CycT1, raising the possibility that it may function by promoting liquid-liquid phase separation. To test this hypothesis in vitro, we first showed that recombinant CycT1 IDR could form liquid-like droplets, whereas the deletion of HRD from IDR produced only very small-size aggregates of irregular shape under the same conditions. The densely distributed histidine residues within the HRD were found to be essential for the droplet formation, as mutating these residues to alanines severely inhibited this process.

Correlating with these in vitro results, CycT1 was shown to localize to the nuclear speckles in an HRD-dependent manner, and this distribution pattern was quickly dissipated by 1,6-hexanediol, a compound known to disrupt phase separation by blocking weak, multivalent interactions [25]. Displaying their liquid-like properties, the CycT1 speckles were shown to undergo frequent spontaneous fusions. Finally, our data indicate that the P-TEFb-mediated CTD hyperphosphorylation and the HRD-induced phase separation were similarly sensitive to inhibition by 1,6-hexanediol, whereas the interaction between the CTD and P-TEFb was considerably more resistant, suggesting that the phase-separated environment established by the HRD plays a key role in promoting the CTD hyperphosphorylation. Taken together, these results allowed us to propose a model that phase separation induced by the CycT1 HRD can direct P-TEFb into compartmentalized transcription loci to promote Pol II hyperphosphorylation on the CTD and transcriptional elongation (Figure1).

Phosphorylation-controlled CTD association/dissociation with distinct phase-separated compartments for optimal transcriptional regulation

A number of recent studies have revealed an intimate association of the Pol II CTD with the phase separation process. First, purified recombinant proteins containing low-complexity domains from the FET (FUS/EWS/TAF15) protein family can form phase-separated droplets/hydrogels that can actively trap the unphosphorylated CTD [26,27]. Recent live-cell single-molecule imaging studies show that transcription factors/coactivators such as Mediator, Brd4, and EWS form local high-concentration interaction hubs inside the nucleus, leading to the enrichment of Pol II and transcriptional activation [28,29]. These results suggest a potential mechanism for Pol II recruitment to the pre-initiation complex (PIC) prior to transcription initiation, whereby certain transcription activators containing IDRs/LCDs form phase-separated compartments to trap Pol II with the unphosphorylated CTD (Figure 1).

Does the CTD also participate in phase separation at the elongation stage of the transcription cycle, and if so, how does its phosphorylation affect this process? In our study, we observed that the CycT1 IDR could recruit the CTD into phase-separated droplets through the direct IDR-CTD interaction, suggesting that the connection between phase separation and the CTD is not limited to just the transcription initiation process. At the pre-initiation stage of the transcription cycle, Pol II is recruited to gene promoters in its IIa form and assembled into the PIC through interactions with the general transcription factors. After that, the CTD is sequentially phosphorylated by CDK7 of TFIIH and then CDK9 of P-TEFb to convert Pol II into the form capable of productive elongation [17]. Notably, we found that pre-phosphorylating the CTD with CDK7 significantly promoted the CTD’s inclusion into the CycT1-IDR droplets [13].

Consistent with our result, a recent study shows that the CTD can undergo phase separation through at least two different mechanisms [30]. When in its unphosphorylated form, the weak homo- and heterotypic hydrophobic interactions are the driving force for CTD phase separation and likely assist in Pol II cluster formation. Notably, we found that the unphosphorylated CTD, although less efficient than its phosphorylated counterpart, can also be incorporated into phase-separated droplets together with CycT1 IDR, suggesting a potential mechanism for recruiting P-TEFb into the Pol II clusters at an early stage of the transcription cycle when the CTD is yet to be hyperphosphorylated. Upon phosphorylation, which leads to the increases in radius, protein accessibility and stiffness of the CTD [31], the multivalent electrostatic interactions with selected factors such as CycT1 become dominant to alter the phase separation behavior of the CTD at a later stage of the transcription cycle (Figure 1). Collectively, these findings suggest that phosphorylation can modulate CTD’s associations with different phase-separated compartments, underscoring the CTD and its differential phosphorylation states as an important nexus for regulating transcription and co-transcriptional RNA processing.

Phase separation controls formation and function of super-enhancers

Recent studies have also revealed a functional link between coactivator-induced phase separation and the activity of super-enhancers [29,32]. As cis-acting gene control elements, super-enhancers are clusters of enhancers that are occupied by an unusually high density of master transcription factors and coactivators [33]. A key feature of super-enhancers is that their formation and function involves many cooperative binding events, leading to the formation of a high-density transcriptional machinery to drive robust transcriptional activation. In addition, the exceptionally sharp transitions during super-enhancers formation and dissolution also highlight the importance of cooperativity among different regulatory components.

Because of these properties, it was proposed that super-enhancers exist in the form of phase-separated condensates [12]. Indeed, recent findings have validated this hypothesis and show that the transcriptional Mediator complex forms stable clusters at super-enhancers, which are largely co-localized with the Pol II clusters in living cells [29,32]. Further biophysical characterizations reveal that these clusters exhibit properties that are consistent with biomolecular condensates resulting from phase separation. In addition to the Mediator, the bromodomain protein Brd4, which is another transcriptional coactivator enriched in the super-enhancers, also forms discrete puncta in the nucleus. Both Brd4 and MED1, a subunit of the Mediator complex, contain extensive IDRs that can form phase-separated droplets in vitro, suggesting that these IDRs may also be responsible for the coactivator-induced condensation in vivo. In agreement with the proposed function of phase separation at super-enhancers, preformed MED-IDR droplets can compartmentalize and concentrate transcriptional machinery from nuclear extracts. Finally, in vivo tracking analysis reveals transient kissing events between the Mediator clusters and their regulated gene locus, suggesting that cooperative phase separation enables transcriptional control at a long-distance range. Thus, in addition to its control of transcription at or near the transcription start site (TSS) and the gene body, these studies have also established phase separation as a key regulator of the formation and function of the super-enhancers that influence gene transcription from a far distance (Figure 1).

Conclusion

In the past, much of our knowledge about the eukaryotic transcriptional control has been acquired through characterizing well-defined and structurally ordered protein domains by conventional biochemical and structural analyzes. These studies have led to the establishment of the widely accepted structure-function paradigm that provides the basic framework for our current understanding of the molecular mechanisms governing gene transcription. However, bioinformatics analyzes indicate that the IDRs are pervasive in the human proteome and that they are especially concentrated in proteins involved in the transcriptional control [9]. As discussed above, observations made by others and us have begun to unveil the roles of the IDRs in transcriptional control by demonstrating their intimate involvement in forming phase-separated functional compartments. However, despite this progress, many key questions remain to be answered.

For example, in addition to the three major steps/categories discussed above and in Figure 1, it is unclear whether other aspects of the transcriptional control such as transcription termination and the coupling of transcription with pre-mRNA processing also involve the formation of phase-separated droplets, if yes, what specific factors contribute to this process and whether the Pol CTD is also involved. Moreover, most of the recent studies have focused on the induction of phase separation, but little is known about how the various phase-separated compartments can be reversibly dissembled in cells in response to specific signals or during transition through the transcription cycle. In addition, do post-translational modifications of the IDRs in the key transcription factors/co-factors and their associated RNA molecules play any roles in the formation and disassembly of the phase-separated compartments? Finally, it is yet to be determined whether and how aberrant phase separation causes misregulation of gene transcription and diseases. Undoubtedly, future studies aiming at addressing these questions will continue to advance our understanding of the mechanisms governing phase separation and the IDR-mediated transcriptional control. The knowledge, tools, and reagents generated in this process will also facilitate the development of novel therapeutic strategies that target the disordered regions in key transcriptional regulators.

Funding Statement

This work is supported by a grant (R01AI041757) from the National Institutes of Health to Q.Z.

Disclosure statement

No potential conflict of interest was reported by the authors.