Phase transition of fibrillarin LC domain regulates localization and protein interaction of fibrillarin

A key nucleolar protein, fibrillarin, has emerged as an important pharmacological target as its aberrant expression and localization are related to tumorigenesis, chemoresistance and poor survival in breast cancer patients. Fibrillarin contains a N-terminal low complexity sequence (LC) domain with a skewed amino acid distribution, which is known to undergo a phase transition to liquid-like droplets. However, the underlying mechanism of the phase transition of the fibrillarin LC domain and its physiological function are still elusive. In this study, we show that the localization of fibrillarin and its association with RNA binding proteins is regulated by this phase transition. Phenylalanine-to-serine substitutions of the phenylalanine:glycine repeats in the fibrillarin LC domain impede its phase transition into liquid-like droplets, as well as the hydrogel-like state composed of polymers, and also its incorporation into hydrogel or liquid-like droplets composed of wild-type LC domains. When expressed in cultured cells, fibrillarin containing the mutant LC domain fails to localize to the dense fibrillar component of nucleoli in the same way as intact fibrillarin. Moreover, the phase transition of the fibrillarin LC domain is required for the interaction of fibrillarin with other RNA binding proteins, such as FUS, TAF15, DDX5 and DHX9. Taken together, the results suggest that the phenylalanine residues in the LC domain are critical for the phase transition of fibrillarin, which in turn regulates the subnucleolar localization of fibrillarin and its interaction with RNA binding proteins, providing a useful framework for regulating the function of fibrillarin. dependent regulatory


Introduction
fibrillarin is still unclear. Although phase separation has been observed and phase behavior characterized for many proteins both in vitro and in vivo, the molecular code that drives phase separation often remains enigmatic [23]. This has been a major barrier to understanding the role of phase separation in the physiological functions of fibrillarin in the context of living cells.
In the present study, we first uncovered the sequence determinants for the phase transition of the fibrillarin LC domain. In vitro assays showed that the fibrillarin LC domain can undergo phase transition into LLDs, as well as the hydrogel-like state composed of polymers. By interrogating the phenylalanine residues within the fibrillarin LC domain, we demonstrated that the F18/40-to-serine (S) substitution is the minimal variant required for interrupting not only the phase transition, but also partitioning into the existing LLDs or hydrogel droplets of the LC domain. Fibrillarin containing the F-to-S mutation in the LC domain also failed to localize in the DFC regions of the nucleoli in the same way as intact fibrillarin. Moreover, the phase transition of the fibrillarin LC domain is required for the interaction of fibrillarin with other RNA binding proteins such as FUS, TAF15, DDX5 and DHX9. Taken together, we can conclude that the phase transition of the fibrillarin LC domain is required for proper nucleolus localization of fibrillarin and its binding to different kinds of regulatory proteins.
Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200847/903281/bcj-2020-0847.pdf by guest on 09 February 2021 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20200847

Cell culture and transfection
HeLa and HEK-293T cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific, USA) and 1% penicillin/streptomycin (PS, Thermo Fisher Scientific, USA). Cells were grown at 37˚C and supplemented with 5% CO 2 . For transient transfection, Lipofectamine 2000 reagent (Thermo Fisher Scientific, USA) was used according to the manufacturer's protocol.

Cloning
To generate plasmids for the bacterial expression of the fibrillarin LC domain, the DNA fragment encoding residues 2-77 of fibrillarin was amplified by PCR from the cDNA library obtained from HEK-293T cells as a template and subcloned into pHis-mCherry or pHis-GFP parallel vectors to add N-terminal mCherry or GFP tags, respectively [24]. For cloning of the LC domain of nucleolin, the DNA fragment encoding residues 661-698 was amplified and subcloned into the pHis-GFP parallel vector. Bacterial expression plasmids for the LC domains of FUS (residues 2-214), TAF15 (residues 2-208), DDX5 (residues 2-167), DHX9 (residues 1151-1270) and hnRNPA2 (residues 181-341) were generated as described in previously published studies [16,18]. For generation of the mammalian expression plasmids, DNA fragments encoding wild-type, F18/40S, FallS or LC fibrillarin were amplified and then subcloned into pDsRed-N1 (C-terminal tag, confocal microscopy) or pCMV10-3XFlag (N-terminal tag, immunoprecipitation) vectors.

Protein purification
The recombinant proteins were purified using the method described in previously published studies [16,18]. E. coli BL21 (DE3) cells were transformed with the bacterial expression plasmids and the transformed bacterial cells were spread onto Luria broth (LB) agar plates containing 100 g ampicillin. A single colony was inoculated in 30 ml of LB Amp media and stored in 37˚C without shaking. After overnight incubation, 5 ml of the precultured bacterial cells were transferred into 1 L of LB Amp media and grown until an OD 600 of 0.7-1.0 was reached. To induce protein expression, 0.5 mM IPTG was added to the bacterial culture and the cells were incubated at 18˚C overnight. After centrifugation, the cell pellets were resuspended in lysis buffer containing 50 mM Tris-HCl pH 7.5, 500 mM NaCl, 20 mM -mercaptoethanol (BME), 1% Triton X-100, protease inhibitor cocktail (Sigma-Aldrich, Germany) and 0.4 mg/ml lysozyme. After incubating on ice for 30 min, the cells were sonicated at 60% power for 3 min (Model 705 Sonic Dismembrator, Fisher, USA) to lyse the bacterial cells. After centrifugation at 20,000 rpm for 1 h at 4˚C, the supernatants were mixed with Ni-NTA resin (Qiagen, Germany) by gentle rocking at 4˚C for 30 min. The Ni-NTA resin was packed in a glass column (Bio-Rad, USA) and then washed with 300 ml of wash buffer containing 20 mM Tris-HCl pH 7.5, 500 mM NaCl, 20 mM imidazole, 20 mM BME and 0.1 mM PMSF. The bound proteins were eluted from the resin using elution buffer containing 20 mM Tris-HCl pH 7.5, 500 mM NaCl, 200 mM imidazole, 20 mM BME and 0.1 mM PMSF. The purified recombinant proteins were stored at -80˚C until further use.

Liquid-like droplet assays
To make the LLDs, the purified fibrillarin LC domain proteins linked to mCherry were dialyzed in gelation buffer containing 20 mM Tris-HCl pH 8.5, 200 mM NaCl, 20 mM BME and 0.1 mM PMSF overnight at room temperature. The dialyzed proteins were diluted to the indicated concentration in gelation buffer without NaCl to obtain the indicated salt concentration. Tev protease (Promega, USA) was added to the reaction mixtures to cleave off the N-terminal Histags. The samples were transferred to a glass-bottomed 96 well plate pre-coated with 3% BSA (Perkin Elmer, USA). After 30 min of incubation at room temperature, the LLDs were visualized using light (DMi8, Leica, Germany) or confocal microscopy (LSM510, Zeiss, Germany). For the liquid-like droplet (LLD) incorporation assays, GFP-linked fibrillarin LC domain proteins (0.2 M) were added to the LLD formation mixtures containing 5 or 2 M of mCherry:FBL-LC. The intensity of the incorporated GFP proteins was analyzed using confocal microscopy (LSM510, Zeiss, Germany).

Hydrogel formation
Hydrogel droplets were generated as described in previously published studies [16,18]. The mCherry-linked LC domains of fibrillarin, FUS, TAF15 or DHX9 were dialyzed in gelation buffer overnight at room temperature. Upon brief sonication (1 s, three times, Model 705 Sonic Dismembrator, Fisher, USA) and centrifugation (3 min at maximum speed) to remove the precipitates, the dialyzed proteins were concentrated to approximately 70-100 mg/ml using Amicon Ultra (Merck Millipore, USA). Small droplets (0.4 l) of the protein solution were made onto the glass-bottomed 35 mm confocal dishes (SPL, Korea) and these were then incubated for 7-8 d at room temperature until gelation occurred. For the hydrogel binding assays, GFP-linked soluble proteins at a concentration of 3 M were added to the hydrogel droplets and stored overnight at 4˚C. The intensity of the trapped GFP-linked proteins was assessed using confocal microscopy.

Cell imaging
HeLa cells expressing dsRed-linked fibrillarin constructs were fixed with 4% paraformaldehyde (PFA) for 15 min at room temperature. Upon permeabilization using 0.5% Triton X-100 in PBS (PBS-T), the cells were incubated overnight with anti-nucleophosmin antibodies at 4˚C. The cells were then washed with 0.1% PBS-T and incubated with Alexa Fluor 488-conjugated secondary antibodies for 1 h at room temperature. The cells were analyzed using confocal microscopy.

Immunoprecipitation and Western blotting
HEK-293T cells were transfected with plasmid constructs of wild-type, F18/40S, FallS or LC fibrillarin genes with N-terminal Flag tags. Following incubation for 48 h at 37˚C, the transfected cells were collected and lysed by sonication in 500 l of IP buffer containing 50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40 and 1 mM EDTA and a protease inhibitor cocktail (Thermo Fisher Scientific, USA). After centrifugation at 4˚C for 10 min at maximum speed, the supernatants were collected and incubated with 2 g of anti-Flag antibodies. Upon overnight rotation at 4˚C, the immune complexes were supplemented with 100 l of the protein G magnetic beads (Bio-Rad, USA) for 2 h and then washed three times with ice-cold PBS. The bound proteins were eluted with 2X SDS sample buffer by boiling at 95˚C for 10 min. For Western blotting, the eluted samples were separated using 10% SDS polyacrylamide gels and the separated proteins were transferred onto the nitrocellulose membrane and detected using the indicated antibodies.

Phase transition of the N-terminal LC domain of fibrillarin
To study the role of the N-terminal LC domain and its phase transition in the nucleolar distribution of fibrillarin, full-length or N-terminally deleted fibrillarin were first expressed in the HeLa cells. As shown in Figure 1A, the full-length fibrillarin with C-terminal dsRed tag (FBL-FL:dsRed) was evenly localized in the nucleolar DFC and FC regions that are surrounded by the GC region in which nucleophosmin is predominantly localized. However, the localization of fibrillarin that was deleted from the N-terminal LC domain (FBL-LC:dsRed) was completely different from the full-length protein. Although still located inside the nucleoli, FBL-LC:dsRed exhibited a dot-like expression pattern ( Figure 1B). Knowing that the LC domain is required for proper nucleolar distribution of fibrillarin, we next validated the phase transition of the fibrillarin LC domain using the recombinant protein of the N-terminal LC domain (amino acids 2-77) of fibrillarin linked to the N-terminal mCherry fluorescent protein (mCherry:FBL). The solution of 5 M of mCherry:FBL with 100 mM of NaCl was clear at room temperature and when observed under light microscopy, only a small amounted of aggregated particles were observed on the surface of the glass slide ( Figure 1C, left panel). Upon dilution of the NaCl concentration to 50 or 25 mM, the mCherry:FBL solution became turbid and the formation of LLDs with a spherical structure was observed by light microscopy ( Figure 1C, middle and right panels). A reverse correlation was observed between the NaCl concentration and the size of the observed LLDs. These results are in agreement with the results of previous research on the phase transition of fibrillarin in vitro [15,22].
Previously, it has been reported that the LC domains of a variety of RNA binding proteins, including FUS, TAF15, and hnRNPA2, can undergo phase transition to become hydrogel droplets that are composed of reversible amyloid-like polymers [16,18,20,25]. Upon increasing the concentration up to 100 mg/ml and incubation at room temperature, the pure solution of the mCherry:FBL LC domain recombinant protein also became hydrogel droplets composed of a bunch of polymers that were visible using fluorescent microscopy (Supplementary Figure S2E). It was also reported in previous studies that the labile amyloid-like polymers and hydrogel droplets of different kinds of LC domains are specifically melted by the aliphatic alcohol, 1,6-hexanediol (HD), but not by the other, 2,5-HD [26]. The mCherry:FBL LC hydrogel droplets were also melted by 1,6-HD, but not by 2,5-HD, suggesting that the hydrogel droplets composed of the fibrillarin LC domain are composed of labile, amyloid-like polymers ( Figure 1D).

Phenylalanine residues are critical to the phase transition of the fibrillarin LC domain
Many LC domains that undergo phase transition often contain repeats of aromatic amino acids, tyrosine and/or phenylalanine residues. In previous studies, it was revealed that repetitive tyrosine or phenylalanine residues are critical to the phase transition and function of LC domains [16,18,20,27]. Tyrosine-to-serine (Y-to-S) substitutions in the LC domain of FUS or TAF15 effectively disrupt the formation of the amyloid-like polymers, as well as the transcriptional activity of the LC domains [18]. In the case of hnRNPA2, a single phenylalanineto-serine (F-to-S) substitution was sufficient to interrupt the incorporation of the LC domain to the pre-existing liquid-like droplets or hydrogel droplets [20]. The LC domain of fibrillarin contains six repeats of the FG (phenylalanine/glycine) motifs (Supplementary Figure S1A). To identify the phenylalanine residues responsible for the phase transition of the fibrillarin LC domain, six mutant LC domains with a single F-to-S (F5S, F12S, F18S, F30S, F40S, and F64S) substitution, two with a double F-to-S (F18/40S and F30/64S) substitution, two with a triple Fto-S (F5/18/40S and F12/30/64S) substitution and one in which all six phenylalanine residues were substituted to serine residues (FallS) were generated. Once recombinantly produced in the form of the fusion protein with the N-terminal GFP, wild-type or mutant LC domains were applied to the chamber slides containing hydrogel droplets composed of the mCherry-linked fibrillarin LC domain. After overnight incubation, the hydrogel trapping of the wild-type or mutant GFP-tagged LC domains of fibrillarin was analyzed using confocal microscopy. As shown in Figure 2A and B, the intensity of the hydrogel binding was decreased in all mutant LC domains when compared to the wild-type, and in particular, LC domains with F18/40S, F5/18/40S, or FallS substitutions exhibited the lowest trapping intensity, suggesting that residues F18 and F40 are critical to trapping by the hydrogel droplets. We then tested the incorporation of two mutant LC domains, F18/40S and FallS, into the LLDs composed of the wild-type fibrillarin LC domain. For this, LLD formation samples containing 5 or 2 M of the mCherry-linked fibrillarin LC domain at 50 or 25 mM NaCl, respectively, were prepared in test tubes ( Figures 2C and D). GFP-only or wild-type, F18/40S or FallS LC domains of fibrillarin linked to GFP (0.2 M) were added to the LLD samples and the mixtures were then incubated in the chamber slides. Using confocal microscopy, as in the hydrogel binding assay, the robust incorporation of the GFP-linked wild-type LC domain into the mCherry LLDs was observed and the incorporation intensity was drastically decreased for the GFP-linked LC domains harboring F18/40S or FallS mutations (Figures 2C and D). No LLD incorporation was observed for the GFPonly ( Figures 2C and D, top panels). We also confirmed that all the GFP-linked proteins tested possess the same ground-level fluorescent properties and purity (Supplementary Figures S1B  and C).
Next, to test the effect of the F-to-S substitutions on the phase transition of the fibrillarin LC domain, the mCherry-linked wild-type LC domain of fibrillarin at varying concentrations ranging from 1 to 20 M was incubated in a buffer containing 100, 50, 25, or 12.5 mM NaCl. LLD formation was observed using microscopy at all protein concentrations (1 to 20 M) in the buffer containing 12.5 mM NaCl, 2 to 20 M of proteins at 25 mM NaCl, 5 to 20 M of proteins at 50 mM NaCl, and only 100 M protein at 100 mM NaCl (Figures 3A, C and Supplementary Figure S2A). In all cases, the lower the NaCl concentration and the higher the protein concentration, the larger the size of the LLDs. In samples with 10 M protein in 12.5 mM NaCl and 20 M protein in 12.5, 25, and 50 mM NaCl, LLD formation was very robust so that the bottom of the chamber slides was completely covered with droplets within 20-30 min of assembling the droplet samples ( Figure 3A and Supplementary Figure S2A). Although still capable of making LLDs, the mCherry-linked LC domain with the F18/40S mutation failed to form droplets at proteins concentrations of 1, 2, 5 or 20 M in buffers containing 12.5, 25, 50 or 100 mM NaCl, respectively (Figures 3B and C and Supplementary Figure S2B). In addition, under all conditions, the droplets formed by the F18/40S LC domain were smaller in size than the ones made by the wild-type LC domain. No droplets were assembled by the LC domain containing the FallS mutation. As shown in Supplementary Figure S2C and D, the wild-type LC domain formed LLDs at protein concentrations of 5 to 20 M under a physiological salt concentration (150 mM NaCl) in the presence of 5% polyethylene glycol (PEG), while the F18/40S mutant LC domain formed smaller droplets at higher protein concentrations (10 to 20 M). When subjected to hydrogel or polymer formation, mCherry-linked fibrillarin LC domains with F18/40S or the FallS mutation failed to make polymers, as well as hydrogel droplets, as did the wild-type LC domain ( Supplementary Figures S2E and F). These results suggest that the two phenylalanine residues, F18 and F40, in the LC domain are critical for the phase transition and also for partitioning of the LLDs or hydrogel droplets.

Role of the LC domain phase transition in nucleolar localization of fibrillarin
We then investigated whether the F-to-S mutation that interrupted the phase transition of the LC domain also has an effect on the nucleolar localization of fibrillarin. To test this, full-length fibrillarin containing F18/40S or FallS LC domains with C-terminal dsRed tags were expressed in HeLa cells. Unexpectedly, fibrillarin containing mutant LC domains was well-localized in the nucleolar DFC regions as wild-type fibrillarin (Supplementary Figure S3). This might be because the mutant LC domains (F18/40S or FallS) are still trapped by the existing phase-separated structure in the nucleoli formed by the endogenous fibrillarin proteins. To test this possibility, HeLa cells were transfected with wild-type or mutant fibrillarin constructs with C-terminal dsRed tags in the presence or absence of the endogenously expressed fibrillarin. Regardless of the presence or absence of endogenous fibrillarin, fibrillarin without a LC domain (LC) were localized to dot-like puncta in the nucleoli (Figures 4A and B, 2 nd rows). As shown in Figure 4A, we again observed an almost normal nucleolar distribution of the fibrillarin containing F18/40S or FallS LC domains when they were expressed in cells that had been pre-transfected with scrambled siRNA. However, when the expression of endogenous fibrillarin was suppressed by the siRNA-mediated knock down (fibrillarin siRNA, Supplementary Figure S4), an aberrant, dotlike expression pattern was observed for the exogenously expressed fibrillarin with mutant LC domains ( Figure 4B). Thus, these results suggest that the phase transition of the LC domain is critical for proper nucleolar distribution of fibrillarin.

Phase transition is required for the interaction of fibrillarin LC domains with RNA binding proteins
Having observed that the phase transition of the LC domain is required for nucleolar distribution of fibrillarin, we next tested whether the phase transition is also critical for fibrillarin to interact with other proteins. To test this, trapping of the GFP-linked LC domains from different kinds of RNA regulatory proteins was first analyzed using mCherry hydrogel droplets composed of the fibrillarin LC domain. As shown in Figure 5A, strong binding of the Nterminal LC domain of FUS was observed and the LC domains of TAF15, DHX9 or nucleolin were trapped at a moderate intensity by the hydrogel droplets. The LC domains of DDX5 or hnRNPA2 were only weakly bound to the hydrogel droplets ( Figure 5A). Next, the trapping of the GFPlinked wild-type or mutant LC domains of fibrillarin by hydrogel droplets composed of FUS, TAF15, DDX5 or DHX9 LC domains fused to mCherry was analyzed ( Figure 5B). Although the wild-type fibrillarin LC domain was readily trapped by all of the hydrogel droplets challenged, the hydrogel binding intensity was decreased for LC domains harboring F18/40S or FallS mutation ( Figure 5B).
Although, the mutations that interrupted the phase transition of the LC domain also interfered with the interaction between the LC domains of fibrillarin and other RNA regulatory proteins in vitro, it was still unclear whether the phase transition capability of the LC domain is required for fibrillarin to interact with other proteins in living cells. To test this, Flag-tagged wild-type, F18/40S, FallS or LC fibrillarin proteins were subjected to the immunoprecipitation assay in HEK-293T cells. As shown in Figure 5C, the interaction between the exogenously expressed wild-type fibrillarin and the RNA regulatory proteins including FUS, TAF15, DDX5 or DHX9 was observed. However, these interactions were disrupted by F18/40S, FallS or LC mutations, indicating that the phase transition of fibrillarin is critical for the fibrillarin to interact with different kinds of RNA dependent regulatory proteins in living cells.

Discussion
Although phase transition has been observed for many proteins, the molecular code that drives the phase separation often remains enigmatic [23]. This has been a major barrier for understanding the role of phase transition in the physiological function of proteins. In the present study, we uncovered the sequence determinants for fibrillarin phase properties in order to study the roles of LC domain phase transition in fibrillarin function (Figure 2). Previous studies have shown that the LC domain of phase-separating proteins often contains repeats of aromatic amino acids, tyrosine or phenylalanine [16][17][18]20]. The tyrosine residues are usually flanked by either glycine or serine residues (Y-motifs) while phenylalanine residues have a glycine residue at their amino or carboxyl sides (FG motifs). For the RNA binding proteins including FUS, TAF15, and heterogeneous nuclear ribonucleoprotein A2 (hnRNPA2), these repetitive tyrosine and phenylalanine residues are critical to phase separation or recruitment into the phase separated structures [16,18,20]. Fibrillarin possesses six repeats of the FG motifs in its LC domain. To determine if the phenylalanine residues in the fibrillarin LC domain are also critical for the phase transition, we generated mutant LC domains harboring one or more of the F-to-S substitutions. As shown in Figure 2, substitution of both F18 and F40 with serine residues was sufficient to interfere with the incorporation of the fibrillarin LC domain into both LLDs and hydrogel droplets. Likewise, the capability of phase separation was reduced by F18/40S and completely abolished by all F-to-S mutations (Figure 3 and supplementary Figure S2). Thus, our findings indicate that the phenylalanine residues in the LC domain are critical to the phase transition of fibrillarin.
One of the main concerns in the present study is whether the phase transition is important for proper nucleolar localization of fibrillarin. Without the N-terminal region, fibrillarin cannot localize to the DFC of the nucleoli, suggesting that the LC domain is required for the normal nucleolar distribution of fibrillarin ( Figure 1). To verify the role of phase transition in fibrillarin localization, we introduced mutant fibrillarin with F-to-S substitutions into cells to replace the endogenous protein. When dsRed-linked wild-type, F18/40S or FallS fibrillarin constructs were expressed in HeLa cells pre-suppressed with the expression of the endogenous fibrillarin, F18/40S and FallS fibrillarin showed aberrant, dot-like expression patterns, while wild-type fibrillarin was evenly localized in the nucleolar DFC regions ( Figure 4B). The correlating effects of the mutations of the FG repeats on the phase transition in vitro and the localization to nucleolar DFC in HeLa cells suggest that the localization of fibrillarin may be driven primarily by the phase transition of the LC domain. When F18/40S and FallS fibrillarin were expressed in HeLa cells without knockdown of the endogenous fibrillarin, however, they can associate with the nucleolar DFC regions as observed in Figure 4A and Supplementary Figure S3. The F18/40S and FallS mutants of the fibrillarin LC domain lack the capability to induce phase transition (Figure 3 and Supplementary Figure S2), but partially retain their ability to bind to either the LLDs or hydrogel droplets composed of wild-type LC domains ( Figure 2). Thus, it is plausible that exogenously expressed mutant fibrillarin proteins may be trapped by the pre-existing phase-separated structure formed by the endogenous fibrillarin within the nucleolar DFC region in cells. These observations are consistent with previous published research showing that the FUS mutants that lack the capability to form hydrogels partially retain their ability to bind to the wild-type FUS LC hydrogel in vitro and to stress granules in living cells [16].
In addition, phase transition appears to be critical to the interaction of fibrillarin with different kinds of RNA binding proteins ( Figure 5). Using a hydrogel binding assay, we demonstrated that the LC domain of fibrillarin and RNA binding proteins including FUS, TAF15, DDX5 and DHX9 recruited each other, while the phase separation mutants (F18/40S or FallS) of fibrillarin showed a reduced binding affinity to the hydrogel droplets of the LC domain for those RNA binding proteins. Consistent with the results from in vitro experiments, pull-down studies performed in living cells showed that wild-type fibrillarin interacted with FUS, TAF15, DDX5 or DHX9, while the phase separation mutants (F18/40S or FallS) did not ( Figure 5C). The results indicate that the ability of fibrillarin to interact with FUS, TAF15, DDX5 or DHX9 correlates with its phase separation capability. The nucleolus is a dynamic structure and its composition can vary dramatically under different cellular conditions [4]. Phase transition is one of the proposed mechanisms for the dynamic movement of RNA-binding proteins in and out of the nucleolus [21]. Liquid-like droplets show selective properties, admitting some proteins and RNAs and excluding others. However, there is a significant lack of information available about the role of phase separation in protein-protein interactions in the nucleolus. Our results suggest that the phase transition of the fibrillarin LC domain can control the organization of the nucleolar composition by recruiting specific RNA binding proteins. Whereas the phase-separated structures of the LC domain can trap specific molecules, it is not known how these proteins are released from fibrillarin in the nucleoli. In a previous study, we demonstrated that the binding of RNA Polymerase II to the FET (FUS/EWS/TAF15) protein LC domains is regulated in a phosphorylation-dependent manner, indicating posttranslational modification as a means of controlling the dynamic behavior of the LC derived phase-separated structures [18]. Moreover, the SR (serine-arginine rich) splicing factors first enter the nucleoli in a hypophosphorylated state, and then re-localize to nuclear speckles upon phosphorylation by CDC2-like kinases (CLKs) [28]. Thus, it is plausible that posttranslational modification of the LC domains in the RNA binding proteins might trigger their release from the phase-separated structures of fibrillarin.
Many of the proteins currently used as models for phase separation are linked to membraneless organelles such as nucleoli, Cajal bodies, and stress granules [29]. hnRNPA1and FUS are the most extensively studied. The RNA binding proteins, hnRNPA1 and FUS, assemble into stress granules via a phase-separation mediated process [16,30]. In hnRNPA1 and FUS, the aromatic residues phenylalanine and tyrosine in the LC domain contribute to the phase separation [16,30]. Further characterization reveals that the phase separation of hnRNPA1 and FUS can be tuned using the environmental conditions; in particular, a lower salt concentration, molecular crowding, and the interaction with RNA promote the phase separation [16,31]. We observed that the phase separation of fibrillarin is dependent on the phenylalanine residues of the LC domain and environmental factors such as the salt concentration and molecular crowding. This indicates that the features of the sequence and the physical properties we have identified for fibrillarin phase separation may have more general implications for the phase separation of membraneless organelle proteins.
The results presented here provide the first direct evidence of the functional role of phase separation in the control of subnucleolar localization and protein interactions of fibrillarin. Fibrillarin plays a key role in the biogenesis of ribosomes, thus, any changes in the expression or localization of fibrillarin can lead to cell abnormalities often connected with cancer [12]. Hence, a mechanistic understanding of fibrillarin regulation may elucidate the role of fibrillarin in health and disease, contributing to novel treatments for diseases, such as cancer.     A and B) Plasmid constructs expressing wild-type, LC, F18/40S or FallS fibrillarin with C-terminal dsRed tags were transfected into the HeLa cells in the presence (A) or absence (B) of the endogenously expressed Fibrillarin. Cells were fixed and stained using anti-nucleophosmin and DAPI to visualize nucleoli and nuclei, respectively. Subcellular localization of exogenously expressed fibrillarin (red signals) and nucleophosmin (green signals) were visualized by confocal microscopy.