Nucleolin modulates compartmentalization and dynamics of histone 2B-ECFP in the nucleolus

ABSTRACT Eukaryotic cells have 2 to 3 discrete nucleoli required for ribosome synthesis. Nucleoli are phase separated nuclear sub-organelles. Here we examined the role of nuclear Lamins and nucleolar factors in modulating the compartmentalization and dynamics of histone 2B (H2B-ECFP) in the nucleolus. Live imaging and Fluorescence Recovery After Photobleaching (FRAP) of labelled H2B, showed that the depletion of Lamin B1, Fibrillarin (FBL) or Nucleostemin (GNL3), enhances H2B-ECFP mobility in the nucleolus. Furthermore, Nucleolin knockdown significantly decreases H2B-ECFP compartmentalization in the nucleolus, while H2B-ECFP residence and mobility in the nucleolus was prolonged upon Nucleolin overexpression. Co-expression of N-terminal and RNA binding domain (RBD) deletion mutants of Nucleolin or inhibiting 45S rRNA synthesis reduces the sequestration of H2B-ECFP in the nucleolus. Taken together, these studies reveal a crucial role of Nucleolin-rRNA complex in modulating the compartmentalization, stability and dynamics of H2B within the nucleolus.


Introduction
The nucleus houses chromatin and several nonmembranous nuclear bodies involved in transcription [1], splicing [2] and nuclear transport [3]. The absence of membranes within the nucleus facilitates dynamic but regulated exchange of molecules between nuclear bodies and chromatin [4]. The import and sequestration of protein and RNA into nuclear bodies modulates their nucleoplasmic concentration and function. The nucleolus is the largest nuclear sub-organelle essential for ribosome biogenesis [5]. The nucleolus also functions as a stress-sensing compartment that sequesters oncoproteins such as BRCA1 and regulators of p53, that are released into the nucleoplasm upon DNA damage [6,7], while, HSP70 and VHL proteins are immobilized in the nucleolus during thermal stress and acidosis respectively [8]. Key mechanisms of protein sequestration into the nucleolus are (i) interaction of proteins with resident nucleolar factors such as Nucleolin and Nucleophosmin [9][10][11] (ii) nucleolar localization signal (NoLS) enriched in lysine and arginine rich repeats [12] and (iii) interaction of proteins with non-coding RNA transcribed from intergenic sequence of the rDNA [8]. Mass spectrometric analyses of nucleolar extracts identified~4500 proteins, which include isoforms of each histone family -H1, H2, H3, H4 and histone-modifying enzymes [13].
Electron microscopy reveals a remarkable tripartite structure of the nucleolus with a central Fibrillar Center (FC), surrounded by the Dense Fibrillar Component (DFC) and the Granular Component (GC). Such an organization facilitates ribosome biogenesis [5,14]. The nucleolus partitions into sub-compartments as a result of the separation of immiscible phases of Fibrillarin and Nucleophosmin [15]. Nucleolar structure is maintained by ongoing rDNA transcription, as its inhibition by Actinomycin D induces nucleolar segregation [16]. Nucleolar structure is also regulated by Nucleolinone of the most abundant proteins of the GC [17]. Nucleolin has diverse roles in rDNA transcription, ribosome biogenesis [18], DNA damage repair [19] and regulation of apoptosis [20]. In vitro studies implicate Nucleolin as a histone chaperone with FACT-like activity, which regulates SWI-SNF function and ACF chromatin remodelers [21]. Nucleolin has a High Mobility Group (HMG)-like N-terminal domain with four acidic stretches of glutamate and aspartate residues, interspersed with basic lysine residues [22]. The acidic stretches interact with histone H1 while the basic residues interact with DNA [22]. Nucleolin also has four central RNA binding domains (RBD1-4) and a C-terminal GAR (Glycine Arginine Rich) domain. The RNA binding domain specifically binds to a 5′ external transcribed sequence (ETS) site on nascent ribosomal RNA. The GAR domain of Nucleolin binds specifically to DNA and non-specifically to RNA, while the RBDs confer specificity to RNA binding [23][24][25]. ChIP-Seq analysis reveals the recruitment of Nucleolin to sites of DNA damage, resulting in the eviction of histones -H2A and H2B thereby allowing access to the DNA double strand break repair machinery [19]. H2B has been detected in the nucleoli of Bovine liver cells and chicken erythrocytes using antibodies raised against its first 58 amino acids [26]. Localization of H2B in the nucleolus is attributed to stretches of basic amino acid residues (KKRKRSRK), similar to the NoLS motifs: (R/K)(R/K)X(RK) or (R/K)X(R/K)(R/ K) [27].
Here we show the RNA-dependent function of Nucleolin in modulating the localization, dynamics and retention of Histone 2B (H2B-ECFP) in the nucleolus.

Histone 2B (H2B) compartmentalizes in the nucleolus
The nucleolus is the largest nuclear sub-organelle and is essential for ribosomal RNA (rRNA) and protein synthesis [28]. However, the mechanisms that regulate the sequestration of proteins within the nucleolus remain unclear. For instance, overexpressed H2B is sequestered in the nucleolus [27].
Here we sought to investigate the mechanisms that modulate the sequestration and dynamics of H2B-ECFP in the nucleolus. We transfected H2B-ECFP into DLD1 colorectal cancer cells and found that although H2B-ECFP localizes in the nucleoplasm of all cells, a significant sub-population of cells (~40%) show H2B-ECFP in the nucleolus (Figure 1  (a,b)). While, the Nuclear Localization Signal (NLS) sequence tagged with CFP localizes in the nucleolus of nearly all transfected cells (~98%) (Figure 1(a,b)). We surmise that the relatively small NLS-CFP freely diffuses into the nucleolus, while the nucleolar localization of H2B-ECFP in a sub-population of~40% cells, is potentially guided by additional interactions with nucleolar factors. H2B-ECFP localizes in the nucleolus of diverse cancer cell lines such as HCT116 (colorectal cancer cell line), MCF7 (breast cancer cell line) as well as DLD1 cells (Figure 1(c)). In addition to visualizing nucleolar localization of overexpressed H2B-ECFP, we found that endogenous H2B also localizes in the nucleolus as revealed by immunofluorescence assays (Figure 1(d)).

Lamin B1 enhances mobility of H2B-ECFP in the nucleolus
We sought to investigate the dynamics of fluorescently labelled H2B in the nucleolus and nucleus by Fluorescence Recovery After Photobleaching (FRAP) (Fig. S1A, B). Interestingly, photobleaching H2B in the nucleolus showed a significantly higher mobile fraction (M.F.~40%) as compared to the nuclear sub-pool (M.F~18%) (Fig. S1C). While NLS-CFP showed complete and immediate recovery further underscoring its ability to freely diffuse into the nucleus as well as the nucleolus (Fig. S1D, E).
We asked if the compartmentalization of H2B-ECFP in the nucleolus, correlates with endogenous levels of Nucleolin across cell lines ( Figure 4(e-g)). Immunoblotting of whole cell extracts across cell lines showed an increase in Nucleolin levels as follows: CRL1790 < DLD1 < HCT116 < MCF7 (Figure 4(f)). Furthermore, increased nucleolar sequestration of H2B-ECFP positively correlates with an increase in the endogenous levels of Nucleolin in these cell lines (Figure 4(e,g)). We further corroborated this by overexpressing Nucleolin in DLD1 cells, which dramatically increased nucleolar compartmentalization of H2B-ECFP in~67% cells, as compared to control cells (~40%) (Figure 4(g)). In summary, an increase in the endogenous or overexpressed levels of Nucleolin, positively correlates with the extent of H2B-ECFP in the nucleolus and Nucleolin therefore functions as a positive regulator of H2B-ECFP sequestration into the nucleolus.
We sought to investigate into the mechanisms of Nucleolin mediated sequestration of H2B-ECFP in the nucleolus. Towards this end, we examined the effect of co-expressing deletion mutants of Nucleolin into DLD1 cells and scored for H2B-ECFP compartments in the nucleolus. We co-expressed H2B-ECFP with (i) full length NCL FL (ii) NCLΔN (N-terminal deleted) (iii) NCLΔRBD (RBD1-4 deleted) and (iv) NCLΔGAR (GAR domain deleted). We observed a comparable localization of NCLΔN in the nucleolus as that of full length NCL, while NCLΔRBD and NCLΔGAR partially mislocalized in the nucleoplasm, consistent with previous studies (Figure 6(b)) [40][41][42]. Co-expression of full length NCL showed a significant increase in nucleolar H2B-ECFP (~74%), as compared to cells transfected with H2B-ECFP alone (~32%) ( Figure  6(c)). Interestingly, co-expression of NCLΔN and NCLΔRBD did not enhance H2B-ECFP localization in the nucleolus, since both conditions showed~31% cells with nucleolar H2B-ECFP ( Figure 6(c)). In contrast, co-expression of NCLΔGAR showed a comparable extent of nucleolar H2B-ECFP as that of full length NCL (~77%) (Figure 6(c)). Taken together, the N-terminal and RNA binding domains of Nucleolin are essential for the enhanced localization of H2B-ECFP in the nucleolus.
Nucleolin mediated nucleolar localization of H2B-ECFP is pre-rRNA dependent Since NCLΔRBD did not enhance nucleolar H2B-ECFP localization, we determined if rRNA was necessary for NCL mediated localization of H2B-ECFP in the nucleolus. We treated DLD1 cells with Actinomycin D (0.05 µg/ml) for 4 hours, which showed a significant decrease in 45S rRNA levels ( Figure 6(d)).
It is noteworthy that upon Act D treatment, Nucleolin speckles in the nucleoplasm do not colocalize with H2B-ECFP ( Figure 6(e), inset). However, Nucleolin shows a distinctive co-localization with H2B-ECFP in the nucleolus, in the presence of 45S pre-rRNA in the nucleolus.  Taken together, Nucleolin and 45S rRNA are required for the compartmentalization of H2B-ECFP in the nucleolus.

Overexpressed H2B localizes in the nucleolus
The nucleolus is a complex milieu of ribosomal DNA, RNA, proteins and non-ribosomal proteins [5]. Sequestration into the nucleolus is an important mode of post translational regulation of proteins such as ARF and Cdc14 that control cell cycle and apoptosis [43]. Histones and histone variants are commonly enriched in the nucleolus. The histone H1 variant H1.0, localizes in the nucleolus and is strongly associated with non-transcribed regions of ribosomal DNA and interacts with nucleolar proteins involved in rRNA processing [44,45]. Another histone variant -macroH2A also localizes at the nucleolus and is directly involved in rDNA repression [46]. Histone 2A methylated by Fibrillarin at Q104 in humans and Q105 in yeast, is exclusively localized in the nucleolus [47]. However, H2B transiently localizes in the nucleolus upon transfection and disperses into the nucleoplasm over time, either integrating or exchanging with nuclear chromatin [27]. In vitro, a higher concentration of histone octamers to DNA (>0.76 mass ratio), aggregates chromatin and inhibits transcription [48]. Furthermore, excess histone expression in budding yeast shows cytotoxicity and is deleterious to these cells [49,50]. We surmise, that the nucleolar sequestration of excess H2B, is a preferred paradigm for preventing the potentially deleterious effects of histone overexpression in the nucleus and toxicity across most cell types.

Lamins as modulators of nuclear histone dynamics
Histones are hyperdynamic in ES cells which have a relatively open chromatin conformation [51]. Histone dynamics is dampened during differentiation and lineage commitment, as chromatin undergoes compaction. Lamin A/C levels are relatively lower in ES cells but increase during differentiation [52]. Consequently, Lamin A overexpression in ES cells, restricts histone H1 mobility [29]. Furthermore, Lamin B1 expression is lower in senescent cells with compact chromatin and Senescence Associated Heterochromatic Foci (SAHF) [53]. Lamin depletion in differentiated DLD1 cells, did not show an appreciable effect on H2B-ECFP dynamics in the nucleoplasm (Figure 2). This is consistent with relatively unaltered chromatin dynamics in differentiated cells upon masking of the histone binding domain of Lamin A/C [54]. We envisage the following scenarios of the role of Lamins in the modulation of histone dynamics -(1) Consistent with previous data, reduced expression levels of Lamin A/C or B-type lamins do not appreciably affect histone mobility in differentiated cells (Figure 2) [29,52] (2) It is likely that the combined depletion of Lamin A/C and B-type Lamins, alter histone mobility, in differentiated cell types (3) Lamin interactors such as Emerin, Lamin B receptor (LBR) and barrier to autointegration factor (BAF) with histone binding domains, maintain histone dynamics in the absence of Lamins [55,56].
On the other hand, Nucleostemin is highly expressed and is a marker of cancer stem cells [57]. Furthermore cancer stem cells show increased DNA accessibility as assessed by formaldehyde-assisted isolation of regulatory elements-sequencing (FAIRE-seq), suggesting open chromatin conformation [58,59]. We surmise that the decrease in H2B-ECFP mobility in the nucleoplasm upon Nucleostemin loss suggests reduced accessibility to chromatin in cancer stem cells. Interestingly, independent knockdowns of Lamin B1, Fibrillarin and Nucleostemin enhance H2B-ECFP mobility in the nucleolus (Figure 3). We surmise that Fibrillarin and Nucleostemin are bonafide nucleolar factors, that control the nucleolar microenvironment, as their depletion enhances H2B-ECFP dynamics to a significantly greater extent than nuclear lamin B1 (Figures 2 and 3). Furthermore, the loss of Fibrillarin, Nucleostemin or Lamin B1, potentially alter the relative stoichiometries of bound and unbound sub-fractions of H2B-ECFP with nucleolar chromatin and consequently enhance histone dynamics in the nucleolus [30,35,60].

Nucleolin modulates H2B-ECFP localization into the nucleolus
Nucleolin exhibits a dominant role in sequestering H2B-ECFP into the nucleolus (Figure 4). Nucleolin is a high mobility group protein and is a major constituent of the granular component of the nucleolus [61]. Nucleolin is involved in rRNA transcription and processing [18,62]. Nucleolin is closely related to another nucleolar phosphoprotein -Nucleophosmin. Phase separation of Nucleophosmin and Fibrillarin to a relatively more viscous nucleolar phase is critical to the maintenance of nucleolar integrity [15,63]. In addition, ribosomal proteins -L3 and S3A and non-ribosomal proteins -Lamin B2 and HIVrev, localize into the nucleolus by virtue of their interaction with Nucleolin and Nucleophosmin [9,31,64]. H2B is localized into the nucleolus through its nucleolar localization signal (NoLS) and electrostatic interaction with nucleolar components [27]. Here, we discovered the requirement of Nucleolin for the sequestration and retention of H2B-ECFP in the nucleolus. Nucleolin plays a more dominant role in the localization of H2B-ECFP in the nucleolus, since the loss of Nucleolin strikingly decreases nucleolar H2B-ECFP, while the co-expression of Nucleolin, retains H2B-ECFP in the nucleolus over a considerably longer duration ( Figure 4). More importantly, the N-terminal domain, previously shown to interact with histones H1 and H2A-H2B dimers and the RNA binding domain of Nucleolin, are indispensable for the nucleolar retention of H2B-ECFP [21,22] (Figure 6). Taken together, the interaction between H2B-ECFP and Nucleolin in the nucleolus serves as a mechanism for the nucleolar localization and retention of overexpressed H2B.

Nucleolin modulates nucleolar H2B-ECFP dynamics
Nucleolin levels modulate H2B-ECFP retention and dynamics in the nucleolus across cell types (Figures 4 and 5). Furthermore, the N-terminal domain and RBD of Nucleolin regulate H2B-ECFP compartmentation in the nucleolus ( Figure 6). Nucleolin functions as a histone chaperone facilitating exchange of H2A-H2B dimers from chromatin [21,39]. However, nucleoplasmic and nucleolar H2B exist in distinct microenvironments. The nucleoplasmic pool of H2B largely associates with DNA, whereas nucleolar subpools of H2B reside in the microenvironment of nucleolar DNA, ribosomal RNA, non-coding RNAs such as snoRNAs, ribosomal and non-ribosomal proteins, which may collectively impinge on H2B dynamics in the nucleolus.
We surmise that the N-terminal domain of Nucleolin rich in acidic amino acid stretches binds to nucleoplasmic H2B-ECFP and transports it to the nucleolus (Figure 7) [21,22]. Thus, with increased Nucleolin expression, there is enhanced H2B-ECFP import into the nucleolus, which correlates with an increase in the recovery of H2B-ECFP ( Figure 5). We surmise that the enhanced retention of H2B-ECFP in the nucleolus upon NCL overexpression is also rRNA dependent. However, Act D treatment redistributes a subpopulation of Nucleolin to the nucleoplasm, potentially resulting in the lowered retention of H2B in the nucleolus. The RNA binding domains of Nucleolin specifically interacts with the 5ʹ-ETS of pre-rRNA while GAR domain of Nucleolin non-specifically binds to any RNA [25,65]. In summary, the nucleolar retention of H2B-ECFP is dependent upon Nucleolin-45S rRNA complex. Therefore, the sub-domains of Nucleolin differentially affect nucleolar H2B-ECFP compartmentation. While the N-terminal domain is potentially required for translocating H2B-ECFP to the nucleolus, the RNA binding domain is also necessary for the retention of H2B-ECFP in the nucleolus.

Implications
While it was previously proposed that overexpressed H2B localizes in the nucleolus, via charge based interactions between the positively charged H2B and the negatively charged nucleic acids within the nucleolar milieu, our studies for the first time unravel a novel Nucleolin guided mechanism that modulates the sequestration, retention and dynamics of H2B in the nucleolus Figure 7. Speculative model of Nucleolin regulating nucleolar compartmentation and dynamics of H2B-ECFP. 1. Nucleolin interacts with H2B-ECFP via its N-terminal domain and shuttles it into the nucleolus. In the nucleolus, Nucleolin binds to pre-rRNA via its RNA binding domain and H2B-ECFP via its N-terminal domain, thus retaining H2B-ECFP in the nucleolus. It is likely that the relative rate of import of H2B-ECFP into the nucleolus is greater in the presence of Nucleolin. 2. In absence of the N-terminal domain, Nucleolin does not bind to H2B-ECFP, thereby reducing nucleolar pools of H2B-ECFP. 3. Nucleolin RBD deletion mutant binds to H2B-ECFP through its N-terminal domain and sequesters H2B-ECFP into the nucleolus. However, in the absence of RBD, H2B-ECFP is not retained in the nucleolus, as the RBD is required for binding to pre-rRNA. 4. GAR domain deletion mutant binds to H2B-ECFP and pre-rRNA and shows enhanced recruitment of H2B-ECFP into the nucleolus, similar to full length Nucleolin. 5. Nucleolin imports H2B-ECFP in the nucleolus but is unable to retain it in the nucleolus in the absence of pre-rRNA transcription inhibited by Act D. [27]. This implicates Nucleolin, 45S rRNA and potentially other bonafide nucleolar factors namely Nucleophosmin in directing the fate of overexpressed and therefore excess nuclear proteins such as histones, into the nucleolus. Considering that the nucleolus is maintained as discrete phase separated entities in the nucleus, the mechanisms involved in targeting nuclear factors into or out of the nucleolus are largely unclear [12]. Histone gene expression is tightly regulated and coupled to DNA replication during S-phase [50,66]. Imbalances in histone expression and its accumulation can induce G1 cell cycle arrest, genomic instability and affect transcription [67][68][69]. It is therefore conceivable that Nucleolin/ Nucleophosmin are specifically involved in dual roles of chaperoning out excess nuclear proteins such as histones into the nucleolus. Furthermore, this study also unravels the key involvement of ribosomal RNA as an essential mediator that facilitates the retention of nucleolar H2B. A combination of nucleolar factors and their interaction with rRNA is potentially involved in the generation of the phase separated nucleolusa unique nonmembranous milieu within the nucleoplasm, for the rapid but regulated entry and exit of factors that potentially facilitate rRNA biogenesis. In summary, this study unravels a unique and novel mechanism whereby proteins are guided and retained into phase separated systems such as the nucleolus. This suggests potential implications towards the targeted therapeutic intervention of dysregulated 'cancer nucleoli'.

Materials and methods
Plasmids H2B-ECFP [70], GFP-nucleolin [71], GFP-NPM1 and NLS-CFP plasmids were kind gifts from Jennifer Lippincott-Schwartz, Sui Huang and Tom Misteli, respectively. NCLΔN, NCLΔRBD and NCLΔGAR plasmids were generated from GFPnucleolin by restriction free (RF) cloning. Primers and DNA transfections were performed sequentially -siRNA transfection was performed as mentioned previously, while DNA transfections were performed after 24 h and cells were imaged by fluorescence microscopy, after 48h of siRNA transfection at 37°C.
For co-immunoprecipitation (Co-IP) assays,~10 7 cells (DLD1) were lysed in co-IP lysis buffer (50 mM Tris [pH 7.4], 150 mM NaCl, 0.5% NP-40, 1× PIC) vortexed and incubated on ice for 15 min, and centrifuged at 12,000 rpm and 4°C for 10 min. The lysate was precleared by incubating with Dynabeads protein A (Invitrogen, 10002D) for 1 h. Anti-Nucleolin antibody (ab22758, 3 µg) or normal rabbit IgG was incubated with lysates overnight at 4°C. Protein A beads, pre-blocked with 0.5% BSA, were incubated with the immunocomplex for~3 h. Beads were washed 6 times with co-IP lysis buffer to minimize non-specific binding. Bound protein was eluted from the beads by boiling in 2× Laemmli buffer for 15 min at 95°C. Samples were resolved on a 15% SDS-PAGE, followed by western blotting.

Fluorescence Recovery after Photobleaching (FRAP) experiments and analysis
A Zeiss LSM710 confocal microscope equipped with a heated stage at 37°C, was used for all photobleaching experiments and fluorescence image acquisitions. For live imaging, cells were grown on a 22 × 22 mm 2 coverslip glued onto a 35 mm petridish coated with 100 μg/ml Collagen (BD Biosciences, 354236), CO 2 independent Leibovitz L-15 medium (Gibco, 21083-027) supplemented with 10% FBS (complete L15), was used during microscopy. Images were acquired using a 63X oil immersion objective, NA 1.4 at 2.5X digital zoom, at 2% laser power to avoid photobleaching. The acquisition parameters were adjusted to avoid bleed-through of ECFP and GFP fluorescence. A 10 pixel X 10 pixel square (1 pixel = 0.11 µm) Region of Interest (ROI) was bleached in both nucleoplasmic and nucleolar H2B-ECFP. Photobleaching was performed using the 405 nm laser line at 100% power. Laser iterations of 120 and 150 were used to photobleach labelled H2B in the nucleus and nucleolus respectively. Images were collected every 3.87 s for a total duration of 5 min. Images were analyzed using Zen 2011 FRAP Analysis module and normalized fluorescence intensity (NFI) was calculated as follows: where, ROI1 is the fluorescence intensity of the 10 px X 10 px ROI that is bleached, ROI2 is the whole nucleus fluorescence intensity and ROI3 is the fluorescence intensity of a 10 px X 10 px background region selected outside the nucleus. ROI1 (t): post-bleach fluorescence intensity at time t. ROI2(t) and ROI3(t): whole nucleus and background, respectively. ROI1(t = 0): average pre-bleach fluorescence intensity. ROI2(t = 0) and ROI3(t = 0): whole nucleus and background, respectively. The NFI was plotted as a function of time to generate double normalized FRAP curves.
Mobile fractions of H2B-ECFP were calculated as follows: Where, Ffinal is the NFI at maximum recovery, Fbleach is the NFI at the instant of bleaching and Fpre-bleach is the NFI before bleaching.

Actinomycin D treatment
Cells transfected for 24 h were treated with 0.05 μg/ ml actinomycin D (Act D) in complete medium for 4 h at 37°C with 5% CO 2 after which they were transferred to complete L-15 medium and imaged live. Equivalent volumes of dimethyl sulfoxide (DMSO) were used as vehicle controls.

Cell cycle analyses
Cells were fixed in 70% ethanol (in 1X PBS) and subjected to RNase treatment and propidium iodide staining for 1 hour at 37°C. Cells were scanned on FACS Calibur (BD Biosciences). Cell cycle analyses was performed by ModFit software.

Statistical analysis and graphs
Two-tailed student's t-test was used to compare the number of cells showing nucleolar H2B-ECFP compartments and mobile fractions of H2B-ECFP, p-value <0.05 was considered significant. Graphs were plotted using GraphPad Prism software.