Nucleolar structure connects with global nuclear organization

The nucleolus is a multifunctional nuclear body. To tease out the roles of nucleolar structure without resorting to the use of multi-action drugs, we knocked down the RNA polymerase I subunit RPA194 in HeLa cells by siRNA. Loss of RPA194 resulted in nucleolar-structural segregation and effects on both nucleolus-proximal and distal-nuclear components. The perinucleolar compartment was disrupted, centromere clustering around nucleoli was significantly reduced, and the intranuclear locations of specific genomic loci were altered. Moreover, Cajal bodies, distal from nucleoli, underwent morphological and some compositional changes. In comparison, when the preribosomal RNA-processing factor, UTP4, was knocked down, neither nucleolar segregation nor the intranuclear effects were observed, demonstrating that the changes of nucleolar proximal and distal nuclear domains in RPA194 knockdown cells unlikely arise from a cessation of ribosome synthesis, rather from the consequence of nucleolar-structure alteration. These findings point to a commutative system that links nucleolar structure to the maintenance and spatial organization of certain nuclear domains and genomic loci.

was one of the key steps in defining the nucleolus as the site of ribosome biogenesis.
It was subsequently found that exposure of mammalian cells to low concentrations of actinomycin D also leads to alterations of nucleolar morphology, in which its three ultrastructurally-defined components, fibrillar center (FC), dense fibrillar components (DFC) and granular components (GC), undergo repositioning, (Clark et al., 1967). At these low concentrations, actinomycin D preferentially intercalates and alkylates DNA at adjacent GC base pairs (Gallego et al., 1997;Sengupta et al., 1988). Since ribosomal RNA genes of eukaryotic cells are GC-rich (at least 65-70% in the case of most mammalian genomes), there has been a tendency to regard low actinomycin as a highly specific experimental intervention. However, there are numerous other GC-rich sites throughout mammalian genomes. Indeed, our bioinformatics survey of the human reference genome revealed 320 sites which are 65-70% GC, including some that are 85% or more (Table S1). Thus, it is possible that nucleolar disruption by actinomycin D and concurrently observed effects could be attributable to the drug's action at one or more of these hundreds of genomic sites or to other yet to be defined actions of the drug.
The circumvent the pitfalls of using small molecule drugs with multi-valent functions, we employed siRNA approach to disrupt rDNA transcription machinery to determine whether the loss of transcription machinery impacts the well-defined nucleolar structure as actinomycin D treatment does. We also address whether nucleolar structure maintenance is dependent upon ribosome synthesis and whether nucleolar structural integrity plays a role in the organization of other nuclear domains.
of the nascent pre-rRNA (Suppl. Fig. 1A and 1B). As expected, there was a resultant reduction of ribosome synthesis as assessed by tracking an inducible GFP-tagged ribosomal protein RPL29 (Wild et al., 2010) (Fig. 1B). In untreated and control siRNA transfected cells, GFP-RPL29 was localized in both nucleoli and the cytoplasm within 24 hours of activation, reflecting its assembly into new ribosomes and localization to the cytoplasm. In comparison, in RPA194 knockdown cells, GFP-RPL29 was detected only in nucleoli and not in the cytoplasm 24 hours after activation, indicating the lack of newly synthesized ribosomal particles bearing GFP tagged RPL29 entering the cytoplasm. When the pre-ribosomal RNA processing factor UTP4 (Freed et al., 2012) was knocked down, ribosome synthesis was similarly blocked (Fig. 1B), as also represented by a similar lack of cytoplasmic GFP-RPL29 signal. These results demonstrate the inhibition of ribosome synthesis by knockdown of either RPA194 or UTP4.
Electron microscopy (EM) revealed that knockdown of RPA194 resulted in a reorganization of nucleolar architecture. The nucleolus normally contains three distinct ultrastructural components: FC ( Fig. 2A, arrows), DFC ( Fig. 2A, arrowheads) and GC ( Fig. 2A, asterisk) (Pederson, 2011;Raska, 2003). Knockdown of RPA194 induced segregation of the fibrillar components from the granular components ( Fig. 2A, left panel). The nucleolar segregation observed by EM corresponds to the redistribution of the fibrillar protein, UBF (Fig. 2B, arrowheads) and the dense fibrillar component marker protein fibrillarin (Suppl. Fig. 2) from their normal nucleolar location into nucleolar caps (Fig. 2B, arrows). Quantitative analyses revealed that the loss of RPA194 in siRNA treated cells closely paralleled nucleolar segregation (Fig. 2C), similar to that observed in cells treated with low concentrations of actinomycin D (Suppl. Fig. 3). In comparison to the knockdown of RPA194, siUTP4 treatment, while also blocking ribosome synthesis, did not lead to nucleolar segregation (Fig. 1B, 2A , B and C). This demonstrated, for the first time, that nucleolar segregation is not the result of ribosome synthesis inhibition, but rather the disruption of rDNA transcription machinery.
The spatial redistribution of the fibrillar centers upon nucleolar segregation implies that the rDNA does so in parallel. To determine if this was the case, we deployed CRISPR-dCas9 mediated labeling of rDNA clusters in siRNA treated and control cells (Ma et al., 2015;Ma et al., 2019;Ma et al., 2018;Ma et al., 2016a;Ma et al., 2016b). Sixteen sets of CRISPR-Cas9 gs RNA were designed to label rDNA. The specificity of the labeling was validated by simultaneous labeling with an anti-UBF antibody on mitotic Nucleolar Organizer Regions (NORs) (Suppl. Fig. 4) as it has been shown that rDNA transcription factors remain bound to active NORs throughout mitosis (Roussel et al., 1996). In untreated cells, control siRNA or siUTP4 transfected cells, the labeled rDNA was localized in the nucleoli in the expected nucleolar pattern. But after RPA194 knockdown, the rDNA signals became redistributed and colocalized with the repositioned fibrillarin caps ( Fig. 3A and C). This was also the case with low actinomycin treatment ( Fig. 3B and C), consistent with findings from a recent study (Mangan and McStay, 2021). In both cases, nearly all the rDNA was associated with the fibrillar protein-localized caps (Fig. 3A, B, and C).
These findings together with those in Fig. 1 show that RPA194 knockdown recapitulates in the classical phenomenon of nucleolar segregation observed with the treatment of low actinomycin.
Nucleolus-proximal nuclear domains are altered by knockdown of RPA194, but not UTP4.
With siRPA194 knockdown as a selective tool to disrupt nucleolar structure without the drawbacks of drug treatment, we asked whether nucleolar structure might play a role in nuclear organization. The perinucleolar compartment (PNC) is a distinct structure in many cancer cell types, including HeLa cells, and is physically associated with the nucleolus (Norton et al., 2008;Pollock and Huang, 2009). In a previous study, knockdown of RPA194 resulted in disassembly of PNCs (Frankowski et al., 2018). To investigate whether PNC disruption is the consequence of ribosome synthesis or rather due to nucleolar structure alteration, we included siUTP4 knockdown as a control. Antibodies against PTB (polypyrimidine binding protein, a fiduciary marker protein of the PNC) were used to immunolabel PNCs. As shown in Fig. 4A and B, while the siRPA194 treatment reduced PNC prevalence, knockdown of UTP4 did not significantly impact PNC bodies. Additionally, most of the remaining PNCs underwent structural alteration, becoming crescent moon shaped (Fig. 4A, arrows) in siRPA194 treated cells. Thus, PNC integrity is dependent on the nucleolar structure or rDNA transcription machinery, rather than ribosome synthesis.
Centromeres are known to cluster around nucleoli (Padeken and Heun, 2013), with an average 60% of the centromeres situated close to nucleoli in the HeLa cells we have used here (Foltz et al., 2009). To evaluate whether nucleolar segregation impacts nucleolus-centromere proximity, . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 we immunolabeled centromeres (of all chromosomes) and nucleoli with anti-CENPA and anti-UBF antibodies respectively. As shown in Figure 5A and B, upon RPA194 knockdown centromeres were less frequently associated with nucleoli and more dispersed throughout the nucleus. In contrast, UTP4 knockdown did not change nucleolus-centromere associations ( Fig.   5A and B). These findings indicate that the nucleolus-proximal centromere arrangement is contingent on normal nucleolar organization or Pol I transcription.

Repositioning of genomic loci after RPA194 knockdown
We next asked whether there are more wide-ranging effects throughout the 3-D nucleome upon RPA194 knockdown and the resulting nucleolar reorganization. As shown in Fig. 6A, the HIST1H2BN locus underwent a repositioning from its mainly nucleolar proximity to a predominantly central nuclear location in siRPA194 treated cells. Conversely, the predominantly central MALL locus become repositioned to more nuclear-peripheral positions (Fig. 6B). These positional changes of genomic loci were not observed in either siUTP4 or control oligo treated cells, nor in untreated cells, and thus are specifically attributable to RPA194 knockdown coupled nucleolar segregation.

Nucleolar organization governs Cajal body integrity
Given these results, we wondered if normal nucleolar organization might play a role not only in the intranuclear spatial location of certain genomic loci, but perhaps also in the maintenance of nuclear bodies. We therefore examined Cajal bodies (Gall, 2003;Nizami et al., 2010;Verheggen et al., 2002;Wang et al., 2016) and found that RPA194 knockdown, but not siUTP4 knockdown, led to their disruption as tracked both by the localization of p80 coilin, a Cajal body marker, and fibrillarin or UBF (Fig. 7A, and Suppl. Fig. 5). The disruption of Cajal bodies was closely correlated with nucleolar segregation (Fig. 7B) like those previously described in actinomycin Dtreated cells (Shav-Tal et al., 2005). The dispersion of Cajal bodies was also observed when examined with the additional marker protein SMN (Fig. 8). In addition to the dramatic structural changes, the Cajal body proteins, SMN and coilin no longer perfectly colocalized as shown in Fig. 8. These findings demonstrate that the nucleolus, in its normal operation, not only impacts genomic organization, but also a class of nuclear bodies usually situated distant from it.
The observed disruption of Cajal bodies prompted us to ask if RPA194 knockdown influences . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.30.534966 doi: bioRxiv preprint another nuclear body, the histone locus body (Nizami et al., 2010). As shown in Suppl. Fig. 6, there was no effect of RPA194 knockdown on histone locus bodies, demonstrating that not all nuclear bodies are regulated by an intact nucleolar structure.

Discussion
To circumvent the lack of specificity of low actinomycin D treatment as a tool to investigate the possible roles of nucleolar structure in global nuclear organization, we deployed knockdown of the Pol I subunit, RPA194, and of the pre-rRNA processing factor, UTP4, both leading to ribosome synthesis inhibition. However, only siRPA194 knockdown, but not UTP4 knockdown, induced nucleolar segregation, indicating that continuous ribosome synthesis is not required for maintenance of the classical tripartite nucleolar ultrastructure and gave us an opportunity to specifically address the impact of nucleolar structure on other nuclear components. We chose to use HeLa cells because they lack the p53 protein, which might otherwise trigger apoptosis upon RPA194 knockdown. We found that nucleolar segregation leads to changes throughout the 3D nucleome, in which nucleolus-proximal entities such as the PNC and centromere clusters reorganize significantly, certain genomic loci change their spatial distribution from nucleolusproximal to central nuclear locations or from mostly nuclear central to the nuclear periphery. In addition, nucleolar segregation disrupts nucleoplasmic Cajal bodies, but not histone locus bodies.
Our findings demonstrate a connection between nucleolar structure and a subset of various intranuclear components or genomic loci, with no apparent connection with others. The nature of the underlying signaling network is not clear. One possibility is that nucleolar segregation alters physical contacts between the nucleolar surface and directly connecting elements, such as centromeres. There is, moreover, extensive evidence for shuttling of proteins between the nucleolus and the nucleoplasm both in normal circumstances (Chen and Huang, 2001;Phair and Misteli, 2000) and in actinomycin D-induced nucleolar segregation (Shav-Tal et al., 2005). Thus, it is also possible that nucleolar segregation restricts the dynamics and exchanges of nucleolar and nuclear components that are part of the regulatory network important to nuclear homeostasis.
Future studies will investigate the mechanics and mechanisms of these connections.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Western blots. These were performed according to manufacture protocols (Millipore). The primary antibodies and dilutions used were RPA194 (see above), 1:500); UBF (see above), 1:500); beta-actin (Sigma, cat.# A5060), 1:3000, as loading controls. LI-COR IRDye secondary antibodies (1:10,000) were used for visualization. Protein bands were detected using LI-COR Odyssey Image Studio (LI-COR Biosciences).
Electron microscopy. Cells were in fixed 2% paraformaldehyde-2.5% glutaraldehyde in 0.1M cacodylate buffer, pH 7.3, processed and examined in the Northwestern University Feinberg School of Medicine Imaging Facility.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made    . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023     . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.30.534966 doi: bioRxiv preprint Suppl. Figure 4: The CRISPR-Cas9 sgRNA labeling of rDNA was validated using mitotic cells where UBF was found to be colocalized with the labeling (arrows), confirming that the sgRNA sets indeed labeled rDNA chromatin on mitotic chromosomes (arrow). Bar= 5 mm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 immunolabeled with antibodies against UBF and coilin. Bar= 5 mm Suppl. Figure 6: While Cajal bodies change significantly in siRPA194 treated cells, histone locus bodies do not show detectable changes. Bar= 5 mm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 (arrows) were segregated from the granular components (*). In comparison, the nucleolus in UTP4 knockdown cells (middle panel) maintained the typical three structural components, fibrillar center (arrow), dense fibrillar components (arrowhead), and granular components (*) as those in the siControl or untreated cells (right panels). (B) Correspondingly, siRPA194 treated, but not siUTP4 treated or control cells, showed segregated caps like localization of UBF as detected at the immunofluorescence microscope level. (C) Quantification showed that the segregation of nucleoli occurs mostly in RPA194 knockdown HeLa cells (n>100 cells, p<0.005 between siRPA194 and all other conditions). Bars =5 mm.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; Figure 3: rDNA co-segregated with the fibrillar proteins (fibrillarin or UBF) present in cap structures in segregated nucleoli. (A) Eleven sets of CRISPR-Cas9 sgRNA specifically complementing the rDNA were labeled with MS2-GFP, showing the localization of the rDNA chromatin, which redistributed with the fibrillar components to the nucleolar caps in siRPA194 treated HeLa cells (arrows) similarly to (B) those treated by ActD (arrows). (C) Quantitative evaluation indicated the co-partitioning of rDNA chromatin with the fibrillar components in nearly all of cells treated by siRPA194 or ActD (n>100 cells). Bars = 5 µm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; Figure 4: PNC structure and prevalence were altered in siRPA194, but not in siUTP4 treated HeLa cells. (A) The shape of most PNCs was changed, becoming extended (arrows) in siRPA194 treated, but not in siUTP4 treated or control cells. (B) Quantitative analyses demonstrated a significant reduction of PNC prevalence in siRPA194 treated cells (n>100 cells). Bars = 5 µm ****p<0.001 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.30.534966 doi: bioRxiv preprint Figure 5: RPA194 knockdown changed nucleoli-centromere interactions (A) The clustering of centromeres (green) around nucleoli (red) decreased in siRPA194 treated, but not in siUTP4 treated and control cells. (B) Quantifying analyses showed changes were statistically significant (n>50 cells) ****p<0,001. Bars = 5 µm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.30.534966 doi: bioRxiv preprint Figure 6: RPA194 knockdown changed spatial distributions of some genes. (A) HIST1H2BN gene loci became more centrally localized in siRPA194 treated HeLa cell nuclei than siUTP4 treated and other control cells (arrowheads) whereas (B) MALL loci became more nuclear peripherally localized in siRPA194 treated cells (arrowheads). Corresponding quantitative analyses showed significant differences in central and peripheral localization of the gene loci between treated and control cells. (n>50 cells) p<0.01. Bars = 5 mm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; Figure 7: siRPA194, but not siUTP4 treatment, induced disruption of Cajal bodies (arrows). (A) Upon siRPA194 treatment, Cajal bodies became irregularly shaped and some juxtaposed segregated nucleoli in siRPA194 treated cells (third panels), as detected using anti-coilin antibodies. However, Cajal bodies remained round-punctuated spots (arrows) in siUTP4 or control treated cells. (B) Quantitative evaluation showed a close coupling between nucleolar segregation and disruption of Cajal bodies (n>100 cells) (p<0.001 between siRPA194 and all others).
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ; https://doi.org/10.1101/2023.03.30.534966 doi: bioRxiv preprint Figure 8: Cajal body components no longer showed nearly complete colocalization in cells with segregated nucleoli from the treatment of siRPA194. Each panel shows a single nucleus. SMN and coilin which normally colocalize to the dotted-like Cajal body structures (Merge) no longer perfectly coincide in the round spots, rather become less colocalized in siRPA194 treated cells (arrowhead). Bar= 5µm . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted March 31, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023