The Nucleolus Gets the Silent Treatment

Abstract

In this issue of Cell Stem Cell, Savić et al. (2014) and Kim et al. (2014) provide insight into how global heterochromatin condensation and LIN28 sequestration in embryonic stem cells are independent mechanisms regulating pluripotency in the oft-overlooked nucleolus.

Even though we have been aware of the nucleolus for 180 years, and it is the largest structure in the nucleus, it gets little attention compared to the nucleus (15,000 versus 374,000 cites). And much less attention is paid to its role in epigenetics (25 Pubmed citations) or heterochromatin (800 citations) compared to its role in ribosomal gene transcription. That’s actually a bit strange for both historical and biological reasons. Historically, because in the 1930s, Emil Heitz described heterochromatic satellite regions near the nucleolus, and Barbara McClintock described the nucleolar-organizing region’s association with heterochromatin. And biologically, because most ribosomal genes are actually not expressed in normal cells, and their silencing by heterochromatin is necessary to prevent chromosomal rearrangements and excessive growth.

In this issue of Cell Stem Cell, two papers now provide insight into independent nucleolar mechanisms regulating pluripotency in embryonic stem cells (ESCs). Much as people now widely accept the importance of the nuclear membrane in organizing heterochromatin structure, and emerging work connects 3D genome organization with regulation of ESC fate (reviewed in Gorkin et al., 2014), the nucleolus has so far gotten the silent treatment in this area. The work by Savić et al. addresses one mechanism for heterochromatin formation during lineage commitment as mouse ESCs (mESCs) differentiate (Savić et al., 2014). They specifically addressed the role of the long noncoding RNA pRNA, which is the mature form of the processed 2 kb long intergenic spacer (IGS)-rRNA, already known to interact with nucleolar repressor factor TIP5 in generating rDNA heterochromatin. They first showed that ribosomal DNA methylation is decreased in induced pluripotent stem cells and heterochromatin marks H3K9me2/3 and H3K27me3 increase during differentiation; and the changes in both directions are associated with expected increases and decreases in transcription, respectively. They then transfected mESCs with pRNA, which led to TIP5 accumulation in nucleoli and induced heterochromatic rDNA with an increase in CpG methylation and H3K9me2 at the rDNA promoter—and mutagenesis experiments showed that this was dependent on an intact 3′-pRNA stem-loop structure, but not the pRNA-TIP5 association itself. Thus pRNA guides TIP5 to rDNA through its hairpin structure and establishes ribosomal gene heterochromatin. Moreover pRNA expression and TIP5 accumulation led to a dramatic general increase in condensed perinucleolar heterochromatin, as well as total levels of H3K9me2 in the nucleus, an indicator of large-scale heterochromatin formation.

Another paper in this same issue, by Kim et al., identifies a nucleolar targeting mechanism by which a histone methyl-transferase—normally these are involved in the epigenetic machinery—affects pluripotency in human ESCs (Kim et al., 2014). The study addresses the mechanism by which the key pluripotency factor LIN28 represses processing of let-7 miRNA (from the pri-let-7 to the pre-let-7) to maintain self-renewal. In the nucleus, LIN28 processing is less well understood than in the cytoplasm, which involves recruitment of TUT4/7 to induce oligouridylation of pre-let-7, which appears to both prevent its Dicer-induced processing into mature let-7 and render it susceptible to digestion by the Dis3L2 exonuclease (reviewed in Shyh-Chang and Daley, 2013). The authors were pursuing a relatively recent thread of studies on the nature and effect of methylation of core pluripotency factors. Specifically they investigated the effect of SET7/9 monomethylase on LIN28A, but not the paralogous LIN28B, and identified the methylated residue as lysine 135. Remarkably, the K135 methylation leads to nucleolar localization of LIN28A and sequestration, preventing LIN28A processing independently of TUTase. Interestingly, the K135 site of LIN28A is homologous to the nucleolar localizing signal region of LIN28B, which was already known to have a role in the nucleolus (Piskounova et al., 2011).

Savic et al. dance with a bee in my own bonnet when they consider that the progressive condensation of euchromatin into heterochromatin in nuclear structures reduces epigenetic and thus gene expression plasticity during differentiation and this process might be modulated during important processes such as the epithelial-mesenchymal transition (EMT) and cancer. The mechanism we have studied is the role of Large Organized Chromatin lysine (K)-modifications (LOCKs), typically H3K9me2/3 in regulating phenotypic plasticity (Wen et al., 2009). These largely overlap structurally defined lamin-associated domains (LADs). LOCKs support formation of organized chromatin structures, or “speckles,” in the nucleus during hematopoietic lineage commitment (Chen et al., 2012). EMT induced by TGF-β causes reversible loss of LOCKs and LADs, and inhibition of this chromatin de-condensation by lysine-specific demethylase inhibitors abrogates downstream phenotypes of EMT such as chemoresistance and cellular migration. Interestingly, the mechanism described by Savić et al. also appears to induce LOCKs on a genome scale. We are only beginning to understand the key role of nuclear structures in organizing heterochromatin domains in the nucleus. Intriguingly, though, heterochromatin domains can shuffle between their nuclear (LAD/LOCK) and nucleolar homes. One should also recall that the first nuclear mapping of CTCF, a key organizer of chromosome territories and a regulator of genomic imprinting, was to the nucleolus (Yusufzai et al., 2004).

Finally, both of the current papers emphasize the importance of noncoding RNAs in the nucleolus’s role in regulating pluripotency and differentiation. This is in keeping with the observations that the X-inactivation RNA XIST is localized to the nucleolar periphery. The long anti-sense RNA LIT1, which regulates one of the imprinted Beckwith-Wiedemann syndrome domains involved in Wilms and other tumors, is also localized to the nucleolus, which appears to participate in the orchestration of large-scale H3K9 and H3K27 methylation and gene silencing (Pandey et al., 2008).

Given that the nucleolus acts in such important structural ways without an actual membrane, it is fascinating that its formation requires only the Pol I machinery and the factors UBF and SL-1 in concert with active genomic nucleolar organizing regions (NORs) (reviewed in Diesch et al., 2014). The current reports support the idea that these processes and their developmental control require the presence of noncoding RNAs that likely facilitate the organization of “anatomical” domains beyond the formation of the nucleolus. The last day of this month marks the 500^th birthday of the Belgian anatomist Andries van Wesel (often Latinized to Vesalius), who defined human anatomy and argued that function follows form. He would be thrilled to learn that we are finally dissecting the nucleus into its components and learning the same lesson.