By far, the most well recognized nuclear biomolecular condensate is the nucleolus. This prominent membrane-less organelle was identified over 100 years ago, and its structure has been continually refined as microscopy techniques have advanced. Nucleoli form around multiple copies of the ribosomal DNA (rDNA) cassette, a genomic region that encodes the ribosomal RNA (rRNA) and an intergenic spacer (Mélèse and Xue, 1995; Scheer and Hock, 1999; Jacob et al., 2012). These condensates appear to be present within all eukaryotic organisms, and are widely known as the site of rRNA transcription and ribosome assembly (Shaw and Doonan, 2004; Hernandez-Verdun, 2005). Despite this functional conservation, the organization of the nucleolus can be notably different among species. For example, higher eukaryotes often have multiple nucleoli per cell. These roughly spherical condensates possess a tripartite organization that subdivides the structure into the fibrillar center, dense fibrillar component, and granular component (Hernandez-Verdun et al., 2010). While each sub-region has a different molecular composition and specialized function, their collaborative goal is the genesis of ribosomal subunits (Thiry and Lafontaine, 2005; Feric et al., 2016). The nucleolus of less complex animals appear structurally simpler than that of humans, possessing only a bipartite composition. These structures contain the fibrillar center and a granular component, but lack the dense fibrillar component found in more complex organisms (Hernandez-Verdun et al., 2010). A survey of nucleoli across eukaryotes revealed that the tripartite structure may have evolved in amniotes, as the bifurcated nucleolus is observed in most non-amniotic species (Thiry and Lafontaine, 2005). This evolutionary shift from a bipartite organization may be related to the length of the rDNA cassette, which is markedly increased in amniotes. It has been theorized that longer rDNA regions allow more space for nucleolar constituents to bind, inducing the formation of this newer subnucleolar compartment (Thiry and Lafontaine, 2005). Meristematic plant cells also contain the tripartite nucleolus, though with far fewer fibrillar centers, and more dense fibrillar components than the animal tripartite nucleolus (Sáez-Vásquez and Medina, 2008). The yeast nucleolus is also notably different from other eukaryotic species. Here, a unique crescent shaped structure forms around the single array of rDNA repeats on the right arm of chromosome 7, and occupies approximately a third of the nuclear volume (Oakes et al., 1998; Matos-Perdomo and Machín 2019). Moreover, the yeast nucleolus exhibits extensive attachments to the nuclear envelope, which are not seen in the analogous human condensates. Despite the differences in structure and composition of the nucleolus between eukaryotic species, the primary function in ribosomal biogenesis is retained. Considering that eukaryotic cell survival is absolutely dependent on effective ribogenesis and downstream protein production, the highly conserved nature of the nucleolus is intuitive. The shifting complexity of the nucleolar structure could be an indication of the need to optimize or enhance ribogenesis in more complex multicellular organisms, compared to their single-celled ancestors.

Regulating protein targeting is critical for nucleolar dynamics and assembly (Hernandez-Verdun, 2006, 2011; Muro et al., 2010; Iarovaia et al., 2019). No universal targeting sequence has been identified for constituent cellular factors, but several studies have suggested that basic amino acids can act as nucleolar localization signals (Scott et al., 2010; Musinova et al., 2015). The positively-charged residues in these client proteins are thought to ionically-interact with the abundant negatively-charged scaffolding rRNA (Table 1) to drive recruitment (Musinova et al., 2015). The basic composition of the nucleolar localization signal is also consistent from insects to humans, further supporting this model for regulating nucleolar protein trafficking (Martin et al., 2015). Growing evidence suggests that nucleolar targeting can also be regulated by stress signaling (Chatterjee and Fisher, 2003; Nalabothula et al., 2010; Latonen et al., 2011; Audas et al., 2012; Iarovaia et al., 2019), as the emergence of the Nucleolar Stress Response pathway has uncovered a novel function for this well-established subnuclear compartment (Mayer and Grummt, 2005; Scott and Oeffinger, 2016; Ogawa and Baserga, 2017; Iarovaia et al., 2019). Here, the nucleolar resident Nucleophosmin (B23), a pre-rRNA processing and cleavage factor, can sequester the tumor suppressor protein ARF within the nucleolus during normal growth conditions (Korgaonkar et al., 2005). In response to DNA damage, ARF is released into the nucleoplasm, where it can promote cell cycle arrest or apoptosis by binding MDM2 and activating p53 (Rubbi and Milner, 2003; Kurki et al., 2004; James et al., 2014; Russo et al., 2021). Activation of the B23/ARF/MDM2/p53 stress-signaling axis has only been observed in vertebrate species. This can likely be attributed to the poor conservation of many of the regulatory molecules, as B23, ARF, MDM2, and p53 appear to be absent in many lower eukaryotic organisms (e.g., D. melanogaster and S. cerevisiae), and plants (Table 2). Despite this lack of conservation, a nucleolar stress response pathway has been identified in the nucleolus of the plant species Arabidospsis thaliana, suggesting that the stress-sensing nature of the nucleolus is not limited to animals (Ohbayashi et al., 2017; Ohbayashi and Sugiyama, 2018). Here, perturbations in ribosome biogenesis due to DNA damage induce the activity of NAC family transcription factors, that seem to act similarly to p53 in regulating gene expression. Though the proteins involved share no sequence similarity and are not considered orthologous, it is notable that the nucleolus is involved in stress response pathways across such diverse species. In this way, the nucleolus can be considered an “assembly line” condensate, for its role in ribosome biogenesis, as well as a molecular “reservoir” for its role in the DNA damage response. Though nucleoli are a defining feature of eukaryotes, these structures have adopted additional functions throughout evolution. This novel stress-sensing functionality highlights that the emergence of a biomolecular condensate is not a static process, but rather that these structures may possess remarkable adaptability to meet the changing needs of increasingly complex organisms.

Though some stressors induce minor changes in nucleolar protein trafficking, severe insults can lead to a complete remodeling of the nucleolar space. This results in the dissolution of the typical tri- or bi-partite structure and the formation of a novel biomolecular condensate called the amyloid body (A-body). Here, in response to extreme environmental stimuli (e.g., heat shock and extracellular acidosis), a diverse set of proteins are targeted to the nucleolus (Audas et al., 2012, 2016; Marijan et al., 2019). The growing protein population dislocates the ribosomal biogenesis machinery, and promotes the assembly of large fibrillar amyloid-like aggregates (Audas et al., 2016). Like many nuclear biomolecular condensates, A-body formation is dependent on a class of lncRNA that act as scaffold molecules during formation (Table 1). Environmental stressors trigger the expression of lncRNAs from the intergenic spacer region of the rDNA cassette, which remain in close proximity to their locus of origin following synthesis (Audas et al., 2012). These seeding molecules are highly enriched in dinucleotide repeats, and are not predicted to form any secondary structure (Wang et al., 2018). This disorder may enhance the surface area of the scaffolding lncRNA, providing more binding sites for A-body constituent proteins that possess inherent amyloidogenic properties. After reaching a micro-concentration threshold within this growing subnuclear domain, client proteins are thought to fibrillate, and the biogenesis of a solid-like functional amyloid aggregate can be observed (Audas et al., 2016; Wang et al., 2018). Though amyloid aggregation is primarily associated with irreversible neurodegenerative disorders, like Alzheimer’s and Parkinson’s diseases (Ross and Poirier, 2004; Gregersen et al., 2005), A-body formation is protective to the cell and rapidly reversed following the termination of stress signaling (Audas et al., 2016).

A-bodies were initially characterized in cultured human cell lines, grown under tightly controlled environmental conditions (Audas et al., 2012; Jacob et al., 2012; Audas et al., 2016). Recently, it has been shown that A-body formation can occur in cells derived from a variety of species, such as monkeys, mice, rats, chickens, fish, and flies (Lacroix et al., 2021). Within an organismal setting, A-bodies are inducible in D. melanogaster (flies) and S. cerevisiae (yeast), demonstrating that this membrane-less organelle is not an artifact of the tissue culture system, and is widespread throughout metazoans and yeast (Lacroix et al., 2021). An analysis of the scaffolding lncRNAs that regulate A-body formation showed no sequence conservation between the human, mouse, and chicken transcripts (Table 1) (Lacroix et al., 2021). However, stress signaling did result in the expression of low-complexity lncRNA critical for A-body formation, from a region directly downstream of the rRNA coding sequence in all three organisms (Lacroix et al., 2021). It would be interesting to assess rDNA cassette transcriptomic profiles across a broader range of organism exposed to A-body-inducing stimuli. Unfortunately, this is technically challenging, as the rDNA region is removed from most bioinformatics analyses. Together, the available data suggests that the genomic location and function of these scaffolding transcripts are retained throughout different species, despite the lack of conservation in their primary sequence (Table 1).

Proteomic analyses of human A-bodies has shown that these structures contain a large array of cellular constituents, including a variety of proteins that regulate the cell cycle (Audas et al., 2016; Marijan et al., 2019). Many of the client proteins are targeted to this subnuclear domain in a stress-specific manner (Marijan et al., 2019), suggesting that cells regulate protein recruitment to tailor the stress-response pathway to each stimulus. To date, no proteomic screens have been performed on A-bodies derived from other organisms, and only the cell cycle regulator CDC73 has been observed as a client within this structure from humans to D. melanogaster (Lacroix et al., 2021). Biologically, A-bodies formation in human cells has been demonstrated to induce a state of cellular dormancy (Audas et al., 2016) and serve as a site for local nuclear protein synthesis (Theodoridis et al., 2021), displaying the nature of both a reservoir and assembly-line condensate. The universal presence of CDC73 within A-bodies derived from humans to flies, suggests that cellular dormancy may be a conserved role for these biomolecular condensates across metazoans. However, local nuclear protein synthesis was absent in lower eukaryotes (i.e., flies and yeast) (Lacroix et al., 2021), suggesting that A-body-mediated translation is a later evolutionary feature for this structure. Comparing the sequence divergence of CDC73 and the critical local nuclear protein synthesis translation factor eIF4B (Theodoridis et al., 2021) in D. melanogaster potentially underlies the mechanism of this functional difference (Table 2).

The amyloid-like biophysical properties and the severe-stress-inducible nature of the A-bodies are the hallmark of this condensate, as these features are observed in all species tested from humans to yeast. Analogous to the dual functionality of the nucleolus (i.e., ribosomal biogenesis and Nucleolar Stress Response), A-bodies appear to have both conserved and adaptive cellular roles. It is likely that local nuclear protein synthesis evolved using a pre-existing A-body framework, as this unique function is not observed in lower eukaryotes. The cause for this emerging functionality can only be speculated; however, the increased complexity of this stress response pathway in progressively more complex organisms is not surprising. Nevertheless, assessing the presence or absence of this structure in more distant species (i.e., additional fungi and plants), as well as future proteomic and transcriptomic studies from organisms across the evolutionary landscape could provide important insights into the conserved/non-conserved roles and regulatory mechanisms of this domain. As a functional and frequently observed solid-like condensate, increased understanding of A-body formation, regulation, and biological relevance could be informative in our understanding of the utility of amyloid aggregation outside of a pathological setting.

Cajal bodies were initially discovered in neurons, but have since been identified in an array of cell types across the eukaryotic domain. Functionally, these condensates are involved in the maturation of spliceosomal RNA and small nuclear ribonucleoprotein complexes (snRNPs). This subnuclear domain serves as the site of small nuclear RNA (snRNA) synthesis and processing, which is regulated by the activity of small Cajal-body-specific RNAs (Ogg and Lamond, 2002; Cioce and Lamond, 2005; Morris, 2008; Hebert, 2010). Unlike many biomolecular condensates, Cajal bodies are built around a protein scaffold that is composed of coilin, rather than a nucleotide-based scaffold (Table 1) (Tucker et al., 2001; Cioce and Lamond, 2005; Lafarga et al., 2016). These structures are also enriched in the SMN protein complex and Fam118b, which are both necessary for Cajal body function (Narayanan et al., 2004; Li et al., 2014). SMN is involved in the recruitment of cytoplasmic snRNPs, where U snRNPs are further modified before spliceosome assembly (Sleeman and Lamond, 1999; Narayanan et al., 2004). The precise function of Fam118b in Cajal bodies remains unclear. It associates with both SMN and coilin and its depletion results in disassembly of this structure, which could suggest a scaffolding role (Li et al., 2014). Although Cajal bodies are important in optimizing the assembly of essential snRNPs, these membrane-less organelles themselves are dispensable for the survival of an organism. Depletion of coilin has been shown to interrupt Cajal body formation in a cultured cell model (Almeida et al., 1998), though it only reduces viability in a knockout mouse setting (Tucker et al., 2001). The non-essential nature of this condensate could be a consequence of its function, as the role of this subnuclear structure is to enhance snRNP maturation kinetics. Here, concentration of factors necessary for snRNP production would markedly increase the molecular efficiency of an organism, but may not dictate whether the generation of these essential complexes occurred entirely.

Although Cajal bodies are not essential to the survival of mice, structures with similar properties are present in humans (Almeida et al., 1998), amphibians (Morgan et al., 2000), flies (Liu et al., 2006, 2009), yeast (Verheggen et al., 2002; Qiu et al., 2008), and a variety of plant species (Chamberland and Lafontaine, 1993; Gall, 2000). These biomolecular condensates are generally considered Cajal bodies if they contain coilin (a marker molecule), possess a subnuclear localization, and play a role in the production of mature snRNPs. Initially, a D. melanogaster ortholog for coilin was elusive (Table 1) and the Cajal bodies were identified by the presence of other marker molecules, such as SMN (Liu et al., 2006). However, a more intense search identified a coilin ortholog that had very small regions of sequence identity at the amino- and carboxyl-termini (Liu et al., 2009). This suggests that the terminal protein regions could be important in the organization of weak multivalent interactions that mediate Cajal body formation. To date, no coilin homolog has been found in yeast, though Cajal body-like structures have been identified. These homologous condensates are referred to as “nucleolar body,” and are believed to represent the early ancestor of a modern Cajal body (Verheggen et al., 2001, 2002). Despite the lack of coilin, over-expression of the human SMN homolog in yeast results in the targeting of this marker protein to the “nucleolar body” (Verheggen et al., 2001). Additionally, the yeast “nucleolar body” serves as the site of U3 snoRNA maturation, a role typical ascribed to Cajal bodies in mammals (Verheggen et al., 2001, 2002; Jády et al., 2003). As many eukaryotes are reliant on extensive amounts of splicing for the production of specific mRNA molecules, the retention of this domain in a variety of species is not surprising. However, it is remarkable that this biomolecular condensate displays remarkable retention despite being non-essential for organismal survival.

Paraspeckles were initially discovered as small dynamic structures that occupy interchromatin spaces within human nuclei (Fox et al., 2002). Recently, they have gained an abundance of attention in the scientific community, as they are an important model for examining lncRNA-protein interactions. Within this subnuclear domain, NEAT1 acts as an architectural RNA to recruit a diverse family of cellular proteins that bind along the length of the transcript (Chujo and Hirose, 2017; Fox et al., 2018; Yamazaki et al., 2018; Wang and Chen, 2020). Like nucleoli, paraspeckles possess liquid-like properties and display remarkable organization and sub-compartmentalization. Each paraspeckle has a distinct core and peripheral sub-domain, characterized by the presence of unique proteins on specific regions of the NEAT1 transcript. The 5′ and 3’ ends of this lncRNA are present in the peripheral shell-like domain (Souquere et al., 2010), while the central region of NEAT1 is localized to the condensate core, along with a group of essential proteins (e.g., NONO, SFPQ, and FUS) (West et al., 2016; Fox et al., 2018; Yamazaki et al., 2018). A variety of other non-essential client molecules have been mapped to this structure, and their biological roles have been linked to numerous cellular pathways (Masuda et al., 2015; Yamazaki and Hirose, 2015; Li et al., 2020; Wang et al., 2021). It has been proposed that formation of this structure may regulate the amount of these client proteins in the cellular milieu, potentially titrating their concentration by acting as a temporary molecular reservoir or sponge (Hirose et al., 2014). The observation that paraspeckle abundance increases in response to environmental perturbation (Mohammad Lellahi et al., 2018) could align with this functionality, as unfavorable growth conditions could lead to more proteins being “soaked-up” into this subnuclear domain to maintain homeostasis by sequestering key regulators of energy-intensive cellular pathways. In addition to regulating client protein concentrations, it has also been proposed that paraspeckles function to regulate NEAT1 availability in the cytosol (Zhang et al., 2019). Inflammatory signals promote the cytosolic release of NEAT1, where it functions to promote inflammasome activation in macrophages, suggesting that paraspeckles may also regulate aspects of the immune response (Zhang et al., 2019).

These biomolecular condensates are currently described as a mammalian-specific structure (Fox et al., 2002; Fox and Lamond, 2010; Fox et al., 2018; McCluggage and Fox, 2021); however, homologs of Paraspeckle proteins form distinct subnuclear domains in other metazoans (Table 2). For example, the D. melanogaster NONO homolog NonA, is a component of Drosophila-specific nuclear condensates, called omega speckles (Prasanth et al., 2000; Singh & Lakhotia, 2015). These foci (described below) diverge from Paraspeckles in several key ways, including the architectural RNA used and the stress-responsive nature of the domain (Singh and Lakhotia, 2015). A recent study in C. elegans (Pham et al., 2021), also identified NONO homolog-containing subnuclear structures, termed “paraspeckle-like” foci. Whether these nematode condensates are more closely related to the omega speckles or paraspeckles is unclear, as more rigorous examination is required. Currently, there is no literature on the existence of yeast paraspeckles, which is not unexpected considering the limited sequence-identity of potential S. cerevisiae homologs of NONO (Pab1p) and SFPQ (Hrp1p) (Table 2). Exogenous expression of the essential human paraspeckle protein FUS in a yeast setting resulted in the formation of cytoplasmic aggregates (Fushimi et al., 2011), further signaling the absence of this subnuclear domain in lower eukaryotes. The NEAT1 scaffolding molecule also appears to be a mammalian-specific lncRNA, thereby limiting the potential for paraspeckle formation outside of this class. Within mammals, there is no evidence of sequence conservation of the NEAT1 lncRNA transcript, suggesting that the nucleotide composition, or structure, is more important for functionality rather than the primary sequence (Table 1). An analysis of NEAT1 knockout mice has demonstrated that these paraspeckle-deficient individuals are viable, but do not develop a functional corpus luteum and fail to establish pregnancy (Nakagawa et al., 2014), implicating a potential role for paraspeckles in the mammalian-specific process of gestation. It is tempting to theorize that paraspeckles evolved to promote fertility in mammals, a functionality that would distinguish these domains further from any possible D. melanogaster, nematode, or yeast condensates that may be identified in the future.

Though they were discovered over 100 years ago (Cajal, 1910), nuclear speckles remain somewhat unclear. Many potential functions have been attributed to these dynamic and irregularly-shaped domains (e.g., transcription, splicing, and mRNA export), yet most of these roles have not been definitively established (Ihara et al., 2008; Fei et al., 2017; Hasenson and Shav-Tal, 2020). In mammalian cells, X. laevis oocytes, and D. melanogaster embryos, structures with very similar characteristics have been observed (Ségalat and Lepesant, 1992; Gall et al., 1999; Spector and Lamond, 2011), though no analogous condensates have been reported within yeast nuclei.

Nuclear speckles themselves are composed of RNA and protein constituents. Numerous poly-adenylated transcripts are found within this subnuclear domain (Fay et al., 1997; Calado and Carmo-Fonseca, 2000; Fox et al., 2002), including several low-complexity RNAs that are generated in repeat expansion disorders (Urbanek et al., 2016). The array of proteins that are targeted to these condensates are also broad, but strongly enriched in RNA-binding proteins that possess mRNA splicing and processing functionalities (Spector et al., 1991). Targeting of these resident proteins does not appear to be as efficient as that seen in other subnuclear domains, with most constituents also possessing a clear nucleoplasmic pool in addition to the nuclear speckle population (Lamond and Spector, 2003). Two proteins, SON and SRRM2, were recently found to be essential for nuclear speckle formation and organization (Ilık et al., 2020). Over evolution, both of these proteins appear to have undergone lengthening in their intrinsically disordered regions, which likely promotes the weak multivalent interactions needed to form this structure (Ilık et al., 2020). SON is absent in yeast, while SRRM2 has a significantly smaller S. cerevisiae homolog (Cwc21) (Table 1). Cwc21 contains the putative spliceosome-associating motif of SRRM2, but lacks the large intrinsically disordered domain that has been linked to liquid-liquid phase separation and condensate assembly (Ilık et al., 2020; Ilık and Aktaş 2021). This poor conservation further validates the absence of nuclear speckles in yeast, and suggests that the role of this structure in splicing may have evolved from a less efficient mRNA processing events that occurred within the nucleoplasm of lower eukaryotes.

Originally termed “perichromatin granules” (Mahl et al., 1989), nuclear stress bodies are stimuli-induced condensates enriched in the well-studied heat shock response protein HSF1 (Sarge et al., 1993). Under conditions of thermal or chemical stress, HSF1 activation induces the transcription of highly repetitive lncRNA from the Satellite III repeat regions present on chromosomes 9, 12, and 15 of the human genome (Denegri et al., 2002). These lncRNA range in length, from 2 to 10 kb, and following expression remain closely associated with their locus of origin (Biamonti and Vourc’h, 2010; Jolly et al., 2004; Rizzi et al., 2004). Here, the transcripts generate phase-separated nuclear stress body by acting as scaffolding molecules that concentrate stress-activated HSF1, RNA-binding proteins, and a broad family of transcription factors (Biamonti and Vourc’h, 2010). The proteomic composition of this subnuclear domain has led many to believe that this structure regulates pre-mRNA maturation (Denegri et al., 2001). However, recent next-generation sequencing results have attributed a role for nuclear stress bodies in the retention of introns by hundreds of pre-mRNA transcripts (Ninomiya et al., 2020). Mechanistically, this involves the activity of the nuclear stress body constituent SRSF9, which is modulated by post-translational modifications. This protein is phosphorylated within the structure during the recovery from thermal stress, which appears to suppress the splicing of hundreds of introns (Ninomiya et al., 2020). Altering intron retention could fine-tune gene expression and aid in the recovery from thermal stress. Despite the presence of HSF1 throughout many eukaryotic species (Table 2), the Satellite III repeats are only present in primates (Table 1), limiting nuclear stress bodies to this taxonomic order (Jarmuz et al., 2007). The late evolution of these condensates will lead to interesting speculation on the function of these structures, raising questions about whether or not nuclear stress bodies play a vital role in primate biology.

In human cells, the formation of large nuclear and cytoplasmic aggregates, termed aggresomes, occurs when the ubiquitin-proteasome system is overwhelmed or impaired (Johnston et al., 1998; Latonen et al., 2011; Latonen 2019). These structures are enriched in poly-adenylated mRNAs and immobilize cell cycle related proteins, such as p53 (Latonen et al., 2011). Aggresomes found within the nucleus form in close proximity to nucleoli, causing a re-arrangement of the nucleolus around the new compartment. Both of these structures remain distinct, as the nucleolus retains its normal role in ribosomal biogenesis, and the aggresomes are devoid of any rRNA or nucleolar marker molecules (Latonen et al., 2011). Nucleolar aggresomes appear to act as reservoirs for misfolded proteins that have exceeded the capacity of the ubiquitin-proteasome system, as disassembly can occur following the addition of excess ubiquitin (Latonen et al., 2011). While these structures have only been observed in human cells, a similar structure has been identified in budding yeast, also in response to proteasome inhibition. These biomolecular condensates, termed intranuclear quality control compartments (INQs), form in close proximity to the nucleolus, and are also composed of misfolded proteins (Gallina et al., 2015; Miller et al., 2015). Though there are considerable similarities between the location, conditions for formation, and composition of nucleolar aggresomes and INQs, more rigorous testing will be necessary to determine if these two structures are evolutionarily related, and if nucleolar aggresomes are present beyond human cells.

The promyelocytic leukemia (PML) protein is a tumor suppressor that is associated with protein-based condensates in the nucleus called PML bodies (Melnick and Licht, 1999; Ching et al., 2005; Torok et al., 2009). Formation of these subnuclear structures is dependent on the presence of PML as a scaffolding protein, which multimerizes upon SUMOylation and recruits additional proteins, such as DAXX, SUMO-1, and Sp100 (Ishov et al., 1999; Zhong et al., 2000; Shen et al., 2006). These biomolecular condensates are not enriched in any particular nucleic acids, but they often form in close proximity to specific genes. The p53 locus is a common genetic location, and formation of PML bodies here is thought to regulate the expression of this tumor suppressor protein (Sun et al., 2003; Ching et al., 2005). Following synthesis PML bodies also play a role p53 stabilization, as the addition of post-translational modifications within this domain is believed to contribute to cell cycle inhibition (Guo et al., 2000; Conrad et al., 2016). This transcriptional and post-translational regulation of p53 has suggested that PML bodies have a tumor suppressive function, though other p53-independent roles in apoptosis, telomere maintenance, and the DNA damage response have also linked this subnuclear domain to cancer (Guo et al., 2000; Bernardi et al., 2008; Draskovic et al., 2009; Conrad et al., 2016). Unsurprisingly, PML^−/− mice are more susceptible to tumorigenesis, though the survival of these knockout mice demonstrates that PML, and by extension PML bodies, are not essential for organismal viability (Wang et al., 1998).

To date, PML bodies have only been observed in mammalian cells (Bernardi and Pandolfi, 2007; Batty et al., 2009). However, PML protein orthologs exist across amniotes, with varying levels of sequence and structure identity (Table 1). This could suggest that non-mammalian amniotes, like birds and turtles, also have the potential to form PML nuclear bodies, but the presence of this structure outside mammals has yet to be identified. The potential for PML body formation appears to be limited to more complex eukaryotes, as there are no identified homologs of the scaffolding protein PML (Table 1) or client molecule Daxx (Table 2) in yeast, consistent with the lack of PML body formation in the single-celled eukaryotes.

At the D. melanogaster 93D gene locus, the hsrω lncRNA is expressed (Lakhotia and Sharma, 1996; Lakhotia, 2011). This transcript acts as an architectural RNA that mediates the recruitment of heterogeneous nuclear ribonucleoproteins (hnRNPs) into biomolecular condensates found within the interchromatin space called omega speckles (Prasanth et al., 2000; Singh and Lakhotia, 2015). At the cellular level, these constitutive foci are present under normal environmental conditions and are thought to regulate molecular pathways by generating a reservoir of inactive hnRNPs (Prasanth et al., 2000; Singh and Lakhotia, 2015). Interestingly, heat stress can trigger the disassembly of omega speckles and the formation of a single large aggregate at the 93D locus (Singh and Lakhotia, 2015). A return to normal growth conditions causes this large aggregate to fragment, releasing smaller condensates that resemble the pre-stress omega speckles back into the nucleoplasm (Singh and Lakhotia, 2015). At the organismal level, 93D-null flies have reduced viability (Mohler and Pardue, 1984), suggesting that omega speckles may be essential for survival. However, whether the reduction in viability is due to a lack of omega speckle formation, or another function of the 93D locus is unclear.

To date, omega speckles have only been observed in D. melanogaster, yet these subnuclear structures bear compositional/functional similarities to the primate-specific nuclear stress bodies and client protein overlap with the mammalian-specific paraspeckle (discussed above). Regardless of the parallels between these domains, in both cases these similarities are not sufficient to consider these structures evolutionarily linked. For example, hnRNPs can be recruited to omega speckles and nuclear stress bodies by hsrω and Satellite III transcripts, respectively. Both of these scaffolding lncRNAs possess repetitive sequences, show stress-induced upregulation, and can remain in close proximity to their gene of origin following expression (Jolly and Lakhotia, 2006). Despite these similarities, there are notable differences between these structures. The Satellite III repeats are present on multiple chromosomes throughout the human genome, while hsrω transcripts are only derived from the 93D gene locus. Also, snRNPs have been found to associate with Satellite III, but not hsrω, lncRNA transcripts (Metz et al., 2004). Moreover, omega speckles are constitutively present in growing cells and altered in heat shock conditions, while nuclear stress bodies are formed only in response to stress stimulation (Sarge et al., 1993; Prasanth et al., 2000). The nuclear stress body marker molecule HSF1 also does not appear to be a component of the omega speckles formed under normal growth conditions, though it can be recruited to the large heat shock-induced aggregates that form at the 93D gene locus (Westwood et al., 1991). This makes it tempting to speculate that while nuclear stress bodies are unrelated to constitutive omega speckles, they may be a distant relative of the large aggregates that form during heat stress. Thus, these large omega speckle-like aggregates may represent a distinct biomolecular condensate, analogous to the constitutive nucleoli and stress-inducible A-bodies (Jacob et al., 2012; Audas et al., 2016; Lacroix et al., 2021). The presence of omega speckles has not been thoroughly examined outside of D. melanogaster, and additional data would help determine the evolutionary stage at which these structures are lost or modified.

Recently, a novel membrane-less subnuclear structure was identified by differential interference contrast microscopy in C. elegans, termed the NUN body (Pham et al., 2021). Here, 4–10 granular phase-separated subnuclear foci could be seen in nematode neuronal and glial cells (Pham et al., 2021). To date, there is limited information about these condensates, as no constituent molecules have been identified (Table 1). An analysis of marker proteins for a diverse array of subnuclear structures, has demonstrated that NUN bodies appear to be distinct from established nuclear domains, including nucleoli, Cajal bodies, paraspeckles, and nuclear speckles (Pham et al., 2021). The narrow cell-type specificity of NUN-bodies will be interesting to study with regard to functionality. Their exclusivity to neurons and glia could suggest that this structure has a neuronal-specific role, though more information is required. Clearly, identification of molecular constituents will be of the utmost importance in assessing the conservation and biological significance of the NUN bodies. Until this is done, it will be almost impossible to establish whether these condensates are a conserved element of neuronal and glial cell biology, or if they represent a nematode-specific domain.

Mei2 is an RNA-binding protein in S. pombe (fission yeast) that is involved in meiosis (Yamashita et al., 1998). This protein functions as an essential inducer of pre-meiotic DNA synthesis and meiosis I by binding to meiRNA, a polyadenylated lncRNA transcript derived from the sme2 gene on chromosome 2 (Yamashita et al., 1998). Similar to Satellite III (nuclear stress bodies), rIGSRNA (A-bodies), and hsrω (omega speckles) lncRNAs, once transcribed the meiRNA transcripts remain in close proximity to their locus of origin (Shimada et al., 2003). When mei2 is de-phosphorylated, its localization shifts from the cytoplasm to the nucleus, where it is able to directly bind meiRNA. Here, meiRNA appears to act as a scaffold molecule (Table 1), concentrating mei2 into a single phase-separated nuclear punctate structure called a mei2 dot. This structure also sequesters the protein Mmi1 (no human homolog), which functions to degrade pro-meiotic mRNA transcripts (Harigaya et al., 2006). The formation of this nuclear condensate appears to be essential for the induction of meiosis I, likely by suppressing Mmi1-mediated degradation of pro-meiosis transcripts (Shichino et al., 2014). As this subnuclear dot regulates the shift of a cell from mitosis to meiosis, the observation of mei2 dots only in the single-celled eukaryotes is not surprising. The transient shifts from sexual to asexual reproduction observed in yeast are not observed in multicellular species. However, future research to identify any similar structures in primordial germ cells could be interesting, as these cells also undergo a shift from mitosis to meiosis. Yet, unlike unicellular organisms, these cells do not shift back to a state of mitosis once they have undergone meiosis, until they have become fertilized. Therefore, a phase separated structure that could rapidly influence this change in reproductive state seems unlikely.