Coilin and Cajal bodies

ABSTRACT

The nucleus of higher eukaryotes contains a number of structures that concentrate specific biomolecules and play distinct roles in nuclear metabolism. In recent years, the molecular mechanisms controlling their formation have been intensively studied. In this brief review, I focus on coilin and Cajal bodies. Coilin is a key scaffolding protein of Cajal bodies that is evolutionarily conserved in metazoans. Cajal bodies are thought to be one of the archetypal nuclear structures involved in the metabolism of several short non-coding nuclear RNAs. Yet surprisingly little is known about the structure and function of coilin, and a comprehensive model to explain the origin of Cajal bodies is also lacking. Here, I summarize recent results on Cajal bodies and coilin and discuss them in the context of the last three decades of research in this field.

KEYWORDS:

• Cajal bodies
• coilin
• snoRNA
• snRNA
• snRNP
• telomerase RNA

Introduction

Cajal bodies (CBs) were first observed 120 years ago by the famous Spanish cytologist Ramón-Y-Cajal as nuclear structures in neurons with high affinity for silver staining [1–3]. However, major research interest in CB was sparked by the discovery of coilin, a major scaffolding protein and marker of CB, in the early 1990s [4,5]. At the same time, spliceosomal small nuclear RNA (snRNA) were found to accumulate in discrete nuclear foci [6,7], which were later identified as CBs [8]. CBs were subsequently found to associate with genes for snRNAs and small nucleolar RNAs (snoRNAs) [9–13]. This finding was further confirmed by chromatin immunoprecipitations that found that coilin associates with snRNA genes [14] and CBs have been proposed as chromosome organizers that arrange snRNA genes and regulate snRNA transcription [15,16]. Consistent with the function of CBs in snRNA expression, the association of CBs with U2 snRNA genes is dependent on transcription of this gene [16], U2 snRNA precursors have been found in CBs [17], and the snRNA nuclear export complex has been localized to CBs [18]. But the function of CBs is not limited to early snRNA biogenesis. Upon return from the cytoplasm, snRNAs, now in the form of small nuclear ribonucleoprotein particles (snRNPs), again visit CBs, where they are post-transcriptionally modified and presumably also truncated at the 3’ end by TOE1 [19–23]. CBs are further involved in snoRNA maturation and these short non-coding RNAs pass through CBs before reaching the nucleolus [14,24–29]. Another important non-coding RNA that is associated with CBs is human telomerase RNA (hTR), and CBs likely play a role in its biogenesis as well.

The molecular mechanism that controls CB assembly has been studied since the discovery of coilin. Early observations linked the occurrence of CB to RNA transcription, the cell cycle, and coilin phosphorylation [8,30–32]. Furthermore, a correlation between cell transcriptional activity and CB number has been observed in neurons and plants [3,33–36]. However, CBs are also present in embryonic cells prior to the onset of transcription, suggesting that CBs may exist even without ongoing RNA synthesis, at least in developing embryos, where significant amounts of maternally provided protein and RNA are stored [37–39]. In recent years, a model that explains the formation of non-membrane organelles based on liquid-liquid phase separation (LLPS) has attracted much attention, and CB has been proposed as one of the archetypal membrane-free structures whose formation is driven by LLPS (reviewed, e.g., in [40,41]. In this review, I would like to discuss recent discoveries about coilin, the origin of the CBs and their function.

Cajal body formation

Using various experimental models, it has been shown beyond any doubt that coilin is an essential CB protein and its depletion leads to CB disappearance [39,42–46]. Upon removal of coilin, some CB components form alternative structures called ‘residual bodies’, but these structures lack snRNPs, indicating an indispensable role for coilin in bringing all CB components together [42]. In the search for coilin domains essential for CB, it has been shown that the N-terminal domain (1–92 amino acids in human coilin) is critical for coilin self-interaction and that inhibition of this interaction by substitution of a single amino acid effectively prevents CB formation [47,48]. Interestingly, replacing the N-terminal domain of coilin with another self-interacting motif did not rescue CB assembly, suggesting that the N-terminus of coilin is specific and may interact with other factors (e.g., NOLC1, see below) in regulating CB formation [48].

However, coilin self-interaction is necessary but not sufficient for CB assembly because coilin self-interacts inside and outside CBs to a similar extend [49]. Soon after the discovery of coilin, post-translational modifications were proposed to significantly affect the ability of coilin to form CBs. First, coilin phosphorylation specific for mitotic cells was identified in human cells and the importance of coilin phosphorylation for CB integrity was articulated [30,50,51]. Since then, the phosphorylation of several threonines (in particular T122 and T303) and serines (in particular S271, S489 and S566) has been identified in several protein-wide screens (https://www.phosphosite.org). More detailed studies in which phosphorylated amino acids were mutated confirmed the link between individual coilin phosphorylation sites and CB integrity [47,51–54] reviewed in [55]. A significant amount of protein kinases and phosphatases have been shown to affect CB integrity and number [56,57]. Finally, several kinases including CDK2-cyclin E, UHKM1 and VRK1 have been proposed to directly phosphorylate coilin [58–61].

While coilin phosphorylation is clearly a critical factor in CB formation, less is known about the molecular mechanism responsible for this phenomenon. Almost no modifications were found in the N-terminal self-interacting domain, suggesting that phosphorylation does not directly regulate coilin oligomerization. Phosphorylation of some amino acids reduces RNA binding, indicating that coilin interaction with RNA is an important factor in maintaining CBs [52]. There is much evidence that coilin directly binds RNA, but the RNA binding motif has not been unambiguously identified [14,62–65]. This lack of knowledge makes it difficult to directly correlate individual phosphorylation sites with RNA binding. The LLPS model predicts that RNA is a fundamental component of membraneless organelles. CBs contain numerous non-coding RNAs, yet the importance of RNA for CB assembly has not been directly tested. The only exception is coilin from Arabidopsis thaliana, whose in vitro aggregation is stimulated by U1 snRNA [63].

Several phosphorylated amino acids are found in the C-terminus of coilin. The most interesting is the phosphorylation of S489, which is located in the evolutionary conserved loop. Mutation of S489 alters the interaction of coilin with several non-coding RNAs found in CBs [52,66]. The C-terminus adopts a Tudor-like structure but does not bind methylated arginine [67]. However, it should be noted that arginine methylation is important for CB maintenance, perhaps via disruption of snRNP biogenesis [68–70]. The C-terminus interacts with snRNPs and Sm proteins, and snRNPs and Sm proteins have been found to promote CB assembly [45,71–75]. In addition, artificial tethering of snRNP components to chromatin seeds CB formation [76]. The importance of snRNAs and snRNPs for CB existence was further documented by inhibition of snRNA/snRNP biogenesis, which resulted in CB disruption [77–84]. It is therefore plausible to speculate that phosphorylation in the C-terminal domain, and in particular S489, affects the interaction with snRNPs, which in turn modulate CB assembly. To test this hypothesis, we need to investigate whether the coilin-snRNP complex is essential for CB assembly, determine a molecular mechanism of how the Tudor-like domain binds snRNPs and Sm proteins, and assess the role of S489 in this interaction.

Another link between CB and phosphorylation is the protein NOLC1/Nopp140, which is one of the most phosphorylated proteins in the cell. NOLC1 is an essential protein that accumulates in nucleoli and CBs (reviewed in [85]. NOLC1 interacts with small nucleolar RNPs (snoRNPs) but is not directly involved in rRNA modifications and has been suggested to act as an snoRNP chaperone [86–88]. NOLC1 interacts with coilin and in particular with its N-terminal self-interacting domain [48,89]. Partial NOLC1 depletion by CRISPR/Cas9 reduces CB size [90], whereas NOLC1 downregulation by RNAi results in complete loss of CBs [48]. NOLC1 has been proposed as a modulator of the N-terminal domain of coilin. When expressed alone, coilin N-terminus forms long filaments in the cytoplasm, while in the nucleus the N-terminal domain assembles into round puncta. NOLC1 has been identified as the nuclear factor that modulates coilin behavior and promotes the formation of rounded puncta [48].

This provides a new view of CB formation in which coilin oligomers are central building blocks of CBs and their N-terminal self-interacting domain is ready to form large aggregates (Figure 1). However, the ability to condensate and form a nuclear body is modulated by additional extrinsic factors, including NOLC1 and protein kinases and phosphatases, as well as other parts of coilin, namely the C-terminal domain. Experiments testing the predictions of the LLPS model showed that CB does not follow simple LLPS and that additional factors are required to explain coilin and CB behavior, which is consistent with this multifactorial model of CB assembly [91]. This arrangement allows coilin and CBs to respond efficiently to changes in the intracellular metabolism of multiple non-coding RNAs as well as to integrate signals coming from various signaling pathways.

Figure 1. Formation of the Cajal body. A coilin-centered view of CB formation. Coilin is the scaffolding protein of CBs, which contain two conserved domains at the N and C termini, termed NTD and CTD. A structure of these domains was approximated by Alphafold (alphafold.Ebi.ac.uk) using human coilin as input. In the first step, coilin self-interacts via the NTD to form oligomers, which are the basic building blocks of CBs. This step is likely spontaneous, but can be modulated by factors that interact with NTD (e.g. NOLC1). In the second step, coilin oligomers condense to form a microscopically visible structure. This step is regulated by several extrinsic factors, including protein kinases and phosphatases that phosphorylate/dephosphorylate coilin and protein interaction partners, such as NOLC1 and snRNPs. The image was created with BioRender.com.

Cajal body and coilin function

Most of the available data point to a function of CB in sn/snoRNP metabolism, namely the enhancement of snRNP assembly [49,92–95] (Figure 2). The strongest evidence came from experiments in developing zebrafish embryos, where the lethal phenotype induced by coilin depletion was rescued by injection of mature snRNPs [39]. Coilin knockout results in fertility problems in mice and recent experiments in plants showed that coilin inactivation negatively affects plant growth [96,97]. These experiments show that, with the exception of Drosophila where no obvious phenotype associated with coilin knockout was observed [43], coilin is important for the proper development of multicellular organisms. It has been suggested that the concentration of snRNPs in the CB enhances their final maturation [92,94]. Similarly, recent experiments revealed that NOLC1 downregulation displaces Cajal body-specific RNAs (scaRNAs) from CBs, resulting in lower modification of snRNAs [90]. Bringing together guide scaRNAs and their snRNA targets in CBs likely promotes the efficiency of the modification process. This is in contrast to Drosophila, where coilin knockout does not alter scaRNA function and snRNAs are modified normally [98]. It should be noted that Drosophila coilin, and in particular the N-terminal sequence, has diverged from the consensus sequence and it is therefore unclear whether coilin and CBs have the same functions in Drosophila as in other metazoans [65].

Figure 2. Function of Cajal bodies. The metabolism of several RNPs is closely associated with CBs. snRNA and some snoRNA genes (U3) are found in the vicinity of CBs. Newly transcribed pre-snRNAs pass through CBs on their way to the cytoplasm where they acquire the Sm ring (yellow bolls). After returning from the cytoplasm, core snRNPs (snRNA+Sm proteins) visit CBs again to complete their biogenesis (trimming of the 3’ extension, ribose methylation and pseudouridylation, and addition of snRNP-specific proteins). Mature snRNPs leave the CB to localize to nuclear speckles and catalyze RNA splicing. Defective and incomplete snRNPs are sequestered in CBs by an unknown mechanism. Human telomerase RNA (hTR) and snoRNAs pass through CBs during their biogenesis to acquire 2,2,7-trimethylation at the 5’ end (hTR, U3 and U8 snRNAs) and possibly assemble here into functional RNPs before reaching their final destination. The image was created with BioRender.com.

My laboratory and others have shown that inhibition of the final steps of snRNP biogenesis results in sequestration of various snRNP assembly intermediates in CB [49,74,93,95,99–102]. Similarly, inhibition of snRNP recycling after splicing increases the accumulation of specific snRNPs in CB [93,103], which is consistent with the view that CB are involved in snRNP reassembly after splicing and quality control of the snRNP (re)assembly process. The molecular mechanism that discriminates between assembly intermediates and fully mature particles is currently unclear. Initially, we proposed that incomplete snRNPs are sequestered in CBs in a SART3-dependent manner [74]. Later studies did not support this conclusion and we found that SART3 associates with incomplete snRNPs within post-splicing complexes [104]. Thus, the proofreading factor that specifically recognizes incomplete snRNPs and sequesters them in CBs is currently unknown.

hTR is a non-coding RNA containing H/ACA box found in the subclass of snoRNAs and scaRNAs. Similar to snoRNA/scaRNAs, hTR has been found in CBs [105,106]. It has been suggested that CBs are important for the proper function of telomerase [107–110]. However, later studies using coilin knockout mice and human cells failed to observe defects in telomere maintenance, calling into question the essential role of coilin and CBs for telomerase function [44,111,112]. But the story of CBs and hTR seems to be more complex. Recent results showed that the 2,2,7-trimethylguanosine cap located at the 5’ end of hTR regulates telomerase function [113]. 5’ cap hypermethylation is catalyzed by TGS1, which is located in CB, is one of the strongest interaction partners of coilin and regulates telomerase function [14,114]. Similar to snRNAs pseudouridylation and methylation, coilin and CB may promote hTR post-transcriptional modifications (cap hypermethylation and pseudouridylation) necessary for proper telomere maintenance [113,115].

Finally, I would like to briefly discuss CBs in plants, where coilin and CBs may have alternative functions to their animal counterparts. In contrast to animal CBs, plant CBs can accumulate large amounts of poly-adenylated RNA, which probably regulates gene expression during germline development [116]. CBs also function in plant defense against various pathogens [97,117]. Several mechanisms have been proposed to mediate antiviral protection. One is based on PARP, which is involved in plant response to various stresses, including defense against pathogens (reviewed in [118]. Upon activation, PARP and other PARylated proteins accumulate in CBs [119]. PARP localization to CB correlates with increased production of salicylic acid, which has also been linked to the antiviral function of coilin [120,121]. In animals, PML bodies are involved in innate antiviral immunity (reviewed in [40]. In plants PML bodies have not been found [122]. It is therefore plausible that CB has replaced PML in the innate immunity response in plants. Or alternatively, innate immune defense was one of the primordial functions of the CB, which in mammals was taken over by the PML bodies.

Perspectives

Considering that their existence was discovered 120 years ago, CBs remain rather enigmatic and are only slowly revealing their secrets. One major obstacle is the lack of an in vitro system that recapitulates CB formation and allows the study of the basic molecular principles that drive CB assembly. A second critical aspect of CB research is to elucidate the molecular function of coilin. It has been proposed that coilin is an RNase, DNase, chaperone, etc., but the experiments that would unambiguously determine coilin function are still lacking. Without these data and new approaches, our progress in unraveling the mysteries of CB will continue to be slow.

Acknowledgments

This work was supported by the Czech Science Foundation (21-04132S) and the institutional funding (RVO68378050 and RVO68378050-KAV-NPUI).

Disclosure statement

No potential conflict of interest was reported by the author.

Data availability statement

There are no primary data associated with this manuscript.

Additional information

Funding

The work was supported by the Akademie Věd České Republiky [RVO68378050]; Grantová Agentura České Republiky [21-04132S].