De Novo Formation of a Subnuclear Body

See allHide authors and affiliations

Science  12 Dec 2008:
Vol. 322, Issue 5908, pp. 1713-1717
DOI: 10.1126/science.1165216


The mammalian cell nucleus contains structurally stable functional compartments. We show here that one of them, the Cajal body (CB), can be formed de novo. Immobilization on chromatin of both CB structural components, such as coilin, and functional components of the CB, such as the SMN complex, spliceosomal small nuclear ribonucleoproteins (RNPs), small nucleolar RNPs, and small Cajal body–specific RNPs, is sufficient for the formation of a morphologically normal and apparently functional CB. Biogenesis of the CB does not follow a hierarchical assembly pathway and exhibits hallmarks of a self-organizing structure.

Compartmentalization of the nucleus contributes significantly to the regulation of nuclear functions (15). The structure of intranuclear compartments is maintained in the absence of defining membranes and despite a highly dynamic exchange of their components with the surrounding nucleoplasm (6, 7). It has been suggested that nuclear bodies are formed by either an ordered assembly pathway or self-organization (3, 5). To discriminate between these two alternatives, we used as a model the Cajal body (CB), a major nuclear body involved in the biogenesis and recycling of several classes of small nuclear ribonucleoproteins (snRNPs).

To assess the ability of CB proteins to nucleate the formation of a CB de novo, we immobilized fusion proteins between the Escherichia coli Lac Repressor (LacI) and green fluorescent protein (GFP)–tagged CB components on chromatin in a previously characterized HeLa cell line (8) in which 256 repeats of the Lac operator (LacO) (9) are stably integrated on chromosome 7 (10). Several CB fusion proteins, including coilin, SMN (survival of motor neuron gene), and Nopp140, were efficiently targeted to the LacO array, and their accumulation appeared as discrete foci (Fig. 1). LacI stained with red fluorescent protein (Cherry-LacI) was coexpressed to distinguish the position of the LacO array from endogenous CBs. Immobilization of one of the two major structural components of CBs, coilin or SMN, was sufficient to trigger the formation of CBs, as indicated by the presence of endogenous coilin (Fig. 1A) and SMN (Fig. 1B) in the newly formed bodies. CBs formed in 52.1% of cells expressing LacI-coilin and in 50.7% of cells expressing LacI-SMN (Fig. 1F). Tethering of the non–CB protein SR (serine-arginine–rich) splicing factor SC35 (Fig. 1C), the major nucleolar protein B23 (Fig. 1D), or nucleoporin NUP62 (fig. S1A) did not lead to formation of a coilin-positive body. Immobilization of the PML body component PML protein did not lead to the formation of CB (fig. S1B); however, PML protein nucleated a PML body, as indicated by recruitment of the PML component Sp100 (Fig. 1E). We conclude that one of the major nuclear compartments, the CB, can be formed de novo. Furthermore, our results with PML suggest that this type of nucleation may be applicable to other nuclear bodies.

Fig. 1.

Immobilization of a single structural component leads to nuclear body formation. Immunofluorescence microscopy on HeLa cells transiently transfected with various GFP-LacI fusion proteins (green), Cherry-LacI (red) and stained with the indicated antibody (A to E). (Insets) High magnification view of formed CB. Arrows, the location of the tethered fusion proteins. Quantification of de novo CB formation (F). Values represent means ± SD (n = 65 to 95 cells) from two independent experiments. Scale bar, 2 μm.

De novo formed CBs were functional by several criteria. They contained components of several functional groups of proteins present in endogenous CBs, including: the SMN protein complex, as indicated by the presence of Gemin2 (Fig. 2A), Gemin5 (fig. S2A); components of spliceosomal snRNPs (SmB/B′ and SmD1 Fig. 2B), the U4-specific protein hPrp31 (fig. S2B); the U5-specific protein hSnu114, and the small nucleolar ribonucleoproteins' (snoRNPs) chaperone Nopp140 (Fig. 2C). In addition, U85 scaRNA, a CB-specific snRNA guide RNA, is present (Fig. 2D). As expected for a functional CB, coimmunoprecipitation experiments demonstrated physical interaction between GFP–LacI-coilin and endogenous coilin, SMN, and Gemin2 (Fig. 2F). Additionally, in the presence of the dominant-negative mutant ΔRG coilin, which lacks the RG domain responsible for the interaction with SMN, de novo CBs display the same physical separation of CB from their twin nuclear gem body as observed for endogenous CBs (Fig. 2E) (7). Inverse fluorescence recovery after photobleaching (iFRAP) in de novo CBs, formed by tethering Cherry–LacI-SMN, demonstrated that GFP-coilin has similar dissociation kinetics in de novo formed CBs as in endogenous CBs (Fig. 2G) (7). Moreover, the size of de novo formed CB on the LacO array corresponds to the size of endogenous CBs present in cells, which suggests a size criticality in the nuclear body formation.

Fig. 2.

Tethering coilin forms bona fide functional CBs. Immunofluorescence microscopy on HeLa cells transiently transfected with GFP–LacI-coilin and/or Cherry-LacI and stained with indicated antibodies (A to C). HeLa cells transiently transfected with GFP–LacI-coilin, blue fluorescent protein–tagged LacI, and U85 scaRNA detected by a specific probe labeled with Cy3 using fluorescence in situ hybridization (D). De novo CBs display the same physical separation of CB from gems as observed for endogenous CBs in the presence of untagged ΔRG coilin (E). (F) Coimmunoprecipitation of GFP fusion proteins with indicated partner. In, input; Sp, supernatant; and IP, + immunoprecipitated material. GFP-coilin has similar dissociation kinetics measured by iFRAP in de novo CBs formed by immobilization of Cherry–LacI-SMN as for endogenous CBs (G). Scale bar, 2 μm.

To establish the kinetics of de novo CB formation, we treated cells with isopropyl-β-d-thiogalactopyranoside (IPTG) for 16 hours, which prevents binding of fusion proteins to the LacO, which, in turn, leads to the disassembly of the de novo formed CB (fig. S3). When IPTG is washed out, LacI-fusion proteins accumulated on the LacO array within 30 min (fig. S3A), with newly forming CBs visible after 2 hours (fig. S3, C, D, and E). No differences in the assembly dynamics of various CB components were detected, which suggested that the components are not recruited in a stepwise manner, but rather, accumulate concurrently.

Coilin and SMN are known to be essential for the structural integrity of CBs (11). To test whether coilin and SMN are also required for CB formation, we selectively depleted the cells of coilin or SMN by RNA interference (fig. S4, A and B) and probed for formation of CBs. Depletion (knockdown) of SMN prevented formation of de novo CBs when coilin was tethered (Fig. 3, A and B). Similarly, no CBs formed when coilin was immobilized in primary fibroblasts derived from a spinal muscular atrophy (SMA) patient with severely reduced SMN protein levels (fig. S4C). However, immobilized coilin is still capable of accumulating fibrillarin (Fig. 3C) and Nopp140 on the array, which indicates that, in the absence of SMN, coilin still interacts with snoRNPs. When the dominant-negative ΔRG coilin mutant was tethered, which disrupts interaction between coilin with SMN (12), in a wild-type SMN background, CB did not form de novo (fig. S5A). Tethering of SMN to chromatin in cells when coilin was knocked down was still sufficient to nucleate de novo residual gem bodies detected by Gemin2 (Fig. 3D), but was insufficient to cause accumulatation of snRNPs and snoRNPs (Fig. 3E) (13). Dominant-negative SMNΔ2b (fig. S5, B and C) and SMNΔC26 (fig. S5, D and E) mutants, which are unable to homopolymerize SMN (14), failed to nucleate CBs (fig. S5, B to E), which suggests that SMN's self interaction is a critical step in CB formation. These data indicate that coilin and SMN are required factors that act cooperatively to facilitate CB formation.

Fig. 3.

Coilin and SMN are required for Cajal body formation. Immobilization of coilin on chromatin in HeLa cells lacking SMN did not lead to formation of de novo CBs (A and B), but lead to accumulation of snoRNPs on the array detected by fibrillarin (C). Immobilization of SMN on chromatin in HeLa cells lacking coilin could nucleate de novo gem bodies detected by Gemin2 (D), but did not lead to accumulation of snRNPs (E). Scale bar, 2 μm.

Given the two CB assembly models, the ordered assembly pathway and self-organization, two distinct predictions about the ability of other CB components to nucleate de novo formation can be made. If CBs form in a hierarchical pathway, only upstream components should be able to nucleate formation. In contrast, if CBs form by self-organization, many, if not all, CB components should be sufficient to nucleate CBs. When, Gemin3, a SMN complex component, was tethered to the LacO array, de novo CBs were nucleated with the same efficiency (51.6%) as coilin (52.1%) or SMN (50.7%) (Fig. 4, B and H). Nopp140, a snoRNP chaperone, which directly interacts with the N terminus of coilin (15), nucleated CB with even higher efficiency (91%) (Fig. 4, A and H). These CBs contained all the same CB components as the de novo CBs formed by tethering of coilin or SMN (fig. S6), which demonstrates that proteins other than coilin and SMN have the ability to form CB.

Fig. 4.

Other Cajal body components are capable of forming Cajal bodies de novo. Immunofluorescence microscopy on HeLa cells transiently transfected with CB components fused with GFP-LacI, Cherry-lacI, and coilin-specific antibody (A to G). Arrows indicate the location of de novo formed CB. Immobilization of U92 scaRNA tagged with the MS2 loop by coexpression of GFP–LacI-NLS-MS2 coat protein on chromatin leads to de novo CB formation (F). In contrast, in the absence of the U92-MS2 scaRNA, GFP–LacI-MS2 coat protein is targeted to the LacO array only (G). Quantitative analysis of de novo CB formation efficiency on the LacO array (H). Values represent the means ± SD (n = 65 to 95 cells) from two independent experiments (H). Scale bar, 2 μm.

Although spliceosomal U snRNPs are major components of the CB, wherein they undergo the final steps of maturation and are recycled after splicing, they are not considered structural CB components (16, 17). Tethering of SmD1, which directly interacts with SMN; SmE and SmG, which interact indirectly with SMN as a part of the Sm ring (18); and Tgs1, the trimethylguanosine synthetase, which interacts with Sm proteins (19), nucleated CB with high efficiency (81.7 to 93.1%) (Fig. 4H). In contrast, immobilization of Snurportin1, which binds the hypermethylated cap of snRNPs and promotes their import to the nucleus (20), nucleated de novo CBs with lower efficiency (48.3%) (Fig. 3H and fig. S7D) than Sm proteins. To exclude the possibility that these effects were due to the association of these Sm proteins with immature snRNPs, which are still part of the SMN complex, we probed the ability of mature snRNPs to nucleate CB. To this end, we tethered the U2 snRNP–specific A′ protein (fig. S7E), the U4 snRNP–specific hPrp3 (Fig. 4D), the U4/U6 snRNP assembly factor SART3 (fig. S7F), and the U5 snRNP–specific human WD repeat domain 57 (WDR57 or snRNP 40 kD) (fig. S7G) to the LacO array. All these snRNP-specific proteins efficiently form de novo CBs with 88.6 to 93.4% efficiency (Fig. 4H).

CBs also play a significant role in the assembly steps of nucleolar snoRNPs (21), and we probed whether snoRNPs are able to nucleate CBs. The snoRNP-specific proteins box C/D U3 snoRNP–specific 55-kD protein (Fig. 4E), box C/D methyltransferase fibrillarin (fig. S7H), and box H/ACA snoRNP–specific protein dyskerin (NAP57) (fig. S7I) were tethered to the LacO array. In all cases, the snoRNP-specific proteins yielded the lowest efficiency of CB formation (from 34.1 to 37.8%) (Fig. 4H). Overall, these data strongly demonstrate that immobilized mature snRNPs have the ability to form CBs, and snRNPs, in general, do so with a significantly higher efficiency than snoRNPs, SMN, or coilin.

Finally, to test whether not only proteins, but also a small CB-associated RNA, could nucleate de novo formation, we tethered the U92scaRNA using the bacteriophage MS2 stem loop–coat protein system (22). Tethered U92 scaRNA, indeed, nucleated CB with 41.6% efficiency (Fig. 4, F and G). Overall, these data indicate that snRNPs, snoRNPs, and scaRNAs can nucleate CBs; of these, snRNPs are the most efficient (Fig. 4H). The fact that these functional groups of CB components have the capacity for de novo CB biogenesis (table S1) indicates that CB formation does not follow a strict linear assembly pathway, but can be triggered by a number of functional CB components.

Here, we demonstrate the formation of a nuclear body de novo. We suggest that CB formation occurs by self-organization based on two observations (3, 5, 23, 24). First, formation of CBs does not involve a hierarchical assembly pathway, because it can be initiated by a large array of CB components, and second, CB de novo assembly occurs temporally by concomitant, not sequential, association of proteins with the newly forming structure. Specifically, it appears that CBs are formed by protein–ribonucleoprotein interactions among CB components, which then directly or indirectly bind coilin and SMN. The previously reported, self-oligomerization of coilin and SMN might facilitate the stabilization of transient interactions among CB components. Our observation that a body forms when PML protein is tethered further suggests that de novo formation may also apply to other nuclear bodies. It seems that the CB does not require a specific locus for nucleation in contrast to the nucleolus, which is formed around specific ribosomal DNA loci. Our findings that snRNPs are the most efficient CB nucleators suggest that CBs primarily form as result of a local snRNP accumulation associated with snRNP function, such as snRNP assembly or regeneration after splicing (16, 17). We suggest that once a threshold of local snRNP concentration is reached, CB formation is initiated by self-organization. This is consistent with previous observations that Sm protein expression enhances the formation of CBs in cells typically lacking CBs, and CB integrity depends on the cellular level of splicing activity and the absolute concentration of nuclear snRNP (25). A concentration of snRNPs could be locally elevated in the proximity of multiple genes at high levels of transcription, when they coalesce to form a transcription center (3, 26).

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Table S1

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article