Distinct pathways of ribonucleoprotein transport exist within the nucleus, connected to their biogenesis and maturation. These occur despite evidence that the major mechanism for their movement within the nucleus is passive diffusion. Using fusions of Sm proteins to YFP, CFP and photoactivatable GFP, I have demonstrated that pathways with uni-directional bulk flow of complexes can be maintained within the nucleus despite multi-directional exchange of individual complexes. Newly imported splicing small nuclear ribonucleoproteins (snRNPs) exchange between Cajal bodies (CBs) within a nucleus and access the cytoplasm, but are unable to accumulate in speckles. By contrast, snRNPs at steady-state exchange freely in any direction between CBs and speckles, but cannot leave the nucleus. In addition to these surprising qualitative observations in the behaviour of nuclear complexes, sensitive live-cell microscopy techniques can detect subtle quantitative disturbances in nuclear dynamics before they have had an effect on overall nuclear organization. Inhibition of the nuclear export factor, CRM1, using leptomycin B results in a change in the dynamics of interaction of newly imported snRNPs with CBs. Together with the detection of interactions of CRM1 with Sm proteins and the survival of motor neurons (SMN) protein, these studies suggest that the export receptor CRM1 is a key player in the molecular mechanism for maintaining these pathways. Its role in snRNP trafficking, however, appears to be distinct from its previously identified role in small nucleolar RNP (snoRNP) maturation.
The nucleus of the mammalian cell contains many distinct structures, enriched in different proteins and ribonucleoproteins (RNPs) (Carmo-Fonseca, 2002; Matera and Shpargel, 2006; Misteli, 2005). The nucleus is a stable structure formed despite the constant, often rapid, flux of molecules through structures within it. Proteins and RNAs have been demonstrated to move within the nucleus by apparently random diffusion (Dundr et al., 2004; Kruhlak et al., 2000; Misteli, 2005; Misteli et al., 1997; Phair and Misteli, 2000; Politz and Pederson, 2000; Snaar et al., 2000). Even proteins involved in pathological formation of nuclear aggregates are in constant exchange between the aggregates and the nucleoplasm (Tavanez et al., 2005). The sub-nuclear structures formed by this constant dynamic exchange are, however, amenable to isolation for biochemical and proteomic analyses, suggesting that the intermolecular interactions taking place within them are not exclusively transient (Andersen et al., 2002; Lam et al., 2002; Saitoh et al., 2004).
The U1, U2, U4, U5 and U6 small nuclear ribonucleoproteins (snRNPs) are essential splicing factors. Each snRNP consists of a uridine-rich small nuclear RNA (snRNA) and seven common (Sm) proteins (B/B′, D1, D2, D3, E, F and G), which form a ring-like structure around the snRNA. In addition to these core proteins, each snRNP also contains particle-specific proteins. At steady state, snRNPs show a complex localization within the cell. The majority of snRNPs localize to speckles, with a small amount localizing to Cajal bodies (CBs). The localization pattern of snRNPs also comprises a diffuse nucleoplasmic component and, sometimes, a diffuse cytoplasmic component. The biogenesis and maturation of splicing snRNPs is a complex process (Lührmann et al., 1990; Will and Luhrmann, 2001), with different stages occurring at different physical locations within the cells. The snRNA is transcribed within the nucleus and then exported to the cytoplasm as part of an export complex containing CRM1 (XpoI), PHAX, CBC and Ran-GTP (Fornerod et al., 1997; Izaurralde et al., 1997; Izaurralde et al., 1995; Ohno et al., 2000). The core Sm proteins are assembled onto the snRNA in the cytoplasm, forming a seven-membered ring around the snRNA (Kambach et al., 1999a; Kambach et al., 1999b; Lehmeier et al., 1994; Raker et al., 1996). The survival of motor neurons (SMN) protein is required for cytoplasmic addition of Sm proteins onto snRNAs (Liu and Dreyfuss, 1996; Pellizzoni et al., 1999). Insufficient expression of SMN results in the inherited neurodegenerative disease, Spinal Muscular Atrophy (SMA) (Lefebvre et al., 1995). The association of Sm proteins with snRNA is a prerequisite for 3′ end trimming and hypermethylation of the cap to form the characteristic 5′ trimethylguanosine (TMG) cap (Will and Luhrmann, 2001). The Sm domain and the TMG cap form a bipartite nuclear localization signal (NLS) that is required for re-import of the newly assembled core snRNP into the nucleus (Fischer et al., 1993; Hamm et al., 1990; Mattaj, 1986). Splicing snRNP import requires snurportin 1, importin-β and SMN (Massenet et al., 2002; Narayanan et al., 2004; Narayanan et al., 2002; Shpargel and Matera, 2005; Sleeman et al., 2001). There is increasing evidence from animal models of SMA that a defect in snRNP processing may be responsible for the cellular pathology of SMA (Eggert et al., 2006; Winkler et al., 2005).
Once re-imported, newly assembled snRNPs accumulate preferentially in CBs (Carvalho et al., 1999; Sleeman and Lamond, 1999). Spliceosomal snRNAs contain numerous pseudouridines and 2′-O-methyl groups (Massenet et al., 1998). Guide RNAs termed scaRNAs (small CB RNAs) have been demonstrated to direct such modifications of the snRNA (Darzacq et al., 2002; Jady et al., 2003) within CBs. The CB has also been implicated as the cellular location for the addition of the snRNP-specific heterotrimeric splicing factor SF3a to U2 snRNPs, completing their maturation (Kramer et al., 2005; Nesic et al., 2004), suggesting that snRNPs leave CBs fully mature. Labelling of newly imported snRNPs with fluorescent protein (FP)-tagged Sm proteins shows a delay of between 6 and 24 hours during which GFP-Sm proteins can be detected in the nucleoplasm and CB, where they colocalize with SMN, but not in nuclear speckles (Sleeman et al., 2001; Sleeman et al., 2003). Later, labelled snRNPs are seen also in speckles, suggesting that the CB phase is a necessary and rate-limiting step in the nuclear snRNP biogenesis pathway. The delays seen experimentally are not a direct measure of the length of time required for each individual snRNP to be matured, but reflect the time needed for a sufficient proportion of the FP-tagged Sm proteins to be assembled into snRNPs and accumulated in speckles to a level that can be detected. Further investigation of the dynamics of snRNP maturation and of the pathways controlling it is vital because of its potential involvement in SMA, as well as for its implications for the further understanding of the formation and maintenance of mammalian nuclear structure.
snRNPs that accumulate in the nucleus after short expression times contain fully assembled Sm cores
Previous studies using FP-tagged Sm proteins have shown that they are incorporated into snRNPs and localize correctly within the nucleus (Sleeman et al., 1998; Sleeman et al., 2001; Sleeman and Lamond, 1999). There is a period of time when cells have recently begun to express FP-tagged Sm protein during which snRNPs tagged with FP-Sm proteins are present diffusely within the cytoplasm and nucleoplasm but concentrate only in CBs, not in splicing speckles. Both the diffuse nucleoplasmic signal and the nucleolar accumulation are more prominent in cells expressing higher levels of the tagged Sm proteins (J.E.S., unpublished), leading to the suggestion that these components of the localization may represent Sm proteins not assembled onto snRNA cores and result from overexpression of the FP-tagged Sm proteins. To address this, acceptor-bleaching fluorescence resonance energy transfer (FRET) experiments were performed using cells expressing YFP-SmB and CFP-SmD1 for different lengths of time. In a FRET experiment, if two fluorophores have overlapping spectra, are close to each other (<10 nm) and in a favourable orientation, then one (the donor, CFP) can transfer energy to the other (the acceptor, YFP) in addition to emitting energy to be detected as a fluorescent signal (Chen et al., 2003; Day et al., 2001; Wouters et al., 2001). This energy transfer can be detected in several ways. In acceptor-bleaching FRET experiments (Chusainow et al., 2005), a laser is used to bleach the population of acceptor (YFP) molecules in a region of interest in a living cell so that the donor molecules can no longer transfer energy to them. This results in a transient increase in the measured fluorescence intensity of the donor (CFP).
Living HeLa cells expressing YFP-SmB and CFP-SmD1 were used for acceptor-photobleaching FRET experiments. The order of assembly of Sm proteins onto the snRNA core of snRNPs is known (Kambach et al., 1999a; Kambach et al., 1999b; Raker et al., 1996) (Fig. 1A). Only after formation of the complete core domain will SmB and SmD1 be found in close proximity, so FRET will only occur between CFP-SmD1 and YFP-SmB in a fully assembled snRNP Sm core (Fig. 1B). Because each Sm protein is only represented once in each core assembly, coexpression of CFP-SmB and YFP-SmB provides a negative control. HeLa cells were transfected with plasmids to express FP-Sm fusion proteins, and used for FRET experiments after 48 hours, when both tagged proteins accumulated in CBs and speckles and, in some cells, the nucleolus. Cells were selected that showed low levels of expression of the tagged proteins, with a similar intensity seen for each of the two. Cells with high expression levels show cytoplasmic accumulations of Sm proteins and were not analysed. In each of the selected cells, a 200-msecond stationary laser pulse at 532 nm was used to bleach the YFP-SmB signal in a region of interest, and the effect of this bleaching on the intensity of the CFP-SmD1 signal was measured over time. Representative plots of signal intensity against time are shown in Fig. 2. A transient increase in the intensity of CFP-SmD1 (donor) was seen on bleaching of YFP-SmB (acceptor) in CBs, nucleoli and speckles, demonstrating that fully assembled Sm cores are present in each of these locations. The FRET efficiencies calculated for each region were similar and approximately 25% of the maximum value obtained using a control construct in which CFP and YFP are separated by only seven amino acids. FRET experiments were also performed in HeLa cells 24 hours after transfection, when the fusion proteins were in CBs and a diffuse nucleoplasmic pool, but not in speckles. Similar FRET efficiencies were detected (Fig. 2), indicating that, although some of the fluorescent signal may result from free Sm proteins, fully formed Sm snRNP cores are present in all nuclear regions examined at both timepoints.
snRNPs show unrestricted exchange between nuclear structures
Splicing snRNPs show a pathway of entry into the nucleus, associated with their biogenesis. A one-way nuclear pathway has been reported for the protein NHPX that binds both to box C/D snoRNAs and to U4 snRNA. NHPX interacts with speckles only when newly expressed before forming a stable interaction with the nucleolus (Leung and Lamond, 2002). It is not clear whether newly imported snRNPs also follow a uni-directional pathway with speckles the final destination. As snRNPs are believed to leave the CB fully mature (Kramer et al., 2005; Nesic et al., 2004), snRNPs that have reached speckles would not be expected to return to CBs during their maturation if the biogenesis pathway is one-way. In order to gain more information about the directionality of the snRNP import pathway, a fusion of the core snRNP protein, SmB to photoactivatable-GFP (PA-GFP), was made. PA-GFP shows very weak fluorescence at 488 nm until activated using a 406 nm laser, following which it gains fluorescence similar to that of enhanced GFP (EGFP). PA-GFP-SmB shows a similar temporal pathway of entry into the nucleus as YFP- and CFP-SmB (data not shown). HeLa cells were transfected with PA-GFP-SmB and incubated for 48 hours, until PA-GFP-tagged snRNPs were seen in CBs and speckles, identified by their size and morphology, using a PA-GFP (405/40 nm) excitation filter. Regions of interest of approximate diameter 1.5 μm within the nuclei were then activated using a fixed laser pulse of 200 mseconds at 406 nm and a short time-lapse sequence taken using a standard fluorescein isothiocyanate (FITC) filter set to visualize the exit of photoactivated snRNPs from the region of interest (Fig. 3). This sequence was repeated 20 times in rapid succession. Activation of PA-GFP-SmB signal in CBs (Fig. 3A,B; see supplementary material Movies 1, 2) or speckles (Fig. 3C; see supplementary material Movie 3) led to the rapid redistribution of signal throughout the entire nucleus, with no clear preference seen for uptake of tagged snRNPs into CBs or speckles, regardless of the type of structure centred in the activated region of interest. This argues against a specific one-way pathway for snRNP transport within the nucleus from CBs to speckles, and suggests that mature snRNPs can return to CBs.
Mature snRNPs do not show a measurable export from the nucleus to the cytoplasm
Splicing snRNPs are large complexes that are perceived as being stable with a slow rate of turnover. If this is the case, Sm proteins incorporated into fully mature snRNPs would not be expected to return to the cytoplasm. To address this question, fluorescence loss in photobleaching (FLIP) experiments were performed using cell line YFP-SmB-E1, a HeLa cell line constitutively expressing YFP-tagged SmB. A region of interest was selected containing a portion of the nucleus of one cell and a portion of the cytoplasm of a neighbouring cell (Fig. 4). This region was repeatedly bleached, using a 532 nm laser, with a short time-lapse sequence being taken between bleaching events to assess the effect of the bleaching (see supplementary material Movie 4).
As anticipated from the preceding studies using PA-GFP-SmB, bleaching a region within the nucleus led to a rapid loss of YFP-SmB-tagged snRNPs from all nuclear structures (Fig. 4). By contrast, no loss of nuclear signal was seen in the nucleus of the neighbouring cell bleached in the cytoplasm. These data demonstrate that, although there is rapid and unrestricted exchange of FP-Sm-tagged snRNPs within the nucleus, there is no significant exit of nuclear snRNPs to the cytoplasm over the 5-minute duration of this experiment.
Recently imported snRNPs are unable to accumulate in speckles, despite the presence of complete Sm cores
Cells that have recently begun to express FP-tagged Sm proteins show accumulation of FP-Sm in CBs but not speckles, with a diffuse cytoplasmic and nucleoplasmic FP-Sm signal also seen. HeLa cells were transfected with a plasmid to express PA-GFP-SmB and used for photoactivation studies 20 hours after transfection. Repeated activation of a region of the nucleus containing a CB demonstrated that, although PA-GFP-SmB could travel across the nucleus and accumulate in other CBs, the rest of the signal remained diffuse within the nucleoplasm (Fig. 5; see supplementary material Movie 5). No accumulation in speckles was seen, even at early stages of the experiment in which there was only a weak diffuse signal that would be insufficient to mask any localization in speckles. Although it is likely that some of the signal represents free Sm protein not assembled onto snRNPs, the FRET studies detailed above (Fig. 2) confirm that snRNPs with fully assembled Sm cores are present in both the CB and the diffuse nucleoplasmic pool at this timepoint. These partially mature snRNPs are clearly unable to localize to speckles, forming interactions of sufficient duration to result in an observed accumulation only within CBs.
At early timepoints, YFP-SmB can exit the nucleus into the cytoplasm
To determine whether recently imported snRNPs can return to the cytoplasm, FLIP experiments were performed as described above. In contrast to the results obtained for steady-state snRNPs, in HeLa cells expressing YFP-SmB for only 20 hours, repeated bleaching of a region encompassing part of the nucleus of one cell and part of the cytoplasm of its neighbour led to a decrease in nuclear signal in both nuclei (Fig. 6; see supplementary material Movies 6, 7).
This suggests that a proportion of the YFP-SmB signal seen in the nucleus at this early timepoint comprises protein that is not assembled into fully mature snRNPs and so can still leave the nucleus and access the cytoplasm. The purpose of this exchange is not clear but it may result from the absence of some molecular signal, either an RNA modification or an associated protein, marking mature snRNPs for retention within the nucleus. Measurement of the loss of YFP-SmB from CBs in cells undergoing bleaching of YFP-SmB signal from the cytoplasm demonstrated a similar rate of loss of signal from the CB as from the nucleus as a whole (Fig. 6). This suggests that the YFP-SmB-containing complexes that are able to access the cytoplasm are interchangeable with those present in the CB.
CRM1 localizes to CBs and is lost rapidly from them following treatment with leptomycin B
Leptomycin B (LMB) is a specific inhibitor of the export factor CRM1/exportin. CRM1 is required for the export of nascent snRNAs from the nucleus and its inactivation by binding to LMB has been used as an indirect way to block snRNP import into the nucleus (Carvalho et al., 1999; Sleeman et al., 2001). HeLa cells treated with LMB show a reduction in the accumulation of snRNPs in CBs. Furthermore, LMB treatment of cells at early stages of FP-Sm protein expression prevents the accumulation of newly imported snRNPs into nuclear speckles. The complexity of the effects seen on snRNP distribution following LMB treatment, together with the detection of CRM1 in CBs (Boulon et al., 2004; Ospina et al., 2005), suggests that CRM1 may have an additional role in transporting species between structures within the nucleus. CRM1 has been implicated in the transport of U3 small nucleolar RNPs (snoRNPs) from CBs to nucleoli (Boulon et al., 2004; Watkins et al., 2004). Treatment of HeLa cells with LMB results in a rapid loss of CRM1 from CBs (Fig. 7), suggesting that the interaction of CRM1 with CBs is dynamic and a result of its activity. Incubation times as short as 15 minutes cause total loss of CRM1 from CBs. Extended incubation with LMB (6 hours or more) is required for effects on other CB markers to be observed. These include the loss of snRNPs from CBs and the relocalization of coilin to the nucleolus (Carvalho et al., 1999; Sleeman et al., 2001).
CRM1 is required for the retention of snRNPs in CBs
In order to investigate the potential role of CRM1 in the interaction between newly imported snRNPs and CBs, HeLa cells were transfected with PA-GFP-SmB and incubated for 20 hours. The positions of 20 cells expressing PA-GFP-SmB and showing clear CBs were marked. One CB in each of these cells was activated using a single pulse from a 406-nm laser and time-lapse recording of the cells made for a 30-second period. LMB was then added to the cells at a concentration of 0.05 μg/ml (LC Laboratories, Woburn, MA). After 2 hours of incubation with LMB, each of the 20 cells was revisited and the single-pulse activation experiments repeated. Where possible, the same CB was used. The average intensity of signal in the CB was measured, corrected for bleaching and image variation by subtracting the average intensity of a region within a neighbouring, non-activated cell, and plotted against time (Fig. 8). Exponential decay curves were fitted to the data using non-linear regression software (Prism 4; GraphPad Software, San Diego, CA). Although measurements taken before LMB addition fitted best to a two-phase exponential decay curve, measurements taken in LMB-treated cells gave a best fit to a single-phase decay curve. To allow comparison of the results, therefore, a simpler one-phase curve was used to calculate an approximate half-life of fluorescence loss for each data set. PA-GFP-SmB was lost from CBs more quickly in cells following LMB treatment (Fig. 8A,B). This suggests that the inhibition of CRM1 and its subsequent loss from CBs prevents the retention of newly imported snRNPs in CBs. This change in kinetics can be measured following 2 hours of LMB treatment: a length of inhibition insufficient to cause any detectable change in CB components other than the loss of CRM1.
CRM1 is not required for the uptake of snRNPs into speckles
LMB treatment of cells newly transfected with FP-tagged snRNP proteins prevents the accumulation of FP-tagged snRNPs in nuclear speckles (Sleeman et al., 2001). To investigate the effect of LMB on the ability of snRNPs to accumulate in speckles, cells were transfected with PA-GFP-SmB and incubated for 72 hours so that PA-GFP-SmB was seen in CBs and speckles. The cells were then treated with LMB for 4 hours, and PA-GFP-SmB-tagged snRNPs activated in different regions of the nucleus using repeated pulses of a 406-nm laser, as in Figs 4 and 5 above. The results obtained were indistinguishable from those seen in cells without LMB (Fig. 9; see supplementary material Movie 8). Whether the activated region contained a speckle or a CB, activated signal was able to access speckles in distant regions of the nucleus. This demonstrates that LMB does not block the uptake of snRNPs into speckles, suggesting that the inability of newly imported snRNPs to accumulate in speckles in LMB-treated cells is connected to the alterations in their kinetics within the CB, probably resulting from a failure of maturation or assembly.
CRM1 associates with newly imported snRNPs in vivo
To investigate whether the requirement for CRM1 in the CB for SmB retention is mediated by interactions between CRM1 and snRNPs, protein lysates were prepared from HeLa cells expressing YFP alone, YFP-SmB for 24 hours, YFP-SmB for 48 hours and GFP-SmD1 for 24 hours. Immunoprecipitations were performed using anti-GFP antibodies (Fig. 10A). Anti-GFP beads efficiently precipitated YFP-SmB, GFP-SmD1 (data not shown) and YFP with a small amount of the target protein remaining in the unbound fraction and none precipitated by control, protein-G sepharose, beads (Fig. 10Ai,ii). On blots immunodetected with anti-CRM1 (Fig. 10Aiii-vi), the majority of CRM1 signal was in the unbound fraction, with a band corresponding to CRM1 also detected in the anti-GFP-bound fractions from cells transfected with GFP-SmD1 (Fig. 10Aiii) or YFP-SmB (Fig. 10Aiv) for 24 hours. No CRM1 was immunoprecipitated from cells transfected with YFP alone (Fig. 10Avii) or with GFP-SmDI (Fig. 10Av) or YFP-SmB (Fig. 10Avi) for 48 hours. Thus, a small amount of endogenous CRM1 can be co-immunoprecipitated with FP-tagged Sm proteins expressed for short periods of time. This suggests that CRM1 interacts with newly imported Sm-containing snRNPs in vivo, and that the loss of this interaction is responsible for the defect in retention of tagged SmB in CBs in LMB-treated cells.
To further investigate the nature of the Sm-containing complexes with which CRM1 interacts, whole-cell lysates were prepared from cells transiently expressing CRM1-GFP (a gift from M. Fornerod, The Netherlands Cancer Institute, Amsterdam, The Netherlands). CRM1-GFP was immunoprecipitated using anti-GFP antibodies (Fig. 10Bi). Duplicate blots were detected with antibodies to SMN, which is associated with new snRNPs in the cytoplasm and accompanies them to CBs (Fig. 10Bii), to Sm proteins (Fig. 10Biii) and to the U2 snRNP-specific protein U2B″ (Fig. 10Biv). The cellular location of addition of U2B″ to U2 snRNPs is not known, but the addition occurs after re-import into the nucleus. CRM1-GFP co-immunoprecipitated SMN and Sm proteins but not U2B″. Furthermore, a small amount of endogenous CRM1 was co-immunoprecipited using GFP antibodies from lysates of HeLa cells expressing GFP-SMN (Fig. 10Ci), but not from cells expressing the U1 snRNP-specific protein, U1A (Fig. 10Cii). These data suggest that CRM1 is associated with newly imported snRNPs in complexes containing SMN, prior to the addition of snRNP-specific proteins.
Steady-state snRNPs show rapid multi-directional exchange between nuclear structures
Despite evidence for a pathway of maturation of splicing snRNPs involving sequential interaction with CBs and speckles, snRNPs show a great deal of flexibility in their movement between nuclear structures. In cells expressing PA-GFP-SmB for 48 hours or more, the steady-state distribution of tagged snRNPs in CBs and speckles is observed. In these cells, regardless of the structure selected for photoactivation of PA-GFP-tagged snRNPs, the signal moves away from the activated region in all directions and is able to accumulate in speckles and CBs. This flexibility of movement is also seen using FLIP, where signal is rapidly lost from all areas of the nucleus during photobleaching of a nuclear region. These data favour a model of movement of mature snRNPs within the nucleus whereby they diffuse randomly, forming interactions with sub-nuclear structures such as speckles and CBs of sufficient duration to result in detectable accumulation if they encounter binding sites. They argue against the presence of a uni-directional pathway of complexes through CBs to speckles connected with snRNP biogenesis. The observed maturation pathway of splicing snRNPs is formed by the ability of snRNPs to interact only with those sub-nuclear structures applicable to their stage of maturation. The molecular basis for this mechanism is currently under investigation. A role for the CB in the regeneration of U4/U6 di-snRNPs between rounds of splicing has been proposed (Schaffert et al., 2004; Stanek et al., 2003). The observed ability of Sm-containing complexes activated in speckles to accumulate in CBs may well be connected to this process.
The presence of a fully assembled Sm core is not sufficient for localization to speckles
In contrast to the behaviour of PA-GFP-SmB-tagged snRNPs following 48 hours of expression, newly expressed PA-GFP-SmB (16 to 24 hours) shows a restricted pattern of interaction with nuclear structures. Activated PA-GFP-SmB accumulates in CBs in distant regions of the nucleus, but is unable to accumulate in speckles. It cannot be ruled out that some of the PA-GFP signal in these nuclei represents PA-GFP-SmB protein not assembled into snRNP particles. Although newly assembled snRNPs require both a TMG cap and an Sm core for their import (Fischer et al., 1993; Hamm et al., 1990), it has not been formally demonstrated that Sm proteins need to be assembled onto snRNAs to be imported. However, FRET analyses of cells expressing CFP-SmD1 and YFP-SmB demonstrate the presence of fully assembled Sm cores in CBs and the nucleoplasm of cells expressing tagged Sm proteins for short periods. These particles are unable to accumulate in speckles, indicating that a complete Sm core domain is insufficient for the accumulation of snRNPs in speckles. Whether it is further modification of the snRNA, which is subject to pseudouridylation and 2′-O-methylation, the presence of additional snRNP proteins, for example SF3a for U2 snRNP, or both of these that mediate the interaction of snRNPs with speckles is not yet clear.
The ability of snRNPs to exchange between the cytoplasm and nucleus is lost with time
FLIP demonstrates that recently imported snRNPs are able to access the cytoplasm, indeed the cytoplasmic and CB fractions appear to be interchangeable, because bleaching of the cytoplasm results in loss of signal from CBs. By contrast, bleaching of the cytoplasm has no effect on the nuclear signal in cells stably expressing YFP-SmB. The ability to accumulate in speckles thus appears to be gained as the ability to access the cytoplasm is lost. This suggests that mature snRNPs are retained exclusively within the nucleus and that, if their Sm cores are subject to turnover or recycling, this occurs at a low rate, or involves nuclear rather than cytoplasmic machinery.
LMB causes a rapid loss of CRM1 from CBs and affects snRNP dynamics within them
CRM1 is required for the export of newly transcribed snRNAs, and LMB has been used previously as an indirect inhibitor of the snRNP import pathway (Carvalho et al., 1999; Sleeman and Lamond, 1999). In HeLa cells, CRM1 is often found in CBs (Boulon et al., 2004; Ospina et al., 2005). A brief incubation with LMB (30 minutes) is sufficient to cause complete loss of CRM1 from CBs, suggesting that CRM1 is localized in CBs as a result of its activity. This loss of CRM1 from CBs is seen before any major disruption of nuclear structure is observed. Longer incubations with LMB (6 hours or more) result in the depletion of snRNPs from CBs and relocalization of the CB marker protein, coilin, to the nucleolus (Carvalho et al., 1999; Sleeman et al., 2001). Short incubations with LMB are sufficient to affect the dynamics of PA-GFP-SmB within CBs. LMB is a specific inhibitor of the nuclear export receptor CRM1 and functions by binding directly to the region of CRM1 required for binding to the nuclear export signals of its cargo (Fornerod et al., 1997). In cells maintained in normal growth conditions, a two-phase or three-phase exponential decay curve is required to model the loss of PA-GFP-SmB from the CB (Fig. 8 and data not shown). The most rapid rate of loss under these conditions may represent free PA-GFP-SmB not assembled onto snRNA. In cells treated with LMB for 2-3 hours, the dynamics of PA-GFP-SmB in CBs are less complex, with a curve of best fit obtained using single-phase exponential decay. Furthermore, the observed average half-lives for PA-GFP-SmB loss in LMB-treated cells are shorter than those seen in untreated cells (P=0.0006). This suggests LMB results in disruption of the interaction of Sm-containing complexes with CBs. Together with the interaction between early snRNP complexes and CRM1 detected by co-immunoprecipitation, this suggests that CRM1 has a role in the interaction of snRNPs with CBs in addition to its established role in snRNA export. The pattern of interactions detected between CRM1 and proteins SmB, SmD1, SMN, U2B″ and U1A suggest that CRM1 associates preferentially with newly imported snRNPs, associated with the SMN protein, prior to the addition of snRNP-specific proteins. The presence of CRM1 and/or the absence of snRNP-specific proteins may form part of a signal preventing accumulation of immature snRNPs in speckles. Furthermore, the presence of CRM1, an export factor, in newly imported, partially mature, snRNP complexes could explain their ability to access the cytoplasm. The current data cannot conclusively rule out the possibility that CRM1 interacts independently with SMN and unassembled Sm proteins. A role for the CB in reclamation of unassembled snRNP components has been proposed (Xu et al., 2005). This would place the role for CRM1 in modulating snRNP assembly earlier in the biogenesis pathway, in ensuring that unassembled Sm proteins are available to the cytoplasmic assembly complex. However, the rationale for the LMB-sensitive retention of free Sm proteins in the CB rather than their rapid export to the cytoplasm is unclear.
Effect of LMB highlights differences between nuclear snoRNP and snRNP transport pathways
snoRNPs are ribonucleoprotein particles found predominantly in the nucleolus that function in rRNA biogenesis (Bachellerie et al., 2002; Kiss, 2002). In contrast to splicing snRNPs, their biogenesis is thought to occur exclusively within the nucleus with no cytoplasmic stage (Terns and Dahlberg, 1994). Like splicing snRNPs, the box C/D snoRNPs localize transiently to CBs before reaching their steady-state localization in nucleoli (Narayanan et al., 1999). Hypermethylation of the 5′ cap of the snoRNA, analogous to one of the cytoplasmic events in snRNP maturation, occurs in the CB (Mouaikel et al., 2002; Verheggen et al., 2002). Studies of U3 snoRNAs (Boulon et al., 2004; Watkins et al., 2004) have demonstrated sequential involvement of PHAX, the snRNA export adaptor, and CRM1, the snRNA export factor, in their transport within the nucleus. PHAX is proposed to recruit TMG-capped U3 precursors into CBs, with CRM1 binding hypermethylated, TMG-capped species in CBs and transporting them away from it. In contrast to this, the current study suggests that the presence of CRM1 in CBs is required for the adequate retention of TMG-capped, Sm core-containing splicing snRNPs within CBs. Previous data have shown that new snRNPs fail to accumulate in speckles in LMB-treated cells. Together with the data presented here, this suggests that CRM1, or associated factors, is required to retain new snRNPs in CBs until they are sufficiently mature to accumulate in speckles. It is possible that the lack of retention of Sm-containing complexes in CBs of LMB-treated cells is an indirect, downstream consequence of blocking snRNA export. However, the alteration in dynamics after relatively short treatments and the interactions detected between CRM1 and Sm-containing complexes by co-immunoprecipitation argue for a direct effect.
In summary, newly imported snRNPs are able to interact with CBs but unable to interact with speckles, despite their ability to access areas of the nucleus containing speckles and the presence of complete Sm core domains. Conversely, snRNPs at steady state are able to form interactions with speckles of sufficient affinity for their rapid accumulation in them to be seen. They are, however, prevented from making any detectable return to the cytoplasm. The physical differences between snRNPs exhibiting these two types of behaviour are yet to be elucidated, but the presence or absence of CRM1 and snRNP-specific proteins such as U2B″ and U1A are likely to be involved. Treatment of HeLa cells with the metabolic inhibitor LMB results in a quantifiable change in the dynamics of interaction of snRNPs with CBs, leading to a lack of retention of snRNPs. Inhibition of CRM1 with LMB results in a dramatic rearrangement of CB proteins over a period of 6 hours, including the depletion of snRNPs from CBs. Using dynamic live-cell microscopy, these changes can be measured at a molecular level at much shorter timepoints, suggesting that these techniques are capable of detecting subtle differences in the behaviour of protein complexes within the nucleus, and thus will prove powerful tools in the analysis of nuclear function.
Materials and Methods
Cell culture and transfection
Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum and 100 U/ml penicillin and streptomycin (Life Technologies). For immunofluorescence assays, cells were grown on coverslips and transfected (if necessary) using Effectene transfection reagent (Qiagen) according to the manufacturer's instructions. For the preparation of cell lysates, cells were grown in 10-cm diameter dishes. For live-cell microscopy, cells were grown on 40 mm coverslips (Intracel).
Plasmid constructs and cell lines
Plasmids pEYFP-SmB, pECFP-SmB, pECFP-SmD1 and stable cell line YFP-SmB-E1 have been described previously (Sleeman et al., 2003). The expression plasmid pPA-GFP-SmB for photoactivatable-GFP-tagged SmB was generated by sub-cloning the SmB sequence from pEYFP-SmB as an EcoRI/SalI fragment into plasmid pPA-GFP-C1 (a gift from J. Lippincott-Schwartz, NIH, Bethesda, MD). CRM1-GFP was a gift from M. Fornerod.
HeLa cells grown on glass coverslips were cotransfected with appropriate constructs. 24 or 48 hours after transfection, cells were mounted in CO2-independent medium (Invitrogen Life Technologies) in an open chamber (Zeiss) within an environmental incubator (Solent Scientific) on an Olympus DeltaVision Spectris Microscope (Applied Precision) with a Quantifiable Laser Module including a 20 mW 532 nm laser to photobleach YFP without bleaching CFP (for FRET and FLIP) and a 20 mW 406 nm laser for photoactivation of PA-GFP. Data analyses were performed using Volocity (Improvision, UK) and Prism 4 (GraphPad).
Kinetic analyses using PA-GFP
HeLa cells were grown on glass coverslips, transfected with pPA-GFP-SmB and scanned for low levels of expression of PA-GFP-SmB using a 405/40-nm excitation filter and a standard FITC emission filter. Selected cells were activated by a 200 msecond 406 nm pulse at 50% laser power focused to a diffraction-limited spot of approximately 1.5 μm diameter. A single z-section of each cell was imaged using a FITC filter at three timepoints before the laser pulse and using an adaptive timecourse 35 seconds after the laser pulse. For analyses of dynamics before and after LMB treatment, cells were marked in a point list so that they could be revisited after incubation with the drug for 1.5 hours. The fluorescence intensity of photoactivated CBs was measured for each timepoint. A region of background fluorescence was defined within a neighbouring non-activated cell and subtracted from these values. The intensity values were normalized against the initial preactivation fluorescence and plotted against time. To compare half-lives before and after LMB treatment, single-phase exponential decay curves were fitted to each data set and the representative half-lives obtained compared using a two-tailed t-test.
For experiments requiring sequential activation of fluorescence in the same region of interest, multiple laser pulses were used with images recorded at three timepoints before each activation and over a 10-second adaptive timecourse after each activation.
Regions of interest in cells expressing YFP-SmB were defined and bleached using a 20 mW 532 mn laser at 50% power, by automatic movement of the stage and repeated pulsing with a diffraction-limited spot. The cells were imaged three times before each bleach event and using a 10-second adaptive timecourse after each bleach event. The bleach sequence was repeated 10 times. The intensity of representative nuclear, cytoplasmic and CB regions was measured, corrected for bleaching during imaging by subtraction of the fluorescence from a background region within an unbleached cell, and plotted against time.
Cell fixation, immunostaining and microscopy
HeLa cells grown on glass coverslips were fixed for 5 minutes at room temperature with 3.7% paraformaldehyde in PHEM buffer [60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl2 (pH 6.9)]. Immunostaining was performed as described previously (Sleeman et al., 2003). Cells were mounted in Citifluor medium (Agar Scientific). Antibodies used were mouse monoclonal antibody (mAb) Ab1 anti-TMG (Calbiochem) (dilution 1:50), mouse Y12 mAb anti-Sm (AbCam) (dilution 1:100), rabbit 5641 anti-CRM1 (a gift from M. Fornerod) (dilution 1:500), rabbit 204,10 anti-coilin (a gift from A. I. Lamond, University of Dundee, Dundee, UK) (dilution 1:500), FITC-conjugated goat anti-mouse and TRITC-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories) (dilution 1:500). Immunostained specimens were examined using a Zeiss 63× 1.4NA Plan-Apochromat objective and recorded using a Hammamatsu C4742-80-12AG camera. Optical sections separated by 200 nm were collected using a binning of 2×2. Images were restored using an iterative deconvolution algorithm using a calculated point-spread function (Volocity; Improvision, UK).
Preparation of cell lysates, immunoblotting and immunoprecipitation
Lysates were prepared as described previously (Sleeman et al., 2003), electrophoresed on a 10% SDS-polyacrylamide gel and transferred to nitrocellulose membranes for immunoblotting. Immunoprecipitation using anti-GFP antibodies (Roche) was performed as described previously (Trinkle-Mulcahy et al., 2001). Primary antibodies used were anti-GFP mouse monoclonal (Roche), dilution 1:1000; 5641 rabbit anti-CRM1 (a gift from M. Fornerod), dilution 1:1000; 4G3 anti-U2B″ (a gift from A. I. Lamond), dilution 1:500; MANSMA1 anti-SMN (a gift from G. Morris, RJAH Orthopaedic Hospital, Oswestry, UK), dilution 1:500; Y12 anti-Sm (AbCam), dilution 1:250; and 456 anti-U1A (a gift from A. I. Lamond), dilution 1:500.
I would like to thank Jennifer Lippincott-Schwartz for PA-GFP-C1, Maarten Fornerod for anti-CRM1, Laura Trinkle-Mulcahy for pCFPYFP, Alan Prescott, Brian McStay and Sam Crouch for comments on the manuscript, and Giles Thomas for help with curve fitting and statistical analysis. J.E.S. is a Royal Society URF. This work was supported by Tenovus Tayside.