Many of the protein factors that play a role in nuclear export of mRNAs have been identified, but still little is known about how mRNAs are transported through the cell nucleus and which nuclear compartments are involved in mRNA transport. Using fluorescent 2'O-methyl oligoribonucleotide probes, we investigated the mobility of poly(A)+ RNA in the nucleoplasm and in nuclear speckles of U2OS cells. Quantitative analysis of diffusion using photobleaching techniques revealed that the majority of poly(A)+ RNA move throughout the nucleus, including in and out of speckles (also called SC-35 domains), which are enriched for splicing factors. Interestingly, in the presence of the transcription inhibitor 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole, the association of poly(A)+ RNA with speckles remained dynamic. Our results show that RNA movement is energy dependent and that the proportion of nuclear poly(A)+ RNA that resides in speckles is a dynamic population that transiently interacts with speckles independent of the transcriptional status of the cell. Rather than the poly(A)+ RNA within speckles serving a stable structural role, our findings support the suggestion of a more active role of these regions in nuclear RNA metabolism and/or transport.
Gene expression is a multistep process that involves transcription, RNA processing, nuclear RNA export, cytoplasmic RNA transport, and translation. Nuclear RNA export has recently been recognized as being an important potential mechanism to regulate gene expression, but has not yet been completely characterized. As a result of gene expression, different classes of RNA, including rRNA, mRNA, snRNA, and tRNA, are produced and transported to the cytoplasm via distinct transport pathways (Jarmolowski et al., 1994; Cullen, 2003). In the case of mRNA, there is evidence that mRNA transport is tightly coupled to mRNA synthesis and processing. Recruitment of nuclear mRNA export factors to transcripts has been coupled to different steps in gene expression including transcription (Lei et al., 2001; Strasser et al., 2002), splicing (Luo and Reed, 1999; Zhou et al., 2000; Le Hir et al., 2001; Luo et al., 2001), and 3′ end formation (Lei and Silver, 2002). Furthermore, the presence of unstructured regions in mRNAs has recently been implicated as mediating recognition by export factors (Ohno et al., 2002). A tight coupling between splicing and nuclear mRNA export would explain why only spliced RNAs are transported to the cytoplasm. The exon junction complex is a protein complex that contains, among other proteins, the mRNA export factor REF1/Aly. It assembles upstream of splice junctions upon pre-mRNA splicing and has been proposed to mediate this coupling through interaction of REF1/Aly with the major receptor for mRNA export, TAP (Le Hir et al., 2001). However, recent works indicate that the exon junction complex proteins may contribute to nuclear mRNA export but are not essential (Gatfield and Izaurralde, 2002). Instead, SR splicing factors that bind to mRNA were shown to interact directly with TAP for efficient export (Huang et al., 2003).
Splicing factors are also considered to be indispensable for retaining pre-mRNAs in the cell nucleus. Pre-mRNAs were shown to accumulate at or near active sites of transcription and to colocalize with splicing factors (Bauren et al., 1996; Dirks et al., 1997; Misteli et al., 1997). Also, transcripts defective in splicing were shown to accumulate at sites of transcription and unable to be exported to the cytoplasm (Custodio et al., 1999). Therefore, it is assumed that most mRNAs are released from their site of synthesis and processing after completion of splicing. These mRNAs move randomly within the cell nucleus with export factors associated with them before they get bound to a nuclear pore complex and enter the cytoplasm. This assumption is supported by fluorescence in situ hybridization and BrUTP labeling studies showing that specific gene transcripts emanate from transcription sites in all directions in cell nuclei (Zachar et al., 1993; Dirks et al., 1995; Macville et al., 1995; Singh et al., 1999). Furthermore, in vivo hybridization studies revealed that poly(A)+ RNA moves randomly in the cell nucleus at a rate compatible with free diffusion (Politz et al., 1998, 1999).
However, various aspects of nuclear RNA export are unclear. These aspects include the role of nuclear compartments in the export process. Speckles, also referred to as SC-35 domains, are nuclear compartments that contain a large number of factors required for mRNA synthesis, processing, and export (Lamond and Spector, 2003). The observation that speckles also contain poly(A)+ RNA led to the suggestion that speckles themselves may play a role in RNA metabolism and export (Carter et al., 1991, 1993). These speculations were supported by the observation that sites of bromouridine incorporation that mark nascent transcripts overlap with speckles (Wei et al., 1999) and that some specific active genes localize at the edges of speckles (Xing et al., 1993, 1995; Smith et al., 1999; Shopland et al., 2003). Furthermore, a number of specific gene transcripts were shown to localize to the inside of speckles (Puvion and Puvion-Dutilleul, 1996; Smith et al., 1999; Johnson et al., 2000; Melcak et al., 2000; Hattinger et al., 2002; Shopland et al., 2002, 2003), indicating that speckles play a direct role in mRNA metabolism and export. However, various lines of evidence argue against a direct role of speckles in gene transcription, RNA processing, and RNA transport. First, in contrast to Wei et al. (1999), several reports indicate that speckles are not labeled after 3H- or bromouridine incorporation (Fakan, 1994; Cmarko et al., 1999). Second, splicing factors were shown to be recruited from speckles to sites of active transcription (Jimenez-Garcia and Spector, 1993; Huang and Spector, 1996; Dirks et al., 1997; Misteli et al., 1997; Zeng et al., 1997; Snaar et al., 1999). Third, poly(A)+ RNA is not exported from speckles when transcription is inhibited and is, therefore, suggested to be a stable population that plays a structural role and acts as a binding site for RNA processing proteins (Huang et al., 1994; Sacco-Bubulya and Spector, 2002). Finally, in vivo hybridization experiments using oligo (dT) probes that hybridize to the poly(A) tails of mRNAs in living cells did not reveal any accumulation of poly(A)+ RNA in speckles at any stage of transport (Politz et al., 1999).
To investigate a possible role for speckles in RNA transport, we analyzed the mobility of poly(A)+ RNA in the nucleoplasm and in nuclear speckles in transcriptionally active and inactive cells. Using 2'O-methyl RNA probes and photobleaching techniques, we demonstrate that poly(A)+ RNA moves throughout the nucleoplasm, though at a much slower rate compared with transport rates determined in previous works using oligodeoxynucleotide probes and compared with proteins that play a role in RNA processing and transport. Furthermore, we present evidence that poly(A)+ RNA transiently interacts with speckle domains independent of transcription but dependent on cellular energy levels.
+ RNA fractions in nucleoplasm and speckles reveal similar kinetic behavior
Poly(A)+ RNA was visualized in nuclei of living U2OS cells by means of a tetramethylrhodamine (TAMRA)-labeled 2'OMe (U)22 probe that is complementary to the poly(A) tail of mRNAs. Shortly after injection into the cytoplasm of U2OS cells, this probe revealed a nuclear localization pattern that is characteristic for poly(A)+ RNA (Carter et al., 1991). In addition to a diffuse staining of the nucleoplasm, excluding nucleoli, a speckled staining was observed. The cytoplasm of cells revealed only a very weak fluorescence staining. The movement of poly(A)+ RNA in the cell nucleus was studied using FRAP and fluorescence loss in photobleaching (FLIP) analysis. For this purpose, we selected cells with moderate levels of fluorescence signals. Images of a typical FRAP experiment are shown in Fig. 1
. In the top cell in Fig. 1 A, a speckle was bleached (green circle), and the cell was subsequently imaged as described in Materials and methods. Using a bleaching pulse of 5 s, it was possible to bleach the area down to only ∼20% of the initial intensity. The recovery of poly(A)+ RNA fluorescence in the speckle, as a consequence of movement of unbleached poly(A)+ RNA from the surroundings into the bleached area, is clearly visible in the pseudo-color images (Fig. 1 A). Next, we determined the average t1/2 of recovery and the diffusion coefficient (D) for poly(A)+ RNA in speckles from the derived FRAP curves (Fig. 1 C): t1/2 ≈ 26 s, D ≈ 0.03 μm2/s (Table I)
. During the time course of the experiment, the fluorescence in the speckles did not recover completely, suggesting that a relatively immobile fraction of poly(A)+ RNA of ∼15% exists in these compartments.
To confirm that, we photobleached 2'OMe (U)22-TAMRA that localized to speckles, we also injected the probe in U2OS cells that express SF2/ASF-GFP. In these cells, speckles, identified by the presence of SF2/ASF-GFP, were selected for photobleaching the TAMRA-labeled probe. The recovery times measured for the 2'OMe (U)22 probe in SF2/ASF-GFP–labeled speckles were similar to the ones measured in the nontransfected cells (unpublished data).
Next, we photobleached nucleoplasmic areas at some distance from a speckle and analyzed the recovery of fluorescence. The recovery appeared to be ∼1.5 times faster (Fig. 1, B and D): t1/2 ≈ 18 s, D ≈ 0.04 μm2/s (Table I). These experiments suggest a difference in mobility between the poly(A)+ RNA fractions inside the speckles and poly(A)+ RNA fractions outside the speckles. However, in cells where only a part of a speckle was bleached, the calculated diffusion constants appeared to be in the order of 0.035 μm2/s. The discrepancy in estimated D-values could result from the difference in the number of molecules present in the bleached area relative to those in the vicinity of the bleached area.
It is likely that the majority of the poly(A)+ RNA that moves into the bleached area initially comes from the immediate vicinity of the bleached spot. To determine the freedom of movement of poly(A)+ RNA throughout the nucleoplasm, including speckles, we applied FLIP. A spot in the nucleoplasm was repeatedly bleached for 3 s with 10-s time intervals in which images of the cell were recorded. Fig. 2
shows the result of a typical FLIP experiment. Loss of total nuclear fluorescence was imaged (Fig. 2 A) and measured over time (Fig. 2 B). After 400 s from the first bleach, ∼85% of the nuclear poly(A)+ RNA fluorescence was lost. Similar results were obtained when a region inside a speckle was repeatedly photobleached, suggesting that the mobility of poly(A)+ RNA in the nucleoplasm and speckles is very similar. The fraction of ∼15% of the total amount of fluorescence that is still present in the nucleus after 400 s bleaching suggests the presence of a relatively immobile population of poly(A)+ RNA, which is in agreement with the result of the FRAP experiments.
To determine if the FLIP curve represents a single population of moving poly(A)+ RNA, the curve was converted to a semi-logarithmic plot and analyzed by curve fitting. The result revealed a single exponential fit, suggesting the existence of a “single” population of moving poly(A)+ RNA (Fig. 2 C). Hence, the fraction of poly(A)+ RNA or unbound probe that diffuses rapidly through the nucleus appears to be very small, if present at all. To confirm this finding, we also measured the fluorescence intensity of regions outside a photobleached spot immediately before and after photobleaching in 10 cells. By comparing the intensities and compensating for photobleaching (using another cell in the same microscopic field as a reference), we measured reductions in intensity between 3–5%, suggesting that there may exist a small free diffusing pool of probe and/or poly(A)+ RNA (unpublished data).
No significant loss of signal was observed when the bleached spot was set in nucleoli or cytoplasm (unpublished data). These results imply that most poly(A)+ RNA is moving throughout the nucleoplasm, except for nucleoli. However, we cannot exclude that low amounts of poly(A)+ RNA move through nucleoli. Prolonged periods of photobleaching a spot in a nucleolus resulted in some loss of fluorescent signal in the nucleoplasm, which could be due to the photobleaching of fluorescent probe either present in nucleoli or in the nucleoplasm above or below nucleoli.
2'OMe (U)22 localization and kinetics are dependent on the presence of poly(A) tails
To confirm that 2'OMe (U)22 probe molecules hybridize specifically to the poly(A) tail of RNAs, cells were treated with cordycepin, which prevents poly(A) addition but does not block RNA synthesis (Darnell et al., 1971; Mendecki et al., 1972; Calado and Carmo-Fonseca, 2000). It was predicted that after cordycepin treatment the 2'OMe (U)22 probe would not localize to poly(A)+ rich speckle domains and would reveal a fast movement consistent with free diffusion. U2OS cells were incubated with cordycepin for 16 h, injected with 2'OMe (U)22 probe, and imaged by confocal microscopy. Consistent with our prediction, the 2'OMe (U)22 probe revealed a diffuse staining in the nucleus excluding nucleoli (Fig. 3, A and B)
. Also, as determined by FRAP (Fig. 3, A and C) and FLIP (Fig. 3, B and D), the mobility of the 2'OMe (U)22 probe was significantly increased in cordycepin-treated cells compared with nontreated cells. These findings demonstrate that the 2'OMe (U)22 probe binds with high specificity to the poly(A) tail of RNAs. Furthermore, the results show that the preferential association of the 2'OMe (U)22 probe with speckles results from their interaction with poly(A)+ RNA in living cells. This result is consistent with our finding that the 2'OMe (U)22 probe is highly specific for poly(A)+ RNA in fixed cells as determined by RNase controls (Molenaar et al., 2001).
Oligodeoxynucleotide (dT)40 moves significantly faster through the nucleoplasm compared with 2'OMe (U)22
The estimated diffusion coefficient for poly(A)+ RNA movement in this work is significantly lower than determined in previous works (Politz et al., 1998, 1999). To explain this discrepancy, we compared the localization and the dynamic behavior of the 2'OMe (U)22 probe with that of a poly(A) tail-specific oligodeoxynucleotide (dT)40 probe, a probe type that has been used in previous papers, a complementary 2'OMe (A)18 negative control probe, and a 2'OMe human cytomegalovirus (HCMV) negative control probe (specific for cytomegalovirus immediate-early mRNA) in living U2OS cells. The (dT)40 probe revealed a diffuse staining of the nucleoplasm and a moderate staining of nucleoli and, significantly, revealed no accumulation in speckles (Fig. 4
A). Furthermore, FRAP analysis of the oligo (dT)40 probe revealed a t1/2 of recovery of 0.35 s and a diffusion coefficient of ∼1.7 μm2/s (Fig. 4, A and B; and Table I). Hence, the oligo (dT)40 probe appears to move ∼50-fold faster than the 2'OMe (U)22 probe, suggesting that only a small proportion of microinjected oligo (dT)40 probe binds to poly(A)+ RNA and that the majority of the probe moves by free diffusion through the nucleoplasm. Also, the probes 2'OMe (A)18 and 2'OMe HCMV revealed a diffuse nuclear staining and rapid movement in U2OS cells. Like the 2'OMe (A)18-TAMRA probe (not depicted), 2'OMe HCMV-TAMRA could not be photobleached in defined areas to <80% of the initial intensity (Fig. 4 C), indicating a high rate of diffusion. The corresponding FRAP curves confirm the fast movement of 2'OMe HCMV probe compared with 2'OMe (U)22 in cell nuclei by showing a rapid recovery of fluorescence to ∼100% after bleaching (Fig. 4 D). For 2'OMe HCMV, a t1/2 recovery time of 0.3 s and a diffusion constant of 1.9 μm2/s was calculated, which is ∼60-fold higher than measured for the 2'OMe (U)22 probe (Table I). To confirm these FRAP results, we analyzed the kinetic behavior of the control probes by FLIP. As shown in Fig. 4 (E–H), the control probes revealed much faster kinetic behavior than 2'OMe (U)22. Together with the cordycepin experiment, these results strongly indicate that the 2'OMe (U)22 probe hybridizes stably and specifically to poly(A)+ RNA in living U2OS cells.
To compare the kinetics of 2'OMe (U)22 movement with that of another probe that is supposed to hybridize to a target sequence in speckles, we designed a 2'OMe RNA probe specific for U1snRNA. U1snRNA is a component of the U1snRNP complex that is involved in RNA splicing and is present throughout the nucleoplasm and in Cajal bodies and, like poly(A)+ RNA, is concentrated in nuclear speckles (Lamond and Carmo-Fonseca, 1993). After injection of the 2'OMe RNA probe specific for U1snRNA in U2OS cells, we observed fluorescence signals emerging in these structures, indicating that this probe specifically hybridized to U1snRNA (Fig. 5
A). Subsequent FRAP analysis revealed that the 2'OMe U1snRNA-specific probe has a 4–5-fold higher rate of movement (Table I, t1/2 ≈ 4.5 s, D ≈ 0.16 μm2/s) than poly(A)+ RNA in speckles (Fig. 5 B). Nevertheless, the rate of 2'OMe U1snRNA movement is considerably slower than measured for the probes 2'OMe (A)18 and 2'OMe HCMV.
+ RNAs enter and leave speckles in 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB)–treated cells
Inhibition of RNA polymerase II transcription by DRB (Sehgal et al., 1976) was shown to result in a redistribution of poly(A)+ RNA in cell nuclei characterized by the disappearance of a nucleoplasmic pool and the maintenance of a stable population of poly(A)+ RNA concentrated in speckles along with pre-mRNA splicing factors (Huang et al., 1994). To investigate the influence of RNA polymerase II transcription inhibition on the mobility of poly(A)+ RNA, we analyzed the dynamic properties of hybridized 2'OMe (U)22 probe in DRB-treated cells by FLIP. Cells were analyzed after treatment with DRB for 4 h either before or after microinjection of the 2'OMe (U)22 probe. As expected, a typical enlargement and rounding-up of the speckles was observed (Fig. 6
A). To measure the mobility of the hybridized 2'OMe (U)22 probe in speckles, a speckle was selected and continuously photobleached using a high-intensity laser beam. Fig. 6 B illustrates the loss of nuclear fluorescence due to continuous photobleaching of a speckle in a DRB-treated cell. The corresponding FLIP curves show a gradual loss of fluorescence in all speckles of DRB-treated cells at a rate that is similar to the one observed for nontreated cells (Fig. 6 C), indicating that nearly all poly(A)+ RNAs present in speckles are mobile and leave speckles independently of the transcriptional activity of the cell.
The uptake of poly(A)+ RNA by speckles is energy dependent
To investigate whether or not the movement of poly(A) RNA toward speckles requires energy, we performed photobleaching experiments using cells maintained at 22 and 37°C. When compared, the typical localization pattern of poly(A)+ RNA was unaltered by the temperature change. In cells maintained at 22°C, speckles were selected and photobleached, and the recovery of fluorescence in the defined areas was imaged and measured. A typical example of this process is shown in Fig. 7
A. A speckle indicated in the prebleach image was photobleached, and the fluorescence recovery is shown at 0, 90, and 270 s after bleaching. Fluorescence did not fully recover in the bleached speckle. Fluorescence recovery values were also measured, and the corresponding recovery curves, each collected from 15 cells, show a slower recovery of fluorescence in speckles at 22°C compared with 37°C (Fig. 7 B). Significantly, not more than 50% fluorescence recovery was measured at 22°C. Similar fluorescence recovery results were obtained from nucleoplasmic areas at some distance from speckles that were photobleached and imaged at increasing time intervals (unpublished data). Together, these results suggest that poly(A)+ RNA transport through the nucleoplasm and into speckles requires energy.
RNA-associating proteins show different kinetic behaviors than poly(A)+ RNA
Next, we investigated how the dynamic behavior of proteins that directly or indirectly associate with mRNA correlates with poly(A)+ RNA dynamics. We analyzed the mobility of the splicing protein SF2/ASF, of the poly(A) tail binding protein PABP2 (poly(A) binding protein II), and of the transport proteins Aly and Tap, all tagged with GFP, in living U2OS cells using FRAP. Previously, these GFP fusion proteins were shown to localize and to function similarly to their endogenous counterparts (Misteli et al., 1997; Katahira et al., 1999; Zhou et al., 2000; Calapez et al., 2002).
To measure the mobility of SF2/ASF-GFP relative to the mobility of poly(A)+ RNA, we microinjected U2OS cells stably expressing SF2/ASF-GFP with the 2'OMe (U)22-TAMRA probe and analyzed them by FRAP. SF2/ASF-GFP and the probe 2'OMe (U)22-TAMRA were photobleached simultaneously in a speckle, and the fluorescence recovery of both fluorophores was imaged separately at each time point in a time series to prevent cross talk. Fig. 8
A illustrates that the fluorescence recovery of SF2/ASF-GFP in a photobleached speckle precedes that of poly(A)+ RNA. Hence, SF2/ASF-GFP seems significantly more mobile than poly-(A)+RNA. This difference in fluorescence recovery was confirmed by comparing the calculated FRAP curves obtained from eight cells (Fig. 8 D).
It was shown previously that the localization of PABP2 in speckles is dependent on binding to poly(A)+ RNA (Calado et al., 2000). Therefore, we were interested to compare the mobility of transiently expressed PABP2-GFP toward speckles with that of poly(A)+ RNA by FRAP in U2OS cells. For this purpose, PABP2-GFP–containing speckles were photobleached and the fluorescence recovery rates were subsequently imaged. The results show that a nearly complete fluorescence recovery is obtained within 50 s (Fig. 8 B) and suggest that SF2/ASF-GFP and PABP-GFP move in and out of speckles faster than poly(A)+ RNA (Fig. 8 D).
Because Aly and Tap have been implicated in playing roles in mRNA export and Aly has been shown to accumulate in nuclear speckles, we expected that the kinetics of Aly and Tap movement would correlate with the kinetic behavior of poly(A)+ movement. Fig. 8 C shows the localization pattern of Aly-GFP as observed in U2OS cells and a speckle that has been photobleached and subsequently imaged at regular time intervals afterwards. As shown, a full recovery of fluorescence is obtained within 1 min after photobleaching (Fig. 8 C). Also, the FRAP curve that has been generated after measuring recovery values in speckles from 10 cells shows that a near full recovery is obtained within 1 min (Fig. 8 E).
Next, we determined the dissociation kinetics of Aly-GFP from speckles by FLIP. Repeated bleaching of Aly-GFP in a defined area using high laser power revealed that the majority of Aly-GFP fluorescence was lost from the nucleus within 80 s (Fig. 8 F). When we performed similar experiments with cells expressing the RNA export factor Tap-GFP and its cofactor p15, which distributes more or less homogeneously throughout the nucleoplasm, we observed a loss of nuclear fluorescence within 60 s (Fig. 8 F). These results show that the transport factors Tap and Aly move more rapidly through the cell nucleus compared with the 2'OMe (U)22 probe hybridized to poly(A)+ RNA (Fig. 2) and suggest that there is a significant fraction of unbound Tap and Aly present in cell nuclei.
Using 2'OMe RNA probes, we analyzed the kinetics of poly(A)+ RNA localization in living U2OS cells. Poly(A)+ RNA was detected by a TAMRA-labeled 2'OMe RNA probe complementary to the poly(A) tail of mRNAs and imaged using a confocal scanning laser microscope. The rationale for using a 2'OMe RNA probe rather than an oligodeoxynucleotide probe is that 2'OMe RNA probes have been reported to possess much higher binding affinities for complementary (RNA) target sequences and to be resistant to nucleases (Majlessi et al., 1998). Furthermore, we have shown previously that microinjection of a TAMRA-labeled 2'OMe (U)22 probe in the cytoplasm of U2OS cells results in a nuclear staining pattern characteristic of endogenous poly(A)+ RNA localization (Molenaar et al., 2001).
Previously, oligo (dT) probes have been used to study the movement of poly(A)+ RNA in living cells and, on the basis of these studies, it was concluded that the majority of poly(A)+ RNA is diffusing freely throughout the interchromatin space in cell nuclei (Politz et al., 1998, 1999). However, dependent on the detection method used, different diffusion coefficients for poly(A)+ RNA movement were estimated. Most recently, by using a caged fluorescein-labeled oligo (dT) probe, a diffusion coefficient of 0.6 μm2/s was estimated for poly(A)+ RNA movement (Politz et al., 1999). However, earlier, a diffusion coefficient of 9 μm2/s was measured by fluorescence correlation spectroscopy (Politz et al., 1998). By measuring fluorescence recovery rates in photobleached areas, we report that poly(A)+ RNA is moving through the nucleoplasm at a significantly slower rate of 0.03 μm2/s. Therefore, we compared the mobility of 2'OMe RNA probes with that of oligodeoxynucleotide probes under identical conditions. We conclude that the discrepancies in rates of poly(A)+ RNA movement that have been measured in this and previous works can be explained by the differences in hybridization properties in living cells between oligodeoxynucleotide and 2'OMe RNA probes (Molenaar et al., 2001). We suggest that our estimate more accurately reflects the in vivo situation. Our data demonstrate that under conditions of in vivo poly(A)+ RNA imaging, a relatively large fraction of oligo (dT) is highly mobile in the cell nucleus and therefore unbound or transiently bound to poly(A)+ RNA. Importantly, we have shown that the localization and kinetics of the 2'OMe (U)22 probe is fully dependent on the presence of a poly(A) tail and that there is at best a very small fraction of unbound probe present that may have led to an overestimation of the poly(A)+ RNA diffusion coefficient. Because hybridization in a living cell is in principle a reversible kinetic process, we cannot exclude that there is some rate of exchange between probe and target molecules during our measurements that can lead to a slight overestimation of the diffusion rate of poly(A)+ RNA.
Recently, it was suggested that the RNA binding proteins PABP2 and TAP move at rates similar to rapidly diffusing poly(A)+ RNA (Calapez et al., 2002). We show that poly(A)+ RNA molecules move at significantly slower rates than previously anticipated, but that PABP2, TAP, and Aly move much more rapidly through the nucleus than poly(A)+ RNA. This finding suggests that a substantial proportion of these proteins is not bound to RNA but is diffusing rapidly throughout the cell nucleus to be available for newly synthesized transcripts.
It should be noted that FRAP analysis provides only an average value for poly(A)+ RNA mobility. Some poly-(A)+ RNAs may move, within a certain range, faster or slower. Previously, we estimated that abundantly synthesized HCMV IE transcripts move through the nucleus at a diffusion rate of 0.13 μm2/s (Snaar et al., 2002), which is fourfold faster than what we measured for poly(A)+ RNA. In any case, our observations are consistent with studies suggesting that poly(A)+ RNA is not transported by free diffusion but, at least at some stages, by an energy-dependent mechanism (Dargemont and Kühn, 1992; Jarmolowski et al., 1994; Calado et al., 2000; Miralles et al., 2000; Snaar et al., 2002).
Consistent with in situ hybridization studies on fixed cells, we observed that poly(A)+ RNA concentrates in speckles in living cells. Notably, this pattern was not observed when we or others (Politz et al., 1998) used an oligodeoxynucleotide probe instead of a 2'OMe RNA probe for detecting poly(A)+ RNA. Microinjected TAMRA-labeled oligo (dT)40 probe revealed a dispersed localization throughout the cell nucleus, but not a staining of speckles. Interestingly, our FRAP and FLIP analyses revealed that the poly(A)+ RNA population inside speckles is mobile and in continuous flux with the nucleoplasm. Only a small amount of the poly(A)+ population that reside in speckles appears to be immobile. Importantly, our observations show that the poly(A)+ population found in speckles is not a stable population of RNAs as suggested previously (Huang et al., 1994). Even when gene transcription is inhibited, poly(A)+ RNA molecules that remain in the nucleus continue to associate and dissociate from speckles and to move throughout the entire nucleus. This observation is consistent with the finding that RNA transport from nucleus to cytoplasm is not dependent on ongoing transcription (Huang and Spector, 1996). Hence, due to their dynamic behavior, it is less likely that the population of poly(A)+ RNA that is localized to speckles plays an essential role in the core organization of speckles by creating binding sites for RNA processing proteins (Sacco-Bubulya and Spector, 2002). In this context, it is worth mentioning that poly(A)+ RNA is not required for the assembly of nuclear speckles in the nuclei of early G1 cells (Ferreira et al., 1994; Gama-Carvalho et al., 1997). Nevertheless, it cannot be excluded that speckle maintenance is a dynamic process and that poly(A)+ RNA plays a role in stabilizing these compartments like mobile heterochromatin protein 1 is responsible for establishing stable heterochromatin domains in cell nuclei (Cheutin et al., 2003; Festenstein et al., 2003).
Another, not necessarily exclusive, possibility is that nuclear poly(A)+ RNA is present in speckles for nonstructural reasons. Because various specific gene transcripts have been found to associate with or to localize to speckles, it has been suggested that speckles play a role in the posttranscriptional processing of RNAs (Xing et al., 1995; Bridge et al., 1996; Smith et al., 1999; Johnson et al., 2000; Hattinger et al., 2002; Shopland et al., 2003). A role for speckles in posttranscriptional splicing is consistent with the observation that splicing factors present in the diffuse nuclear pool of cells lacking speckles are not competent to perform pre-mRNA splicing (Sacco-Bubulya and Spector, 2002). However, there is compelling evidence that the majority of pre-mRNAs are processed cotranscriptionally (Bauren and Wieslander, 1994; Wuarin and Schibler, 1994; Tennyson et al., 1995), and therefore, it is not expected that a high concentration of non- or partially spliced transcripts is present in speckles. Nevertheless, our FRAP and FLIP results show that nearly all nuclear poly(A)+ RNAs move through speckle domains and are therefore consistent with the suggestion that a quality control for correctly spliced or export-competent RNAs takes place in speckles (Johnson et al., 2000). Also, it cannot be excluded that many of the most active, rapidly transcribed genes may have substantial posttransciptional processing. For these purposes, transcripts may transiently interact with some speckle components, and then be released to the nucleoplasm. Interestingly, the import of poly(A)+ RNA into speckles appears to be temperature, and thus energy dependent, which is consistent with the observation that the uptake of microinjected adenovirus pre-mRNAs by speckles was precluded when cells were incubated at 4°C or ATP depleted (Melcak et al., 2001; Kopsky et al., 2002).
In conclusion, speckles may fulfill different functions in the cell nucleus. In addition to playing a role in the assembly, supply, storage, and, possibly, recycling of RNA processing complexes, speckles may represent a checkpoint for whether or not RNAs are appropriately processed and assembled in transport-competent complexes.
The apparently immobile pool of poly(A)+ RNA residing in speckles and nucleoplasm may represent very slow moving RNAs, structural RNAs, as well as incorrectly or slowly processed RNAs. However, the immobile pool may also reflect storage of specific mRNAs. Many mature mRNAs were observed to accumulate in cell nuclei to higher levels than the corresponding precursors (Gondran et al., 1999), and it has been suggested that the nucleus may function as a reservoir for these mRNAs until they are required in the cytoplasm and released by some stimulus. The mechanism by which these mRNAs are retained in the nucleus has not yet been determined though it was shown that some mRNAs are tightly associated with a nuclear matrix structure that remained after nuclear extraction (Gondran et al., 1999). Future work may shed some light on the role that immobile poly(A)+ RNA plays in the cell nucleus.
Materials And Methods
The DNA probe (dT)40 and the 2'OMe RNA probes (U)22, (A)18, U1snRNA (ccugccagguaaguaug), and HCMV-IE (aaacauccucccauca) were synthesized by M. Lemaitre, M. Dechamps, and D. Largana (Eurogentec, Seraing, Belgium). The DNA oligonucleotide (dT)40 was synthesized using standard phosphoramidite chemistry and purified by HPLC. 2'OMe RNA probes were synthesized using standard 2'OMe phosphoramidite monomers. TAMRA was covalently linked to the 5′-end of probes via a succinimidyl ester derivative (Molecular Probes). All 2'OMe RNA probes were purified twice by reverse phase HPLC with a Waters 600E instrument. Ion molecular weights of purified probes were determined by mass spectrometry using a Time-Of-Flight instrument (Dynamo).
Construction and expression of GFP fusion proteins
The cDNA encoding SF2/ASF was generated by RT-PCR and cloned into the pEGFP-C1 vector (CLONTECH Laboratories, Inc.) using the EcoRI and BamHI restriction sites as described previously (Molenaar et al., 2001). The cDNA encoding PABP2 was subcloned from a construct provided by E. Wahle (Martin-Luther-Universitat Halle, Halle, Germany) in the pEGFP-C1 vector. The constructs coding for TAP-GFP and p15 were provided by E. Izaurralde (European Molecular Biology Laboratory, Heidelberg, Germany; Braun et al., 2001), and the construct ALY-GFP was a gift from J. Katahira (Osaka University, Osaka, Japan; Zhou et al., 2000). SF2/ASF-GFP was transfected stably into U2OS cells. All other constructs were transiently expressed in U2OS cells using DOTAP (Roche Diagnostics GmbH). Cells were analyzed 24–48 h after transfection and were selected for moderate expression and protein-specific localization.
Cell culture and microinjection
U2OS (human osteosarcoma) cells were cultured on coverslips in 3.5-cm petri dishes (Mattek) in RPMI 1640, without phenol red supplemented with 5% FCS, 0.03% glutamine, and 1000 U/ml penicillin/streptomycin and buffered with 25 mM Hepes buffer to pH 7.2 (all from Life Technologies). Cordycepin (Sigma-Aldrich) was used at 50 μg/ml. Microinjection of probes was performed as described previously (Molenaar et al., 2001). Cells showing moderate levels of fluorescence were selected and analyzed by digital fluorescence microscopy on the day of microinjection.
Live cell imaging
Cells were monitored using a confocal microscope system (model TCS/SP2; Leica). Cells were scanned in 2D in time, with a pinhole setting of 2.5 Airy. During the experiment, the temperature of the cells was maintained at 37°C using a heated ring surrounding the culture chamber (Harvard App. Inc.) and a microscope objective heater (Bioptechs) unless indicated otherwise. The 543-nm He Ne laser was used for TAMRA excitation with the emission window set between 560–630 nm. GFP was scanned with the 488-nm line of an Argon laser with the emission window set between 500–540 nm. In the double-labeling experiment, GFP and TAMRA were sequentially scanned to avoid cross talk. Images were acquired using a 100× NA 1.4PL APO lens and analyzed with Leica software. Images were further analyzed using Leica software and Adobe Photoshop. To show colocalization, masked images were obtained using the Leica multi-color software package.
Photobleaching experiments and quantitative analysis
For spot bleaching in FRAP and FLIP analysis, the laser beam parking option on the confocal microscope was used. The 543- and the 488-nm lasers were set at 100%, and the duration of the spot bleaching was set such that bleaching resulted in a nearly complete loss of fluorescence in the defined area. In practice, TAMRA-labeled probes were bleached for 5 s and GFP fusion proteins for 3 s. Subsequent images where recorded before, just after, and at different time intervals after bleaching. The length of the time intervals was established depending on the speed of recovery of the fluorescence. For example, for imaging poly(A)+ RNA, 24 images were acquired in three series with increasing time intervals (10 images every 2 s, 10 images every 10 s, and then 4 images every 30 s). For imaging GFPs and control probes, the time intervals were shorter (indicated in the FRAP curves in Results). Quantitative analysis of fluorescence intensities was performed using Leica software and Excel. FRAP recovery intensities were corrected for background intensities. The total cellular fluorescence was measured before and immediately after the bleach pulse to correct for loss of fluorescence during the bleach pulse and during imaging. For determination of the t1/2 values and for the immobile fractions, the intensity values before bleaching were set at 100% and the intensity values directly after bleaching were set at 0%. All intensities measured during recovery were transformed to relative intensities. The t1/2 values, indicating the time points at which 50% of the end-fluorescence intensity (Fend) was reached, were determined from the FRAP curves. Estimation of the effective diffusion coefficients (Deff) from FRAP experiments was performed as described by Yguerabide et al. (1982) using the formula D = βw2/4t1/2, where w is the radius of the bleached area at e−2 intensity and β is a parameter that depends on the percent bleach. The values for β were determined for each experiment from a theoretical plot generated from data presented by Yguerabide et al. (1982), and w was estimated using fixed cells expressing GFP. For FLIP experiments, cells were repeatedly imaged and bleached at intervals of 2 s for measuring loss of GFP signal and of 3 s for measuring loss of TAMRA fluorescence. For curve fitting, the program CurveExpert 1.3 has been used.
We are grateful to E. Izaurralde for providing TAP-GFP and p15, to J. Katahira for providing Aly-GFP, and to S. van Koningsbruggen for generating the PABP2-GFP construct. We thank K. Wiesmeijer for excellent technical assistance and M. Lemaitre, M. Dechamps, and D. Largana for the synthesis of the probes used in this work. We are also thankful to David Baker for critical reading of the manuscript.
This work was supported by the Netherlands Organization of Scientific Research program 4D Imaging of Living Cells and Tissues (grant 901-34-144).
Abbreviations list: DRB, 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole; FLIP, fluorescence loss in photobleaching; HCMV, human cytomegalovirus; PABP2, poly(A) binding protein II; TAMRA, tetramethylrhodamine.