of mammalian cells - NCBI

The EMBO Journal vol. 1 0 no.7 pp. 1 863 - 1873, 1991

In vivo detection of snRNP-rich organelles in the nuclei of mammalian cells

Maria Carmo-Fonseca, Rainer Pepperkok, Brian S.Sproat, Wilhelm Ansorge, Maurice S.Swanson' and Angus 1.Lamond European Molecular Biology Laboratory, Meyerhofstrasse 1, Postfach 102209, D-6900 Heidelberg, Germany and 'Department of Immunology and Medical Microbiology, College of Medicine, Box J-266, JHMHC, University of Florida, Gainesville, FL 32610, USA Communicated by I.W.Mattaj

The in vivo distribution of snRNPs has been analysed by microinjecting fluorochrome-labelled antisense probes into the nuclei of live HeLa and 3T3 cells. Probes for U2 and U5 snRNAs specifically label the same discrete nuclear foci while a probe for Ul snRNA shows widespread nucleoplasmic labelling, excluding nucleoli, in addition to labelling foci. A probe for U3 snRNA specifically labels nucleoli. These in vivo data confirm that mammalian cells have nuclear foci which contain spliceosomal snRNPs. Co-localization studies, both in vivo and in situ, demonstrate that the spliceosomal snRNAs are present in the same nuclear foci. These foci are also stained by antibodies which recognize snRNP proteins, m3G-cap structures and the splicing factor U2AF but are not stained by anti-SC-35 or anti-La antibodies. Ul snRNP and the splicing factor U2AF closely co-localize in the nucleus, both before and after actinomycin D treatment, suggesting that they may both be part of the same complex in vivo. Key words: in vivo antisense labelling/mammalian nucleus/ microinjection/2'-0-alkyl RNA/snRNA

Introduction A major step in the expression of most eukaryotic protein coding genes involves the excision of introns from nuclear mRNA precursors (pre-mRNAs) in a process termed splicing. An understanding of the splicing mechanism has been greatly aided by the availability of in vitro systems from both yeast and human cells which accurately splice exogenous pre-mRNAs. The splicing mechanism involves two sequential transesterification reactions (Padgett et al., 1984; Ruskin et al., 1984), in a pathway similar to that seen with group II self-splicing introns (reviewed by Jacquier, 1990). A large number of studies have shown that nuclear pre-mRNA splicing takes place in a multi-component structure termed a spliceosome. Major subunits of spliceosomes are the U 1, U2, U4/U6 and U5 snRNPs. In addition, a number of non-snRNP protein factors are required for spliceosome assembly and/or pre-mRNA splicing (for recent reviews see Steitz et al., 1988; Guthrie and Patterson, 1988; Lamond et al., 1990; Bindereif and Green, 1990). Oxford University Press

There is broad agreement from studies in both yeast and mammalian splicing systems that spliceosomes assemble through the sequential binding to pre-mRNA of snRNP and protein components in an ordered pathway. Despite the considerable progress made in recent years in elaborating the details of the splicing mechanism in vitro, comparatively little is known about how the splicing machinery is organized in vivo, or how splicing is integrated in the nucleus with other RNA processing reactions such as 3' polyadenylation. Previous studies have shown that antibodies specific for snRNP proteins give a characteristic 'speckled' nuclear staining pattern when analysed by indirect immunofluorescence (Spector, 1984, 1990; Smith et al., 1985; Nyman et al., 1986; Verheijen et al., 1986; Habets et al., 1989; reviewed in Mattaj, 1988). More recently, a monoclonal antibody raised against partially purified spliceosomes, called anti-SC-35, was shown to label a related pattern of speckled structures in the nucleus (Fu and Maniatis, 1990). However, a different picture of the distribution of snRNPs within the nucleus has emerged from a new in situ hybridization study using antisense oligonucleotide probes specific for individual snRNAs (Carmo-Fonseca et al., 1991). The U2, U4, U5 and U6 snRNAs were found localized in a small number of discrete nucleoplasmic foci, while the Ul snRNA was widely distributed throughout the nucleoplasm, excluding nucleoli. Interestingly, an antibody specific for the splicing factor U2AF gave a widespread nucleoplasmic staining similar to Ul snRNP, in addition to concentrated spots which co-localized with the snRNPcontaining foci (Zamore and Green, 1991; Carmo-Fonseca et al., 1991). Recent studies on the distribution of snRNPs in amphibian oocytes show a different pattern of labelling (reviewed by Gall, 1991). Antibodies specific for snRNP proteins stain nascent transcripts on lampbrush chromosomes and at least three classes of extrachromosomal granules, termed A, B and C 'snurposomes'. Type A snurposomes predominantly contain U1 snRNP while B snurposomes contain all the spliceosomal snRNPs. C snurposomes can be as large as 20 4tM, stain with anti-Sm antibodies and can have several B snurposomes on their surface. The relationship between the mammalian and amphibian oocyte staining patterns is at present completely unclear. In mammalian cells the situation is complicated still further by the observation that certain snRNP-specific antibodies can produce quite different labelling patterns on cells fixed using different procedures (Carmo-Fonseca et al., 1991). In this study we have sought to analyse the in vivo distribution of snRNPs in live mammalian cells in the hope of avoiding problems associated with fixation procedures. To do this, we have made use of a battery of antisense probes made of 2'-O-alkyl RNA. These probes are highly specific for individual snRNAs, hybridize stably with targeted complementary sequences and are resistant to nuclease degradation (Iribarren et al., 1990). These in vivo data

1863

M.Carmo-Fonseca et al.

demonstrate that the discrete foci containing spliceosomal snRNPs previously detected in fixed cells are also present in the nuclei of non-fixed, living cells.

Results In situ labelling of snRNP-containing foci In our previous study, snRNAs were labelled by in situ hybridization in cells fixed after a brief Triton X-100 preextraction treatment. This procedure had been empirically found to give the most efficient labelling with antisense probes. Considerable effort has been made to ensure that the labelling patterns seen with the anti-snRNA oligonucleotide probes are not generated by either the protocol used to prepare cells for in situ hybridization or

by the biotin-avidin detection method. This is important since the in situ labelling pattern of anti-snRNA protein antibodies can be altered by different fixation procedures (Carmo-Fonseca et al., 1991). To demonstrate that the pattern of snRNP labelling was not produced by the Triton pre-extraction treatment, HeLa cells fixed with paraformaldehyde and permeabilized with methanol were labelled in situ using monoclonal antibodies which recognize Sm proteins (Figure IA) and m3G-cap structures (Figure 1B). Both antibodies strongly label the nuclear foci described by Carmo-Fonseca et al. (1991). These foci are also stained by an anti-U2 snRNA antisense

probe (cf. Figure 6, D -F). We conclude that the foci are not generated by Triton treatment. The authentic distribution of Ul snRNPs in the nucleus has not been clearly established since the staining pattern of U 1-specific, anti-70K antibodies can be either widespread in the nucleoplasm, or speckled, depending upon the fixation procedure used (Carmo-Fonseca et al., 1991). To investigate this further, the anti-70K antibody was allowed to bind in living cells by microinjecting it into HeLa cell nuclei (Figure IC). This resulted in widespread nucleoplasmic labelling and not speckled staining. The microinjected anti-70K antibody also labels foci which co-localize with those labelled by an anti-U2 snRNA antisense probe (Figure 1, C and D). These results confirm that Ul is present in the snRNP-rich nuclear foci and support the previous in situ hybridization results showing that U I snRNP is widely distributed throughout the nucleoplasm. Co-localization of snRNAs in situ Antisense probes used to label snRNAs in situ were biotinylated and detected by secondary labelling with fluorochrome-coupled avidin (Carmo-Fonseca et al., 1991). To demonstrate that the snRNP labelling pattern is independent of this detection sysem, anti-snRNA probes were made containing 2,4-dinitrophenylaminopentyl (DNP) residues instead of biotin. These probes are detected using an anti-DNP antibody and fluorochrome-coupled anti-IgG (see Materials and methods). The DNP-containing, anti-

Fig. 1. In situ detection of snRNP-rich foci. HeLa cells fixed with paraformaldehyde and permeabilized with methanol were labelled with monoclonal antibodies directed against Sm proteins, 'Y-12' (A) and m3G-cap structures (B). In both cases bright, discrete nuclear foci are observed (arrowheads). In addition to the foci, the Y-12 antibody stains nucleoplasmic speckles and the anti-m3G-cap antibody gives a punctate cytoplasmic staining. (C) A HeLa cell microinjected with monoclonal antibody anti-U1 snRNP specific protein 70K. One hour after injection the cell was briefly extracted with Triton X-100, fixed and labelled with anti-U2 snRNA biotinylated probe (D). The anti-70K antibody gives a widespread staining of the nucleoplasm with local concentrations at foci which are labelled by the anti-U2 snRNA probe. Bar indicates 10 zm.

1864

In vivo detection of snRNAs

snRNA probes gave an identical in situ labelling pattern to the biotin-linked probes (Figure 2). In both cases, Ul snRNA was widely distributed throughout the nucleoplasm while U2 snRNA was concentrated in discrete nuclear foci. We conclude that the snRNP labelling patterns are not dependent on the biotin-avidin detection system. The twin biotin and DNP detection systems have been exploited to co-localize separate snRNAs directly in situ (Figure 3). HeLa cells were double labelled using a DNPlinked anti-U2 snRNA probe (Figure 3, A-C), and biotinylated probes specific for either Ul (Figure 3D), U5 (Figure 3E) or U6 (Figure 3F) snRNAs. In each case

the foci labelled by both probes co-localize. These data demonstrate that the spliceosomal snRNAs are present in the same nuclear foci. In vivo labelling of snRNAs The nuclease resistance and favourable antisense properties of the 2'-O-alkyl RNA probes (Iribarren et al., 1990) have been exploited to detect snRNAs in live cells using an in vivo hybridization procedure. Anti-snRNA probes coupled to fluorochromes were microinjected into the nuclei of HeLa and 3T3 cells (Figure 4). The cells, still in culture medium, were then directly observed in the fluorescence microscope

Fig. 2. In situ detection of Ul (A, C) and U2 snRNAs (B, D) using antisense 2'-O-allyloligoribonucleotides coupled to either biotin (A, B) or dinitrophenyl (DNP) (C, D). In both cases the Ul snRNA is observed to have a nucleoplasmic widespread localization, while the U2 snRNA appears concentrated at foci. Bar indicates 10 1im.

Fig. 3. In situ co-localization of snRNAs. HeLa cells were double labelled with anti-U2 snRNA probe coupled to DNP (A, B, C) and, respectively, anti-Ul (D), anti-U5 (E) and anti-U6 snRNA (F) probes coupled to biotin. The nuclear foci (arrows) are shown to contain each of the spliceosomal snRNAs. Bar indicates 10

jtm. 1865


J Fig. 4. In vivo labelling of snRNAs. HeLa (panels A-C; G-L) and 3T3 cells (panels D-F) were microinjected with fluorescent antisense probes complementary to U2 (A, D), Ul (B, E), U5 (L) and U3 (C, F) snRNAs. Arrowheads point to nuclear foci. In panel A the cell was injected with a fluorescein-labelled anti-U2 snRNA probe together with an excess of non-fluorescent oligonucleotide probe not complementary to any of the snRNAs. Injected cells show no apparent morphological changes when compared with non-injected cells visualized by phase contrast microscopy (G, J; * indicates the injected cell). Panels E and F depict the same cell co-injected with anti-Ul snRNA coupled to rhodamine X and anti-U3 snRNA coupled to fluorescein. Panels H-L depict co-injection of either anti-U2 snRNA coupled to fluorescein (H) and anti-Ul snRNA coupled to rhodamine X (K) or anti-U2 snRNA coupled to rhodamine X (I) and anti-U5 snRNA coupled to fluorescein (L). Note that in addition to the foci, irregular aggregates of fluorescent maerial are occasionally observed (small arrow in panel I). These structures are not competed with non-fluorescent anti-snRNA probes and are more frequently observed after co-injection of two different oligonucleotides. Bar indicates 10 ltm.

1866


and photographed without any fixation procedures or further manipulations. In both cell types, the anti-U2 snRNA probe specifically labelled discrete nuclear foci (Figure 4, A and D); the anti-U1 snRNA probe showed widespread nucleoplasmic labelling, excluding nucleoli, as well as labelling foci (Figure 4, B, E and K); the anti-U3 snRNA probe specifically labelled nucleoli (Figure 4, C and F). The same specific labelling patterns were observed upon co-injecting two anti-snRNA probes coupled to different fluorochromes into the same cell (Figure 4, compare E with F and H with K). In the case of Ul snRNA, the intense widespread nucleoplasmic fluorescence frequently makes it difficult to distinguish the foci (Figure 4, B and K). It is striking that, in vivo, the anti-U2 snRNA probe labels only discrete foci. Increasing the amount of fluorescent oligonucleotide injected results in a general increase in the level of background nuclear staining but does not label additional foci. The labelling of foci, but not the background nucleoplasmic staining, could be specifically competed by co-injection of an excess of non-fluorescent U2 probe (data not shown). Co-injection of a non-fluorescent oligonucleotide not complementary to U2 snRNA did not compete for labelling of foci but rather increased the staining of foci relative to the general nuclear background, presumably by blocking non-specific binding sites (Figure 4, compare A with G). Injection of a non-specific antisense probe showed only a diffuse nuclear staining and did not preferentially label

either foci or nucleoli (data not shown). Cells microinjected with oligonucleotide probes showed no discernible morphological changes compared with uninjected cells (Figure 4, G and J). A fluorescent antisense probe specific for U5 snRNA also labelled only discrete foci when microinjected into HeLa cell nuclei (Figure 4L). Co-injection of anti-U5 and anti-U2 snRNA probes coupled to different fluorochromes showed that both U2 and U5 snRNAs are present in the same foci in vivo (Figure 4, I and L). Similarly, co-injection of anti-Ul and anti-U2 snRNA probes coupled to different fluorochromes showed that U1 is also in the same foci in vivo (Figure 4, H and K). We conclude that this antisense approach accurately localizes snRNAs in vivo. The in vivo distribution of snRNAs observed closely parallels that seen by in situ hybridization using fixed, permeabilized cells (Carmo-Fonseca et al., 1991). Antibodies against the splicing factors U2AF and SC-35 stain distinct nuclear structures Fu and Maniatis have recently described the characterization of a monoclonal antibody (anti-SC-35) raised against partially

purified mammalian spliceosomes (Fu and Maniatis, 1990). This antibody inhibits pre-mRNA splicing in vitro and recognizes a protein of -35 kd on Western blots. In situ labelling with anti-SC-35 showed a pattern of 20-50 discrete nuclear speckles. We have observed a similar speckled

Fig. 5. HeLa cells permeabilized with Triton X-100 and fixed with paraformaldehyde were incubated with antibodies directed against the splicing factors SC-35 (A, B, D) and U2AF (C) and monoclonal antibody LaiB5 directed against the La antigen (E, F). The cells depicted in panel B were microinjected with anti-SC-35 antibody prior to fixation. Panels C, F and D, E represent double labelling experiments analysed with the confocal fluorescence microscope. Both anti-SC-35 and LalB5 antibodies stain 20-50 nuclear speckles which are clearly distinct from the foci labelled by anti-U2AF (arrows in panel C). Bar indicates 10 Mm.

1867


nuclear staining pattern using the anti-SC-35 antibody to stain HeLa cells fixed under different conditions (Figure 5A and data not shown). When the anti-SC-35 antibody is microinjected into HeLa cell nuclei prior to fixation, a related labelling pattern is observed, although the stained structures appear more rounded (Figure 5B). This labelling pattern is quite different from that reported for the mammalian splicing factor U2AF (Zamore and Green, 1991; Carmo-Fonseca et al., 1991). As shown in Figure SC, antibodies specific for U2AF give widespread nucleoplasmic staining with concentrations in the snRNP-rich foci. By chance, we noticed that the anti-SC-35 antibody staining pattern is exactly the same as the nuclear staining observed with another monoclonal antibody, called LaIBS (Bachmann et al., 1986), which was raised against purified La antigen (Figure 5,

compare D with E). This antibody recognizes a 50 kd protein on Western blots and appears not to inhibit pre-mRNA splicing in vitro (our unpublished observations). We have confirmed by Western blotting that anti-SC-35 and LalBS monoclonal antibodies primarily label proteins of 35 kd and SO kd as reported, although both additionally label multiple other bands more weakly (data not shown). Like anti-SC-35, the staining pattern observed with the LaiBS monoclonal antibody did not co-localize with U2AF (Figure 6, compare C with F). Additional experiments show that the anti-SC-35 staining pattern does not co-localize with that of other antibodies reported to recognize La antigen, including patient autoimmune sera, affinity purified human autoimmune antibodies and monoclonal antibody SWS (data not shown, see Materials and methods). Therefore, as not -

m

Fig. 6. HeLa cells were double labelled in situ with antibodies directed against Sm proteins ('Kung', A) and the splicing factor SC-35 (B). The overlay (C) demonstrates that the foci labelled by anti-Sm antibodies (arrowheads, green staining) are not labelled by anti-SC-35. Most nucleoplasmic speckles stained by anti-SC-35 antibody co-localize with those labelled by anti-Sm (yellow staining). Additional structures labelled by anti-SC-35 are not stained by anti-Sm (red staining indicated by small arrow). Cells double labelled with anti-U2 snRNA probe (D) and anti-Sm Y-12 antibodies (E) show that the snRNP-rich foci are stained by anti-Sm antibodies (arrowheads, yellow staining in F). In this experiment the cells were fixed with paraformaldehyde and permeabilized with methanol. Overlays of cells microinjected with anti-SC-35 antibody and double labelled with either anti-U2 snRNA probe (G) or anti-m3G-cap antibody (H) demonstrate that SC-35 is not present in the snRNP-containing foci (arrowheads, green staining). Bar indicates 10 Am.

1868


all antibodies reported to be specific for La showed the same pattern of in situ labelling, the relationship between the SC-35 and La antigens is unclear. What is clear however is that a monoclonal antibody specific for an antigen of 50 kd, which has no obvious relation to the splicing machinery, shows an identical nuclear labelling pattern in situ to that of anti-SC-35. SC-35 is not present in the snRNP-containing nuclear foci The anti-SC-35 staining pattern has been analysed further for co-localization with other components of the splicing machinery (Figure 6). In agreement with Fu and Maniatis (1990), we observe a partial overlap between the in situ staining patterns of anti-SC-35 and anti-Sm antibodies (Figure 6, A-C). The regions where the staining patterns overlap appear yellow in Figure 6C. Structures labelled by anti-Sm antibodies which are not stained by anti-SC-35 include the snRNP-rich foci (green staining in Figure 6C). It is also apparent that the anti-SC-35 antibody labels structures not stained by the anti-Sm antibodies, as shown

by the red staining in Figure 6C. A parallel double labelling proves that the foci stained by anti-Sm antibodies co-localize with those stained by a U2 snRNA-specific antisense probe (Figure 6, D -F; see yellow staining in F). Either a U2 snRNA-specific antisense probe (Figure 6G) or an antim3G-cap monoclonal antibody (Figure 6H) also show that SC-35 is not present in the snRNP-containing foci. The anti-SC-35 stained structures are also not labelled by a monoclonal antibody specific for m3G-cap (Figure 6H). This implies that either snRNPs are not present in these structures (although most of them react with anti-Sm antibodies) or else the snRNPs present have their m3G-caps masked. hnRNP proteins are not concentrated in the snRNP-rich foci Antibodies specific for the A, C, L and M classes of hnRNP proteins were used in double labelling experiments with the anti-U2 snRNA antisense probe to test for their presence in foci (Figure 7). In agreement with previous studies (Choi and Dreyfuss, 1984; Pihol-Roma et al., 1989) we observe

Fig. 7. HeLa cells were double labelled with antibodies specific for the A, C, L and M classes of hnRNP proteins (A, B, C, G, H) and either antiU2 snRNA probe (D, F, I) or anti-U2AF antibodies (E). None of the hnRNPs are apparently concentrated at the snRNP-containing foci (arrowheads). Also the large nuclear structures stained specifically by the anti-hnRNP L antibodies (arrows in panel H) do not co-localize with the foci. Bar indicates 10 /m.

1869


the A, C and L hnRNP staining widely distributed throughout the nucleoplasm, excluding nucleoli (Figure 7, A, B, G and H). A similar pattern is seen with the newly characterized M class of hnRNPs (Figure 7C). None of the hnRNPs are apparently concentrated in the snRNP-containing foci, contrasting with the staining pattern of anti-U2AF antibodies (cf. Figure 7E). The large, spot-like nuclear structures stained specifically by the anti-hnRNP L antibodies (Pihol-Roma et al., 1989) do not co-localize with the snRNPcontaining foci (Figure 7H and I). Optical sectioning using the confocal microscope showed no evidence for exclusion of any of the classes of hnRNP proteins from the foci, as might be expected if the foci represented snRNP storage particles. However, although the data are consistent with hnRNPs being present in the foci, the widespread, intense nuclear staining shown by all the anti-hnRNP antibodies makes it difficult to be sure of this.

Splicing factor U2AF and U1 snRNP remain co-localized after RNA synthesis is inhibited by actinomycin D The nuclear distribution of U 1 snRNP was previously shown to alter dramatically after inhibition of RNA synthesis by actinomycin D. The widespread nucleoplasmic distribution of Ul was changed to a concentrated, petal-like pattern around the remnants of nucleoli, while the other spliceosomal snRNPs remained associated with foci (Carmo-Fonseca et al., 1991). This effect is also observed in unfixed cells which have been exposed to actinomycin and then microinjected with a fluorescently labelled anti-U1 snRNA probe (Figure 8A). Similar results are obtained when the fluorescently labelled anti-U 1 snRNA probe is microinjected prior to the actinomycin treatment (data not shown). This change in the Ul labelling pattern is thus clearly not the result of fixation procedures.

Fig. 8. HeLa cells treated for 1 h with 5 Lg/ml actinomycin D were microinjected with anti-Ul snRNA probe coupled to fluorescein (A). Arrowheads point to the petal-like structure that forms around remnants of the nucleolus. Similar structures are labelled in situ by anti-U2AF (B) and Ul-specific anti-70K (C) antibodies in actinomycin treated 3T3 cells. Longer exposure to the drug (up to 2 h) results in disintegration of nucleoli and appearance of isolated 'petals' scattered throughout the nucleus (small arrows in panel B). Double labelling of actinomycin-treated HeLa cells with antibodies directed against hnRNP proteins (D-G) shows no concentration of staining around nucleoli, in contrast with the pattern revealed by anti-U2AF antibodies in the same cells (arrowheads in panels H and I). Note that the large nuclear structures stained specifically by the anti-hnRNP L antibodies do not appear to be affected by actinomycin treatment (arrow in panel F). Bar indicates 10 ptm.

1870


An identical effect of actinomycin D is seen on the nuclear distribution of the splicing factor U2AF in both 3T3 (Figure 8B) and HeLa (Figure 8H and I) cells. In both cases, U2AF and U 1 snRNP are present around nucleolar remnants in the characteristic petal-like pattern (Figure 8B and C). Double labelling shows that U 1 snRNP and U2AF precisely co-localize after actinomycin treatment (Figure 9, A-C). We have never observed a difference in the kinetics with which Ul and U2AF respond to actinomycin D. These results are therefore consistent with U1 snRNP and U2AF being present in a complex in vivo. Having observed that U2AF responds to actinomycin D in the same way as Ul snRNP, experiments were performed to assess whether this is also true for hnRNP proteins or SC-35 (Figures 8 and 9). Cells treated with actinomycin were therefore labelled with antibodies specific for the A, C, L and M classes of hnRNP proteins (Figure 8D -G). In all cases the widespread nuclear staining seen before actinomycin treatment remains (cf. Figure 7), but takes on a more granular appearance. There is no concentration of labelling around nucleoli. Double labelling experiments show that U2AF adopts the characteristic petal-like staining after actinomycin treatment in the same cells where hnRNP M and L proteins are still widely distributed throughout the nucleoplasm (Figure 8, compare E with H and F with I). The anti-SC-35 antibody continues to label nuclear

speckles after actinomycin treatment and also does not show concentrated labelling around nucleoli (Figure 9D). Double labelling with anti-SC-35 and anti-Sm antibodies again shows a partially overlapping staining pattern (cf. Figure 6), but with only the anti-Sm antibody staining the petal-like structures (Figure 9D-F, see red staining in F). This Sm staining is presumably due to the presence of Ul snRNP. These data establish that the dramatic concentration of labelling around nucleoli is only observed using probes which recognize either U1 snRNP or U2AF. It is not seen with the other spliceosomal snRNPs (Carmo-Fonseca et al., 1991), and is not a general property of RNA binding proteins (this work). As both the U1 snRNP and U2AF labelling patterns not only specifically change after actinomycin treatment, but do so with apparently identical kinetics, it is likely, though not proven, that these factors are associated.

Discussion By microinjecting fluorochrome-coupled antisense probes specific for spliceosomal snRNAs into the nuclei of mammalian cells, we have demonstrated that the snRNPcontaining foci previously identified by in situ hybridization analyses are authentic nuclear structures present in living cells. Co-localization studies confirm that the spliceosomal snRNAs are present in the same foci, along with snRNP

Fig. 9. Double labelling of HeLa cells treated with actinomycin D for 1 (D, E, F) or 2 h (A, B, C). The labelling pattern observed after actinomycin treatment with the anti-U2AF antibody (A) completely co-localizes with that of the Ul-specific anti-70K antibody (B), as demonstrated by the yellow staining in panel (C). The overlay depicted in (F) shows only a partial overlap in the staining pattern of anti-SC-35 (D) and anti-Sm 'Kung' (E) antibodies after actinomycin treatment. In particular, only anti-Sm antibodies stain the 'petal-like' perinucleolar rim (panel F, arrowheads, red staining). Bar indicates 10 itm.

1871


proteins, m3G-cap structures and the splicing factor U2AF. The in vivo data also support the distinction between the labelling pattern of Ul snRNA, which is widely distributed throughout the nucleoplasm as well as being in foci, and the other spliceosomal snRNAs, which are predominantly localized in the same discrete foci. We propose that the snRNP-containing foci correspond to nuclear organelles and anticipate that they play a specific role in the processing of pre-mRNA transcripts. Based on their composition, possible functions in the RNA processing pathway which could be localized in the foci include the pre-assembly of multi-snRNP particles, splicing itself, re-cycling of snRNPs from post-splicing complexes, degradation of excised introns or possibly sorting of excised introns from spliced mRNA as part of a nuclear export pathway. It is known that cytoplasmic organelles such as the Golgi complex are involved in molecular sorting processes. A nuclear organelle might also be required to sort out RNAs that are to leave the nucleus from those that are to be retained and could possibly perform a proof-reading function to ensure that exiting RNAs have been correctly processed. In this regard it is interesting that the protein product of the yeast PRP22 gene has recently been reported to have a specific role in promoting release of spliced mRNA from the spliceosome (Company et al., 1991). It is conceivable that this activity could be compartmentalized in vivo. Alternatively, it is possible that a separate compartment is used within the nucleus for intron degradation and snRNP recycling, maybe to minimize the chance of unspliced pre-mRNA being accidentally degraded in the spliceosome. While we cannot presently exclude the possibility that the foci represent some form of storage particle, we consider this unlikely. There is no reason to expect that the splicing apparatus is present in such large excess over substrate that part of it must be stored. Indeed, U1, the most abundant snRNP in somatic cells, is mostly not confined to the foci. We are also unaware of any precedents for nuclear storage particles in somatic cells. Understanding what function the snRNP-containing organelles carry out clearly now requires determination of whether intron and/or exon sequences are also localized in these structures. Due to the intense, widespread nucleoplasmic staining seen with all classes of hnRNP proteins it is not yet possible to draw firm conclusions as to whether pre-mRNAs are in the snRNP-containing organelles, although the data are certainly consistent with this possibility. Experiments to localize specific pre-mRNAs within the nucleus are currently in progress. It is unclear why the immunolabelling with anti-SC-35 antibodies shows no correlation with either the splicing factor U2AF or the spliceosomal snRNAs detected using antisense probes. The most interesting explanation for this difference would be that SC-35 acts at a distinct step in splicing and is therefore located in a separate intranuclear compartment. There may conceivably also be a subset of spliceosomal snRNPs, which are not accessible to antisense probes or to the anti-m3G-cap monoclonal antibody, present in the same structures that contain SC-35. An alternative explanation could be that the major component of the in situ staining pattern seen with the anti-SC-35 antibody corresponds to labelling of a different protein to the splicing factor detected by Western blotting of nuclear extracts. It is hoped this will be clarified by future studies. 1872

The striking co-localization of U1 snRNP and splicing factor U2AF, and their specific and simultaneous change in labelling pattern after actinomycin D treatment, is highly suggestive of an association between the two in vivo. Although U2AF has been purified to homogeneity and clearly fractionates independently of either U1 or U2 snRNPs (Zamore and Green, 1989, 1991), it is possible that the purification procedure employed in these studies disrupts an interaction between U2AF and Ul which occurs in vivo. Ul snRNP and U2AF are the first factors shown to bind to the 5' and 3' splice site regions during pre-mRNA splicing in vitro. Their putative in vivo association could thus form a complex capable of recognizing both ends of an intron and so, upon binding to a nascent transcript, serve to bring together the separate splice sites. The widespread nucleoplasmic distribution of Ul and U2AF could be a means to ensure the rapid identification of introns within any new transcript appearing in the nucleus. Whether the rest of the splicing machinery would then assemble on the nascent transcript, or the pre-mRNA - U1 -U2AF complex would migrate instead to the snRNP-containing organelles for splicing, is an open question. An interesting corollary of all the models discussed above, however, is that specific transport mechanisms must exist within the nucleus to localize and relocate pre-mRNA substrates and components of the processing machinery.

Materials and methods Oligonucleotide synthesis and labelling Oligonucleotides were synthesized as described by Sproat et al. (1989). All oligonucleotides were made of 2'-O-allyl RNA as described by Iribarren et al. (1990). Sequences of the oligonucleotides are listed in Table I. Biotinylation of antisense probes was performed during the solid phase synthesis (Pieles et al., 1990). Incorporation of a modified deoxycytidine monomer carrying a 2,4-dinitrophenylamino moiety was based on the procedure of Roget et al. (1989). Post-labelling of antisense probes with fluorophores was enabled by prior incorporation of either a modified 2'-deoxycytidine building block bearing an N4-(5-trifluoroacetylaminopentyl) spacer (Sproat et al., 1989), or a 5'-amino linker (Connolly, 1987), during the solid phase synthesis. A suitable amino linker is available from Applied Biosystems (Foster City, CA, USA). Fluorescent labelling of aminolinked antisense probes was performed in DMSO/0. 1 M aqueous borate buffer, pH 8.5, using the succinimidyl esters of 5-(and -6)-carboxyfluorescein or of 5-(and -6)-carboxy-X-rhodamine. The latter compounds were purchased from Molecular Probes, Inc. (Eugene, OR, USA). Post-labelling reactions were typically done for 14-18 h at room temperature. Fluorophore-labelled oligonucleotides were separated from free dye on a Sephadex G15 column eluted in 15% ethanol and then purified by electrophoresis through a 15% denaturing polyacrylamide gel run in 1 x TBE buffer. Table I. Sequence of oligonucleotides

Oligonucleotide

Sequence (5'-3')

anti-Ul anti-U2 anti-U3 anti-U5 anti-U6

CUICCAIIUAAIUAU

UA*GUA*AA*AGGCG AUGGAACGCUUCACAAU

non-snRNA competitor

IACCAIAUIIACICIICC

IAACAIAUACUACACUU UUUCIIUICUC

All oligonucleotides were made of 2'-0-allyl RNA (Iribarren et al., 1990). For the various experiments described, the oligonucleotides were 5' end-labelled with either biotin, DNP, fluorescein or rhodamine X. The bases in the U5-specific probe marked A* correspond to 2-aminoadenosine (Sproat et al., 1991a,b; Lamm et al., 1991). Other bases used were uridine (U), adenosine (A), inosine (I), guanosine (G) and cytidine (C).


In situ hybridization and immunofluorescence HeLa and 3T3 cells were grown on glass coverslips and processed as described (Carmo-Fonseca et al., 1991). Cells were either (i) extracted with 0.5% Triton X-100 in CSK buffer for 30 s on ice and fixed in 3.7% paraformaldehyde in the same buffer for 10 min at room temperature; or (ii) fixed with 3.7% paraformaldehyde in CSK buffer for 10 min and subsequently permeabilized in cold (-20°C) methanol for 10 min. Fixed cells were hybridized with anti-snRNA probes or incubated with antibodies as previously described (Carmo-Fonseca et al., 1991). Hybridization sites of antisense 2'-O-alkyl RNA probes labelled with DNP were detected using a monoclonal anti-DNP antibody (Sigma, St Louis, MO, USA) and a secondary anti-mouse IgG antibody coupled to fluorescein (Sigma). The following antibodies were used: human autoimmune sera directed against Sm ('Kung') and La antigens ('Mar' and 'Le', kindly provided by Dr J.Steitz); affinity purified human autoantibodies and monoclonal antibody anti-La ('SW5', kindly provided by Dr W.Van Venrooij); monoclonal anti-La antibody (Bachmann et al., 1986); monoclonal antibodies anti-Sm ('Y 12') (Pettersson et al., 1984); anti-70K protein, 70K (Billings et al., 1982); anti-2,2,7-trimethylguanosine cap structures (Bochnig et al., 1987); anti-SC-35 splicing factor (Fu and Maniatis, 1990); anti-hnRNP peptides Al ('4B10'), C1-C2 ('4F4'), L ('4DII') (Choi and Dreyfuss, 1984; Pinol-Roma et al., 1989), and M1-M2 ('1D8' and '2A6'; Swanson et al., manuscript in preparation); and affinity purified rabbit antibodies raised against a peptide from the 65 kd subunit of U2AF (Zamore and Green, 1991). Conventional fluorescence microscopy and confocal fluorescence microscopy were performed as described (Carmo-Fonseca et al., 1991).

Microinjection of antibodies Microinjection into the nucleus of HeLa cells was performed using the AIS microinjection system previously described (Ansorge and Pepperkok, 1988). Cells were incubated for 1-2 h at 37°C, washed with PBS, briefly extracted with 0.5% Triton X-100 (30 s on ice) and fixed with 3.7% paraformaldehyde (10 min at room temperature). After extensive washing in PBS, the cells were directly labelled with fluorescent secondary antibodies or pre-incubated with a biotinylated anti-snRNA probe and avidin-fluorescein.

Bochnig,P., Reuter,R., Bringmann,P. and Luhrmann,R. (1987) Eur. J. Biochem., 168, 461-467. Carmo-Fonesca,M., Tollervey,D., Pepperkok,R., Barabino,M.L., Merdes,A., Brunner,C., Zamore,P.D., Green,M.R., Hurt,E. and Lamond,A.I. (1991) EMBO J., 10, 195-206. Choi,Y.D. and Dreyfuss,G. (1984) J. Cell Biol., 99, 1997-2004. Company,M., Arenas,J. and Abelson,J. (1991) Nature, 349, 487-493. Connolly,B.A. (1987) Nucleic Acids Res., 15, 3131-3139. Fu,X.-D. and Maniatis,T. (1990) Nature, 343, 437-441. Gall,J.G. (1991) Adv. Dev. Biochem., in press. Guthrie,C. and Patterson,B. (1988) Annu. Rev. Genet., 22, 387-419. Habets,W.J., Hoet,M.H., DeJong,B.A.W., VanDer Kemp,A. and Van Venrooij,W.J. (1989) J. Immunol., 143, 2560-2566. Iribarren,A.M., Sproat,B.S., Neuner,P., Sulston,I., Ryder,U. and Lamond,A.I. (1990) Proc. Natl. Acad. Sci. USA, 87, 7747-7751. Jacquier,A. (1990) Trends Biochem. Sci., 15, 351-354. Lamm,G., Blencowe,B.J., Sproat,B.S., Iribarren,A.M., Ryder,U. and Lamond,A.I. (1991), submitted. Lamond,A.I., Barabino,S. and Blencowe,B.J. (1990) In Eckstein,F. and Lilley,D.M.J. (eds), Nucleic Acids and Molecular Biology. SpringerVerlag, Berlin , Vol. 4, pp. 243-257. Mattaj,I.W. (1988) In Birnstiel,M.L. (ed.), Structure and Function of Major and Minor Small Nuclear Ribonucleoprotein Particles. Springer-Verlag, Berlin, pp. 100-114. Nymann,U., Hallman,H., Hadlaczky,G., Pettersson,I., Sharp,G. and Ringertz,N.R. (1986) J. Cell Biol., 102, 137-144. Padgett,R.A., Konarska,M.M., Grabowski,P.J., Hardy,S.F. and Sharp,P.A. (1984) Science, 225, 898-903. Pettersson,I., Hinterberger,M., Mimori,T., Gottlieb,E. and Steitz,J.A. (1984) J. Biol. Chem., 259, 5907-5914. Pieles,U., Sproat,B.S. and Lamm,G.M. (1990) Nucleic Acids Res., 18, 4355-4360.

Pihol-Roma,S., Swanson,M.S., Gall,J.G. and Dreyfuss,G. (1989) J. Cell Biol., 109, 2575-2587. Roget,A., Bazin,H. and Teoule,R. (1989) Nucleic Acids Res., 17, 7643 -7651.

Ruskin,B., Krainer,A.R., ManiatisT. and Green,M.R. (1984) Cell, 38, In vivo hybridization Anti-snRNA 2'-O-alkyl oligonucleotides linked to fluorescein or rhodamine were injected into the nucleus of HeLa and 3T3 cells grown on coverslips. Note that due to the unavoidable inherent variation in the volume injected per cell by the AIS apparatus it is not possible to quantify precisely the amount of probe injected into any given cell. Cells were incubated at 37°C for 5-120 min. The coverslips were mounted in a Plexiglass chamber containing 400 tsl culture medium and transferred to the microscope stage (IM 35; Zeiss). Photographs were taken on Tmax 400 or 3200 film (Kodak).

Acknowledgements The authors are grateful to Susan Weston and Samantha O'Loughlin for expert assistance in the preparation of oligonucleotides and to Ramon Guimil and Uwe Pieles for synthesizing modified nucleoside phosphoramidites bearing 2,3-dinitrophenylamino and biotin moieties respectively. We thank Dr Iain Mattaj for carefully reading the manuscript and making helpful comments, and Dr Ernst Stelzer for help in the use of the confocal fluorescence microscope. We also thank the following laboratories for generously providing antibodies used in this study: Professor Tom Maniatis for anti-SC-35, Professor Gideon Dreyfuss for antibodies specific for the hnRNP A, C and L proteins, Professor Reinhard Luhrmann for anti-m3G-cap monoclonal antibody, Professor Michael Green for antiU2AF, Professor Joan Steitz and Professor Walter van Venrooij for antibodies against La antigen and Dr M.Bachmann for monoclonal antibody La1B5. M.C.-F. was supported by a European Molecular Biology Organization long-term fellowship.

317-331.

Smith,H.C., Spector,D.L., Woodcock,C.L.F., Ochs,R.L. and Bhorjee,J. (1985) J. Cell Biol., 101, 560-567. Spector,D.L. (1984) J. Biol. Cell, 51, 109-111. Spector,D.L. (1990) Proc. Natl. Acad. Sci. USA, 87, 147-151. Sproat,B.S., Lamond,A., Beijer,B., Neuner,P. and Ryder,U. (1989) Nucleic Acids Res., 17, 3373 - 3386. Sproat,B.S., Iribarren,A.M., Beijer,B., Peiles,U. and Lamond,A.I. (1991a) Nucleosides and Nucleotides, 10, in press. Sproat,B.S., Iribarren,A.M., Guimal,R. and Beijer,B. (1991b) Nucleic Acids Res., 19, 733-738. Steitz,J.A., Black,D.A., Gerke,V., Parker,K.A., Kramer,A., Frendeway,D. and Keller,W. (1988) In Birnstiel,M.L. (ed.), Structure and Function of Major and Minor Small Nuclear Ribonucleoprotein Particles. SpringerVerlag, Berlin, pp. 71-99. Verheijen,R., Kuijpes,H., Vooijs,P., Van Venrooij,W. and Ramaekers,F. (1986) J. Cell Sci., 86, 173-190. Zamore,P.D. and Green,M.R. (1989) Proc. Natl. Acad. Sci. USA, 86, 9243 -9247. Zamore,P.D. and Green,M.R. (1991) EMBO J., 10, 207-2 14. Received on March 18, 1991

References Ansorge,W. and Pepperkok,R. (1988) J.

Biochem.

Biophvs. Methods, 16,

283-292.

Bachmann,M., Mayet,W.J., Schroder,H.C., Pfeifer,K., Meyer zum Buschenfeld,K.-H. and Muiller,W.E.G. (1986) Proc. Natl. Acad. Sci. USA, 83, 7770-7774. Billings,P.B., Allen,R.W., Jenssen,F.C. and Hoch,S.O. (1982) J. Imnunol., 128, 1176-1180. Bindereif,A. and Green,M.R. (1990) In Setlow,J.K. (ed.), Genetic Engineering. Plenum Press, New York, Vol. 12, pp. 201-224.

1873