Interaction of the human autoantigen p150 with splicing snRNPs

Messenger RNA precursor (pre-mRNA) splicing takes place within a multi-subunit structure, termed the spliceosome (Grabowski et al., 1985; Brody and Abelson, 1985). Assembly of functional spliceosomes involves the step-wise addition of small nuclear ribonucleoprotein particles (snRNPs) together with non-snRNP protein factors on the pre-mRNA (reviewed by Steitz et al. 1988; Lamond et al., 1990; Green, 1991). These splicing factors co-operate to fold the pre-mRNA into a catalytically active complex, in which the precise cleavage and ligation steps take place, resulting in the production of mature mRNA. Much effort is currently being directed at understanding the role of individual components of the spliceosome in the splicing process and, in particular, how these components serve to accurately locate splice sites, bring about catalysis, and also regulate splice site selection (for a recent review see Rio, 1992).

The best-characterised pre-mRNA splicing factors are the ‘U’ (for uracil-rich) snRNPs, in particular U1, U2, U4/U6 and U5. Each snRNP particle is composed of one (U1, U2 and U5) or two (U4/U6) small nuclear RNAs (snRNAs), complexed to a set of proteins (reviewed by Lührmann 1988; Lührmann et al., 1990). These proteins can be broadly divided into two groups: the common (Sm-class) proteins, which are associated with all four snRNP particles, and the specific snRNP proteins, associated with only one snRNP particle. At an early stage of spliceosome formation, U1 snRNP binds to the 5′ splice site region and forms a ‘commitment complex’ (Ruby and Abelson, 1988; Séraphin and Rosbash, 1989; Michaud and Reed, 1991; Jamison et al., 1992). Subsequent binding of U2 snRNP results in the formation of a kinetically distinct, pre-splicing complex, termed complex A. This is followed by the addition of the U4/U6 and U5 snRNPs to form a spliceosome (Pikielny et al., 1986; Konarska and Sharp, 1987; Cheng and Abelson, 1987; Lamond et al., 1988). As well as the snRNPs, nonsnRNP protein factors also interact with pre-mRNA at different stages of the assembly pathway. For example, PTB (Garcia-Blanco et al., 1989), SC-35 (Fu and Maniatis, 1990), SF1,3 (Krämer and Utans, 1991), SF2/ASF (Ge and Manley, 1990; Krainer et al., 1990) and U2AF (Zamore and Green, 1989) all function early during spliceosome formation, and SF4 (Utans and Krämer, 1990) is required for the conversion of an assembled, inactive complex containing all four snRNPs, to an active spliceosome. There is also increasing evidence that the binding of heteregeneous nuclear ribonucleoproteins (hnRNPs) to pre-mRNA plays an important role in the splicing reaction (Choi et al., 1986; Mayeda and Krainer, 1992).

As described above, the later stages of spliceosome formation involve the addition of the U4/U6 and U5 snRNPs after previous binding of U1 and U2 snRNPs. In both mammalian and yeast extracts the U4/U6 and U5 snRNPs have been shown to assemble into a U4/U6.U5 tri-snRNP particle in the absence of pre-mRNA (Konarska and Sharp, 1987; Cheng and Abelson, 1987). Several lines of evidence indicate that the U4/U6 and U5 snRNPs join the spliceosome in the form of this pre-associated U4/U6.U5 trisnRNP particle. First, the kinetics of snRNP binding to premRNA during spliceosome formation shows that U4/U6 and U5 snRNPs bind at the same time (Konarska and Sharp, 1987; Cheng and Abelson, 1987; Bindereif and Green, 1987). Second, selective depletion of the U4/U6 snRNP from splicing extracts prevents stable binding of the U5 snRNP to pre-mRNA and selective depletion of U5 snRNP similarly prevents stable binding of U4/U6 (Barabino et al., 1990; Lamm et al., 1991). Third, a heat-labile factor required for pre-mRNA splicing and spliceosome formation has been shown to correspond to a protein that is both specifically associated with the U4/U6.U5 tri-snRNP and needed for tri-snRNP assembly (Utans et al., 1992). This latter study also provides one of the clearest examples of the functional role of a specific mammalian snRNP protein in the splicing mechanism.

In order to make further progress in elucidating functional roles for the snRNP proteins it is essential to determine the exact protein composition of each snRNP species as a basis for more detailed structural and functional studies. In mammalian systems, purification of spliceosomal snRNPs by a combination of biochemical fractionation and immune-affinity chromatography has identified many of the snRNP-specific protein components belonging to U1 snRNP, 17 S U2 snRNP, 20 S U5 snRNP and U4/U6.U5 tri-snRNP particle (Pettersson et al., 1984; Bringmann and Lührmann, 1986; Lelay-Taha et al., 1986; Bach et al., 1989; Behrens and Lührmann, 1991; Behrens et al., 1993). In the case of the mammalian U4/U6 snRNP, however, no specific proteins have been identified, despite reports of U4/U6-specific proteins in other organisms. For example, genetic studies in yeast have revealed the product of the PRP24 gene to be a U4/U6-specific protein (Shannon and Guthrie, 1991), and proteins that are specific for trypanosomal U4/U6 snRNP have been detected using antisense-affinity selection (Palfi et al., 1991). In this regard it is interesting that Okano and Medsger (1991) recently reported a rare patient serum (designated ‘MaS’) containing autoantibodies that specifically immunoprecipitate human U4/U6 snRNP. These authors showed that MaS serum specifically recognises a 150 kDa antigen (here referred to as p150) in total nuclear extracts; antibodies affinity purified from the p150 antigen could immunoprecipitate U4/U6 snRNP. In addition to p150 and U4/U6 snRNAs, MaS serum was also shown to co-immunoprecipitate proteins corresponding to 120, 80, 46 and 34 kDa. Thus it was proposed that at least p150, and possibly one or more of these additional proteins, is specific for the U4/U6 snRNP. However, it was not clear from these studies how the proteins that are immunoprecipitated by the MaS serum relate to proteins recently identified in the U4/U6.U5 tri-snRNP particle (Behrens and Lührmann, 1991). Nevertheless, the MaS serum is the first reagent that specifically recognises mammalian U4/U6 through an associated protein component, and therefore is an important tool for characterising the biochemistry and cell biology of this snRNP particle.

In this paper we report new data on the structure, function and nuclear organisation of the snRNPs that assemble late in the spliceosome pathway, i.e. U4/U6 and U5. Specifically, we present a combination of biochemical and cytological evidence indicating that the autoantigen recognised by MaS serum, p150, is, at least in part, a snRNP-associated factor that can interact with multiple splicing snRNPs and has a nuclear distribution similar to that of the nonsnRNP splicing factor U2AF. In addition, we confirm that the purified mammalian U4/U6.U5 tri-snRNP particle is a functional spliceosome subunit, by using a biochemical complementation assay.

2′-O-alkyl RNA oligonucleotides were synthesised as described by Sproat et al. (1989). Sequences of the oligonucleotides are shown in Table 1. Biotin residues were coupled to the oligonucleotides during solid-phase synthesis as previously described (Pieles et al., 1990).

Preparation of snRNP-depleted HeLa cell nuclear splicing extracts was carried out using 2′-O-alkyl RNA oligonucleotides as previously described (Barabino et al., 1990; Blencowe et al., 1992). Depletion of U6 snRNP was performed using an oligonucleotide complementary to nucleotides 42-61 of U6 snRNA, and U4 snRNP was depleted using an oligonucleotide complementary to nucleotides 64-83 of U4 snRNA (see Table 1). Approximately 1-2 nmole of oligonucleotide was used per ml of nuclear extract. The biotinylated oligonucleotides were incubated in nuclear extracts for 1 h at 30°C, prior to affinity selection with streptavidin-agarose beads (Sigma, Germany); incubation of nuclear extracts with streptavidin-agarose was carried out twice for 45 min each time. U4/U6 and U1 snRNP-depleted nuclear extracts were prepared exactly as previously described (Barabino et al., 1990).

RNAs were analysed by electrophoresis in 10% polyacrylamide (1:30, bisacrylamide:acrylamide)/8 M urea denaturing gels run in 1× TBE buffer. snRNAs were transfered to Hybond membrane (Amersham, UK) and detected using snRNA-specific riboprobes, as described by Blencowe et al. (1989). Blots and polyacrylamide gels were exposed to Kodak XAR-5 film using a single intensifying screen.

Standard (non-snRNP-depleted) nuclear splicing extracts were prepared essentially as described by Dignam et al. (1983), but using the modified conditions described by Barabino et al. (1990). Transcription of uniformly labeled, capped pre-mRNA was carried out as previously described (Lamond et al., 1987). Unlabeled pre-mRNA used for affinity selection of splicing complexes was transcribed using 5 mM ribonucleoside triphosphates (rNTPs), instead of 1 mM rNTPs used for labeled pre-mRNA. A pre-mRNA derived from the major late transcript of adenovirus (adeno premRNA) was made from plasmid pBSAd1 (Konarska and Sharp, 1987) cut with Sau3A (New England Biolabs, MA, USA). All splicing assays were incubated at 30°C for the times indicated in the figure legends. Splicing complementation assays with purified snRNP fractions were performed in 30 μl reactions containing approximately 11 μl of depleted nuclear extract and 400-600 ng of purified snRNP fraction per assay (contained in Dignam ‘D’ Buffer; Dignam et al., 1983). Equal volumes of snRNP-depleted nuclear extracts were mixed in the control complemention reaction. All other splicing reaction components were as described by Lamond et al. (1987).

12 S U4/U6 snRNPs, U1 snRNP and 20 S U5 snRNP particles were purified by mono-Q chromatography of snRNPs isolated upon an anti-2,2,7-trimethylguanosine cap immunoaffinity column (Bach et al., 1990). The U4/U6.U5 tri-snRNP complex and 17 S U2 snRNP were purified using the procedures of Behrens and Lührmann (1991) and Behrens et al. (1993), respectively.

Immunoprecipitations were carried out in IPP150 buffer (10 mM Tris-HCl, pH 7.6, 150 mM KCl, 0.1% NP40 and 0.1% azide) using 50 μl of HeLa cell nuclear extract (10 mg/ml). Antibodies were coupled to 40 μl (packed bead volume) of Protein A-Sepharose beads (Pharmacia) for 1-2 h in 150 μl of IPP150. Before coupling monoclonal antibodies, beads were coated with goat antimouse Ig antibodies (Dyanova, Germany) for 1 h. Beads were washed three times in IPP150, resuspended in 150 μl of fresh IPP150, then incubated with nuclear extract for 2-3 h. Beads were washed three times in IPP150, then boiled in 2× Laemmli loading buffer (for protein analysis). For analysis of immunoprecipitated RNAs, beads were incubated for 40 min at 65°C in 300 μl of proteinase K buffer (0.5% SDS, 100 mM NaCl, 10 mM Tris-HCl, pH 7.6) and 30 μl of proteinase K (10 mg/ml) (Boehringer, Germany). RNAs were recovered by ethanol precipitation using 3 μl glycogen (10 mg/ml) as a carrier.

Proteins were separated in 10% or 12% SDS-polyacrylamide gels using the method of Lehmeier et al. (1990). For western blot analysis the proteins were transferred onto nitrocellulose membrane (Schleicher and Schuell, Dassel, Germany) using semi-dry electro-transfer. The membranes were blocked and washed in 2% non-fat milk powder in phosphate buffered saline (PBS). Blotted membranes were incubated overnight with primary antibodies, washed, then incubated for one hour with secondary antibodies. Immunoblots shown in Figs 3 and 4 (below) were developed using an ECL chemiluminescence detection kit (Amersham), in combination with horseradish peroxidase secondary antibodies (Amersham), as per the manufacturer’s instructions. Immunoblots shown in Fig. 5 (below) were developed using a secondary antibody coupled to alkaline phosphatase (Bio-Rad, CA, USA) as described by Reuter and Lührmann (1986).

The following primary antibodies were used: anti-U1-70 kDa (Billings et al., 1982), anti-U2-B′′ (Habets et al., 1989), anti-2,2,7-trimethylguanosine cap (m₃G cap) (Bochnig et al., 1987), anti-Sm (Y12) (Lerner et al., 1981) and MaS patient serum (Okano and Medsger, 1991). Anti-p80-coilin antibodies were raised in rabbits using a β-galactosidase fusion protein, containing a C-terminal fragment of p80-coilin (Andrade et al., 1991).

The cells were grown on glass coverslips and analysed at 70-80% confluency. For drug treatment the cells were incubated in fresh medium containing 50 μg/ml α-amanitin (Sigma) for 5 h prior to fixation. The coverslips with attached cells were washed in PBS and incubated with 3.7% paraformaldehyde in 100 mM NaCl, 300 mM sucrose, 10 mM PIPES, pH 6.8, 3 mM MgCl₂, 1 mM EGTA (CSK buffer; Fey et al., 1986) for 10 min at room temperature. After fixation the cells were permeabilised with either 0.5% Triton X-100 in CSK buffer or 0.2% SDS in 20 mM Tris-HCl, pH 7.4, 30 mM NaCl and 2.5 mM EDTA for 15 min at room temperature.

Immunofluorescence was performed as previously described (Carmo-Fonseca et al., 1991b). The compact confocal microscope, developed and constructed at EMBL (Stelzer et al., 1992) was used. Excitation wavelengths of 488 nm (for fluorescein fluorescence) and 529 nm (for Texas Red fluorescence) were selected from an argon-ion laser. For double-labeling experiments each fluorochrome was independently recorded. Psuedo-coloured images of both signals were generated and superimposed. Images were photographed on Fujichrome 100 film, using a Polaroid Freeze Frame Recorder.

In agreement with the original observation of Okano and Medsger (1991), we find that the patient autoimmune serum MaS specifically immunoprecipitates U4/U6 snRNP from HeLa nuclear extracts (see Fig. 1B). In order to extend this observation we have analysed the interaction between p150, the autoantigen recognised by MaS serum, and the separate U4 and U6 components of the U4/U6 snRNP (Fig. 1A). To do this we have prepared HeLa extracts specifically depleted of either U4 snRNA, U6 snRNA or both U4 and U6 snRNAs by an Antisense Affinity Depletion (AAD) strategy based on the use of biotinylated 2′-O-alkyl oligoribonucleotides (Barabino et al., 1990; Lamm et al., 1991; see Materials and Methods). This has allowed us to address the question of whether the MaS patient serum can still immunoprecipitate U4 from a U6-depleted nuclear extract and/or vice versa.

Fig. 1A shows a northern hybridisation analysis of the snRNA composition of different depleted nuclear extracts.

Extracts depleted separately of U4 (lane 3) or U6 (lane 4) were prepared in parallel with a U4/U6 snRNP-depleted extract (lane 2) and a mock-depleted nuclear extract (lane 1), confirming that the AAD protocol can be used specifically and efficiently to deplete U4 or U6 without significantly changing the levels of non-targeted snRNA species (Fig. 1A, cf. lanes 3 and 4 with lane 1). MaS serum was used to perform immunoprecipitations in each of these depleted nuclear extracts (Fig. 1B). As a positive control, parallel immunoprecipitations were carried out from a mock-depleted nuclear extract using two distinct antisnRNP antibodies (Fig. 1B, lanes 1 and 2). These two antisnRNP antibodies are specific for either the common (Sm class) snRNP proteins (anti-Sm antibodies), or the 2,2,7-trimethylguanosine cap (m3G cap) structure present at the 5′ termini of U1, U2, U4 and U5 snRNAs (anti-m3G cap antibodies). This shows that all four spliceosomal snRNP particles, U1, U2, U4/U6 and U5 can be immunoprecipitated from a nuclear extract that has been through the AAD procedure. Lane 3 shows the snRNAs immunoprecipitated from a mock-depleted nuclear extract using the MaS serum. A lower level of U4/U6 snRNAs are immunoprecipitated compared to the level immunoprecipitated by the anti-Sm and anti-m3G cap antibodies from the same extract (Fig. 1B, cf. lanes 1 and 2 with lane 3). However, only very minor levels of U1, U2 or U5 snRNAs were co-immunoprecipitated by MaS, confirming the results of Okano and Medsger (1991) that this serum is specific for U4/U6 snRNP. The data show clearly that in a U4-depleted nuclear extract, MaS serum still immunoprecipitates U6 snRNA (lane 4), whereas no U4 snRNA is immunoprecipitated from a U6-depleted nuclear extract (lane 5). As expected, no snRNA is immunoprecipitated from a U4/U6 snRNP-depleted nuclear extract (lane 6).

Since there is an excess of U6 over U4 snRNA in HeLa nuclear extracts, there exists a pool of free U6 snRNA that is not assembled into the U4/U6 snRNP particle. This free U6 snRNA does not appear to be immunoprecipitated by the MaS serum. Note that depletion of U4/U6 snRNP (Fig. 1A, lane 2) leaves behind a low level of free U6 that is not immunoprecipitated by MaS (Fig. 1B, lane 6). A likely explanation for this is that p150 is only associated with U6 assembled into the U4/U6 snRNP, although we cannot exclude the possibility that p150 is also associated with free U6, but is not accessible to serum antibodies in this particle.

We conclude from this analysis that the autoantigen p150 is preferentially associated with U6 snRNA in the U4/U6 snRNP particle. Furthermore, we also conclude that the association of p150 with U6 snRNA does not require the presence of U4 snRNA, since the level of U6 immunoprecipitated by MaS serum in the absence of U4 is not diminished relative to the level of U6 immunoprecipitated in the mock-depleted nuclear extract (Fig. 1A, cf. lanes 3 and 4).

Although MaS serum specifically immunoprecipitates U4/U6 snRNP, it is noteworthy that it consistently immunoprecipitates a relatively low level of the total U4/U6 snRNAs present in HeLa nuclear extracts (Fig. 1B). The level of U4/U6 snRNAs immunoprecipitated is not increased by addition of more serum, or by reducing the ratio of nuclear extract:serum (data not shown). This suggests that the MaS serum may be recognising a subpopulation of the U4/U6 snRNP in the nuclear extract. HeLa U4/U6 snRNAs are known to be assembled into at least two structurally distinct populations of snRNP particle. For example, separation of HeLa cell nuclear extracts on non-denaturing gels reveals that the majority of U4/U6 snRNAs are present in a complex with U5 snRNP, corresponding to a U4/U6.U5 tri-snRNP particle (Konarska and Sharp, 1987). A lower level of ‘free’ U4/U6 snRNP complex is also observed. In order to characterise the population of U4/U6 snRNP recognised by the MaS serum, we have exploited conditions that result in the differential extraction of snRNP particles from HeLa cell nuclei.

Fig. 2 shows a northern hybridisation analysis of snRNPs immunoprecipitated from two different nuclear extracts, one of which has been prepared using 0.42 M KCl during nuclear extraction (designated 0.42 M NE); the other nuclear extract was prepared in parallel using 0.05 M KCl instead during extraction of nuclei (designated 0.05 M NE). We observe that there are both quantitative and qualitative differences in the composition of snRNP particles extracted from HeLa cell nuclei under these two different conditions (Fig. 2A and B; and see below).

The snRNAs in the 0.42 M KCl and 0.05 M KCl nuclear extracts before immunoprecipitation are shown in Fig. 2A, lanes 1 and 2, respectively. A lower level of snRNA is extracted using 0.05 M KCl compared with 0.42 M KCl. To characterise the snRNPs in the two different extracts, immunoprecipitations were carried out using monoclonal antibodies specific for m3G cap (Fig. 2A, lanes 3 and 6), Sm protein (lanes 4 and 7), and also using MaS patient serum (lanes 5 and 8). As already shown in Fig. 1B, there is a clear difference in the level of U4/U6 snRNAs immunoprecipitated from a 0.42 M NE by MaS, compared with the level immunoprecipitated by either anti-m3G cap or Sm antibodies (Fig. 2A, cf. lanes 3 and 4 with lane 5). By contrast, in the 0.05 M NE, the level of U4/U6 snRNAs immunoprecipitated by MaS and Sm antibodies is approximately the same (cf. lanes 7 and 8) and corresponds to essentially all of the U4/U6 snRNP present in the 0.05 M NE (cf. lanes 7 and 8 with lane 2). We conclude that a subpopulation of U4/U6 snRNP particles containing both Sm and MaS antigens is preferentially released from HeLa cell nuclei under 0.05 M KCl extraction conditions. Using 0.42 M KCl extraction conditions, additional U4/U6 snRNAs are released from nuclei, which can be immunoprecipitated by anti-Sm and anti-m3G cap antibodies, but not by MaS serum antibodies. Analysis of the two nuclear extracts by nondenaturing gel electrophoresis reveals that all the U4/U6 snRNAs present in the 0.05 M NE corresponds to free U4/U6 snRNP and no complex corresponding to the U4/U6.U5 tri-snRNP particle is detected (data not shown). Thus U4/U6 snRNP that is immunoprecipitated from the 0.05 M NE by MaS serum most likely corresponds to U4/U6 that is not assembled into the U4/U6.U5 tri-snRNP. This is also supported by the observation that U5 snRNA is not co-precipitated by MaS serum from either the 0.05 M or 0.42 M KCl NEs (Fig. 2A and B, lanes 5 and 8). We cannot, however, rule out the possibility that factors present in the MaS serum cause a disruption of the U4/U6.U5 trisnRNP, thereby preventing U5 snRNA co-immunoprecipitation.

It is apparent that snRNPs in the 0.05 M NE are poorly immunoprecipitated by anti-m3G cap antibodies compared with anti-Sm antibodies (Fig. 2A, cf. lanes 6 and 7). This difference is less pronounced in the 0.42 M NE (Fig. 2A, cf. lanes 6 and 7 with lanes 3 and 4). There is also a clear difference in the relative levels of individual snRNPs that are immunoprecipitated by anti-m3G cap antibodies in the 0.05 M NE extract. Upon longer exposure of the gel shown in Fig. 2A, lanes 6-8 (see Fig. 2B), low levels of U1, U2 and U5 snRNAs are observed, whereas no U4/U6 snRNAs are detected (Fig. 2B, lane 6). U4/U6 snRNP, which is not immunoprecipitated by anti-m3G cap antibodies, is, however, clearly immunoprecipitated by both anti-Sm and MaS serum antibodies (Fig. 2B, cf. lane 6 with lanes 7 and 8). U4/U6 snRNAs could be immunoprecipitated by anti-m3G cap antibodies from a 0.05 M NE in which proteins had been removed by phenol/chloroform extraction (data not shown). In particular, the majority of U4 snRNA can be immunoprecipitated by anti-m3G cap antibodies from a deproteinised HeLa nuclear extract (data not shown). This excludes the possibility that the failure of anti-m3G cap antibodies to immunoprecipitate U4/U6 snRNP is due to the presence of U4 snRNA that lacks an m3G cap structure.

In summary, we conclude that the MaS patient serum recognises a structurally distinct sub-population of U4/U6 snRNP particles in HeLa nuclear extracts, which seem to have a masked m3G cap structure. The data further show that this sub-population of U4/U6 snRNP is less tightly associated with nuclei than other complexes containing U4/U6 snRNAs and is therefore preferentially enriched by extraction of nuclei with a low-salt buffer.

Surprisingly, even though the free U4/U6 snRNP recognised by MaS serum appears not to have an accessible m3G cap structure, antibodies specific for m3G cap immunoprecipitate the p150 autoantigen from HeLa nuclear extract.

Fig. 3 shows a western blot of proteins immunoprecipitated with different snRNP-specific antibodies and probed with MaS serum. Note that p150 is detected in immunoprecipitates prepared using both anti-Sm and anti-m3G antibodies (Fig. 3A and B, lanes 2 and 3). Similar results were obtained for both the 0.05 and 0.42 M KCl NEs (data not shown). This suggests that p150 is not only associated with the m3G cap-inaccessible U4/U6 snRNP, but also with other snRNP particles that can be immunoprecipitated by the anti-m3G cap antibody. To address this question, monoclonal antibodies specific for either U1 or U2 snRNPs were used to screen for p150-snRNP interactions (Fig. 3A and B, lanes 4 and 5). As a control for specificity, a parallel northern blot analysis of immunoprecipitated snRNAs was carried out for all of the antibodies used (Fig. 3C). This confirms the specificity of immunoprecipitation for U1 (lane 4), U2 (lane 5) and U4/U6 (lane 6). The MaS immunoprecipitate shown in Fig. 3A and B (lanes 6) was included as a positive control for p150 detection. Interestingly, after a long exposure of the western blot shown in Fig. 3A, a low level of p150 is detected in the U2 snRNP immunoprecipiate (Fig. 3B, lane 5), whereas no p150 was detected in the U1 snRNP immunoprecipitate prepared in parallel under identical conditions (Fig. 3B, lane 4).

We conclude from this analysis that the p150 autoantigen can associate with snRNP particles that can be immunoprecipitated by both anti-Sm and anti-m3G cap antibodies. The data also show that a low level of p150 can be immunoprecipitated by a monoclonal antibody specific for the U2 snRNP protein B′′, suggesting that p150 may also interact, albeit weakly, with U2 snRNP.

To investigate the association between p150 and spliceosomal snRNPs in more detail, a western blot analysis was performed on proteins that are present in highly purified preparations of snRNP particles (Fig. 4). The purified snRNP preparations correspond to U1(lane 3), U2 (lane 4), U5 (lane 5), and U4/U6.U5 tri-snRNP (lane 6), in addition to a fraction of total m3G snRNPs (lane 1). The purification and characterisation of these particles has been previously documented (Bach et al., 1989, 1990; Behrens and Lührmann, 1991). The U2 snRNP fraction used in this study corresponds to the recently isolated 17 S U2 particle shown to contain at least nine additional proteins, not previously detected in association with U2 snRNA (Behrens et al., 1993). Probing the blot with MaS serum shows that p150 is detected co-purifying with total m3G, U2, U5 and U4/U6.U5 tri-snRNPs (Fig. 4, lanes 4-6). In agreement with the previous immunoprecipitation data (Fig. 3), the MaS patient serum did not detect p150, or any other proteins, in the purified U1 snRNP particles. The same blot, as shown in Fig. 4, was also probed separately with anti-Sm antibodies, which confirmed that approximately equivalent levels of each snRNP preparation were loaded (data not shown).

Since p150 is associated with multiple snRNP particles as detected by immunoblotting (Fig. 4, cf. lanes 4-6), it was important to compare its relative abundance with other snRNP proteins and to determine whether it corresponded to a previously identified snRNP protein. Therefore, the proteins present in purified U4/U6.U5 tri-snRNP complex (Fig. 5A), 17 S U2 snRNP (Fig. 5B) and the U1 and 20 S U5 snRNPs (data not shown), were separated on SDS-PAGE. Part of each sample was then transferred to nitrocellulose membrane and the resulting blots were probed with MaS serum, while for comparison the other half of each sample was stained with Coomassie blue (Fig. 5A and B, immunoblots are shown in lanes marked ‘MAS’ and Coomassie blue-stained samples in lanes 1). This shows that the p150 antigen is detected in the purified U4/U6.U5 tri-snRNP complex, the 20 S U5 snRNP and the 17 S U2 snRNP. However, it was completely absent from the purified U1 snRNP fractions. On the basis of its abundance, the p150 is clearly not one of the previously identified snRNP proteins (Bach et al., 1989; Behrens and Lührmann, 1991; Behrens et al., 1993). We conclude from this analysis that the MaS antigen p150 can associate with multiple spliceosomal snRNPs, excluding U1 snRNP, but is present in purified snRNP particles at lower stoichiometry than the major U snRNP-associated proteins. The lower stoichiometry of p150 could either reflect its association with a specific subpopulation of each snRNP species (as shown in the case of U4/U6) or, alternatively, could be caused by its having a lower stability than the major snRNP proteins and hence being under-represented through losses that occur during the purification.

Having established that p150 can associate with multiple splicing snRNPs, we were interested to characterise its nuclear distribution in relation to other components of the pre-mRNA splicing machinery. To compare the nuclear distribution of p150 with that of splicing snRNPs, HeLa cells were double-labeled with MaS serum and with an anti-Sm monoclonal antibody (Fig. 6A and B). The MaS serum (Fig. 6A) gives a widespread nucleoplasmic staining, excluding nucleoli, with several bright foci strongly labeled (Fig. 6A, arrows). The anti-Sm antibody shows a related, widespread nucleoplasmic distribution, excluding nucleoli (Fig. 6B). However, although the anti-Sm antibody stains the same foci as the MaS serum (Fig. 6, cf. foci indicated by arrows in A and B), the anti-Sm antibody also stains additional nucleoplasmic speckled structures (Fig. 6B, note speckles indicated by arrowheads). In contrast, the MaS serum is less clearly concentrated in these additional nucleoplasmic speckled structures. The staining pattern obtained with the MaS serum is similar to that of the non-snRNP splicing factor U2AF (Carmo-Fonseca, 1991a,b; Zamore and Green, 1991; Zhang et al., 1992), which has a widespread nucle-oplasmic distribution, excluding nucleoli, with elevated concentration in snRNP-containing foci, which have been shown to correspond to sub-nuclear organelles called coiled bodies (CBs) (Raska et al. 1991; Carmo-Fonseca et al., 1992; Zhang et al., 1992). To investigate whether the bright foci seen with the MaS serum are also CBs, HeLa cells were double-labeled with MaS serum and with antibodies specific for a coiled body protein, termed p80-coilin (Raska et al., 1991; Andrade et al., 1991), and images were recorded in the confocal fluorescence laser scanning microscope (Fig. 6C-E). Both antibodies reveal prominent nuclear foci (Fig. 6C and D, arrows). An overlay of the separate images shown in C and D confirms that the foci stained in each case are identical (Fig. 6E, note yellow foci resulting from co-localisation of the red and green staining).

We conclude that the p150 antigen, as detected by stain-ing with MaS serum, is present throughout the nucleoplasm, except nucleoli, and is enriched in CBs together with splicing snRNPs.

The widespread nuclear distribution of splicing snRNPs and their concentration in CBs is transcription-dependent (Carmo-Fonseca et al., 1992). We therefore investigated whether the nuclear organisation of the p150 antigen is similarly transcription-dependent by comparing the MaS staining pattern in HeLa cells treated with the transcriptional inhibitor α-amanitin (Fig. 7A and B). This shows that the staining pattern obtained with the MaS serum changes markedly after α-amanitin treatment from the widespread nucleoplasmic distribution including CBs (Fig. 7A) to a pronounced punctate pattern with all the staining confined to 20-30 clumps or speckles. A parallel staining of HeLa cells with an anti-Sm antibody confirms that a similar effect of α-amanitin is seen on the distribution of splicing snRNPs, which also become concentrated exclusively in 20-30 nucleoplasmic speckles (Fig. 7B). These punctate structures correspond to clusters of interchromatin granules, which are the same speckled domains that are preferentially labeled by the anti-Sm antibodies in transcriptionally active cells (cf. Fig. 6A and B).

An additional experiment was performed to test whether the p150 antigen remained associated with the coiled body protein, p80 coilin, after transcription was blocked by αamanitin (Fig. 7E-F). HeLa cells were treated with α-amanitin and then double-labeled with MaS serum and with antip80 coilin antibodies, and analysed in the confocal fluorescence laser scanning microscope (Fig. 7E-G). The staining patterns for both antibodies are markedly changed compared with cells that have not been treated with αamanitin (cf. Fig. 6C-E and Fig. 7E-G). An overlay of the two confocal images obtained in α-amanitin-treated HeLa cells shows that the MaS and p80 coilin staining patterns no longer co-localise (Fig. 7G). This is consistent with previous data showing a similar separation of snRNP staining from p80 coilin in α-amanitin-treated cells (Carmo-Fonseca et al., 1992).

We conclude that the nuclear distribution of the p150 antigen and its co-localisation with p80 coilin in CBs is transcription-dependent. p150 behaves like other snRNP antigens and becomes highly concentrated in 20-30 nucleoplasmic clusters when transcription is inhibited. The immunolocalisation data therefore support the biochemical evidence that p150 is a snRNP-associated protein in vivo.

As previously discussed, a large fraction of U4/U6 snRNP in HeLa nuclear extracts is complexed with U5 snRNP. The discovery that the separate U4/U6 and U5 snRNPs can associate in the absence of pre-mRNA to form a stable U4/U6.U5 tri-snRNP particle in both mammalian and yeast extracts immediately suggested that this tri-snRNP may represent a subunit of the spliceosome (Konarska and Sharp, 1987; Cheng and Abelson, 1987). Following the recent success in purifying and characterising the protein composition of mammalian U4/U6.U5 tri-snRNP particles (Behrens and Lührmann, 1991), it is important to confirm biochemically that this isolated tri-snRNP species corresponds to a genuine functional complex that is competent to participate in spliceosome assembly and pre-mRNA splicing. We have therefore assayed the ability of the purified U4/U6.U5 trisnRNP particles to functionally reconstitute spliceosome assembly and splicing in nuclear extracts that have been specifically depleted of either the U4/U6 or U5 snRNPs using the AAD procedure (Fig. 8).

The specificity of snRNP depletion (Fig. 8A), was confirmed by comparing total RNA isolated from either mockdepleted (lane 1), U5-depleted (lane 2) or U4/U6-depleted (lane 3) extracts by silver staining and by northern hybridisation (data not shown). In assays for pre-mRNA splicing (Fig. 8B), neither the U5 snRNP-depleted nor the U4/U6 snRNP-depleted nuclear extract is able to splice a premRNA derived from the major late transcript of adenovirus, whereas the same precursor is spliced in control, undepleted nuclear extract (Fig. 8B, cf. lane 1 with lanes 2 and 6). In both depleted extracts, adding back purified U4/U6.U5 trisnRNP particles is sufficient to efficiently restore splicing (lanes 5 and 9, note restored formation of spliced mRNA and excised intron). This reconstitution of splicing is specific, since addition of purified U4/U6 snRNP to U5-depleted extract, or U5 snRNP to U4/U6-depleted extract, did not rescue splicing (lanes 4 and 7). A low level of splicing was recovered by adding purified U5 snRNP particles to U5-depleted extract (lane 3) and by adding U4/U6 snRNP to U4/U6-depleted extract (lane 8). In both these cases, however, the efficiency of splicing was considerably lower than that obtained by adding back purified U4/U6.U5 tri-snRNP particles (Fig. 8B, cf. lanes 5 and 9 with lanes 3 and 8). A parallel analysis of splicing complex formation by native gel electrophoresis confirmed the previous observation that only the pre-spliceosome ‘A’ complex formed in the U5 or U4/U6-depleted extracts (Barabino et al., 1990; Lamm et al., 1991) and further demonstrated that spliceosome formation was restored to the U5 and U4/U6-depleted extracts by addition of the purified U4/U6.U5 tri-snRNP particle (data not shown).

We conclude from this analysis that the highly purified U4/U6.U5 tri-snRNP is a functional particle. The ability of the purified U4/U6.U5 tri-snRNP particles to specifically restore spliceosome formation and splicing to either U4/U6 or U5 snRNP-depleted nuclear extracts directly confirms earlier predictions that the U4/U6.U5 tri-snRNP is a functional intermediate in the spliceosome assembly pathway (Konarska and Sharp, 1987; Cheng and Abelson, 1987).

In this study we have presented new biochemical and cytological data that address the structure and function of the mammalian U4/U6 snRNP. The patient autoimmune serum MaS, which specifically immunoprecipitates mammalian U4/U6 snRNP (Okano and Medsger, 1991), is shown here to recognise a sub-population of the U4/U6 snRNPs present in HeLa nuclear splicing extract. This sub-population of U4/U6 snRNPs has its m3G cap structure masked and is preferentially extracted from HeLa nuclei by treatment with low-salt buffer. In the U4/U6 snRNP particle the p150 autoantigen that is recognised by the MaS serum is preferentially associated with U6 snRNA. While the excess, free U6 snRNA in HeLa nuclear extract is not immunoprecipitated by MaS serum, the association of p150 with U6 snRNA in the U4/U6 snRNP particle is stable and apparently independent of the presence of U4 snRNA. This was clearly demonstrated by an experiment showing that the MaS serum could still immunoprecipitate the U6 snRNA remaining in a HeLa nuclear extract after the U4/U6 snRNP was disrupted and the U4 snRNA component selectively removed by incubation with a biotinylated antisense oligonucleotide complementary to U4. In contrast, the U4 snRNA remaining after selective depletion of U6 snRNA from HeLa nuclear extract could not be immunoprecipitated by MaS serum. These data indicate that the autoantigen p150 interacts, either directly or indirectly, with mammalian U6 snRNA that has assembled into U4/U6 snRNP particles and that the antigen either is only present in a sub-population of U4/U6 snRNPs or else is differentially accessible to serum antibodies in different forms of U4/U6 snRNP particles.

Although the MaS serum only immunoprecipitates U4/U6 snRNP, we have also detected a low level of p150 associated with 20 S U5, U4/U6.U5 and 17 S U2 snRNPs, but not with U1. This is the likely explanation as to why antibodies specific for the m3G cap structure are still able to immunoprecipitate p150 when the sub-population of U4/U6 snRNP recognised by the MaS serum is not immunoprecipitated by anti-m3G cap antibodies (Fig. 2). While it is clear that p150 is present in lower stoichiometry than major snRNP proteins, both the biochemical and cytological data obtained in this study support the view that p150, at least in part, is a snRNP-associated factor. It therefore remains to be established whether the low levels of p150 recovered in purified snRNPs reflect a lower stability as compared to core snRNP proteins or instead result from p150 being associated with only discrete sub-populations of snRNPs. Whatever the explanation, the association of p150 with multiple snRNP particles indicates that it is actually not a true U4/U6-specific protein and raises a number of interesting possibilities concerning its function. For example, p150 may act as a ‘bridging factor’ to stabilise or hold together individual snRNPs within multi-snRNP complexes. This is consistent with the observation that U2 and U6 snRNAs interact within the spliceosome (Hausner et al., 1990; Wu and Manley 1991; Datta and Weiner 1991). Alternatively, p150 may be involved in transport or assembly of snRNPs. The data also leave open the question as to whether the mammalian U4/U6 snRNP particle contains any unique protein components. Previous studies have identified proteins that are specific for U4/U6 snRNP in other organisms such as yeast and trypanosomes (Shannon and Guthrie, 1991; Palfi et al., 1991). Proteins that co-purify with free U6 snRNP have also been identified, but these do not appear to assemble into the U4/U6 snRNP particle (Hamm et al., 1990; Grönig et al., 1991). Conventional purification procedures, involving biochemical fractionation and immune-affinity chromatography, have failed to detect mammalian U4/U6 snRNP-associated proteins other than Sm proteins, possibly due to their relative instability under the isolation conditions used (Bach et al., 1989; Kastner et al., 1991). However, recent purification of U4/U6 snRNP using antisense 2′-O-methyl RNA oligonucleotides has revealed several candidate-specific proteins, and also other proteins that appear (by size) to be common to U5 and/or U4/U6.U5 tri-snRNP (Blencowe, 1991). These studies, together with the data on p150 presented here, suggest that U4/U6 snRNP may contain proteins that are distinct from the common Sm proteins but are still shared with one or more other snRNP particles.

It is particularly interesting that p150 can associate, albeit at low levels, with each of the splicesomal snRNPs except for U1, even though U1 is the most abundant snRNP in HeLa nuclear extracts. Several previous studies have revealed differences between U1 and the other spliceosomal snRNP particles. For example, only U1 snRNP will bind to pre-mRNA in the absence of ATP (Black et al., 1985; Bindereif and Green, 1987; Ruby and Abelson, 1988) and under certain in vitro conditions a multi-snRNP complex containing U2, U4/U6 and U5 snRNPs has been shown to assemble in the absence of pre-mRNA (Konarska and Sharp, 1988). This complex does not contain U1 snRNP, indicating that the other snRNPs can form stable interactions with each other independently of U1. Analyses of splicing complexes by either velocity gradient centrifugation or non-denaturing gel electrophoresis have also shown that U1 snRNP is considerably less stably bound in splicing complexes than the other snRNP components (Grabowski and Sharp, 1986; Konarska and Sharp, 1987; Bindereif and Green, 1987). Differences between U1 and the other spliceosomal snRNP particles are also apparent at the level of their intracellular organisation. For example, localisation studies carried out on either fixed or liveinjected HeLa cells have revealed that U1 snRNP shows a more widespread nucleoplasmic distribution relative to the other spliceosomal snRNPs (Carmo-Fonseca et al., 1991a,b; Carmo-Fonseca et al., 1992). Similarly, U1 shows a distinct distribution pattern in Xenopus oocytes where only U1 is found in ‘A snurposomes’ while each of the snRNPs are present in ‘B snurposomes’ (Gall, 1991). The present data are thus consistent with the view emerging from these earlier studies that U1 snRNP plays a functionally distinct role from that of the other spliceosomal snRNPs.

It is also of interest that p150 has a nuclear distribution like that of the non-snRNP splicing factor U2AF (Zamore and Green 1991; Carmo-Fonseca et al., 1991b; Zhang et al., 1992). In particular, although it otherwise has a very similar distribution to Sm proteins, p150 shows less concentration in nucleoplasmic speckles than most other snRNP antigens. The speckled structures primarily correspond to clusters of interchromatin granules (Fakan et al., 1984). This could mean either that p150 is not present in the sub-population of total splicing snRNPs that concentrate in the speckles, or, alternatively, that the p150 antigen is not accessible to antibodies when snRNPs associate with these structures. However, it appears less likely that the anti-p150 antibodies simply cannot penetrate these structures, since clusters of interchromatin granules are strongly labelled by the MaS serum in HeLa cells that have been treated with α-amanitin. It has been previously proposed that coiled bodies and interchromatin granules represent different sub-nuclear structures in which distinct metabolic activities connected with the assembly/disassembly or function of splicing snRNPs take place (Carmo-Fonseca et al., 1992, 1993). The differential labelling of coiled bodies and interchromatin granules by the MaS serum is consistent with this view and supports the idea that different (though possibly overlapping) sub-populations of snRNPs may be present in each structure. A differential association of distinct sub-populations of U4/U6 snRNP with different nuclear structures could also account for the observation that extraction of HeLa nuclei with low-salt buffer preferentially removes a sub-population of U4/U6 snRNPs that have a masked m3G cap structure and are efficiently precipitated by MaS serum (Fig. 2).

The present data show that the highly purified U4/U6.U5 tri-snRNP particle efficiently restores splicing activity to HeLa nuclear extracts that have been selectively depleted of either U4/U6 or U5 snRNPs. This confirms the view that the tri-snRNP is a functional spliceosome subunit. Incorporation of U4/U6 snRNP into a tri-snRNP complex together with U5 thus appears to be an integral step in a ‘snRNP cycle’ leading to the formation of an active spliceosome. Apart from the p150 antigen studied in this paper, it has previously been shown that the U4/U6.U5 tri-snRNP contains five unique protein components of 90, 60, 27, 20 and 15.5 kDa that are not present in either the free U5 or U4/U6 snRNP particles (Behrens and Lührmann, 1991). All these five proteins are present at much higher levels than p150 in the purified U4/U6.U5 tri-snRNP particle. Since this tri-snRNP is active in splicing, these proteins may play important roles in the splicing mechanism. Our data support and extend the observations of Utans et al. (1992) that a heat-labile component of the U4/U6.U5 tri-snRNP particle is required for pre-mRNA splicing and spliceosome formation. We therefore anticipate that future studies on both these tri-snRNP specific proteins and on the autoantigen p150 may reveal important new insights into the mechanism of pre-mRNA splicing in mammalian cells.

The authors are particularly grateful to Drs Yutaka Okano and Thomas Medsger for generously providing MaS patient serum and to Gabor Lamm for providing HeLa nuclear extract efficiently depleted of U5 snRNP. We also thank Professor Eng H. Tan for anti-p80-coilin antibodies and Professor Walter van Venrooij for monoclonal antibodies against U2 B′′ and U1 70kDa proteins. We thank the EMBL photolab for help in preparing Figs 1-4, 6 and 7. M. C.-F. was supported by a European Molecular Biology Organisation (EMBO) Long-term Fellowship. R.L. gratefully acknowledges that part of this work was supported by a grant from the Deutsche Forschungsgemeinschaft (SFB 272/A3).