The Human U5-220kD Protein (hPrp8) Forms a Stable RNA-Free Complex with Several U5-Specific Proteins, Including an RNA Unwindase, a Homologue of Ribosomal Elongation Factor EF-2, and a Novel WD-40 Protein

Abstract

The human small nuclear ribonucleoprotein (snRNP) U5 is biochemically the most complex of the snRNP particles, containing not only the Sm core proteins but also 10 particle-specific proteins. Several of these proteins have sequence motifs which suggest that they participate in conformational changes of RNA and protein. Together, the specific proteins comprise 85% of the mass of the U5 snRNP particle. Therefore, protein-protein interactions should be highly important for both the architecture and the function of this particle. We investigated protein-protein interactions using both native and recombinant U5-specific proteins. Native U5 proteins were obtained by dissociation of U5 snRNP particles with the chaotropic salt sodium thiocyanate. A stable, RNA-free complex containing the 116-kDa EF-2 homologue (116kD), the 200kD RNA unwindase, the 220kD protein, which is the orthologue of the yeast Prp8p protein, and the U5-40kD protein was detected by sedimentation analysis of the dissociated proteins. By cDNA cloning, we show that the 40kD protein is a novel WD-40 repeat protein and is thus likely to mediate regulated protein-protein interactions. Additional biochemical analyses demonstrated that the 220kD protein binds simultaneously to the 40- and the 116kD proteins and probably also to the 200kD protein. Since the 220kD protein is also known to contact both the pre-mRNA and the U5 snRNA, it is in a position to relay the functional state of the spliceosome to the other proteins in the complex and thus modulate their activity.

Nuclear pre-mRNA splicing is a dynamic process in which intervening sequences (introns) are excised from pre-mRNAs in a two-step mechanism that is catalyzed by the spliceosome (for review, see references 15 and 28). The spliceosome assembles on the pre-mRNA by the stepwise integration of the small nuclear ribonucleoproteins (snRNP), U1, U2, U4/U6, and U5, and an as-yet-undefined number of non-snRNP proteins. In the early phase of spliceosome formation, U1 snRNA base pairs with the 5′ splice site, and U2 snRNA interacts with the branch site to form the prespliceosomal E and A complexes, respectively. Spliceosomal assembly is completed by addition of the 25S U4/U6.U5 tri-snRNP complex, which is formed under splicing conditions from the U4/U6 and U5 snRNP particles (reviewed in reference 28). After assembly, the spliceosome undergoes several RNA and protein conformational rearrangements. For example, the U4/U6 duplex, which is present in the tri-snRNP, is unwound, and the U6 snRNA interacts instead with the 5′ end of the U2 snRNA (7, 26, 47, 58) and with intron sequences at the 5′ splice site (14, 21, 40, 41, 44, 56). The rearranged U2/U6 snRNA network is thought to be involved in the catalytic step of splicing (25), and U5 snRNA is involved in aligning the two exons for ligation (32). After the splicing reaction, the spliceosome is dissolved, and the tri-snRNP is generally believed to be assembled on the U5 snRNP again. The U5 snRNP therefore plays an important role in both spliceosome assembly and splicing.

The protein composition of the U5 snRNP complex has been most thoroughly investigated for HeLa cells; this complex contains at least 17 proteins, many of which have been characterized by cDNA cloning. These proteins can be classified into two groups. Like all other snRNPs of the spliceosome, U5 snRNP contains a set of the so-called Sm proteins: B/B′, D1, D2, D3, E, F, and G. These proteins bind to a conserved site on the snRNA, thus forming the “core complex.” It has been shown that the Sm proteins play an essential part in the biogenesis of the snRNPs (33) and that they also have an important task in the stabilization of the snRNPs and their integration into the spliceosome. In addition to the Sm proteins, the U5 snRNPs contain a number of particle-specific proteins with molecular masses of 15, 40, 52, 100, 102, 110, 116, 200, and 220 kDa. In accordance with the conservation of the splicing mechanism, the yeast Saccharomyces cerevisiae possesses a very similar set of proteins (10): not only are the Sm proteins all strongly conserved (42), but clear homology could also be established for many of the U5-specific proteins.

Interestingly, many of the U5-specific proteins are essential for splicing and have important tasks in the dynamics of the spliceosome (reviewed in references 45 and 57). Some examples of these are as follows. (i) The 200-kDa protein (200kD) and its yeast homologue, Snu246p (18)—also termed Brr2p, Slt22p, or Rss1p (22, 34, 60)—contain both DEXH-box RNA helicase motifs, and the isolated human 200kD protein possesses RNA unwindase activity (16). (ii) The 100-kDa protein of U5 (U5-100kD) also possesses an RNA helicase domain, in this case of the DEAD box type (50). Its yeast orthologue, Prp28p, is needed for the maturation of the spliceosome before the first step of the splicing reaction; however, the target of its action is not known (46). (iii) The U5-116kD protein probably catalyzes rearrangements during the spliceosomal cycle. Both this protein and its orthologue in yeast, Snu114p, show extensive homology with the translational elongation factor EF-2, a GTPase that is needed for the translocation of mRNA on the ribosome (11). The human U5-116kD protein has been shown to bind GTP in vitro, and a point mutation in the GTP-binding motif of Snu114p yields a lethal phenotype. (iv) The 220kD protein and its yeast orthologue, Prp8p, have also been associated with rearrangements on the spliceosome. In both systems, this protein has been shown to contact pre-mRNA. Interestingly, the 220kD protein (or Prp8p) contacts the pre-mRNA at the 5′ splicing site before the first catalytic step of splicing (38, 52, 59) and then moves over and contacts the 3′ splicing site after the first step (6, 51, 53). Furthermore, this protein has been shown to be close to the branching point (24, 52). Thus, the 220kD protein (or Prp8p) is in contact with all three elements of the pre-mRNA involved in splicing, and it clearly plays a key role in the spliceosome. This central role is underlined by the strikingly high degree of conservation of this protein (12, 18, 23).

As the 220kD protein binds to different parts of the pre-mRNA, there arises the interesting possibility that it plays a central role in the communication between the pre-mRNA and the enzymes in the U5 snRNP—for example, by triggering their activity. This would require these proteins to interact with one another, directly or indirectly. However, nothing is known at present about the interactions between the U5-specific proteins. In the work described here, we studied protein-protein interactions in the U5 snRNP. In one approach, U5-specific proteins were dissociated from the U5 core particle by using the chaotropic salt sodium thiocyanate, allowing the investigation of snRNA-free subcomplexes of these proteins. As described below, we have been able to show that the 220kD protein, the 200kD RNA unwindase, the 116kD G protein, and the U5-40kD protein (here shown to be a new member of the WD-40 repeat family) together make up an snRNA-free complex that comprises more than one-half of the total mass of the U5 snRNPs. The core of this complex is the 220kD protein, which binds directly to the 40- and 116kD proteins and probably also to the 200kD RNA unwindase.

MATERIALS AND METHODS

Preparation of 20S U5 snRNPs from HeLa nuclear extracts.

Nuclear extracts were prepared from HeLa cells (Computer Cell Culture Company, Mons, Belgium) according to the method of Dignam et al. (8). Total snRNPs containing U5 and U4/U6 snRNPs were purified from nuclear extract by affinity chromatography at 420 mM NaCl with the monoclonal antibody H20 as described by Laggerbauer et al. (17). U5 snRNPs and U4/U6.U5 tri-snRNPs were obtained by the same procedure at 250 mM NaCl. 20S U5 snRNPs were obtained by subsequent fractionation of the snRNP particles on a 10-to-30% (wt/wt) glycerol gradient containing 150 mM KCl according to Laggerbauer et al. (17).

Fractionation of U5 snRNP components in thiocyanate-containing glycerol gradients.

In a typical experiment, 750 μg of purified U5 snRNP was concentrated by centrifugation in a Beckman TLA 100.3 rotor at 70,000 rpm for 17 h at 4°C. The supernatant was discarded, and the U5 snRNP-containing pellet was resuspended in 800 μl of buffer R400 (20 mM HEPES-KOH, [pH 7.9], 400 mM sodium thiocyanate [NaSCN], 1.5 mM MgCl₂). For fractionation, the U5 snRNPs were layered onto a linear 12-ml, 5-to-20% (wt/wt) glycerol gradient prepared with buffer R400. The gradients were centrifuged in a Beckman SW40 Ti rotor at 33,000 rpm for 21 h at 4°C, and 30 420-μl fractions were harvested manually from top to bottom. Proteins were recovered from 50-μl aliquots by acetone precipitation, separated by sodium dodecyl sulfate (SDS)–13% polyacrylamide gel electrophoresis (PAGE) with a high concentration of TEMED (N,N,N′,N′-tetramethylethylenediamine) (20), and visualized by Coomassie blue staining. RNA was recovered from 30-μl aliquots by phenol extraction and ethanol precipitation, separated by urea–10% PAGE, and visualized by silver staining. Alternatively, the U5 snRNPs in 800 μl of R400 buffer were layered onto a 10-to-30% (wt/wt) glycerol gradient in buffer R200 (20 mM HEPES-KOH [pH 7.9], 200 mM NaSCN, 1.5 mM MgCl₂) and centrifuged in an SW40 Ti rotor at 30,500 rpm for 18.5 h at 4°C. The gradient was harvested as before. In reassociation experiments, fractions from glycerol gradients in buffer R400 were mixed and dialyzed against one change of 200 volumes of R200 buffer without glycerol. Four hundred microliters of the dialysate was layered onto a 4.2-ml, 5-to-20% (wt/wt) glycerol gradient prepared in buffer R200 and centrifuged in a Beckman SW60 Ti rotor for 12 h at 40,000 rpm at 4°C. Twenty-eight 140-μl fractions were collected from the top.

Microsequencing and database search.

snRNP proteins were extracted from 20 mg of affinity-purified snRNPs with 1 volume of phenol, and proteins were precipitated from the phenolic phase with 5 volumes of acetone. The precipitated proteins were separated by preparative SDS-PAGE. Coomassie-stained bands of individual snRNP proteins were excised from the acrylamide gel, concentrated by funnel-well SDS-PAGE, and electroblotted onto a sequencing-grade polyvinylidene difluoride membrane (Bio-Rad). Partial amino acid sequences of tryptic peptides of the snRNP proteins were obtained by microsequencing on an ABI 477A protein sequencer. Sequence analysis was performed at the microchemical facility of Harvard University, Cambridge, Mass.

The peptide sequences obtained from the U5-40kD protein were KGPELPLVPVKR, QRHELLLGAGSGPGAGQQQATPGALLQAGPPR, and LWDIR.

Database searches carried out on the National Institutes of Health mail server by using the TBLASTN and FASTA programs (2) identified an expressed sequence tag (EST) with an open reading frame (ORF) matching all three partial peptide sequences of the U5-40kD protein (GenBank accession no. R20187). This EST was used for further studies.

Expression of His-tagged 40kD protein and immunization of rabbits.

To clone the 40kD cDNA into the pET 28b vector (Novagen), NcoI and XhoI restriction sites were added via PCR amplification to the N and C termini, respectively, of the 40kD EST. The PCR fragment was cloned into the pET expression vector, thus encoding the full-length 40kD protein fused to a C-terminal His tag (six histidines). Correct orientation and the presence of an ORF were verified by DNA sequencing. The fusion protein was expressed in Escherichia coli BL21 cells for 3 h at 30°C after induction with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). The protein was found to accumulate in inclusion bodies, from which it was extracted and purified with Ni-agarose under denaturing conditions essentially as described by the supplier (Novagen). This fusion protein was used for the immunization of rabbits.

For Western blot analysis, proteins were separated by SDS-PAGE, electroblotted onto nitrocellulose, and immunostained with alkaline phosphatase-conjugated secondary antibodies as described previously (20).

Translation and radioimmunoprecipitation in vitro.

For translation in vitro, the StuI-NotI fragment containing the entire ORF of the 40kD protein was subcloned into the pBluescript SK(−) vector. Coupled transcription and translation reactions in vitro were performed by using the Promega TNT (T7) system, applying procedures recommended by the manufacturer.

Radioimmunoprecipitations with antibodies against the 40-, 116- (11), 200- (18), and 220kD (18) proteins were performed essentially as described by Raker et al. (36). In short, the radiolabelled protein(s) was preincubated with the respective proteins or particles and subsequently added to 10 μl of preblocked antibody–protein A-Sepharose beads in 400 μl of IPP₁₅₀ (20 mM Tris-HCl [pH 8.0], 150 mM NaCl, 0.05% Nonidet P-40) and incubated for 1 to 12 h at 4°C with constant rotation. Subsequently, the reactions were washed five times with IPP₁₅₀ (the reaction tube was changed once) and then dried. Precipitated proteins were resuspended in SDS sample buffer and separated on SDS–12% polyacrylamide gels. The gels were dried and autoradiographed with Kodak XAR film with an exposure period of 1 to 5 days.

Far-Western analysis.

Protein fractions were separated on an SDS–9% polyacrylamide gel and transferred to nitrocellulose. Proteins were denatured and subsequently renatured by washing (for 15 min at 4°C) the blot in NET₁₅₀ (50 mM Tris-HCl [pH 7.9], 150 mM NaCl, 0.1% Triton X-100) containing 6 M guanidine-HCl, then in NET₁₅₀ containing 2 M guanidine-HCl, and finally in NET₁₅₀ containing 0.66 M guanidine-HCl. The nitrocellulose was then blocked for 3 h with NET₁₅₀ containing 5% fat-free dried milk, washed with NET₁₅₀, and incubated overnight at 4°C with 10⁶ cpm of [³⁵S]Met-labelled protein prepared by in vitro translation (40kD or 116kD) in NET₁₅₀. The blot was washed three times for 15 min with NET₁₅₀ and autoradiographed with X-ray film.

Nucleotide sequence accession number.

The nucleotide and amino acid sequences for the 40kD protein have been deposited in the GenBank database under accession no. AF090988.

RESULTS

Dissociation of 20S U5 snRNPs with NaSCN.

To study possible protein-protein interactions, we first attempted to dissociate purified U5 snRNPs by treatment with various salts, with the aim of isolating stable protein heteromers. Interpretation of the results obtained by this approach relies on the assumption that strong protein-protein interactions, once established in RNA-free assembly intermediates, may persist in the intact particle and may also survive RNP dissociation under mild conditions. This method has been successfully employed to elucidate protein-protein interactions in the ribosome (9) and the Sm core (36) and to characterize the complex of the 20kD, 60kD, and 90kD proteins in the U4/U6 snRNP (49). With the U5 snRNP, this approach is especially useful because the high molecular weight of several of the U5-specific proteins renders recombinant expression of full-length proteins difficult (our unpublished results). Treatment of U5 snRNPs with a high concentration of salt, such as KCl, causes only a few proteins, including the 100kD putative RNA helicase, to dissociate almost quantitatively (16). The remaining, salt-resistant U5 snRNP particle contains, in addition to the Sm core proteins, the U5-specific 40kD, 102kD, 116kD, 200kD, and 220kD proteins, which are therefore good candidates for studying stable protein-protein interactions.

As an alternative, we used chaotropic ions, which are known to disrupt hydrophobic interactions (27). Because we were looking especially for interactions between U5-specific proteins, we first aimed at finding conditions that allow the U5-specific proteins to dissociate while leaving the Sm core particle intact. This was achieved by treatment with NaSCN at a concentration of 0.4 M. Figure 1 shows the results for U5 particles treated with 0.4 M NaSCN and then separated by centrifugation on a glycerol gradient (for conditions, see Materials and Methods). The protein analysis of the gradient (Fig. 1) shows that all Sm core proteins peak in precisely the same fractions (7 to 9), which also coincides with the peak of U5 snRNA (Fig. 1, lower panel). We therefore conclude that the U5 core particle containing Sm proteins and snRNA stays intact under these conditions. This conclusion is further supported by our finding that monoclonal anti-Sm antibodies (Y12) coprecipitate U5 snRNA from fraction 8 (not shown). Minor amounts of other U5-specific proteins are also found in fraction 8, but they are not likely to be stably associated with either U5 snRNA or the U5 core particle. Rather, the presence of these proteins results from a partial overlap of the peaks of proteins in the 100-kDa range, the Sm core and the 200kD protein, which agrees with their molecular weights (see below). Furthermore, homogeneous U5 core particles can be separated from the cosedimenting proteins by anion-exchange chromatography at NaSCN concentrations considerably lower than 0.4 M (16). We conclude that 0.4 M NaSCN causes all U5-specific proteins to dissociate from the intact U5 core particle.

FIG. 1.

Dissociation of U5 snRNP particles by using sodium thiocyanate. U5 snRNPs were disrupted in 0.4 M NaSCN and separated on a 5-to-20% glycerol gradient containing the same buffer as described in Materials and Methods. Proteins were recovered from aliquots of each fraction by acetone precipitation, separated on an SDS–13% polyacrylamide gel and visualized by staining with Coomassie blue. The top 24 fractions are shown (no protein was seen in the bottom 6 fractions). The positions of the U5 proteins are indicated on the right, and those of molecular mass markers are indicated on the left. The positions to which ovalbumin (3.5S), albumin (4.6S), and immunoglobin G antibodies (7.0S) sedimented on an identical gradient are shown at the top. The bottom panel shows the RNA that was recovered by phenol extraction and ethanol precipitation, separated by urea–10% PAGE, and visualized by silver staining.

With 0.4 M NaSCN in the gradient, most of the U5-specific proteins sediment as monomers, resulting in a good correlation between the sedimentation rate and the molecular weights of the individual proteins. For example, the smaller, 52kD protein remains at the top of the gradient (fractions 1 to 3), and the proteins in the 100-kDa range are found in fractions 4 to 7, well separated from the 200kD protein in fractions 8 to 11. The conclusion that these proteins sediment as monomers is further supported by the finding that marker proteins of comparable size, which were run on a parallel gradient, show similar sedimentation behavior: ovalbumin (45 kDa, 3.5S) was found to peak in fraction 4, bovine serum albumin (66 kDa, 4.6S) peaked in fractions 5 and 6, and immunoglobulin G antibodies (150 kDa, 7S) peaked in fractions 8 and 9. Most interestingly, both the 116kD and 220kD proteins sedimented significantly faster than other proteins of comparable size and peaked in precisely the same fractions (Fig. 1, fractions 15 to 17), indicating that the two proteins form a stable complex. No significant amounts of U5 snRNA were found to cosediment with this complex, and the only other U5 component observed in these fractions is the 40kD protein. While part of this protein clearly peaks in the same fractions as the 116kD and 220kD proteins, the bulk fraction of the 40kD protein streaks over the top third of the gradient (fractions 2 to 8). This behavior can be most easily explained by assuming that at 0.4 M NaSCN the 40kD protein associates semistably with the dimer consisting of the 116kD and 220kD proteins (the 116/220 dimer) (see also Fig. 2).

FIG. 2.

Dissociated U5 snRNP particles fractionated at reduced sodium thiocyanate concentration. U5 snRNPs were disrupted in 0.4 M NaSCN as in Fig. 1 and then layered onto a 10-to-30% glycerol gradient containing 0.2 M NaSCN in the same buffer. Proteins (top panel) and RNA (bottom panel) were recovered and visualized as described for Fig. 1. The positions of the U5 proteins are indicated on the right, and those of molecular mass markers are shown on the left.

The 220kD protein forms an RNA-free, heteromeric complex with the 40kD, 116kD, and 200kD proteins.

Figure 1 shows that the 116kD protein directly and stably associates with the 220kD protein and provides initial evidence that the 40kD protein associates with this dimer. To investigate whether further U5-specific proteins also interact directly with the 220kD protein, we lowered the stringency of the dissociation conditions under which the U5 components were separated. The proteins and complexes obtained by resuspending U5 RNP particles in 0.4 M NaSCN were separated on a glycerol gradient containing only 0.2 M NaSCN, so that protein-protein interactions that are stable under these conditions could be reestablished upon entry into the glycerol gradient. As shown in Fig. 2, under these conditions the bulk of the 40kD protein cosedimented with the 220kD and 116kD proteins (fractions 13 to 17), while a minor part of the 40kD protein stayed on top (fractions 1 to 3), as expected for a monomeric protein of this size. This result substantiates the hypothesis, posited above, that the 40kD protein forms a specific complex with the 116kD and 220kD proteins and that this complex is semistable in 0.4 M NaSCN.

Interestingly, under these conditions the 200kD protein also comigrates with the 116/220 complex. As seen with the 40kD protein, a minor part of the 200kD population sediments as expected for the monomeric protein (fractions 6 to 8), whereas the major part cosediments with the 40kD, 116kD, and 220kD proteins (fractions 12 to 17), thus providing strong evidence that the 200kD protein is part of the complex at the lower NaSCN concentration. The U5 snRNP particles used in this experiment were purified from a pool of snRNPs prepared at 250 rather than 420 mM KCl and therefore contain a loosely associated protein of 65 kDa (17a); this protein migrates as a monomer (fractions 1 to 4). Further, the U5 preparation contained a minor contamination with U1 snRNP. At least the U1-70kD and U1-A proteins stay associated under these conditions with the U1 core particle and can be seen as weak bands in fractions 7 to 9, where U1 snRNA is also observed (lower panel).

While the comigration of the four proteins strongly hints at an association of the 200kD protein with the 40/116/220 complex, the possibility that the 200kD protein fortuitously comigrates with the other proteins, e.g., by forming homooligomers, could not be excluded. This possibility was ruled out by the following experiment. Fractions enriched in the 200kD protein were taken from a gradient run at 0.4 M NaSCN and pooled. Part of this pool was dialyzed to 0.2 M NaSCN in the same buffer and fractionated on a 5-to-20% glycerol gradient in 0.2 M NaSCN buffer (Fig. 3, middle panel). As a reference, a pool of the 116/220 dimer with some 40kD protein was dialyzed and fractionated as before (upper panel). As Fig. 3 (upper and middle panels) shows, the 40/116/220 complex sediments significantly faster than the 200kD protein alone. Only when aliquots of both pools were combined during dialysis and then run on the same gradient was an additional complex formed that contained the 40-, 116-, 200-, and 220kD proteins and sedimented faster than both the free 200kD protein and the 40/116/220 complex (Fig. 3, lower panel). This shows that the 200kD protein associates with the other three proteins at the lower NaSCN concentrations. When this result is compared with that in Fig. 2, it is seen that significantly less of the 200kD protein associates with the complex. Possibly, a part of the 200kD protein could have been denatured by the prolonged exposure to 0.4 M NaSCN. Notably, the 40kD protein, which is underrepresented in the pool used here, is not enriched in fractions 22 and 23, which contain the tetrameric complex (Fig. 3, lower panel). From this, we conclude that the 40kD protein is not necessary for the association of the 200kD protein with the 116/220 protein heteromer. This is corroborated by another finding, from an experiment analogous to the one shown in Fig. 3, that the monomeric 40kD protein present in the top fractions of a 0.4 M NaSCN gradient (fraction 3) (Fig. 1) does not cosediment with the 200kD protein (data not shown).

FIG. 3.

The U5-200kD RNA helicase associates at 0.2 M sodium thiocyanate with the 40/116/200 complex. Fractions enriched in the 40kD, 116kD, and 220kD proteins (similar to fractions 16 and 17 in Fig. 1) or enriched in the 200kD protein (corresponding to fractions 9 to 11) were pooled from a gradient similar to that shown in Fig. 1, dialyzed separately (top and middle panels) or together (bottom panel) against buffer containing 0.2 M NaSCN, and then run on separate glycerol gradients prepared with the dialysis buffer as described in Materials and Methods. Proteins from 40 μl of 140 μl of each fraction, or from 20 μl of the sample loaded onto the gradients (first lane), were separated by SDS–10% PAGE and visualized by staining with Coomassie blue. The positions of the relevant proteins are indicated on the left.

The U5-40kD protein is a novel WD-40 repeat protein.

The experiments discussed above demonstrate that the 40-, 116-, and 200kD proteins form a stable, snRNA-free complex with the 220kD protein. While the 116kD and 200kD proteins have been characterized by cDNA cloning and are implicated in the dynamics of the spliceosome (see the introduction), the 40kD protein has not yet been characterized at a molecular level. To obtain further details of the protein-protein interactions in this interesting heteromer, it was first necessary to characterize the 40kD protein by cDNA cloning.

For this purpose, the 40kD protein present in purified HeLa snRNPs was microsequenced. Three partial peptide sequences were obtained and used in database searches, which led to the identification of several EST entries representing overlapping cDNAs. The longest EST (GenBank accession no. R20187, derived from a human fetal brain library) was obtained and sequenced. It is 1,507 bp long, and the largest ORF encodes a protein with 357 amino acids and a predicted molecular mass of 39.3 kDa, which is in good agreement with the apparent size of the protein. All three peptide sequences obtained from microsequencing are present in the deduced amino acid sequence (underlined in Fig. 4A).

FIG. 4.

The U5-40kD protein is a novel WD-40 repeat protein. (A) Predicted amino acid sequence of the 40kD protein. Peptide sequences obtained by microsequencing are underlined; amino acids highly conserved in WD-40 repeats are shown in bold. (B) Alignment of the seven WD-40 repeats. Amino acids that agree with the consensus at the five highly conserved positions are boxed in black, and those meeting the requirements in the less conserved positions are shaded in grey. The bottom lines list all amino acids that may occur in a given position of the consensus sequence. Lowercase letters indicate groups of amino acids: h, ACMFWYVIL; t, DGNP; s, GSTACY. The brackets indicate how many nonconserved residues can be found in that region. These data are adapted from reference 31.

To confirm the identity of the cDNA, we used biochemical and immunological methods. First, ³⁵S-labelled 40kD protein prepared by translation in vitro comigrates with the native protein present in purified snRNPs, confirming that the EST encodes the full-length protein (data not shown, but see Fig. 7). Next, antisera were produced by immunizing rabbits with a His-tagged fusion protein expressed in E. coli. This immune serum, but not the preimmune serum derived from the same rabbit, strongly and specifically reacts on Western blots with the 40kD protein present in affinity-purified snRNP particles and in crude nuclear extract (Fig. 5, compare lanes 3 and 5 and lanes 4 and 6, respectively). These results provide strong evidence that the product of the identified EST and the 40kD protein from HeLa U5 are indeed identical proteins.

FIG. 7.

³⁵S-labelled 40kD protein prepared by translation in vitro associates specifically with the 116/220 dimer. The translated 40kD protein was incubated in the presence of U5 snRNP particles (lanes 1 to 4) or with fractions enriched in proteins in the 1000-kDa range (lanes 5 to 8), the 200kD protein (lanes 9 to 12), or the 40kD, 116kD, and 220kD proteins (lanes 13 to 16). The protein composition of these fractions is shown in Fig. 8, left panel. Proteins precipitated with protein A-Sepharose-bound antibodies specific for the 100kD, 116kD, 200kD, or 220kD protein, as indicated above each lane, were separated by SDS-PAGE, and the 40kD protein was visualized by fluorography.

FIG. 5.

Characterization of the cDNA-encoded 40kD protein. Antibodies raised against the recombinant 40kD protein recognize the native U5 snRNP protein. Proteins present in 20 μl of nuclear extract (lanes 3 and 5) or in 5 μg of purified snRNP (lanes 4 and 6) were separated by SDS-PAGE and blotted onto nitrocellulose. The membrane was immunostained with antibodies directed against the 40kD protein (lanes 5 and 6) or preimmune serum derived from the same rabbit (lanes 3 and 4). Lanes 1 and 2 show marker proteins and snRNP proteins, respectively, which were separated on the same gel and visualized by Coomassie blue staining. Molecular mass markers are shown on the left, and the position of the 40kD protein is shown on the right.

Database searches showed that the 40kD protein is a novel protein. Close inspection of the sequence revealed that this protein contains seven WD-40 repeats which were first identified in the β subunits of the heterotrimeric G proteins. Figure 4B shows an alignment of the seven repeats with the consensus sequence as reported by Neer et al. (31). All highly conserved residues are present, except for a GH-to-QN substitution in the fourth repeat and a WD-to-GE substitution in the last repeat. This is consistent with the observation by Neer et al. (31) that WD-40 proteins, defined by the presence of at least one highly conserved repeat, frequently contain other repeats with up to three mismatches with respect to the consensus. Thus, the entire 40kD protein, except for just 63 residues at the amino terminus, consists of WD-40 repeat structures.

Consistent with their domain structure, many other WD-40 proteins show significant homology (20 to 25% identity) with the 40kD protein. The matches are, however, mostly limited to the residues defining the WD-40 repeats, and these proteins are not likely to be functional homologues. In contrast, three ORFs in the database display much clearer homology with the 40kD protein: a 38-kDa protein from Arabidopsis thaliana (GenBank accession no. AC002333) displays 55% amino acid identity with the 40kD sequence. Two ORFs which were found in Caenorhabditis elegans cosmid clones Z66561 and AF000265, respectively, also exhibited extensive homology with the 40kD protein. In these entries, the assignment of intron-exon boundaries had to be corrected by comparison of the sequences with the EST database. The corrected ORFs (Fig. 6) encode proteins of 38.7 and 38.8 kDa, respectively. As shown in Fig. 6, all three homologues contain a short N-terminal extension of lesser homology to the 40kD protein followed by the seven highly homologous WD-40 domains, sharing even the WD-to-GE substitution in the last repeat. Which of the two C. elegans proteins is the true functional homologue, or whether both proteins are associated with different U5 populations in the nematode, is presently not clear. Additional fragments from highly homologous protein sequences from rat, mouse, Drosophila melanogaster, and rice were identified in the EST data bank (details not shown). Intensive homology searches in the yeast database revealed the presence of several putative WD-40 repeat proteins which display 20 to 25% sequence identity with the 40kD protein. Based only on the sequence alignments, it is difficult to decide which of these proteins, if any, is the functional homologue of the 40kD protein. Biochemical characterization of the proteins present in U4/U6/U5 tri-snRNPs purified from yeast should help to answer this question.

FIG. 6.

Alignment of the U5-40kD protein with highly homologous WD proteins. The U5-40kD protein was aligned with a hypothetical 38kD protein from A. thaliana (accession no. AC002333), and hypothetical proteins of 38.8 and 38.7 kDa from C. elegans (accession no. AF000265 and Z66561, respectively). The position of the WD domains is indicated by the numbered double arrows. Amino acids that are identical in at least three sequences are boxed in black, and those that are conserved in at least three sequences are shaded grey. Conserved amino acids are grouped as follows: DE, HKR, CILMVFYW, ST, and AG.

The 220kD protein interacts directly with the 40kD protein and presumably also with the 200kD protein.

Our sedimentation studies demonstrate that the 40kD, 116kD, and 200kD proteins form a specific, snRNA-free complex with the 220kD protein. However, only the 116kD protein has been shown so far to interact directly with the 220kD protein, because this is the only protein stably and stoichiometrically associated with it at 0.4 M NaSCN. Since the 220kD protein plays a key role in the spliceosome, we were interested to see whether the other two proteins also interact directly with the 220kD protein. For this purpose, we first tested which protein 40kD interacts with. Fractions enriched in the 100kD and 200kD proteins as well as those containing the 116/220 complex were taken from an NaSCN gradient similar to the one shown in Fig. 1 and used to coprecipitate ³⁵S-labelled 40kD protein prepared by translation in vitro. In the presence of the U5-specific 100kD protein (Fig. 7, lanes 5 to 8) or the 200kD protein (lanes 9 to 12), none of the sera directed against the 100kD, 116kD, 200kD, or 220kD protein coprecipitated the 40kD protein, whereas all four sera coprecipitated the 40kD protein when it was incorporated into U5 snRNPs (lanes 1 to 4). This indicates that the 40kD protein interacts neither with the 200kD protein (at least not directly) nor with the 100kD protein. In contrast, radioactively labelled 40kD protein coprecipitated in significant amounts with the 116/220 dimer (Fig. 7, lanes 14 and 16). This interaction is specific, since no precipitation was observed in the absence of the 116/220 complex (lanes 6, 8, 10, and 12).

While these data demonstrate that the 40kD protein specifically interacts with the 116/220 complex, they do not distinguish which of the two proteins the 40kD protein interacts with. Since we were unable to separate the native 116- and 220kD proteins, we employed far-Western analysis (overlay blots) to address this question. Total U5 snRNP proteins or partially purified proteins in the 100-kDa range, as well as partially purified 200kD protein and the 40/116/220 protein complex (the same fractions as used in Fig. 7), were fractionated by SDS-PAGE, blotted onto nitrocellulose, and probed with ³⁵S-labelled 40kD protein prepared by translation in vitro (Fig. 8, middle panel). A second, identical blot was probed with the 116kD protein (right panel). Both proteins were found to bind to the 220kD protein on the blot (lanes 5, 8, 9, and 12). The signal can be assigned by comparison with the protein pattern after Coomassie staining (left panel) and because it is observed only in those lanes that contain the 220kD protein, not in those containing partially purified 200kD protein. No interaction was observed between the 40kD and 116kD proteins, irrespective of which of the proteins was used as the probe (middle and right panels). We note, however, that the 116kD protein additionally binds to one of the proteins in the 100-kDa range (lane 10); the exact identity of this protein is currently not known.

FIG. 8.

The 40kD and 116kD proteins both interact with the 220kD protein on far-Western blots. U5 snRNP proteins and gradient fractions containing proteins in the 100-kDa range, the 200kD protein, or the 40kD, 116kD, and 220kD proteins were separated on three identical SDS–9% polyacrylamide gels. The proteins were either visualized by staining with Coomassie blue (left panel) or blotted onto nitrocellulose and probed with the 40kD protein prepared by translation in vitro (middle panel) or the 116kD protein (right panel) as described in Materials and Methods. The autoradiograms of the membranes are shown.

In summary, we have shown by radioimmunoprecipitation that the 40kD protein specifically binds to the 116/220 dimer, and on far-Western blots it interacts solely with the 220kD protein, not with the 116kD protein. The 200kD protein is not recognized in the far-Western blots by the 40kD or the 116kD protein, and radioimmunoprecipitation yields the same result (Fig. 7, lane 11, and data not shown). The 220kD protein is thus the direct binding partner for both the 40kD and 116kD proteins. Direct interactions between the 220kD and 200kD proteins are also very likely. The 220kD protein is therefore the centerpiece in the architecture of the U5 snRNP, interacting directly with three proteins to give an snRNA-free protein complex that accounts for more than half of the total mass of the particle.

DISCUSSION

We have shown here that four of the U5-specific proteins, together comprising more than one-half of the mass of the U5 snRNP particle, form a complex which is stable in the absence of snRNA or the Sm proteins. Two of these proteins, the U5- 116kD G protein and the U5-200kD RNA unwindase possess domain structures which suggest that they participate in rearrangement steps during spliceosome assembly or the splicing process itself. The third protein in the complex is the 220kD human orthologue of the yeast protein Prp8p, which binds on both cutting sites of the pre-mRNA and thus plays a central part in the functioning of the spliceosome. Finally, the protein complex contains the U5-40kD protein, which we here show to be a novel WD-40 repeat protein. WD-40 repeat proteins are frequently involved in the regulated association and dissociation of protein complexes; it can therefore be assumed that the 40kD protein also plays a part in the assembly of the spliceosome.

As an initial approach to investigating protein-protein interactions in the U5 snRNP particle, we dissociated purified U5 snRNP particles in the presence of 0.4 M NaSCN. Under these conditions, the U5-specific proteins are completely released while the snRNP core particle is left intact. The resulting pool of U5-specific proteins yields a reconstituted complex of the 40kD, 116kD, 200kD, and 220kD proteins when the NaSCN concentration is reduced to 0.2 M (Fig. 2). Even though all other U5 components were present during this step, no interaction between these four proteins and the other U5-specific proteins or the core complex was observed. This shows that the other proteins do not compete for the interactions that keep the tetrameric complex intact, and it can therefore be inferred that these four proteins also interact with one another in the intact U5 snRNP. Further evidence for this is found in the fact that none of these four proteins can be made to dissociate from the U5 snRNP by treatment with high concentrations of potassium chloride whereas other U5-specific proteins do dissociate under these conditions (16). Finally, our reconstitution method allows the preparation of the complex in pure form (Fig. 2 and 3) so that the participation of other proteins can be ruled out.

The detailed investigation showed that the 220kD protein must be centrally placed in the complex because it interacts with at least two of the three other proteins. Its interaction with the 116kD protein is strong enough that the 116/220 dimer remains completely intact even in the presence of 0.4 M NaSCN and can be isolated on a glycerol density gradient (Fig. 1). The 40kD protein is found by radioimmunoprecipitation to associate specifically with the 116/220 dimer (Fig. 7), and far-Western blots show that it interacts directly with the 220kD protein (Fig. 8). Finally, the 200kD protein also interacts with the 220/116 complex (Fig. 3); an interaction with the isolated 40kD or 116kD protein could not be shown, either by radioimmunoprecipitation (Fig. 7 and unpublished data) or by far-Western blotting. We conclude, therefore, that the 200kD RNA unwindase also binds directly to the 220kD protein, although the possibility remains that a composite binding site for the 200kD protein is created only upon formation of the 116/220 protein complex. Clearly, further experiments are required to provide us with a more comprehensive map of all interactions occurring among the four proteins in the 40/116/200/220 heteromer. In any case, the large number of specific interactions of the 220kD protein with other proteins of the U5 snRNP and also with other components of the spliceosome (see below) could also explain why this protein is so highly conserved among all eucaryotic organisms (12, 18, 23).

Apart from its obvious importance for the structural organization of the U5 snRNP particle, the four-protein complex may also play a major part in the biogenesis of the U5 snRNP, by first assembling on its own and then becoming integrated into the nascent particle. The remarkably strong association among the four proteins raises the possibility that at least some of them may already associate with one another in the cytoplasm, directly after synthesis, and are then cotransported into the nucleus. From this idea, it would be predicted that the proteins of the complex (i.e., half of the total snRNP mass) cannot be integrated into the snRNP particle in the absence of the 220kD protein. In support of this, it has been observed that genetic depletion of the Prp8p protein in yeast results in a dramatic decrease in mass of the U5 snRNP particle (shown by sedimentation) (5).

The 220kD protein not only interacts with the components of the U5 snRNP; like its yeast homologue Prp8p, it also contacts both splicing sites of the pre-mRNA (38, 52, 59) and thus has not only a central location but also probably a central role in the spliceosome. This appears to reflect the dynamics of the spliceosome: before the first step of splicing it is in contact with the 5′ splicing site, and after this step it contacts the 3′ splice site (6, 51, 53). Furthermore, the PRP8 gene in yeast shows genetic interaction with a group of splicing factors that interact with the spliceosome only after the first step of the splicing reaction, which may indicate the existence of further, splicing-step-dependent contacts involving the Prp8p protein (54). At the same time, the 220kD (or Prp8p) protein interacts strongly, as we have shown here, with the 200kD and 116kD proteins, which are probably involved in the dynamic processes of the spliceosome. The former contains, as does its homologue in yeast, two DEXH RNA-helicase motifs (18, 22, 34, 60), and mutations in the yeast gene show that at least one of these helicase domains is essential. Moreover, it was shown that the highly purified 200kD protein is able to unwind the duplex of U4 and U6 RNAs (16). The 116kD protein is a putative GTPase with strong homology to the elongation factor EF-2, which catalyzes the rearrangement of the ribosome during the translocation step (11). Both of these proteins belong to families that generally require other proteins as coactivators—for example, the GTPase-activating proteins and the guanyl exchange factors in the case of the GTPases, or the protein eIF-4B in the case of the best-studied RNA unwindase, eIF-4A (1, 37, 39). There is thus the interesting possibility that the 220kD protein may be an activating factor for the 200kD RNA unwindase and/or for the 116kD EF-2 homologue. Considering that the 220kD protein also contacts in a sequential manner the 5′ and 3′ splice sites, its potential activation of the 200kD RNA unwindase or the 116kD protein might be triggered by certain functional states of the spliceosome. It will be interesting to see which, if any, of the protein-protein contacts described in this paper change in the course of the splicing reaction and how this affects the activity of the RNA unwindase and the GTPase.

In this connection, the domain structure of the 40kD protein also present in the heteromeric complex raises some interesting questions. As we have shown here, the 40kD protein is a new member of the family of WD-40 repeat proteins. The principal function of the WD-40 proteins is the mediation of reversible protein-protein interactions. For example, the β subunits of the heterotrimeric G proteins bind to the α subunits to which GTP is bound and dissociate after hydrolysis of the GTP. In fact, there is also a G protein in the U5 snRNP, the 116kD protein; however, it is not very likely that these two proteins form a functional pair, both because no interaction between them has been detected and because the 116kD protein does not belong to the same subfamily of GTPases as the G-α proteins do. WD-40 proteins are not only found in connection with the heterotrimeric GTPases; they are also present among the transcription factors (reviewed in reference 31). Furthermore, these proteins have also recently been identified as common constituents of large RNA-protein complexes—for example, in the spliceosome (4, 6a, 13, 19, 35, 43, 55, 61) (also this work), in the polyadenylation complex (48), in the initiation complex of translation (3, 30), and in telomerase (29). Common to all these complexes is the fact that they are continually built up de novo on an appropriate substrate and then dismantled in a later reaction step. It is possible that the WD-40 proteins in the multifactoral RNP complexes, as in the trimeric G proteins, may contribute to the assembly of the relevant complex. Since these interactions are reversible, the components can be reused after the dissolution of the complex. A similar scenario may be envisioned for the 40kD WD-40 protein as part of the U4/U6.U5 tri-snRNP. During the splicing of pre-mRNA, the U4/U6.U5 tri-snRNP proceeds through a cycle in which it enters into various transient interactions: first of all, the U5 and U4/U6 snRNPs bind to one another; then the tri-snRNP thus formed becomes part of the prespliceosome; then the U4 and U6 RNA duplexes are unwound, and the U4 RNA is released along with an as-yet-undefined protein complement. After the splicing reaction, the spliceosome dissociates, and it is generally believed that the tri-snRNP must then be reassembled out of its components. Interestingly, the human U4/U6.U5 tri-snRNP contains, in addition to the 40kD protein, a second WD-40 repeat protein, the U4/U6 60kD protein (Prp4p in yeast) (4, 13, 19, 35, 55). Thus, the U4/U6.U5 tri-snRNP contains at least two proteins capable of participating in changing protein-protein contacts. It will be of great interest to see which of the steps described involve the U5-40kD protein and which other proteins it interacts with during these steps.

It is clear that the protein-protein interactions described here reveal the existence of a large number of functional interactions between proteins in the spliceosome that may have a decisive influence upon the timing of the assembly of the spliceosome and the rearrangements that occur during its maturation.