Advertisement

Identification of a human protein that recognizes the 3′ splice site during the second step of pre‐mRNA splicing

Shaoping Wu, Michael R. Green

Author Affiliations

  1. Shaoping Wu1 and
  2. Michael R. Green1
  1. 1 Howard Hughes Medical Institute, Program in Molecular Medicine, University of Massachusetts Medical Center, Worcester, MA, 01605, USA
View Full Text

Abstract

Accurate splicing of precursor mRNAs (pre‐mRNAs) requires recognition of the 5′ and 3′ splice sites at the intron boundaries. Interactions between several splicing factors and the 5′ splice site, which occur prior to the first step of splicing, have been well described. In contrast, recognition of the 3′ splice site, which is cleaved during the second catalytic step, is poorly understood, particularly in higher eukaryotes. Here, using site‐specific photo‐crosslinking, we find that the conserved AG dinucleotide at the 3′ splice site is contacted specifically by a 70 kDa polypeptide (p70). The p70–3′ splice site crosslink has kinetics and biochemical requirements similar to those of splicing, was detected only in the mature spliceosome and occurs subsequent to the first step. Thus, p70 has all the properties expected of a factor that functionally interacts with the 3′ splice site during the second step of splicing. Using antisera to various known splicing factors, we find that p70 corresponds to a previously reported 69 kDa protein of unknown function associated with the Sm core domain of spliceosomal small nuclear ribonucleoproteins.

Introduction

Splicing of precursor mRNAs (pre‐mRNAs) involves the assembly of a multicomponent complex, the spliceosome, which contains four small nuclear ribonucleoprotein (snRNP) particles (U1, U2, U4/U6 and U5) and many non‐snRNP proteins (Green, 1991; Guthrie, 1991; Bennett et al., 1992; Lamm and Lamond, 1993; Moore et al., 1993; Madhani and Guthrie, 1994; Neilsen, 1994; Kramer, 1996). An important function of the spliceosome is to bring into proximity the chemical groups on the pre‐mRNA that participate in the two trans‐esterification reactions that result in intron excision and exon ligation.

In higher eukaryotes, three major signals direct accurate splicing: the 5′ splice site, the branch point and the 3′ splice site. The 5′ splice site is recognized early during spliceosome assembly by intermolecular base‐pairing with U1 snRNA (Zhuang and Weiner, 1986; Seraphin et al., 1988; Siliciano and Guthrie, 1988). At later times, this base‐pairing is disrupted and the 5′ splice site is bound by U6 snRNA (Sawa and Abelson, 1992; Sawa and Shimura, 1992; Wassarman and Steitz, 1992; Kandel‐Lewis and Seraphin, 1993; Konforti et al., 1993; Lesser and Guthrie, 1993; Sontheimer and Steitz, 1993). The branch point region is also recognized by intermolecular base‐pairing, in this case by the U2 snRNA component of U2 snRNP (Parker et al., 1987; Zhuang and Weiner, 1989; Wu and Manley, 1989).

In higher eukaryotes, the 3′ end of the intron comprises a polypyrimidine (Py) tract followed by an AG dinucleotide at which cleavage occurs (for review, see Umen and Guthrie, 1995c). Several factors are known to interact with the Py tract early during spliceosome assembly, including U2AF65, IBP, PTB/hnRNP I, PSF and a 200 kDa U5 snRNP protein (reviewed in Lamm and Lamond, 1993; Moore et al., 1993; Umen and Guthrie, 1995c; Kramer, 1996). One such Py tract‐binding protein, PSF, is thought to be a second step splicing factor (Gozani et al., 1994). However, in higher eukaryotes, little is known about splicing components that interact with the highly conserved AG dinucleotide during the second step of splicing. The AG dinucleotide is essential for the second step of splicing, and 3′ splice site cleavage occurs, in general, at the first AG downstream of the branch point (Smith et al., 1989; Zhuang and Weiner, 1990). These observations imply that during the second step of splicing an as yet to be identified factor functionally interacts with the AG dinucleotide.

Here we use site‐directed photo‐crosslinking procedures to identify a 70 kDa polypeptide that interacts with the conserved AG dinucleotide at the 3′ splice site and whose properties strongly suggest a role in the second step of the splicing reaction. We go on to show that p70 corresponds to a previously identified 69 kDa of unknown function that reversibly interacts with Sm core domain of snRNPs (Hackl et al., 1994).

Results

Identification of polypeptides that interact with the AG dinucleotide at the 3′ splice site

To identify polypeptides that interact with the AG dinucleotide at the 3′ splice site, we prepared an adenovirus major late pre‐mRNA derivative (MINX) containing a single radioactive phosphate within the AG dinucleotide (A32pG) (see Figure 1A). The splicing assay of Figure 1B shows that in a HeLa cell nuclear extract this pre‐mRNA substrate was spliced with typical kinetics: the lariat–exon 2 and excised intron appeared after ∼15 and 20 min, respectively. Because the single radioactive phosphate lies within the intron, the 5′ exon and the ligated exons were not visualized.

Figure 1.

Crosslinking of polypeptides to the AG dinucleotide. (A) Experimental strategy. 32P was incorporated between the AG dinucleotide of an adenovirus major late pre‐mRNA substrate (MINX) based on the procedure of Moore and Sharp (1992). The substrate was incubated in a HeLa cell nuclear extract under standard splicing conditions. The reaction mixture was irradiated with UV light followed by treatment with RNase A. 32P‐tagged polypeptides were detected by electrophoresis on a 10% SDS–polyacrylamide gel followed by autoradiography. (B) Splicing time course and UV crosslinking assay. Splicing reactions (top) were performed for the times indicated. The lariat, lariat–exon 2 intermediate and the pre‐mRNA substrate are marked. The exon 1 and the exon 1–exon 2 ligated product were not detected because the 32P‐label was located within the intron. Aliquots were removed from the splicing reactions and subjected to the UV crosslinking assay (bottom) described in (A). Crosslinked polypeptides are marked on the right and molecular weight standards on the left.

At various times, aliquots of the splicing reaction mixture shown in Figure 1B were irradiated with UV light and treated with RNase A. 32P‐Tagged polypeptides were detected by SDS–PAGE and autoradiography (see Figure 1A). The crosslinking assay of Figure 1B shows that four groups of polypeptides were detected. A polypeptide of ∼50 kDa was evident at time zero but disappeared upon incubation at 30°C. A 35–37 kDa doublet (p35/37) was observed at time zero and remained throughout the timecourse, although its intensity was diminished at later times. In some experiments, a weak band of ∼200 kDa could be detected as early as 15 min and reached a maximal level at 20–30 min. Based upon our data and previous studies (Teigelkamp et al., 1995a, b; Umen and Guthrie, 1995a, b; Chiara et al., 1996), it is likely that this band corresponds to a 200 kDa U5 snRNP polypeptide that is the probable homolog of yeast Prp8. Most importantly, a 70 kDa polypeptide (p70) was first detected at 15 min, concomitant with the appearance of splicing intermediates and products (compare splicing and crosslinking in Figure 1B). Crosslinking of this polypeptide reached a maximal level between 20 and 45 min, and its intensity diminished at later times.

Figure 2 presents several controls to verify the authenticity of the crosslinked products and their relevance to splicing. In the absence of UV irradiation, no crosslinked polypeptides were detected (crosslinking, lane 1). As predicted from the kinetics of Figure 1B, crosslinking of p70 was ATP dependent, whereas crosslinking of p35/37 was not (compare lanes 2 and 3). To determine whether crosslinking of p70 was specific to RNAs that contained authentic splicing signals, we performed competition experiments. The homologous competitor, unlabeled MINX pre‐mRNA, competed with crosslinking of p70 to the 3′ splice site (lanes 4–6), whereas tRNA had little effect (lanes 7–9). Pre‐treatment of the nuclear extract with micrococcal nuclease, which inactivates splicing factors containing essential RNA components, abolished crosslinking of p70 but not of p35/37 (lane 11), whereas a control reaction containing both micrococcal nuclease and EGTA had no effect (lane 12). The splicing assay of Figure 2 analyzes the identical reaction mixtures shown in the crosslinking experiment.

Figure 2.

Requirements for crosslinking of polypeptides to the 3′ splice site. UV crosslinking assays were performed as described in Figure 1A. Lane 1, no UV irradiation (No UV); lane 2, in the absence of ATP and creatine phosphate (−ATP); lane 3, in the presence of ATP and creatine phosphate (+ATP); lanes 4–6, 1000‐, 500‐ and 300‐fold molar excess of unlabeled pre‐mRNA substrate (Unlabeled MINX); lanes 7–9, 3000‐, 1500‐ and 300‐fold molar excess of unlabeled tRNA (tRNA); lane 10, with a HeLa cell nuclear extract (NE) under standard splicing conditions; lane 11, in the presence of micrococcal nuclease (MN/NE); lane 12, in the presence of micrococcal nuclease and EGTA (MN/NE/EGTA). Crosslinked polypeptides are marked on the right. The splicing assay analyzes the identical reaction mixtures shown in the crosslinking experiment.

p70 is bound to pre‐mRNA only in the catalytically active spliceosome

The ATP dependence and microccocal nuclease sensitivity of the p70 crosslink suggested that interaction of p70 with the 3′ splice site would occur only under conditions that supported spliceosome assembly. The experiment shown in Figure 3A addresses this issue in greater detail. Addition of EDTA to the splicing reaction mixture allows formation of the pre‐spliceosome (complex A) and spliceosome (complex B) but blocks formation of the catalytically active spliceosome (complex C) (Abmayr et al., 1988; our unpublished data). This block can be reversed by addition of Mg2+. Figure 3A shows that 4 mM EDTA blocked crosslinking of p70, which was restored by addition of 7.5 mM Mg2+. These results indicate that p70 contacts the AG dinucleotide following spliceosome assembly, and that this binding is associated with formation of the catalytically active spliceosome (C complex).

Figure 3.

Presence of the crosslinked polypeptides in spliceosomal complexes. (A) Crosslinking of p70 is blocked by EDTA. UV crosslinking time course in the absence of MgCl2 plus 4 mM EDTA (top). UV crosslinking time course in which 7.2 mM MgCl2 was added following a 20 min pre‐incubation with 4 mM EDTA (bottom). Control, standard splicing conditions. (B) Spliceosomal complexes were separated by native gel electrophoresis, the gel slice was irradiated with UV light, treated with RNase A in situ and then placed on top of a 10% SDS–polyacrylamide gel for analysis in the second dimension. The positions of the A, B and C spliceosomal complexes are marked. The polypeptides present in the various spliceosomal complexes are indicated. Molecular weight markers are shown.

Next, we asked at what point of spliceosome assembly do these polypeptides contact the pre‐mRNA. In Figure 3B, a reaction mixture was irradiated with UV light, splicing complexes were separated by native gel electrophoresis, digested in situ with RNase A, and crosslinked polypeptides were detected by SDS–PAGE and autoradiography. The results indicate that crosslinking of p70 was observed only in the mature spliceosomal complex C. In contrast, the p35/37 crosslink was present in the pre‐spliceosome (complex A) and the B complex but was absent from complex C.

Interaction of p70 with the AG dinucleotide occurs during the second step of splicing

To delineate further the role of p70, we analyzed several previously characterized pre‐mRNA mutants. Pre‐mRNA derivatives bearing specific mutations within the GU dinucleotide at the 5′ splice site (Newman et al., 1985; Aebi et al., 1986, 1987; Seraphin and Rosbash, 1990) or the AG dinucleotide at the 3′ splice site (Reed and Maniatis, 1985) can undergo the first step of splicing, giving rise to a lariat intermediate that is blocked for the second step (reviewed in Green, 1991).

We constructed a pre‐mRNA substrate containing a GU→AU substitution at the 5′ splice site and a single radioactive phosphate within the AG dinucleotide at the 3′ splice site (Figure 4A). As expected from previous results (Aebi et al., 1986; Lamond et al., 1987), the first step of splicing was decreased and the second step was almost completely blocked (Figure 4B). The results of Figure 4C show that crosslinking of p70 was significantly impaired by the GU→AU substitution (lanes 1–5). Similar results were obtained with a GU→GA substitution mutant (data not shown). Collectively, these data suggest that binding of p70 to the 3′ splice site requires a normal RNA branch.

Figure 4.

Crosslinking of p70 to the AG dinucleotide in a 5′ splice site mutant pre‐mRNA. (A) Diagram of wild‐type (wt) and mutant (5′ ss mutant) pre‐mRNA substrates. (B) Splicing time course with the wild‐type and 5′ ss mutant substrate. (C) Crosslinking assay with the wild‐type and the mutant substrate.

To characterize further the interaction between p70 and the AG dinucleotide, we constructed a pre‐mRNA containing an AG→GG substitution and a radioactive phosphate at this position (Figure 5A). Consistent with previous results (Reed and Maniatis, 1985), splicing of this mutant pre‐mRNA proceeded normally through the first step of splicing, but was completely blocked in the second step and thus accumulated high levels of the lariat‐containing intermediate (Figure 5B). Significantly, this mutation also abolished crosslinking of p70 (Figure 5C, lanes 1–5). Taken together, these results suggest that interaction of p70 with the 3′ splice site requires both a normal RNA branch (Figure 4) and an intact AG dinucleotide (Figure 5). The fact that these mutant pre‐mRNAs can undergo the first step of splicing without binding p70 implies that the p70–3′ splice site interaction occurs after the first step.

Figure 5.

Crosslinking of p70 to the AG dinucleotide in a 3′ splice site mutant pre‐mRNA. (A) Diagram of wild‐type (wt) and mutant (3′ ss mutant) pre‐mRNA substrates. (B) Splicing time course with the wild‐type and the 3′ ss mutant substrate. (C) Crosslinking assay with the wild‐type and the mutant substrate.

To confirm this conclusion, we took advantage of the fact that a nuclear extract can be heat treated to inhibit the second step selectively (Krainer and Maniatis, 1985; Reed et al., 1988; Sawa and Shimura, 1991; Figure 6, Splicing, lane 4). Complete splicing can be restored by addition of an S100 extract, which by itself is inactive but contains the heat‐labile second step splicing factor (Sawa and Shimura, 1991; Figure 6, Splicing, lanes 2 and 3).

Figure 6.

Crosslinking of p70 in heat‐treated nuclear extract. Left: an in vitro splicing complementation assay. Splicing reaction with HeLa nuclear extract (lane 1, NE); cytoplasmic S100 (lane 2, S100); ΔNE and S100 (lane 3, ΔNE/S100); and heat‐treated nuclear extract (lane 4, ΔNE). Right: crosslinking assay. Aliquots were removed from the splicing reactions performed on the left and subjected to the UV crosslinking assay. Crosslinked polypeptides are indicated on the right.

Crosslinking assays were performed with a wild‐type pre‐mRNA substrate in either mock‐ or heat‐treated nuclear extract. Crosslinking of p70 was severely reduced in the heat‐treated nuclear extract (lane 4), and was restored by addition of S100 (lane 3). These results confirm that p70 is not crosslinked to the 3′ splice site under conditions in which only the first step of splicing occurs.

Delineating the boundaries of p70 interaction

To map more precisely the p70–3′ splice site interaction, we analyzed a series of pre‐mRNAs containing a single 32P‐label at either four or six nucleotides upstream (−4 or −6), or four or seven nucleotides downstream (+4 or +7), from the 3′ splice site (Figure 7A). Because T7 and SP6 RNA polymerase efficiently initiate transcription only with a guanosine (Milligan et al., 1987), the adenosine at −4 and cytosine at −6 were substituted with a guanosine (Figure 7A). Pre‐mRNAs carrying these single guanosine substitutions were spliced normally (see Figure 7D, lanes 9–12). Figure 7B shows that p70 was not crosslinked to RNA substrates site‐specifically labeled at +7 (lane 2), +4 (lane 4) or −6 (lane 6), but a 70 kDa polypeptide was crosslinked to the −4 position (lane 3). We note that an ∼100 kDa polypeptide was detected with the RNA substrate labeled at the −6 position, and may correspond to a 100 kDa polypeptide that crosslinked to the 3′ splice site reported by Chiara et al. (1996).

Figure 7.

Delineating the boundaries of p70 interaction. (A) Diagram of MINX pre‐mRNA. The sequence of a portion around the 3′ splice site is shown with the positions of the 32P‐label indicated. Sites are numbered using the 3′ cleavage site as position 0. The cytosine at −6 and the adenosine at −4 were substituted by guanosines to transcribe the 3′ half RNAs. (B) Polypeptides crosslinked to the substrates site‐specifically labeled at the position indicated. Crosslinking assays were performed as described in Figure 1A. p70 and molecular weight markers are marked. (C) Crosslinking assay. The polypeptide crosslinked to the −4 position had the same properties as p70 crosslinked to the AG dinucleotide (−1 position). Lane 1, polypeptides crosslinked to the −1 position; lane 2, in the absence of ATP with the substrate labeled at the −1 position; lane 3, in the absence of Mg2+ and presence of 4 mM of EDTA with the substrate labeled at the −1 position; lane 4, polypeptides crosslinked to the −4 32P‐labeled substrate; lane 5, in the absence of ATP with the substrate labeled at the −4 position; lane 6, in the absence of Mg2+ and presence of 4 mM of EDTA with the substrate labeled at the −4 position; lane 7, polypeptides crosslinked to the −1 position; lane 8, polypeptides crosslinked to the −1 position with heat‐treated nuclear extract; lane 9, polypeptides crosslinked to the −4 position; lane 10, polypeptides crosslinked to the −4 position with heat‐treated nuclear extract. (D) Splicing assay. Lane 1, with substrate labeled at the −1 position; lane 2, in the absence of ATP with the substrate labeled at the −1 position; lane 3, in the absence of Mg2+ and presence of 4 mM of EDTA with the substrate labeled at the −1 position; lane 4, with substrate labeled at the −4 position; lane 5, in the absence of ATP with the substrate labeled at the −4 position; lane 6, in the absence of Mg2+ and presence of 4 mM of EDTA with the substrate labeled at the −4 position; lane 7, with substrate labeled at the −1 position; lane 8, substrate labeled at the −1 position with heat‐treated nuclear extract; lane 9, with substrate labeled at the −4 position; lane 10, substrate labeled at the −4 position with heat‐treated nuclear extract; lane 11, with substrate labeled at the −6 position; lane 12, substrate labeled at the −6 position with heat‐treated nuclear extract. (E) Summary. The sequence of a region surrounding the 3′ splice site is shown. The AG dinucleotide is underlined. The maximum boundaries of p70 crosslinking are illustrated.

To determine whether the 70 kDa polypeptide that crosslinked to the −4 position was in fact p70, we analyzed the requirements for its interaction with the pre‐mRNA. First, the results of Figure 7C show that crosslinking of the 70 kDa polypeptide to position −4 required ATP and Mg2+ (lanes 4–6) as does crosslinking of p70 to the AG dinucleotide (lanes 1–3). Second, heat treatment of the nuclear extract abolished crosslinking of the 70 kDa polypeptide to position −4 (lanes 9 and 10) and crosslinking of p70 to the AG dinucleotide (lanes 7 and 8). Finally, immunoprecipitation analysis also indicated that the 70 kDa polypeptide that crosslinked to −4 was p70 (see below and data not shown). On the basis of these data, we conclude, first, that crosslinking of p70 to the 3′ splice site is specific and, second, that the maximal p70‐binding site extends from five nucleotides upstream (−5) to three nucleotides downstream (+3) of the 3′ cleavage site (see Figure 7E).

Identification of p70

To determine whether p70 corresponded to a known splicing factor, we tested whether 32P‐tagged p70 could be immunoprecipitated by previously characterized antisera directed against various splicing factors of size ∼70 kDa [U1 70K (Query et al., 1989), the SF3A 66 kDa subunit (Brosi et al., 1993) and a 69 kDa protein (Hackl et al., 1994)], previously identified Py tract‐binding proteins [PTB (Garcia‐Blanco et al., 1989), PSF (Patton et al., 1993) and IBP (Gerke and Steiz, 1986; Tazi et al., 1986)] or proteins otherwise implicated in the second step of splicing [PRP17 (Frank and Guthrie, 1992) and PSF]. Figure 8A shows that the 32P‐tagged p70 was immunoprecipitated by a monoclonal antibody against U1 snRNP 70 kDa protein (2.73; Billings et al., 1982) and by a polyclonal antibody directed against a 69 kDa protein reversibly associated with Sm core proteins (Hackl et al., 1994), but not by antibodies directed against PTB, PSF, U1A, the SF3A 66 kDa subunit, Sm or PRP17.

Figure 8.

Immunoprecipitation analysis. (A) Immunoprecipitation of 32P‐tagged p70 with antibodies to splicing factors. Lane 1, total input of UV‐irradiated, RNase A‐treated splicing reaction mixture (Total Input); lane 2, 2.73 monoclonal antibody against human U1 70K [U1–70K(2.73)]; lane 3, polyclonal antiserum against human PTB (PTB); lane 4, polyclonal antiserum against human PSF (PSF); lane 5, polyclonal antiserum against human U1‐A protein (U1A); lane 6, polyclonal antiserum against human 69 kDa Sm core‐associated protein (69kD); lane 7, total input of UV‐irradiated, RNase A‐treated splicing reaction mixture (Total Input); lane 8, 2.73 monoclonal antibody against human U1 70K [U1–70K(2.73)]; lane 9, monoclonal antibody against human SF3a66 (mAb66); lane 10, monoclonal antibody against human Sm epitope, Y12 (Sm); lane 11, polyclonal antiserum against yeast PRP17 protein (PRP17). (B) Immunoprecipitation of 32P‐tagged p70 with U1 70K monoclonal antibodies. Lane 1, total input of reaction mixture (Total Input); lane 2, 2.73 U1 70K monoclonal antibody [U1–70K(2.73)]; lane 3, H111 U1 70K monoclonal antibody [U1–70K(H111)].

It has been demonstrated previously that U1 70K and the 69 kDa protein are distinct (Hackl et al., 1994). We performed four experiments to determine whether p70 corresponded to the 69 kDa protein or to U1 70K. First, we carried out immunoprecipitation analysis using another antibody to U1 70K. Figure 8B shows that the U1 70K‐specific HIII monoclonal antibody (Reuter and Luhrmann, 1986) failed to immunoprecipitate 32P‐tagged p70 (lane 3), suggesting that p70 was not U1 70K.

Second, we carefully compared the electrophoretic mobility of the polypeptides. In Figure 9A a reaction mixture containing 32P‐tagged p70 was fractionated by SDS–PAGE, transferred to nitrocellulose and analyzed by immunoblotting. Comparison of the autoradiogram (lane 1) and immunoblots (lanes 2–4) indicate that 32P‐tagged p70 co‐migrated with the 69 kDa protein (compare lanes 1 and 3) and not U1 70K (compare lanes 1 and 2). In agreement with previous studies (Staknis and Reed, 1995), the 2.73 U1 70K monoclonal antibody cross‐reacted with many polypeptides, including a major polypeptide co‐migrating with U1 70K and the 69 kDa protein (lane 4).

Figure 9.

Electrophoretic mobilities of p70, 69 kDa and U1 70K. (A) One‐dimensional SDS–PAGE. The UV‐irradiated, RNase A‐treated splicing reaction mixture was separated on an 8% SDS–polyacrylamide gel, transferred to nitrocellulose and the strips of the filter were either autoradiographed (lane 1) or immunoblotted with H111 U1 70K monoclonal antibody (lane 2), with antiserum against 69 kDa (lane 3) and with 2.73 U1 70K monoclonal antibody (lane 4). (B) Two‐dimensional gel analysis. A crosslinking reaction mixture containing 32P‐tagged p70 and p35/37 was separated by two‐dimensional gel electrophoresis, transferred to a nitrocellulose filter and visualized by autoradiography (a). The identities of the crosslinked polypeptides are indicated. The same nitrocellulose filter was immunoblotted with the α‐69 kDa protein antibody (b). The position of 69 kDa protein is indicated.

Third, to rule out the unlikely possibility that 32P‐tagged p70 was not the 69 kDa protein but instead was immunoprecipitated by virtue of association with it, we carried out the experiment shown in Figure 9B. A reaction mixture containing 32P‐tagged p70 was fractionated by two‐dimensional gel electrophoresis and the polypeptides were transferred to nitrocellulose. The nitrocellulose filters were analyzed by autoradiography (i) or were probed with α‐69 kDa antibody (ii). The results show that 32P‐tagged p70 migrated almost exactly with immunoreactive 69 kDa protein. However, upon careful inspection, 32P‐tagged p70 was slightly more acidic than immunoreactive 69 kDa, presumably as a result of the negatively charged RNA tag (see, for example, Gozani et al., 1994; Staknis and Reed, 1994).

Finally, we compared the chemical digestion patterns of 32P‐tagged p70 with the 69 kDa protein (Figure 10). 32P‐Tagged p70 was treated with N‐chlorosuccinimide (NCS), which cleaves at tryptophan residues (Mirfakhrai and Weiner, 1993), fractionated by SDS–PAGE and transferred to nitrocellulose strips followed by autoradiography (lane 2). HeLa nuclear extract was fractionated by SDS–PAGE as in Figure 9A, and regions corresponding to the 69 kDa protein were excised, treated with NCS, fractionated by SDS–PAGE and transferred to nitrocellulose strips followed by immunoblotting. The NCS partial digestion of 32P‐tagged p70 generated products of ∼62 and ∼52 kDa. NCS digestion gave rise to a 62 kDa digestion product that was detected by the α‐69 kDa antibody and which corresponded precisely to the 62 kDa fragment derived from digestion of 32P‐tagged p70 (compare lanes 2 and 4). The 52 kDa NCS digestion product of 32P‐tagged p70 was not detected, most likely because it lacked the epitope recognized by the α‐69 kDa antibody. In contrast, the NCS digestion pattern of U1 70K was clearly distinct from that of 32P‐tagged p70: the 2.73 monoclonal antibody detected NCS digestion products of ∼55 and ∼40 kDa, resulting from cleavage at two tryptophans within U1 70K (see Figure 10, right). These data provide independent evidence that p70 is the 69 kDa protein.

Figure 10.

N‐chlorosuccinimide (NCS) cleavage patterns of p70, 69 kDa protein and U1 70K. Left: undigested 32P‐tagged p70 (lane 1), NCS‐digested 32P‐tagged p70 (lane 2), immunoblot of undigested 69 kDa protein (lane 3), immunoblot of NCS‐digested 69 kDa protein (lane 4). Right upper: immunoblot of undigested and NCS‐digested U1 70K by 2.73 αU1‐70K antibody. Right lower: diagram of U1 70K and NCS partial cleavage products of U1 70K. The two tryptophans and the epitope of mAb 2.73 are indicated (Spritz et al., 1987; Query et al., 1989).

Discussion

Higher eukaryotic pre‐mRNAs contain several sequence elements that are involved in splicing, including the 5′ splice site, the branch point and the Py tract/3′ splice site. These serve as recognition signals for splicing factors and participate in the two trans‐esterification reactions that occur during the first and second steps of splicing. Methods for preparing pre‐mRNAs containing site‐specific modifications provide a powerful approach to identify splicing components that contact these sequence elements (Moore and Sharp, 1992; Wyatt et al., 1992; Gaur et al., 1995; Teigelkamp et al., 1995a, b; Umen and Guthrie, 1995a, b; Chiara et al., 1996). Here we have performed UV crosslinking experiments with pre‐mRNAs containing a single radioactive phosphate between the A and G residues of the conserved AG dinucleotide. This approach has enabled us to identify p70, and several other polypeptides, that specifically contact the 3′ splice site.

Evidence that p70 is a second step splicing factor

Several lines of evidence argue strongly that the p70–3′ splice site interaction is functionally relevant: crosslinking of p70 is highly specific for the conserved AG dinucleotide (Figure 5); crosslinking of p70 and splicing have similar biochemical requirements (Figures 2 and 3A); the kinetics of p70 crosslinking parallel those of the second step of splicing (Figures 1B); both the second step of splicing and crosslinking of p70 require a normal RNA branch structure (Figure 4); p70 is bound to the 3′ splice site in the mature spliceosomal complex (Figure 3); certain cis‐acting mutations impair both the binding of p70 and the second step of splicing (Figures 4 and 5); and interaction of p70 with the 3′ splice site is severely reduced in a heat‐treated nuclear extract that can only support the first step of splicing (Figure 6). Taken together, the most straightforward interpretation of these results is that the conserved AG dinucleotide at the 3′ splice site is recognized by p70 during the second step of splicing, and that this interaction is involved in the second step. However, isolation of a p70 cDNA clone and subsequent biochemical complementation experiments will be required to confirm this interpretation.

Sequential recognition of the AG dinucleotide

In several instances, pre‐mRNA splicing signals appear to be recognized many times during a single splicing reaction. For example, the 5′ splice site is first bound by U1 snRNA, and subsequently by U6 snRNA (reviewed in Moore et al., 1993; Madhani and Guthrie, 1994; Neilsen, 1994). The interpretation of previous analyses of pre‐mRNA mutants was that the conserved AG dinucleotide is recognized more than once, and may also participate in the first step of splicing (Reed, 1989; Zhuang and Weiner, 1990). A number of factors, in addition to p70, have been reported to bind at or near the 3′ splice site (reviewed in Umen and Guthrie, 1995c). Although the precise role of these factors is not yet clear, one or more of them may be involved in sequential 3′ splice site recognition. Such factors include: a human 35/37 kDa protein that we have described (see Figure 1); a 100 kDa human polypeptide reported by Chiara et al. (1996); a 200 kDa U5 snRNP polypeptide that is the probable homolog of yeast Prp8 (Chiara et al., 1996); U6 snRNA (Lesser and Guthrie, 1993); and several yeast proteins including Prp8, Prp16 and Slu7 (Teigelkamp et al., 1995a, b; Umen and Guthrie, 1995a, b). Finally, at some point during the splicing process, there is probably an interaction between the AG dinucleotide and the 5′ splice site (Parker and Siliciano, 1993).

We have shown that under several different circumstances in which splicing failed to progress to the second step, p70 was not crosslinked to the AG dinucleotide. Thus, during spliceosome assembly, crosslinking of p70 to the AG dinucleotide must be regulated temporally. There are several possible mechanisms by which the p70–3′ splice site interaction may be controlled. For example, at early times, the AG dinucleotide may be occupied by another factor, such as p35/37, which blocks binding of p70. Alternatively, interaction of p70 with the AG dinucleotide could be facilitated by conformational changes or by other spliceosomal components that enter the complex at later times. For example, p70 binds poorly to an abnormal RNA intermediate that contains an A(2′–5′)A branch and is blocked in the second step (Figure 4). Thus, an additional factor could selectively bind to a normal RNA branch and promote binding of p70 to the AG dinucleotide. Finally, p70 is a phosphoprotein (data not shown) and its phosphorylation state could regulate the 3′ splice site interaction.

In the transition from the first to second step, the spliceosome undergoes a major conformational change (Sawa and Shimura, 1991; Schwer and Guthrie, 1992). Our results suggest that this conformational change may involve binding of p70 to the 3′ splice site. The splicing components that recognize the splice site and comprise the catalytic site are primarily snRNAs (reviewed in Lesser and Guthrie, 1993; Madhani and Guthrie, 1994; Nielsen, 1994), but this does not preclude participation of protein factors. Further experiments will be required to determine the role of p70 in splice site recognition and catalysis.

p70 is the 69 kDa protein

We have provided several lines of evidence that p70 corresponds to a previously reported 69 kDa protein of unknown function. First, p70 was specifically immunoprecipitated by an α‐69 kDa antibody (Figure 8). Second, the equivalence of p70 and the 69 kDa protein is supported by their electrophoretic mobilities in both one‐ and two‐dimensional protein gels (Figure 9). Third, the chemical digestion patterns of p70 and the 69 kDa protein support the conclusion that the two proteins are the same (Figure 10).

Conversely, our results allow us to conclude that several other proteins implicated in 3′ splice site recognition and splicing are not p70. For example, unlike p70, IBP and PTB bind to the Py tract, not the AG dinucleotide, in a manner that does not require an AG dinucleotide or ATP (Gerke and Steitz, 1986; Tazi et al., 1986; Ruskin et al., 1988; Garcia‐Blanco et al., 1989; Singh et al., 1995). The Py tract‐binding protein PSF is a second step splicing factor, but our immunoprecipitation experiments indicate that p70 is not PSF (Figure 8). Moreover, binding of PSF to the Py tract does not require the AG dinucleotide (Patton et al., 1993; Gozani et al., 1994), again distinguishing PSF from p70. In similar UV crosslinking assays, Chiara et al. (1996) detected binding of 75 and 100 kDa polypeptides to the 3′ splice site. The properties of the 75 kDa polypeptide were similar to those of our p70, suggesting that the two proteins may be the same. Although we did not detect a 100 kDa polypeptide binding to the AG dinucleotide, a polypeptide of this size was crosslinked to a pre‐mRNA labeled at the −6 position (Figure 7B, lane 6).

The 69 kDa protein was identified originally in 12S U1 snRNP (Hackl et al., 1994), but was also found to associate reversibly with the Sm core domains present in all spliceosomal snRNPs (Hackl et al., 1994). The function of the 69 kDa protein was completely obscure and in fact was speculated to involve assembly and/or nucleocytoplasmic transport of snRNP particles. Here, we show that this protein specifically recognizes the 3′ splice site during the second step of splicing. These results provide a rationale for isolating a p70 cDNA clone to study the role of this protein in 3′ splice site recognition and splicing.

Materials and methods

Pre‐mRNA substrates

Plasmid SpP65‐AdML (MINX) derived from the adenovirus major late promoter (Zillmann et al., 1988) was used to transcribe pre‐mRNA substrates in vitro. To prepare a pre‐mRNA substrate containing a radioactive phosphate between the A and G residues of the AG dinucleotide, two DNA templates were synthesized by PCR from the MINX plasmid: one of the DNA templates (5′ DNA template) was designed to encompass the 5′ exon and the entire intron except for the last G residue, and the other DNA template (3′ DNA template) to encode the last residue of the intron and the 3′ exon. The 5′ primer for making the 5′ DNA template was 5′TAATACGACTCACTATAGAATTCGAGCTCGC3′, and the 3′ primer was 5′GGATGACCGCGAGCTGTGGAAAAAAAAG3′. The resulting PCR product contained a T7 RNA polymerase promoter, the 5′ exon, the intron and a FokI restriction enzyme site at the 3′ end. Digestion of the PCR product with FokI created the 5′ DNA template containing a T7 RNA polymerase promoter, the 5′ exon and the intron except for the last G residue. The 5′ primer for making the 3′ DNA template was 5′GGAATTCTAATACGACTCACTATAGCTCGCCCTTGAGGACAAA3′, and the 3′ primer was 5′CTTGGGCTGCAGGTCGACTCTAGAGGATCCCCACTGG3′. The resulting PCR product contained a T7 RNA polymerase promoter, the last nucleotide of the intron (the G residue) and the 3′ exon, and was used as the 3′ DNA template.

To prepare a pre‐mRNA substrate containing a GU→AU mutation at the 5′ splice site, the 5′ primer for the 5′ DNA template was 5′CGATTTAGGTGACACTATAGGAATTCTTGGATCGGAAACCCG TCGGCCTCCGAACGATAAGAGCTAGC3′, a GT→AT change was introduced in the primer; the 3′ primer was the same as that for the wild‐type substrate.

To prepare a substrate containing an AG→GG mutation at the 3′ splice site, the 3′ primer for the 5′ DNA template was 5′CGCGGATCCGGATGACCGCGAGCCGTGGAAAAAAAAGGG3′, and the 5′ primer was the same as that for the wild‐type substrate. The FokI‐digested PCR product contained a T7 RNA polymerase promoter, the 5′ exon and the intron with the second last A residue changed to G and not containing the last guanosine residue, and was used as the 5′ DNA template.

Pre‐mRNA substrates carrying the 32P‐label at the position several nucleotides downstream or upstream of the 3′ splice site were constructed similarly to the substrate side‐specifically labeled at the AG dinucleotide described above. Sequences of oligonucleotides for synthesis of DNA templates of these substrates by PCR are available upon request.

RNA was synthesized in vitro by T7 or SP6 RNA polymerase at 37°C for 2 h. Transcription reactions for the 5′ RNA fragment contained 1× transcription buffer (Promega Inc.), 10 mM dithiothreitol (DTT), 0.05 mM GTP, 0.5 mM of each nucleotide except GTP, 0.5 mM G(5′)ppp(5′)G (Promega Inc.), 0.1 mg/ml of the DNA template and 10 U/ml SP6 or T7 RNA polymerase. The 3′ RNA fragment was synthesized in a reaction mixture containing 1× transcription buffer (Promega, Inc.), 10 mM DTT, 0.5 mM of each of the four NTPs and 0.1 mg/ml of the DNA template. The 3′ RNA fragment was treated with alkaline phosphatase (1 U/ml) (Boehringer Mannheim Biochemicals) in 1× phosphatase buffer for 30 min at 37°C, followed by addition of 2 μl of 500 mM EGTA and incubated at 62°C for 20 min. The 3′ RNA fragment was then 5′‐end‐labeled with [γ‐32P]ATP and polynucleotide kinase. The two RNAs were annealed to a DNA bridging oligonucleotide complementary to the 20 nucleotides at the 3′ end of the 5′ RNA fragment and the 20 nucleotides at the 5′ end of the 3′ RNA fragment and ligated using T4 DNA ligase as described (Moore and Sharp, 1992; Wyatt et al., 1992). The ligation product was purified following electrophoresis on an 8% denaturing polyacrylamide gel.

UV crosslinking assay

Splicing reactions (12.5 μl) were performed with ∼7000–70 000 c.p.m. (sp. act. ∼2×106 to ∼4×106 c.p.m./μg) of pre‐mRNA substrate containing single radioactive phosphate at the 3′ splice site in 50% HeLa nuclear extract for various times at 30°C (Zamore and Green, 1991). For −ATP reactions, ATP, creatine phosphate and creatine phosphate kinase were omitted, and the nuclear extract was pre‐incubated at 30°C for 20 min. For micrococcal nuclease‐treated reactions, micrococcal nuclease was added to nuclear extract with 0.02 volumes of 0.05 M CaCl2 and 0.02 volumes of micrococcal nuclease (5000 U/ml), and the splicing reaction was incubated for 30 min at 30°C. For the micrococcal nuclease/EGTA control reaction, 0.02 volumes of 0.1 M EGTA/K+ (pH 8.0) was added together with micrococcal nuclease. For +EDTA reactions, MgCl2 was replaced by 4 mM EDTA. To restore the splicing activity in the +EDTA reaction, 7.5 mM of MgCl2 was added and the reaction mixtures were incubated for the times indicated. To analyze splicing products, proteinase K was added to the reaction mixture and RNA fractionated on a 13% sequencing gel.

To analyze proteins that interact with the 3′ splice site, the reaction mixtures were irradiated with UV light (wavelength 254 nm) in a Stratalinker (Stratagene) at 800 000 μJ/cm2 (∼5 min) on ice at a distance of 10 cm from the source. RNase A was added to 2 mg/ml and the reaction mixtures were incubated for 30 min at 37°C. SDS gel loading buffer was added and the reaction mixtures were boiled for 5 min and fractionated on 10% SDS–polyacrylamide gels. The gel was dried and the 32P‐tagged polypeptides were detected by autoradiography.

Native gel analysis

After incubation at 30°C for various times, the splicing reaction mixtures were loaded on a 4% polyacrylamide (80:1 acrylamide/bis‐acrylamide)/0.5% agarose composite gel containing 50 mM glycine/50 mM Tris base. Electrophoresis was carried out at 250 V for 5 h at 4°C. 32P‐labeled RNP complexes were detected by autoradiography. A gel slice containing the complexes from a 30 min splicing reaction was then excised, UV irradiated on ice and incubated with 2 mg/ml RNase A at 37°C for 20 min, followed by incubation in 4× SDS gel loading buffer at 37°C for 20 min. The gel slice was embedded in the stacking portion of an SDS–polyacrylamide gel and subjected to electrophoresis. Crosslinked polypeptides were visualized by autoradiography.

Splicing complementation in heat‐treated nuclear extract

The heat‐treated nuclear extract (ΔNE) was prepared by warming at 43°C for 20 min (Krainer and Maniatis, 1985). Splicing reactions were performed with 50% nuclear extract (NE), 50% heat‐treated nuclear extract (ΔNE) or 50% cytoplasmic S100 extract (S100) at 30°C for 30 min. In the splicing complementation assay with ΔNE and S100, 25% ΔNE and 25% S100 were used and incubated at 30°C for 30 min.

Immunoprecipitation and immunoblotting analysis

Splicing and UV crosslinking assays were performed as described above in a reaction mixture containing 30% nuclear extract. The reaction mixtures were treated with RNase A (2 mg/ml) for 30 min at 37°C. Antibody (12 μl) was added to 12 μl aliquots of the reaction mixtures followed by incubation at 4°C for 1 h. Protein A–Sephorose 4B (fast flow, Sigma) beads were then added and incubated for another 1 h at 4°C. The beads were washed four times with 1 ml of NET‐2 (150 mM NaCl, 0.05% NP‐40, and 50 mM Tris–HCl, pH 7.5). SDS gel loading buffer was added and the reaction mixtures were boiled for 10 min, fractionated by SDS–PAGE, and 32P‐labeled polypeptides detected by autoradiography.

The following immunological reagents were used: the monoclonal antibody against U1 70K (2.73, described by Billings et al., 1982); the monoclonal antibody against U1 70K (H111, provided by W.‐Y.Tarn and J.A.Steitz, Yale University, described by Reuter and Luhrmann, 1986); the polyclonal antiserum against human U1‐A protein (pAB856, provided by I.W.Mattaj, EMBL, Heidelberg, Germany); the polyclonal antiserum against 69 kDa Sm core‐associated protein (provided by W.Hackl, Philipps Universitat Marburg, Hackl et al., 1994); the polyclonal antiserum against PSF and PTB (provided by J.G.Patton, Vanderbilt University, Patton et al., 1993); the monoclonal antibody against SF3a66 (mAB66, provided by A.Kramer, Universite de Geneve, described by Bennett et al., 1992; Brosi et al., 1993); the monoclonal antibody against Sm epitope (Y12, provided by J.A.Steitz, Yale University, described by Gerke and Steitz, 1986); and the polyclonal antiserum against yeast PRP17 protein (provided by D.Frank and C.Guthrie, University of California, San Francisco, described by Frank and Guthrie, 1992).

For immunoblotting analysis, the UV‐irradiated, RNase A‐treated splicing reaction mixture was fractionated in SDS–PAGE or by two‐dimensional gel electrophoresis [Bio‐Lyte 3/10 Ampholyte (Bio‐Rad was used for the isoelectrofocusing in the first dimension)], transferred to nitrocellulose filter (Millipore) and subjected to autoradiography. The same membrane was then probed with the primary antibody at room temperature for 1 h. Secondary antibody was conjugated to horseradish peroxidase and detection was with the ECL system (Amersham).

N‐chlorosuccinimide cleavage

The crosslink mixture was separated by SDS–PAGE. The gel slides containing 32P‐tagged p70 and proteins of size ∼70 kDa were treated with NCS as described (Mirfakhrai and Weiner, 1993). Briefly, the labeled crosslinked p70 band was identified by autoradiography, excised and equilibrated with H2O, followed by a solution containing 6.6 M urea and 40% acetic acid. After chemical cleavage in the same buffer with 15 mM NCS, the gel slides were washed with H2O, equilibrated in a stacking buffer (10% glycerol, 15% β‐mercaptoethanol, 3% SDS, 0.0625 M Tris–HCl, pH 6.8) and loaded into 12 or 15% SDS–polyacrylamide gels. Proteins in the gels were transferred to nitrocellulose filter (Millipore), and subjected to autoradiography or immunoblotting.

Acknowledgements

We are particularly grateful to J.Valcarcel for advice during the course of the work and on the manuscript. We thank R.Singh, M.L.Zapp, J.Umen, C.Guthrie, and M.Rosbash for critical comments; J.L.Patton, I.W.Mattaj, A.Kramer, W.Hackl, W.‐Y.Tarn, J.A.Steitz, D.Frank and C.Guthrie for antibodies; T.O'Toole for secretarial assistance; and L.Chiang for technical assistance. This work was supported by a postdoctoral fellowship from the American Cancer Society to S.W. and by a grant from the National Institutes of Health to M.R.G. M.R.G. is an investigator of the Howard Hughes Medical Institute.

References

View Abstract