Alterations in the Polypyrimidine Sequence Affect the in Vitro Splicing Reactions Catalyzed by HeLa Cell-free Preparations*

.-The polypyrimidine tract, located at the 3’ end of intron 1 of the adenovirus major late transcript, was studied for its role in splicing using cell-free prepara-tions isolated from HeLa cells. A plasmid (pIz) was constructed in which seven purine bases were substituted for pyrimidine bases within the 14-nucleotide polypyrimidine sequence. Runoff transcripts extend-ing to the middle of intron 2 were tested for their ability to support in vitro splicing. The efficiency of these reactions was compared with pre-mRNA tran- scripts made from the wild-type nonmutated plasmid (pl-2). Neither spliced products nor splicing intermediates were detected in reactions with the pIz pre- mRNA. The formation of the nucleoprotein complexes involved in splicing was examined with this altered pre-mRNA. No 55 S splicing complex was detected and only low levels of the 30 S presplicing complex formed (30-fold less than with wild-type pre-mRNA). How-ever, when a longer runoff transcript was prepared from the polypyrimidine mutated plasmid pIz, spliced RNA was formed. This activity required specific downstream sequences, since transcripts produced from pIz which contained substituted downstream sequences were not spliced. Although intron 2 of the adenovirus major late transcript does not contain a discernible 3‘ polypyrimidine sequence, pre-mRNA (p2-3) containing this intron was efficiently spliced. However, dried autora-diography Formation and Analysis of Splicing Complexes-The partially pu- rified fractions used for generating nucleoprotein complexes have been described previously (20). 3ZP-Labeled transcripts were incu- bated under the same conditions as described above for the standard splicing reaction, except that appropriate fractions (10 pl of fraction I1 (3-4 mg/ml) and fraction I1 plus 5 pl of fraction Ib (4 mg/ml)), isolated from nuclear extract of HeLa cells, were substituted for unfractionated nuclear extracts. Reactions were incubated at 30 “C for 2 h and then loaded directly onto 10-30% sucrose gradients, prepared, and centrifuged as described previously (20). Nucleoprotein complexes were analyzed neutral polyacrylamide-agarose com- posite gel electrophoresis (3.5% polyacry1amide:bisacrylamide 8O:l and 0.5% agarose using 112 Tris-borate-EDTA running buffer).

The polypyrimidine tract, located at the 3' end of intron 1 of the adenovirus major late transcript, was studied for its role in splicing using cell-free preparations isolated from HeLa cells. A plasmid (pIz) was constructed in which seven purine bases were substituted for pyrimidine bases within the 14-nucleotide polypyrimidine sequence. Runoff transcripts extending to the middle of intron 2 were tested for their ability to support in vitro splicing. The efficiency of these reactions was compared with pre-mRNA transcripts made from the wild-type nonmutated plasmid (pl-2). Neither spliced products nor splicing intermediates were detected in reactions with the pIz pre-mRNA. The formation of the nucleoprotein complexes involved in splicing was examined with this altered pre-mRNA. No 55 S splicing complex was detected and only low levels of the 30 S presplicing complex formed (30-fold less than with wild-type pre-mRNA). However, when a longer runoff transcript was prepared from the polypyrimidine mutated plasmid pIz, spliced RNA was formed. This activity required specific downstream sequences, since transcripts produced from pIz which contained substituted downstream sequences were not spliced.
Although intron 2 of the adenovirus major late transcript does not contain a discernible 3' polypyrimidine sequence, pre-mRNA (p2-3) containing this intron was efficiently spliced. However, when the 3' region of intron 2 was substituted for the polypyrimidine sequence of intron 1, the resulting pre-mRNA did not support efficient splicing in vitro. However, when the polypyrimidine sequence of intron 1 was substituted for the sequence at the 3' end of intron 2, efficient splicing occurred, and the rate of formation of splicing intermediates and the accumulation of nucleoprotein complexes was greater than with the wild-type pre-mRNA (p2-3).
The consensus sequence of the 5' and 3' splice sites of pre-mRNAs has been determined from a large number of known intron sequences (1-3). These compilations identified the 5' consensus sequence (AC)AG/GU(AG)AGU, whereas the 3' consensus sequence contains a run of pyrimidmes followed by a nonconserved sequence, a pyrimidine, and then the invariant AG dinucleotide at the intron-exon border ((U/C),N(C/ U)AG/G).
Alterations of the 3' conserved sequence and their effects on splicing have been studied both i n vivo and in vitro. In vivo studies have shown that deletion of the 3' AG dinucleotide or reduction of the size of the polypyrimidine sequence reduced the synthesis of spliced RNA (4,5) or resulted in the utilization of cryptic 3' splice sites (6). I n vitro studies have shown that although deletion or mutation of the AG dinucleotide partially reduced the efficiency of the first step in splicing (formation of the 5' exon and the intron-exon lariat), removal of the polypyrimidine region markedly reduced this step (7, 8). When the 3' AG dinucleotide was changed to AC, splicing was abolished and reduced levels of 5' exon and intron-exon lariat were detected (9). Thus, changes in the 3' intron sequence markedly affect both steps in the splicing reaction. Interestingly, Fu et al. (10) have shown that changes in the polypyrimidine sequence of the SV40 viral early region altered the selection of the small tumor antigen and large tumor antigen 5' splice sites without affecting the level of spliced products formed. These results suggest that the polypyrimidine site should be recognized by a protein component of the overall splicing and indeed a small nuclear ribonucleoprotein-associated protein with this property has been detected (11-13). Alterations in the polypyrimidine sequence reduced the binding of this factor (11). However, this factor has not been purified or further characterized.
A number of introns lack a polypyrimidine sequence at their 3' ends (2). In several cases, such introns are involved in alternative splicing reactions (14)(15)(16). This is the case with intron 2 of the adenovirus major late transcript which separates exons 2 and 3. In the middle of intron 2 an alternatively spliced exon is present, which has been called the "i leader." This exon is found only in late mRNAs whose main nucleotide sequences lie just downstream of the tripartite leader, such as the 52-and 55-kDa mRNAs (17, 18). However, most late mRNAs of adenovirus do not contain the i leader.
In the studies presented here, we have investigated the role of the polypyrimidine sequence in splicing, using base substitutions in this region rather than deletions. The plasmid p l -2 (previously named pKT.l) was used as the wild-type pre-mRNA because it is small (having an intron of 86 nucleotides) and well characterized (9,(19)(20)(21)(22). It has been used extensively by our laboratory as a substrate for examining the effects of sequence changes in pre-mRNA on splicing (9,191.' In all of these mutational studies, no cryptic 3' or 5' splice sites were activated, and only a single alternative branch point has been detected (19).' The experiments presented here focus on the introns of the adenovirus 2 tripartite leader of the major late transcription unit. Intron 1 contains a prominent polypyrim-idine tract, whereas intron 2 lacks a discernible polypyrimidine sequence. The effects of alterations in the 3' regions of these two introns were examined in pre-mRNAs containing both introns as well as pre-mRNAs containing only one intron.

MATERIALS AND METHODS~
Plasmid Construction and Mutagenesis-The construction of plasmid pl-2 (previously known as pKT.l) has been described (22). A detailed description of the construction of other plasmids is presented in the Miniprint section at the end of this paper.
In Vitro Splicing Reacti~ns-~~P-Labeled transcripts were prepared by linearizing plasmids with the appropriate restriction enzyme (ScaI for cleavage in exon 2 and either PstI or Hind111 for cleavage in exon 3), followed by incubation with SP6 RNA polymerase as described previously (22). The formation of spliced products and splicing complexes was measured in reaction mixtures (50 pl) containing 20 mM Hepes3-KOH buffer (pH 7.6), 3 mM MgC12,lO mM creatine phosphate, 0.4 mM ATP, 2% polyethylene glycol 8000, 2 mM dithiothreitol, 3zPlabeled pre-mRNA (1-5 pmol, 2 X lo5 cpm/pmol), and nuclear extract (40-60 pg of protein) (23) incubated at 30 "C for the time indicated. Reactions were stopped by the addition of 7 volumes of a solution containing 1% sodium dodecyl sulfate, 0.3 M NaCl, and 2 mM EDTA. RNA was extracted with phenol and chloroform and ethanol-precipitated. The RNA was collected by centrifugation, resuspended in 3 pl of 80% formamide, heated to 80 "C for 2 min, and electrophoresed on polyacrylamide-urea gels using Tris-borate-EDTA as the running buffer (22). Splicing products and intermediates were quantitated by excising the appropriate bands from the dried gel following autoradiography and measuring their radioactive content by scintillation counting.
Formation and Analysis of Splicing Complexes-The partially purified fractions used for generating nucleoprotein complexes have been described previously (20). 3ZP-Labeled transcripts were incubated under the same conditions as described above for the standard splicing reaction, except that appropriate fractions (10 pl of fraction I1 (3-4 mg/ml) and fraction I1 plus 5 pl of fraction Ib (4 mg/ml)), isolated from nuclear extract of HeLa cells, were substituted for unfractionated nuclear extracts. Reactions were incubated at 30 "C for 2 h and then loaded directly onto 10-30% sucrose gradients, prepared, and centrifuged as described previously (20). Nucleoprotein complexes were also analyzed by neutral polyacrylamide-agarose composite gel electrophoresis (3.5% polyacry1amide:bisacrylamide 8O:l and 0.5% agarose using 112 Tris-borate-EDTA as the running buffer).

RESULTS
The Polypyrimidine Tract Is Required for Splicing-It has been shown previously that the polypyrimidine sequence located at the 3' end of introns is required for efficient splicing in vitro (7,s). These studies were carried out with pre-mRNAs in which all or part of this sequence was deleted. Such deletions resulted in a reduction in the level of spliced RNA and splicing intermediates. In the studies described here, the influence of altered polypyrimidine sequences on in vitro splicing was examined using pre-mRNAs in which base substitutions rather than deletions were used. By this method, the observed effects can be attributed to the perturbation of the sequence and not the positional changes caused by the deletion of nucleotides.
The wild-type pre-mRNA used in these studies was transcribed from plasmid pl-2 and contains intron 1 of the adenovirus major late pre-mRNA. This substrate, which was described previously as plasmid pKT.l (lS), is shown schematically in Fig. 1A. Plasmid pIz was constructed so that  Analysis of products and splicing complexes formed with pre-mRNAs p l -2 (wild-type) and pIz (polypyrimidinealtered). A, schematic summary of the structures of pre-mRNAs pl-2 and pIz. In this figure, boxes represent exons, solid lines denote introns, and hatched boxes indicate plasmid sequences. The sequences above and below the drawings are the nucleotides present in the polypyrimidine region of each transcript. The numbers below the line indicate the size of each exon or intron. B. the transcripts pl-2 and pIz were synthesized from ScaI-digested plasmid DNA producing runoff products of 180 nucleotides. Following incubation with nuclear extract, as described under "Materials and Methods" for the indicated period, the RNA was isolated and the products separated on 18% polyacrylamide-urea gels. Lanes 1-4 show the products formed from pre-mRNApl-2 after incubation for 0,30,60, or 120 min, respectively. Lanes 5-8 show the products produced by incubation of pIz pre-mRNA with nuclear extract under splicing conditions for 0, 30, 60, or 120 min, respectively. The diagrams at the left of the gel represent the structures of the RNAs present in the adjacent gel bands. C, pre-mRNAs prepared from pl-2 and pIz were incubated for 2 h with fractions derived from nuclear extract (fractions I1 and Ib (20)) as described under "Materials and Methods" and the complexes separated by electrophoresis on native acrylamide-agarose gels; gels were then dried and autoradiographed. The sedimentation values of the complexes determined by sucrose gradient centrifugation are shown on the left. The positions of complexes formed in the absence of ATP are also shown. The fractions used to generate the various complexes are indicated below each lane. several of the pyrimidine residues within the polypyrimidine tract of the intron 1 were changed to purines (Fig. JA). When transcripts prepared from plasmid pIz were used as substrates in the in uitro splicing reaction, no intermediates or products were detected (Fig. 1B). The amount of spliced RNA formed after 2 h of incubation was at least 40-fold greater with the wild-type pre-mRNA (lane 4 ) than with the mutated pre-mRNA ( l a n e 8).
In order to determine which reaction in the splicing process was affected by the alteration of the polypyrimidine sequence, wild-type and mutated transcripts were tested for their ability to support the formation of nucleoprotein complexes. pl-2 and pIz pre-mRNAs were incubated with separated fractions prepared from nuclear extracts. Earlier studies showed that incubation of wild-type pre-mRNA with these fractions resulted in the production of either a 30 S prespliceosome complex (fraction I1 alone) or the 55 S spliceosome complex (fractions I1 plus Ib) (20). Following incubation for 2 h at 30 "C, the reaction mixtures were loaded onto a nondenaturing polyacrylamide-agarose composite gel and the complexes separated by electrophoresis (Fig. IC). Pre-mRNA pIz sup-ported the synthesis of low levels of both 55 S splicing complex (Fig. IC, lane 4 ) and 30 S presplicing complex compared to complexes formed with the wild-type pre-mRNA (approximately 30-fold; compare Fig. IC, lanes 1 and 3 ) . This result suggests that the polypyrimidine tract is required for efficient complex formation and indicates that it plays an important role in an early step in splicing.
Intron 2 Is Different Than Intron 1-Intron 2 of the adenovirus 2 major late mRNA lacks a discernible 3' polypyrimidine sequence. The plasmid p2-3 was constructed in order to generate a pre-mRNA substrate containing this intron ( Fig.  2A). When pre-mRNA p2-3 was incubated in the in vitro splicing reaction mixture, spliced RNA was produced (Fig.  2B). Quantitation of splicing products and intermediates formed in these reactions revealed that 20% of the pre-mRNA was spliced after 2 h at 30 "C.
To examine the influence of a polypyrimidine sequence at the 3' end of intron 2, a new plasmid, pIIx, was constructed in which the sequence between the branch site and the 3' splice junction of plasmid p2-3 was removed and replaced with the corresponding region of intron 1 (see Fig. 2 A ) . In order to retain the branch point of intron 2 in modified pre-mRNA substrates the position of this site was determined. For this purpose, the intron-exon lariat formed from pre-mRNA p2-3 was isolated and the position of the 2' to 5' phosphodiester structure was determined by primer extension analysis. The branch point was mapped to a single adenosine residue located 21 nucleotides upstream of the 3' splice junction (data not shown; see Miniprint section for this sequence).
Pre-mRNA pIIx and pre-mRNA p2-3 were compared as substrates in splicing reactions (Fig. 2B). After 30 min of incubation, the level of spliced RNA formed with either pre-mRNA was approximately the same. However, the level of 5' exon and intron-exon lariat formed was 6-to 10-fold greater with pIIx than with p2-3 transcript (Fig. 2B, lanes 2 and 6). After 2 h of incubation the level of final spliced product accumulated from pIIx pre-mRNA was only slightly higher (30%) than that found with p2-3 pre-mRNA (Fig. 2B, lanes 4  and 8). These results suggest that the presence of the polypyrimidine sequence in intron 2 markedly increased the rate of the first step of the splicing reaction but did not significantly increase the yield of the final product. This suggests that the second step may be the rate-limiting step in the overall splicing reaction.
The formation of nucleoprotein complexes was also examined with these RNA substrates. Pre-mRNAs from p2-3 and pIIx were incubated with fractions isolated from nuclear extracts, and the complexes were separated by sucrose gradient centrifugation rather than by gel electrophoresis since nucleoprotein complexes formed with these longer transcripts were not resolved by electrophoresis on neutral gels. As shown in Fig. X , both substrates yielded about the same level of 30 S presplicing complex after 2 h of incubation, whereas the level of 55 S complex formed was 2-fold higher with pIIx pre-mRNA than with the p2-3 transcript.
Effects of the Presence of Intron 2 Sequence in Place of the Polypyrimidine Tract of Intron 1-To determine whether the short uridine-rich region, . . . UUGUUGU . . ., present at the 3' end of intron 2 functioned as a polypyrimidine tract, the plasmid pIy was constructed from plasmid pl-2. Plasmid pIy contains the sequence between the branch point and the 3' splice junction of intron 2 inserted in place of the corresponding sequence of intron 1. This plasmid was used to generate transcripts which were then tested in the in vitro splicing reaction. As shown in Fig. 3B, low levels of spliced RNA were formed, indicating that this sequence can partially substitute A. The transcripts were generated with SP6 RNA polymerase using HindIII-digested plasmid DNA as the template. A , a schematic summary of the two transcripts is shown. The boxes represent exons, the solid lines introns, and the hatched areas plasmid sequences. The length of each exon and intron is indicated below each drawing. The sequences at the 3' end shown above pl-2 and below pIy indicate differences between these two pre-mRNAs. The first A residue in each inserted sequence (reading left to right) is the site at which the branch structure is formed. B, p2-3 and pIIx pre-mRNAs were incubated with nuclear extracts, and the products were separated by electrophoresis on 18% polyacrylamideurea gels. Reaction mixtures incubated for 0, 30, 60, or 120 min with pre-mRNA p2-3 and pre-mRNA pIIx are shown in lanes 1-4 and 5-8, respectively. A diagram representing the structures of the RNA found in the adjacent gel bands is shown to the left of the figure. C, ribonucleoprotein complexes formed after incubation of p2-3 pre-mRNA or pIIx pre-mRNA with fractions derived from nuclear extracts as described under "Materials and Methods" separated by sucrose gradient centrifugation. The closed circle symbols indicate the sedimentation profile generated when the pre-mRNAs were incubated with fraction I1 alone, whereas the open circle symbols indicate the sedimentation profile observed when the pre-mRNA was incubated with fractions I1 plus Ib.
for the polypyrimidine tract. The amount of spliced product formed after 2 h of incubation was considerably lower (&fold) than the level of spliced RNA formed with the wild-type transcript pl-2 (Fig. 3B, lanes 4 and 8).
The ATP-dependent synthesis of the 30 and 55 S complexes with pIy and pl-2 transcripts were examined by gel electrophoresis (Fig. 3C). The level of each complex formed with the non-polypyrimidine transcript (pIy) was greatly reduced compared with the amount produced from the wild-type pre- FIG. 3. Analysis of splicing reactions carried o u t w i t h pre-mRNA pIy (containing the 3' region of intron 2 inserted in place of the polypyrimidine tract of plasmid pl-2) or with pre-mRNA pl-2 (wild-type). The labeled pre-mRNAs were synthesized with SP6 RNA polymerase using ScaI linearized plasmid DNA as the template as described previously (20). A , a schematic summary of the pIy and pl-2 pre-mRNAs is shown. The boxes indicate exons, solid lines introns, and hatched areas are plasmid sequences. The length of each exon and intron is indicated below each drawing. The sequences written above pl-2 and below pIy indicate the differences between these two pre-mRNAs. The first A residue in each inserted sequence (reading left to right) is the site at which the branch structure is formed. R, pl-2 and pIy pre-mRNAs were incubated with nuclear extract and the products separated by electrophoresis on 18% polyacrylamide-urea gel. Incubation of pl-2 pre-mRNA and pIy pre-mRNA for 0, 30, 60, or 120 min are shown in lunes 1-4 and 5-8, respectively. The diagrams at the left of the figure indicate the structure of the RNA present in each adjacent gel band. C, nucleoprotein complexes, formed after incubating pl-2 or pIy pre-mRNA with separated fractions as described under "Materials and Methods," were separated by electrophoresis on native polyacrylamide-agarose gels after which the gels were dried and autoradiographed. The sedimentation values of the complexes, determined by sucrose gradient sedimentation, are shown at the right of the autoradiograph. The position of ATP-independent complexes is also shown. The fractions used to generate the various complexes are indicated below each lane. mRNA (20-and 30-fold less 30 and 55 S complex, respectively).
Downstream Sequences Influence the Requirement for a Polypyrimidine Tract-To examine the role of the polypyrimidine sequence on splicing of pre-mRNAs containing multiple introns, plasmids were constructed in order to generate the five different pre-mRNAs described in Fig. 4A. Each transcript was analyzed for its ability to support splicing. The results indicated that pre-mRNAs containing an altered polypyrimidine sequence of intron 1 (as shown with pre-mRNAs pIyIIx, pIyIIy, and pIzIIx), as well as a second intron, were efficiently spliced. This was in contrast to results obtained with pre-mRNAs containing a single mutated intron. A quantitative analysis of the products resolved by gel electrophoresis showed that the level of spliced RNA formed with pre-mRNA pIyIIy or pIyIIx was 90% of the level observed with the wildtype transcript synthesized from pl-2-3 (Fig. 4, lanes 9 and  12 compared with lane 3 ) . This represented an 8-fold increase in the level of spliced RNA formed with the multiple intron pre-mRNA compared with the pre-mRNA pIy which contained the single intron. In the case of the transcript, pIzIIx, the level of spliced RNA was 40% of the spliced product formed from the wild-type pre-mRNA pl-2-3. In contrast, the ' 1 2 3"4 5 6"7 8 9"lO 11 d l 3 1415' ." drawing. Roman numerals I and I1 refer to introns 1 and 2, respectively. The letters within the introns refer to the 3' end sequence; X is the polypyrimidine tract derived from intron 1, Y is the nonpolypyrimidine sequence from intron 2, and 2 is the mutated sequence from plasmid pIz. B, each R2P-labeled pre-mRNA was produced from PstI linearized plasmids using the SP6 RNA polymerase and splicing reactions with crude nuclear extracts were as described under "Materials and Methods." Splicing products were separated by electrophoresis on 8% polyacrylamide-urea gels. The five double intron pre-mRNAs which were incubated for 0, 60, or 120 min are as shown. The diagrams at the right of the figure represent the products present in the adjacent gel bands. The length of incubation (minutes) is indicated at the bottom of each lane. single intron pre-mRNA pIz yielded no spliced products. These results suggested that the efficiency of splicingof intron 1 containing an altered polypyrimidine site was influenced by downstream sequences.

5' Sequences from a Downstream Intron Can Overcome the Effects of an Upstream Polypyrimidine Mutant-To
determine whether a complete second intron was required for the activation of an altered polypyrimidine sequence of intron 1, longer runoff transcripts from plasmids pIz and pl-2 (referred to as pIzHc and pl-2Hc, respectively) were prepared and  5. Analysis of the splicing products formed from longer HincII runoff pre-mRNAs. A , schematic summary of the structure of the four pre-mRNAs. The boxes represent exons, the solid lines introns, and the hatched areas are plasmid sequences. The numbers below each drawing indicate the length in nucleotides of each exon and intron. B, the 32P-labeled pre-mRNAs were generated with SP6 RNA polymerase using the HincII-digested plasmid DNA as templates. Following incubation with nuclear extracts, the products were isolated and separated by electrophoresis on an 18% polyacrylamideurea gel. The RNAs were incubated for 0, 60, or 120 min, as shown. The arrows in lane 6 indicate the positions of alternatively branched lariat RNAs. analyzed in the in uitro splicing reaction. These longer runoff transcripts contained a complete exon 2 and the first 27 nucleotides of intron 2. As shown in Fig. 5B, the pIzHc transcript supported splicing at a level that was 40% of that observed with the wild-type transcript, pl-2Hc. The presence of this limited sequence added to pIz (to form pIzHc pre-mRNA) resulted in an increased level of excision of intron 1 as was found with the pre-mRNA pIzIIx which contained two introns.
Examination of Fig. 5B, lanes 5 and 6, revealed that additional intron-exon lariat and intron lariat products were formed with pl-2Hc pre-mRNA (indicated by arrows in Fig.  5B). Based on their altered gel migration, these products most likely represent a lariat formed at an alternative branch site located six nucleotides upstream of the normal branch site. This branch site was shown previously to be activated when the normal branch site of pl-2 was changed from an adenosine to a guanosine residue (19). Pre-mRNAs that formed a branch at this alternative adenosine site spliced normally (19).
The possibility that a specific sequence was required for the activation of pIz was also examined. For this purpose, plasmids were constructed in which pl-2 and pIz DNAs were cut at the ScaI site in exon 2 and ligated to plasmid sequences. The sequences involved are indicated in the Miniprint section at the end of this paper. The resulting plasmids, pl-2X and pIzX, were then linearized with HincII so that RNA runoff products from these DNAs contained 55 nucleotides of plasmid origin at their 3' ends replacing the 63 nucleotides present at the 3' ends of pIzHc and pl-2Hc pre-mRNAs. As shown in Fig. 5B, pIzHc pre-mRNA did not support the synthesis of spliced RNA, whereas pl-2XHc pre-mRNA did. These results demonstrate that specific sequences are present in pIzHc pre-mRNA that are responsible for the activation of this altered intron.

DISCUSSION
In this study we have described experiments that examined the role of the polypyrimidine sequence in pre-mRNA splicing in uitro. The polypyrimidine tract has been shown to be required in pre-mRNA splicing (4-8) and is bound by com-ponents of the splicing apparatus (11-13). Using a series of mutated pre-mRNAs we examined the function of the polypyrimidine region at various stages in the splicing process. These results show that although the polypyrimidine tract is important, it is not always essential for splicing.
The wild-type transcripts used in these studies, pre-mRNAs pl-2 and p2-3, contained introns 1 and 2 of the Ad2 major late tripartite leader sequence, respectively. Although pre-mRNA p2-3 lacks a discernible polypyrimidine sequence, both pre-mRNAs were spliced efficiently by nuclear extracts. Consistent with the findings of Ruskin and Green (7) and Frendewey and Keller (8), we found that extensive alteration of the polypyrimidine tract of intron 1, as shown with pre-mRNA pIz, blocked splicing. Neither spliced RNA nor spliting intermediates were formed when pIz pre-mRNA was incubated with nuclear extract.
We have previously reported the use of fractions derived from nuclear extracts to study the formation of splicing complexes (20). Following incubation of pre-mRNA pIz with these fractions, little 30 S presplicing complex or 55 S splicing complex formed. This finding indicates that the polypyrimidine tract affects an early step in the splicing process.
These results suggest a strong requirement for the presence of a polypyrimidine tract in splicing of intron 1 and an apparent lack of such a requirement for intron 2 of the Ad2 major late pre-mRNA. One possible explanation is that the 3' end of intron two contains a sequence that can function in place of the polypyrimidine sequence. To test this possibility, the plasmid pIy was constructed, in which the sequence between the branch point and the 3' splice junction of intron two replaced the polypyrimidine sequences of pl-2. Low levels (16% of that observed with pre-mRNA pl-2) of spliced RNA were formed when pIy pre-mRNA was incubated with nuclear extracts as well as reduced levels of 30 and 55 S splicing complexes. Thus, the 3' end sequence of intron two can only partially substitute for the polypyrimidine tract of intron 1.
These findings prompted the construction of plasmid pIIx in which the polypyrimidine tract of intron 1 was substituted for the 3' sequences of p2-3. A marked increase @-fold) in the level of splicing intermediates (5' exon and intron-exon lariat) resulted when pIIx pre-mRNA was incubated with nuclear extracts as compared with p2-3 pre-mRNA. However, only a slight increase in the rate of formation of the final product was observed. One possible explanation for this difference is that a factor required for the second step of the splicing reaction is limiting.
The effect of altering the polypyrimidine tract in intron 1 was dependent on the length of the transcript. This was discovered when pIy or pIz was linked to intron two sequences (as in pIyIIx, pIyIIy, and pIzIIx pre-mRNAs). Intron 1 of the double intron transcripts produced from these plasmids was efficiently excised in all cases. This result demonstrated that the presence of sequences downstream of the ScaI site, in the middle of exon 2, reduced the need for a polypyrimidine tract.
To determine the length of downstream sequences required for this effect, runoff transcripts that extended past the ScaI site were prepared from plasmid pIz. Digestion with HincII produced a transcript, pre-mRNA pIzHc, that was 63 nucleotides longer than the original pre-mRNA pIz. When this transcript was incubated with nuclear extracts, intron 1 was excised with the same efficiency as found with pre-mRNA pIzIIx (40% of pl-2). Thus, the presence of all of exon 2 and the first 27 nucleotides of intron 2 in pre-mRNA pIz obviated the requirement for a polypyrimidine sequence in intron 1. These results suggest that the polypyrimidine tract is necessary for the high levels of splicing but is not absolutely in Splicing required. To show that this effect was due to specific sequences, we constructed a new plasmid, pIzXHc, in which the 63 additional nucleotides present in pre-mRNA pIzHc were replaced with 55 nucleotides of plasmid sequence. When this pre-mRNA was incubated in nuclear extract no spliced RNA formed. These same 55 nucleotides had no effect when added onto the 3' end of pre-mRNA pl-2 (pre-mRNA pl-2XHc).
Splicing reactions carried out with pre-mRNA pIzHc resulted in the formation of two different branched lariat products. It is possible that a second branch point located at the adenosine residue located six nucleotides upstream of the normal branch site was used with this pre-mRNA. We have described previously the activation of this branch site with pre-mRNAs in which a guanosine residue replaced adenosine at the normal branch point (19). This result suggests that the polypyrimidine tract may play a role in branch point selection.
The effect of changing the length of the transcript on the requirement for the polypyrimidine tract in splicing indicates that the role of these sequences must be complex. Although the absence of a polypyrimidine tract in intron 2 suggests that this sequence is not required, the presence of this sequence increased the rate of the formation of 5' exon and intronexon lariats. The third exon of the Ad2 tripartite leader has two potential 5' donor splice sites (the '5" leader and exon 2).
It would be of interest to determine whether the presence of the polypyrimidine tract influences the selection of the 5' donor site as observed for the SV40 early transcript (10).
The features governing splicing must include both specific sequence recognition and structural characteristics contained in ribonucleoprotein complexes. At present, only the two dinucleotide sequences at the 5' and 3' ends of introns are absolutely essential for splicing in higher eukaryotes. The sequence UACUAAC present in the small number of yeast introns is an essential third signal for splicing but only in that system. In view of the large number of introns present in higher eukaryotes and the added complexity of alternative splicing, it seems likely that additional structural features are required for the accuracy of the splicing reaction. It is evident that the formation of prespliceosome and spliceosome complexes, essential intermediates for splicing, do not guarantee that spliced RNA will ultimately form. We, and others, have shown that protein components (present in our fraction Ib) are required to further activate the spliceosome to generate the first covalent modification of the pre-mRNA (20, 24, 25). We have shown that formation of an intron-exon GpG (5'-2') lariat structure rather than an intron-exon GpA (5'-2') lariat structure did not lead to splicing, despite the fact that the branch site was formed at the normal distance from the 3' splice site (19). Similar observations have been reported by Hornig et al. (26). How ribonucleoproteins interact with pre-mRNAs, juxtaposing and joining exons with the accuracy that is essential, remains to be elucidated.