INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Splice site selection has been most extensively studied in higher eukaryotes (reviewed in reference 11), where abundant evidence indicates that the unit initially recognized by the splicing machinery is the exon, as proposed by Robberson et al. nearly a decade ago (53). Particularly compelling in this regard is the observation that the most common effect of a 5' splice site mutation is skipping of the preceding exon rather than inclusion of the mutant intron (61; reviewed in reference 6). Moreover, in the subset of cases in which a 5' junction mutation causes activation of a cryptic splice site rather than exon skipping, the new exon-intron boundary is almost invariably located within the preceding exon, again supporting the view that communication occurs across the exon rather than the intron. Finally, there are significant constraints on exon length in vertebrate pre-mRNAs, consistent with the proposal that the 3' and 5' splice sites on opposite sides of the exon must be recognized concurrently. Not only are the vast majority of natural internal exons in vertebrate pre-mRNAs <300 nucleotides in length (6), but expanding an exon beyond this size causes it to be skipped (53), particularly if it is surrounded by large introns (60). In contrast to the limitations on exon length, the introns in vertebrate pre-mRNAs can be extremely large (tens of kilobases [29]).

Although many questions remain to be answered, several components of the machinery responsible for exon definition have been identified. First, UV cross-linking experiments revealed that binding of the U1 snRNP to the downstream 5' splice site stabilizes the association of U2AF⁶⁵ with the polypyrimidine tract of the upstream intron (32). Likely candidates to form a bridge between these components were identified by protein-protein interaction assays, which indicated that the 70,000-Da protein of the U1 snRNP binds to members of the serine-arginine-rich (SR) family of splicing factors, which in turn bind to the small subunit of the U2AF heterodimer (5, 38, 70). The U1-70K/SR/U2AF³⁵/U2AF⁶⁵ network has also been proposed to play a role in communication across introns (70). However, because one of these components (U2AF³⁵) is absent in Saccharomyces cerevisiae and two others (U2AF⁶⁵ and SR proteins) are not highly conserved, this mode of connecting splice sites may not be ubiquitous (1, 2). A distinct network of intron-spanning interactions forms at an early stage of the splicing pathway in yeast and most likely in mammals as well (2, 7). In addition to this network, which extends from the large subunit of U2AF to the branchpoint bridging protein to a different component of the U1 snRNP, Prp40p, recent work with Drosophila melanogaster points to a third set of early intron-bridging interactions involving a divergent member of the SR protein family, SRp54 (36). The relationships among these networks of protein-protein interactions remain to be elucidated.

Extremely small exons also pose recognition problems for the vertebrate splicing machinery, leading to a default splicing pattern in which the microexon is skipped (e.g., 9, 18, 59). This phenomenon was originally proposed to result from steric interference between closely juxtaposed 3' and 5' splice sites (9), but it is now attributed primarily to a lack of positive interactions across the small exon (10, 13, 59). In the three examples studied most extensively, incorporation of the microexon is promoted by complex enhancer elements located in the downstream intron (10, 13, 66). In the case of c-src, it has been shown that a large assemblage of proteins, including hnRNP F (44), K-SRP (45), and hnRNP H (14), binds to the intronic enhancer and regulates microexon inclusion, possibly by promoting use of the abutting 5' splice site.

In both budding yeast and fission yeast, as well as other unicellular eukaryotes, small introns predominate, and exon size does not appear to be constrained (15, 56, 72). These observations prompted Talerico and Berget (62) to propose that, in simple eukaryotes, the intron rather than the exon serves as the initial unit of recognition during spliceosome assembly (62; reviewed in reference 6). Consistent with this proposal, alternative exon usage has not yet been demonstrated in either yeast species. However, two well-documented instances of regulated splicing have been described in S. cerevisiae, both utilizing intron retention as an on-off switch for protein expression (19, 20). The situation is less clear in Schizosaccharomyces pombe, but a similar form of regulation at the level of splicing has been proposed for mes1 pre-mRNA during meiosis (37).

While small introns are also common in certain metazoa including Caenorhabditis elegans and D. melanogaster, these species contain large vertebratelike introns as well (22, 46). In the fruit fly, there is experimental evidence for initial splice site pairing via "intron definition," since expansion of small introns leads either to their retention or to activation of a cryptic 3' splice site (27, 62). On the other hand, several examples of exon skipping have been reported in Drosophila, both naturally occurring, as in the sex determination regulatory cascade (reviewed in reference 41) and experimentally induced (e.g., 47, 57), consistent with splice site pairing via exon definition. In S. cerevisiae, only a handful of pre-mRNAs harbor more than one intron (58), and the trans-acting factors implicated in exon-spanning interactions are either absent or highly divergent (1, 2, 8). In contrast, S. pombe contains all of the factors implicated in forming bridges between exons, including at least two canonical SR proteins and highly conserved homologs of both subunits of U2AF (26, 42, 49, 67). This fact, together with the ability of the Drosophila splicing machinery to utilize both the exon and intron definition modes, prompted us to ask whether communication can occur across exons in S. pombe.

To address this question, we first engineered constructs containing splice site mutations which, in mammals, would produce the outcome that is most diagnostic for this mode of initial splice site pairing, namely, exon skipping. In S. pombe, mutating the downstream 5' splice site produced exclusively intron retention, and even in a pre-mRNA carrying severe mutations in both flanking splice sites, exon skipping was rare. To address the possibility that the lack of skipping was due to the large size of the internal exon, we turned to a different S. pombe pre-mRNA which contains a microexon. Again, the profile of products was as predicted by the intron definition model. A final indication that splice site pairing proceeds via intron definition in fission yeast is the location of several cryptic 5' splice sites within an intron. The competition between these and the natural 5' junction provided an opportunity to explore parameters that influence splice site pairing in S. pombe. In alleles containing deletions and insertions within the intron, as well as those with wild-type splice site spacing, we found that the pattern of cryptic splice site usage not only conformed to the predictions of the intron definition model but suggested that the fission yeast splicing machinery has a strong preference for excising the smallest intron possible.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid construction and mutagenesis. Construction and analysis of polypyrimidine tract variants of cdc2/pREP2, which carry the second intron of the S. pombe cdc2 gene together with flanking exon sequences under control of the nmt1 promoter, were described elsewhere (54). For the exon-skipping experiments reported here, the remainder of the third exon, as well as the third intron and the fourth exon, were incorporated into the BamHI sites of the wild-type, R-short, and R-long alleles as a PCR fragment amplified from a genomic clone (31) with Taq DNA polymerase, using the procedure suggested by the manufacturer (Gibco-BRL) (35) and the primers cdc2-Ex3-5' and cdc2-Ex4-3' (Table 1); these constructs are designated cdc2-Long. Site-directed mutagenesis to inactivate the 5' splice site of intron 3 was carried out with reagents supplied commercially (Amersham Corp., Arlington Heights, Ill.), using the oligonucleotides cdc2-I3G1A, cdc2-I32nd5'SS, and cdc2I3Random (Table 1).

S. pombe transformation, RNA preparation, and primer extension analysis. The recipient S. pombe strain for assaying splicing of cdc2 and cgs2 variants was DS2 (h⁺ ade6-210 leu1-32 ura4-d18). Transformation and RNA preparation were as previously described (52). Primer extension reactions to assay cdc2 and cgs2 splicing were also described previously (4, 54). Quantitation was performed on a Molecular Dynamics PhosphorImager using ImageQuant software (version 3.1).

Computer-assisted RNA secondary structure analysis. Folding patterns for various RNAs mentioned in the text were analyzed using the MFold secondary structure prediction program developed by Zuker and colleagues. The program, which was run using standard parameters, is available at http://www.ibc.wustl.edu/~zuker/.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Exon skipping is a rare event in pre-mRNAs derived from cdc2. As noted in the introduction, both exon and intron definition modes of splice site pairing have been observed in Drosophila, and we wanted to test whether the S. pombe splicing machinery could also switch back and forth despite the preponderance of small introns in the genes characterized to date. The most salient prediction of the exon definition model is that an exon surrounded by weak splice sites will be ignored. Thus, to seek evidence for this mode of splice site pairing in fission yeast, we first attempted to induce exon skipping. Because strong pyrimidine tracts favor the exon mode of substrate recognition in vertebrates (62), we chose to analyze the second intron of the cdc2 gene in these experiments, since it contains the most extensive run of pyrimidines of any fission yeast intron experimentally verified to date (J. A. Wise and C. M. Romfo, unpublished observations). In earlier work, we analyzed splicing of cdc2 intron 2 alleles which contained polypyrimidine tracts of various strengths, using constructs containing a single intron (54). To provide substrates suitable for assessing exon skipping, we incorporated the third intron and fourth exon into each of our intron 2 polypyrimidine tract variants to produce the cdc2-Long constructs shown in Fig. 1A. If S. pombe recognizes any of these pre-mRNAs via exon definition, then mutating the 5' splice site following the internal exon will result in either exon skipping or activation of an upstream cryptic 5' splice site. On the other hand, if intron definition applies, the predicted outcome is inclusion of the downstream intron or activation of a cryptic 5' splice site located within intron 3. 

View larger version (28K): [in this window] [in a new window]   FIG. 1.   Analysis of exon skipping in cdc2 pre-mRNAs containing an internal exon flanked by mutant splicing signals. (A) Schematic representation of cdc2-Long polypyrimidine tract and 5' splice site variants. The transcripts analyzed contained the second and third introns of the cdc2 gene and flanking exons embedded within transcription signals from the nmt1 gene (43) (see Materials and Methods for details). Sequences between the branchpoint and 3' splice site of intron 2, as well as mutations introduced to inactivate the 5' splice site of intron 3, are indicated. The arrow designates the position where the oligonucleotide used for primer extension [nmt1-poly(A), Table 1] hybridizes. (B) Primer extension splicing assays on cdc2-Long pre-mRNAs. Total RNA was isolated from S. pombe cells transformed with the indicated plasmid, and the relative levels of precursor, partially spliced, fully spliced, and exon-skipped RNAs were determined using primer extension analysis with an nmt1-specific oligonucleotide as described previously (4, 51). (B, top panel) Gel electrophoretic analysis. The identities and mobilities of the observed cDNA products are indicated schematically alongside the gel. The predicted sizes of the primer extension products derived from cdc2-Long are precursor, 951 nucleotides (nt); Int1 (splicing of intron 1 only), 880 nt; Int2 (splicing of intron 2 only), 851 nt; M (mature), 780 nt; ES (exon 2 skipping), 480 nt; intron 1 lariat, 679 nt; and intron 2 lariat, 274 nt. The positions where the lariats are expected to migrate are devoid of signal and, in the case of the intron 2 species, not shown. Lane 1, wild-type cdc2-Long; lane 2, wild-type cdc2-Long with a mutant 5' splice site in intron 3; lane 3, R-short variant of intron 2 with wild-type intron 3; lane 4, R-short variant of intron 2 with a mutant 5' splice site in intron 3; lanes 5 and 6, as in lanes 3 and 4 except that the constructs contain the R-long allele of intron 2; M, molecular size markers. (B, bottom panel) Quantitation of primer extension data. The levels of precursor and mature message were determined by PhosphorImager analysis and are displayed as a bar graph in which the y axis shows the percentage of each species. For each sample, pre-mRNA + mRNA totals 100%.

A fission yeast intron that lies downstream from a microexon is inefficiently spliced. One possible explanation for the dearth of exon skipping in fission yeast cdc2 is the size of the internal exon in this pre-mRNA (301 nucleotides), which exceeds that of most vertebrate internal exons (6). However, because statistical analyses indicate that it is the lengths of introns, not exons, that are constrained in fission yeast (see the introduction), the size of exon 3 is more likely to pose a problem as the longest segment of the intron excised via skipping (extending from the 5' splice site of intron 2 to the 3' splice site of intron 3). To identify a potentially more favorable context for observing exon skipping in fission yeast, we searched a database of published gene sequences for pre-mRNAs containing a small internal exon sandwiched between introns that are also relatively small. Among several candidate genes, we selected cgs2⁺ (17), which contains a second intron of slightly above average size for S. pombe (see Fig. 8 below), a nine-nucleotide second exon, and an average-size third intron (72) (Fig. 2A). In addition to its potential for exon skipping, analysis of cgs2 offered an opportunity to examine similarities and differences between vertebrate and fission yeast cells in the processing of microexons.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Analysis of the profile of products from a pre-mRNA containing an internal microexon. (A) Schematic representation of cgs2 transcripts containing intron 1, intron 2, or both. The construct designated cgs2-Long (top) contains both the first and second introns of the cgs2 gene and flanking exons embedded within transcription signals from the nmt1 gene (43) (see Materials and Methods for details). The constructs designated cgs2-Int2 (middle) and cgs2-Int1 (bottom) contain either the second or the first intron from the cgs2 gene and flanking exons, respectively. (B) Primer extension splicing assays on cgs2-Long, cgs2-Int2, and cgs2-Int1 pre-mRNAs. RNA extraction and analysis were performed as described in the legend to Fig. 1B. Because the products from the cgs2-Int1 pre-mRNA are much smaller than those from cgs2-Int2 and cgs2-Long, they are shown in a separate panel even though they were run on the same gel. The identities and mobilities of observed as well as some potential cDNAs derived from all three pre-mRNAs are indicated schematically alongside each gel. Lane 1, cgs2-Long; lane 2, cgs2-Int2; M, molecular size markers (1-kb ladder); lane 3, cgs2-Int1. The predicted sizes of the possible primer extension products derived from cgs2-Long are as follows: precursor, 510 nucleotides (nt); intermediate in which only intron 1 is spliced, 449 nt; intermediate in which only intron 2 is spliced, 464 nt (not indicated); mature mRNA, 403 nt; product derived from exon 2 skipping, 394 nt; intron 1 lariat, 295 nt; intron 2 lariat, 233 nt. The predicted sizes of the primer extension products derived from cgs2-Int1 are as follows: precursor, 270 nt; mature, 209 nt; lariat intermediate, 60 nt. Those from cgs2-Int2 are as follows: precursor, 359 nt; mature, 313 nt; lariat intermediate, 233 nt.

View larger version (37K): [in this window] [in a new window]   FIG. 3.   (A) RT-PCR assay comparing the profiles of splicing products from chromosomal and plasmid-borne cgs2 genes. (A, top panel) Schematic representation of the relevant region of cgs2 pre-mRNA expressed from the endogenous locus, with arrows indicating the cgs2Int1-5' and cgs2Int2-3' primers (Table 1) used for reverse transcription and PCR amplification. For RNA from strains harboring a plasmid, the same 5' primer was employed but nmt1-poly(A) (see Fig. 2A; Table 1) was used as the 3' primer to prevent endogenous cgs2 from contributing to the signal. (A, bottom panel) Total RNA was subjected to RT-PCR as described in Materials and Methods, and the products were displayed on a 4% Nu-Sieve agarose gel stained with ethidium bromide. To allow a semiquantitative comparison of the different species, the number of cycles was limited to 22 for all samples. The identity of each species is indicated schematically alongside the gel. Lane 1, RNA from untransformed strain DS2 cells was subjected to RT-PCR using cgs2Int2-3' as the 3' primer; lane 2, RNA from cells harboring the cgs2-Long plasmid was subjected to RT-PCR using nmt1-poly(A) as the 3' primer; lane 3, as in lane 2 except that the cells harbored the empty vector; lane 4, as in lane 2 except that the RNA was from untransformed DS2 cells. The predicted sizes of the possible RT-PCR products derived from chromosomal cgs2 are as follows: precursor, 420 nucleotides (nt); intermediate in which only intron 2 is spliced, 374 nt; intermediate in which only intron 1 is spliced, 359 nt; mature mRNA, 313 nt. The predicted sizes of the possible RT-PCR products derived from cgs2-Long are as follows: precursor, 447 nt; intermediate in which only intron 2 is spliced, 401 nt; intermediate in which only intron 1 is spliced, 386 nt; mature mRNA, 340 nt. (B) Design and RT-PCR analysis of cgs2-Long mutants. (B, top panel) Schematic representation of cgs2-Long mutants in which the indicated base substitutions have been introduced at the downstream 5' splice site or the microexon and surrounding sequences have been deleted (indicated by an arrowhead). (B, bottom panel) RT-PCR assays of wild-type and mutant cgs2 splicing were performed as in panel A, using cgs2-Int1-5' and nmt1-poly(A) as primers. The identity of each species is indicated schematically alongside the gel, with an asterisk denoting the 5' splice site mutations. Lane 1, untransformed DS2; lane 2, empty vector control; lane 3, wild-type cgs2-Long; lane 4, 5'-splice-site mutant; lane 5, deletion mutant. The predicted sizes of the possible RT-PCR products derived from cgs2-Int25' are the same as for wild-type cgs2-Long. The predicted sizes of the possible RT-PCR products derived from cgs2- are as follows: precursor, 380 nt; mature mRNA, 331 nt.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Effect of expanding the microexon in cgs2. (A) Model to account for the inefficient splicing of the second intron in cgs2-Long pre-mRNA. See text for details. (B) Schematic representations of cgs2-Long RNAs in which exon 2 (E2) is expanded. The fragments introduced to increase the size of exon 2 (see Materials and Methods for details) are indicated by thick horizontal lines above E2. (C) Primer extension splicing assays on expansion alleles of cgs2-Long. RNA extraction and analysis were performed as described in the legend to Fig. 1B. M, molecular size markers (X174-HinfI); lane 1, wild-type cgs2-Long pre-mRNA; lane 2, cgs2-Long-XhoI; lane 3, cgs2-Long-31; lane 4, cgs2-Long-58; lane 5, cgs2-Long-75. In each lane, the top band corresponds to the intron 2 retention product and the bottom band to mature mRNA; the segments inserted into exon 2 are indicated as in panel B. The predicted sizes of the primer extension products derived from cgs2-Long-XhoI are the same as for wild-type cgs2-Long (see the legend to Fig. 2); for cgs2-Long-31, the predicted cDNA sizes are as follows: precursor, 532 nucleotides (nt); intermediate in which only intron 1 is spliced, 471 nt; and mature mRNA, 425 nt. For cgs2-Long-58, the sizes are as follows: precursor, 559 nt; intermediate in which only intron 1 is spliced, 498 nt; and mature mRNA, 452 nt. For cgs2-Long-75, the sizes are as follows: precursor, 576 nt; intermediate in which only intron 1 is spliced, 515 nt; and mature mRNA, 469 nt. (Lower panel) Quantitation as in Fig. 1B.

Expanding the microexon allows efficient splicing of cgs2 pre-mRNA. The findings presented in the preceding section suggest that the inefficient removal of the second intron in the cgs2-Long pre-mRNA is primarily due to the close juxtaposition of the first intron. A plausible model to explain this observation is steric occlusion of the downstream 5' splice site by an upstream spliceosome (Fig. 4A), as originally proposed to explain the exclusion of the mouse c-src microexon in non-neuronal cells (9). To test the resulting hypothesis that increasing the distance separating the relevant 3' and 5' splice sites will allow efficient splicing, we first engineered an XhoI site in exon 2. Unexpectedly, the substitution of four bases at positions 3, 5, 7, and 8 relative to the 5' splice site of intron 2 to create the restriction site virtually abolished splicing (Fig. 4C, compare lanes 1 and 2). These changes affect nucleotides which are not highly conserved through evolution (33), and thus they seem unlikely to diminish snRNA binding. Computer-assisted RNA secondary structure analysis indicates that the new sequence can form a local hairpin (W. J. van Heeckeren and J. A. Wise, unpublished data) which may interfere with splicing, as observed previously in cdc2 (4). Another possibility is that the sequence changes might affect binding of a protein factor to the microexon. For example, in rat 2 pre-mRNA, it was shown that some of the nucleotides required for optimal inclusion of a small internal neuron-specific exon reside within the microexon itself and function in an unpaired state (71).

Activation of an unusual cryptic 5' splice in cdc2 intron 2 is likely to reflect its favorable location for splicing. In addition to the incidence of exon skipping versus intron retention, the exon and intron definition models predict different locations for cryptic 5' splice sites. In the course of analyzing the contribution of U1 snRNA to 5' splice site selection in S. pombe (C. J. Alvarez and J. A. Wise, unpublished data), we discovered that an unusual cryptic 5' splice site is activated by certain mutations at the standard exon 1-intron 2 boundary in fission yeast cdc2. The location of this cryptic 5' splice site, which lies within the intron, 27 nucleotides downstream from the standard 5' junction (Fig. 5A) provides a third indication that splice site pairing in S. pombe proceeds via intron definition. As noted in the introduction, cryptic 5' splice sites in vertebrates generally lie upstream, within the preceding exon (6). The cryptic 5' splice site in cdc2 intron 2 is used to a significant extent, accounting for 8% of the total primer extension products when the standard 5' junction is mutated at position +6 (Fig. 5B, lane 3) (C. J. Alvarez and J. A. Wise, unpublished data) despite a deviation from consensus (an A at position +2) that would normally render it inactive (3, 64).

View larger version (29K): [in this window] [in a new window]   FIG. 5.   Effects of reciprocal mutations at the standard and cryptic 5' splice sites on splicing of cdc2 intron 2. (A) Sequence of cdc2 intron 2 (31) with the splicing signals, including the nonconsensus cryptic 5' junction discussed in the text, highlighted in bold. The 5' and 3' splice sites are demarcated by arrows and the branchpoint A is indicated by an asterisk. The nonconsensus nucleotide in the cryptic 5' splice site is shown in outline. Also indicated are the locations of deletions and insertions analyzed in subsequent experiments. Shown above the sequence is a 27-nucleotide (-nt) segment derived from just downstream of the 5' splice site in rabbit -globin IVS 1 (3), which was inserted at the positions indicated by open triangles for the experiments shown in Fig. 7. The 18 nt bracketed by the triangle beneath the sequence were deleted for the experiment shown in Fig. 6. (B) Primer extension splicing assay on 5' splice site mutants. The products are indicated schematically alongside the gel, with the portion of the intron retained when the cryptic 5' splice site is used indicated by a thick line. Lane 1, control demonstrating the absence of primer extension products in RNA from the untransformed recipient strain (DS2); lane 2, primer extension products from wild-type cdc2-Int2; lane 3, primer extension products from an allele containing a U+6G substitution at the standard 5' junction to provide a marker for the position of the cryptic band; lane 4, primer extension products from an allele containing a U+2A substitution at the standard 5' junction; lane 5, primer extension products from an allele containing an A+2U substitution at the cryptic 5' junction; M, molecular-size markers. The predicted sizes of the cDNA species, visualized by autoradiography, are as follows: precursor, 388 nt; mature, 317 nt; and lariat intermediate (not shown because no product was visible), 121 nt. (Lower panel) Quantitation as in Fig. 1B.

View larger version (32K): [in this window] [in a new window]   FIG. 6.   Effect of decreasing the size of cdc2 intron 2 in alleles containing position +6 changes at the standard 5' junction. Top: M, molecular size markers; lane 1, recipient strain control; lane 2, wild-type cdc2-Int2; lane 3, U+6G allele with normal spacing between the standard and cryptic 5' splice sites. The last three lanes show primer extension products from alleles containing an 18-nucleotide deletion between the standard and cryptic 5' splice sites (see Fig. 5A; designated 18) and either a U+6A (lane 4), U+6C (lane 5), or U+6G (lane 6) substitution at the standard 5' junction. Products are shown schematically as in Fig. 5B. (Lower panel) Quantitation as in Fig. 1B.

The effects of expanding cdc2 intron 2 are also consistent with the intron definition model. The fact that decreasing the size of the second intron in cdc2 stimulates its removal, consistent with splice site pairing via intron definition, prompted us to test the converse prediction that an increase in size will diminish splicing. To this end, we doubled the interval between the standard and cryptic 5' splice sites from 27 to 54 nucleotides via the insertion of a segment from intron 1 of rabbit -globin (3; see Fig. 5A, top left; designated 27-5'); as for the deletion constructs, the first alleles examined also contained substitutions at position +6 of the standard 5' splice junction. Primer extension splicing assays (Fig. 7A, lanes 4 to 6) indicate that increasing the size of cdc2 intron 2 dramatically reduces splicing at the standard 5' junction relative to an allele in which the spacing is normal; in the expanded alleles, most (from 82 to 88% in the three +6 mutants examined) of the RNA detected is linear precursor, whereas approximately equal quantities of precursor and mature mRNA are again observed for an allele with normal spacing (Fig. 7A, lane 3). In an expanded allele containing the wild-type sequence at the standard 5' junction, splicing at the original exon-intron boundary is readily detectable, but still quite inefficient (20%; Fig. 7B, lane 2) compared to either a fully wild-type (74%; lane 1) or a contracted allele (100%; lane 3).

View larger version (89K): [in this window] [in a new window]   FIG. 7.   Effect of increasing the size of cdc2 intron 2. (A) Primer extension analysis of alleles containing position +6 mutations at the standard 5' splice site. Products are shown schematically as in Fig. 4B, with the inserted nucleotides derived from rabbit -globin IVS 1 indicated by a wavy line. Lane 1, recipient strain control; lane 2, wild-type cdc2-Int2; lane 3, U+6G allele with standard spacing. The last three lanes show products derived from alleles containing a 27-nucleotide insertion between the standard and cryptic 5' splice sites (Fig. 5A; designated 27-5') and either a U+6A (lane 4), U+6C (lane 5), or U+6G (lane 6) substitution at the standard 5' junction. M, molecular-size markers. The more rapid mobility of the bands in lane 5 is due to salt in the sample. (B) Primer extension analysis of alleles containing the wild-type sequence at the standard 5' junction. M, molecular-size markers; lane 1, wild-type cdc2-Int2; lane 2, primer extension products from an otherwise wild-type allele containing a 27-nucleotide insertion between the standard and cryptic 5' splice sites; lane 3, primer extension products from an otherwise wild-type allele containing an 18-nucleotide deletion between the standard and cryptic 5' splice sites. (C) Primer extension analysis of an allele in which the distance between the cryptic 5' splice sites and the branch point is increased. M, molecular size markers; lane 1, U+6G allele with standard spacing; lane 2, products from an allele containing the 27-nucleotide insertion between the cryptic 5' junctions and the branch point (Fig. 5A; designated 27-3') and a U+6G substitution at the standard 5' junction. (Lower panel) Quantitation as in Fig. 1B.

The pattern of cryptic splice site utilization in cdc2 reflects the natural distribution of intron sizes in S. pombe. The foregoing analyses of cryptic splice site utilization are consistent with the notion that intron size is an important determinant of splicing efficiency in S. pombe. In Fig. 8, the sizes of the introns excised via use of each cryptic 5' splice site cdc2 intron 2 are superimposed on a histogram displaying the distribution of lengths for 156 naturally occurring introns from fission yeasts. Intriguingly, the length of the intron excised via use of the unusual (GA-containing) 5' splice site in cdc2 (44 nucleotides) coincides with the peak of this histogram, and the intron removed via use of the standard 5' splice site in the contracted alleles (53 nucleotides) also lies in a zone that is quite densely populated. In contrast, the intron removed via splicing at the standard 5' junction lies in a relatively barren region of the graph, and even fewer naturally occurring introns correspond in size to the very inefficiently spliced expansion alleles. Also consistent with the strong preference for small introns in S. pombe is our finding that the proximal 3' splice site is strongly preferred in alleles of cdc2 intron 2 carrying duplications of the 3' splicing signals (C. M. Romfo and J. A. Wise, unpublished data). As in our analysis of 5' splice site utilization, deletions of intronic sequences improved splicing efficiency in these pre-mRNAs (data not shown).

View larger version (17K): [in this window] [in a new window]   FIG. 8.   Sizes of the introns removed via use of cryptic 5' splice sites in cdc2 intron 2 in relation to the overall distribution of intron lengths in fission yeast. The histogram shows the sizes of 156 naturally occurring fission yeast introns arranged in bins of three. The arrows denote the lengths of the segments spliced out of the alleles examined here.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The fission yeast splicing machinery is most likely restricted to the intron definition mode of splice site pairing. In this report, we present three lines of evidence which, in aggregate, strongly suggest that the splicing machinery in fission yeast pairs splice sites exclusively across introns, in contrast to vertebrate cells. First, even when the splicing signals on both sides of an exon were mutated, the exon skipping product represented only a minor fraction of the total RNA (Fig. 1), while mutating the downstream 5' splice site alone was sufficient to produce exon skipping in vertebrate pre-mRNAs (reviewed in reference 6). Second, exon skipping did not occur at a detectable level during splicing of a wild-type fission yeast pre-mRNA that contains a microexon (Fig. 2), whereas this was the default splicing pattern in pre-mRNAs containing internal exons of comparable size in mammals (e.g., 9, 18, 59). Moreover, it was not possible to induce skipping of the microexon by mutating the downstream 5' splice site. Finally, the locations of cryptic splice sites, as well as the effects of expanding and contracting an intron (Fig. 5 through 7), are as predicted by the intron definition model, and contrast with the patterns of cryptic splice site usage observed in vertebrate pre-mRNAs (reviewed in reference 6). Our finding that increasing the size of cdc2 intron 2 compromises splicing confirms and extends the results of earlier experiments with an artificial intron in S. pombe, in which expansions also reduced splicing efficiency (24).

Do microexons serve a regulatory role in fission yeast? While microexon recognition has been subjected to intense experimental scrutiny in vertebrates, splicing of pre-mRNAs containing extremely small exons has not been previously investigated in a unicellular eukaryote. We could envision a priori, three possible profiles of splicing products from the cgs2-Long pre-mRNA: efficient removal of both introns, skipping of the microexon, or retention of one intron. The fact that only the third outcome was observed is consistent with the view that the fission yeast splicing machinery is restricted to the intron definition mode of splice site pairing. The apparent inability of the S. pombe splicing machinery to switch back and forth between intron and exon definition modes of splice site pairing contrasts with the situation in Drosophila, and indicates that fission yeast is unlikely to provide a simpler model system in which to study alternative splicing of the exon-skipping variety. On the other hand, the data presented here are consistent with the idea that S. pombe may be capable of modulating splicing efficiency to regulate the amount, if not the precise structure, of the mRNA produced from a particular gene.