127

Mechanisms of Development, 37 (1992) 127-140 © 1992 Elsevier Scientific Publishers Ireland, Ltd. 0925-4773/92/$05.00 MOD 00087

Suboptimal 5' and 3' splice sites regulate alternative splicing of Drosophila rnelanogaster myosin heavy chain transcripts in vitro D i a n n e H o d g e s a n d S a n f o r d I. B e r n s t e i n Biology Department and Molecular Biology Institute, San Diego State University, San Diego, CA 92182, U.S.A. (Received 22 November 1991; revision received and accepted 20 January 1992)

Using a Drosophila cell-free system, we have analyzed the regulation of alternative splicing of Drosophila muscle myosin heavy chain (MHC) transcripts. Splicing of MHC 3' end transcripts results in exclusion of adult-specific alternative exon 18, as is observed in embryonic and larval muscle in vivo. Mutations that strengthen either the 5' or the 3' splice sites of exon 18 do not promote inclusion of this exon. However, strengthening both splice junctions results in efficient removal of both introns and completely inhibits skip splicing. Our data suggest that the affinity of exons 17 and 19, as well as failure of constitutive splicing factors to recognize exon 18 splice sites, causes the exclusion of exon 18 in wild-type transcripts processed in vitro. Myosin heavy chain; Drosophila; Alternative RNA splicing; In vitro RNA splicing

Introduction

Drosophila muscles, like those of most higher eukaryotes, are structurally complex (Miller, 1950). This complexity derives, in part, from a large number of contractile protein isoforms which are generated by multigene families or through the process of alternative RNA splicing. The most abundant protein in the thick filaments of the muscle sarcomere is myosin, a multimeric protein consisting of two heavy chains (MHCs) and four light chains. In vertebrates MHC is encoded by multigene families (Emerson and Bernstein, 1987) whereas muscle MHC isoforms of Drosophila are encoded by a single copy gene in the haploid genome (Bernstein et al., 1983; Rozek and Davidson, 1983). The MHC isoform diversity required by the functionally and morphologically distinct embryonic, larval and adult muscles of Drosophila is generated by tissue-specific alternative splicing of the single primary transcript (Bernstein et al., 1986; Rozek and Davidson, 1986; Wassenberg et al., 1987; George et al.,

Correspondence to: S.I. Bernstein, Biology Department and Molecular Biology Institute, San Diego State University, San Diego, CA 92182, U.S.A.

1989; Hess et al., 1989; Collier et al., 1990; Hess and Bernstein, 1991). The 22 kb muscle MHC gene of Drosophila has a complex transcription unit containing 19 exons. There are three major mRNA size classes (6.1, 6.6 and 7.1 kb) that result from use of two polyadenylation sites as well as inclusion or exclusion of the penultimate exon 18 in a stage and tissue-specific manner (Bernstein et al., 1986; Rozek and Davidson, 1986; George et al., 1989). Exon 18 is included in pupal and adult messages expressed in the major muscles of the thorax, the indirect flight muscles (Bernstein et al., 1986; Kazzaz and Rozek, 1989). However, it is excluded in all embryonic, larval and some adult muscle transcripts. The MHC primary transcript also contains 5 sets of mutually exclusive exons which are alternatively spliced so that only one exon from each set is included in mature mRNAs (George et al., 1989; Hess et al., 1989; Collier et al., 1990; Kronert et al., 1991). We are interested in determining how this complex tissue-specific regulation of muscle MHC isoform diversity is mediated by alternative splicing. Alternative RNA splicing is an important mode of generating protein diversity (Left et al., 1986; Breitbart et al., 1987; Bingham et al., 1988; Bandziulis et al., 1989; Smith et al., 1989a; McKeown, 1990; Maniatis, 1991), yet little is known about the mechanisms in-

128 volved in the regulation of alternative splicing pathways. With the exception of the location of the branchpoints and polypyrimidine tracts (Helfman and Ricci, 1989; Smith and Nadal-Ginard, 1989; Goux-Pelletan et al., 1990; Helfman et al., 1990) no major differences in conserved sequence elements have been observed between constitutive and alternatively spliced transcripts. Specific transoacting factors are essential for tissuespecific alternative splicing of a single transcript. Breitbart and Nadal-Ginard (1987) have shown that transacting factors required for correct alternative splicing are induced during myogenesis. A striking example of trans-acting factor involvement in alternative splicing occurs in Drosophila where sex-specific trans-acting factors regulate the sex determination pathway (Bandziulis et al., 1989; Hodgkin, 1989; Mattaj, 1989; McKeown, 1990; Maniatis, 1991). Recent reports indicate that cell-specific differences in the abundance or activity of general splicing factors (Ge and Manley, 1990; Krainer et al., 1990a,b, 1991; Ge et al., 1991; Harper and Manley, 1991) may play a role in alternative splicing as well. Our laboratory has taken two parallel and complementary approaches to studying alternative splicing of MHC transcripts: in vivo and in vitro analyses. Using the in vivo approach we have identified and characterized mutations that affect splicing of the mutually exclusive exons (Collier et al., 1990; Kronert et al., 1991). Our in vivo analysis also involves characterization of spliced products generated from P element minigenes containing the exon 17 to exon 19 region (Hess and Bernstein, 1991). The in vitro analysis reported in this communication provides a relatively rapid method for identifying cis-acting elements involved in the regulated splicing of exon 18. We have opted to use homologous (Drosophila) Kc cell nuclear extracts rather than heterologous (human) HeLa extracts in order to mimic as closely as possible the in vivo mechanisms. Although successful processing of an alternatively spliced transcript has not previously been reported in the Drosophila in vitro system, we find that precursor RNAs prepared from the alternatively spliced 3' end of the MHC gene are processed in the embryonic/larval mode (exon 18 is skipped). Insertion of a constitutively used 3' splice site (including the branchpoint) into the junction of intron 17 and exon 18 results in the ligation of exon 17 to exon 18 if the test transcripts lack the downstream intron. Skip splicing (17 to 19) is still observed when the downstream intron is present. However, when the 5' splice site of exon 18 is mutated to the consensus sequence in transcripts that also contain the improved 3' splice site, efficient splicing of all three exons is observed. Our results indicate that the MHC exon 18 splice junction signals themselves are key elements in recognition of this alternative exon. Since the constitutive Drosophila Kc cell splicing machinery does

not efficiently recognize the wild-type MHC splice sites, tissue-specific factors are likely required for their activation.

Results

Ideally we would like to compare in vitro splicing of MHC transcripts in muscle and non-muscle nuclear extracts. Unfortunately immortalized Drosophila myoblast cell lines are not available. We therefore tested nuclear extracts prepared from 0-12 h Drosophila embryos, Schneider 2 and Kc Yale Drosophila cells using a number of extraction procedures and splicing protocols. We encountered the most efficient splicing and the least difficulty with non-specific nuclease digestion using Kc cell extracts prepared by the procedure of Dignam et al. (1983). For splicing reactions a modification of the method by Rio (1988) worked best with Kc extracts.

Splicing of muscle MHC 3' end transcripts in Drosophila Kc cell extracts Muscle MHC transcripts are alternatively processed in vivo to include exon 18 in mRNA of thorax muscle (predominantly in the indirect flight muscle) but to exclude it in some other adult muscles as well as all embryonic and larval muscle (Bernstein et al., 1986; Kazzaz and Rozek, 1989). MHC minigene constructs, tested in vivo using P element-mediated transformation (Hess and Bernstein, 1991), indicated that only sequences located 3' to the HindlII site in exon 17 are required for tissue-specific alternative splicing. Therefore, to analyze the in vitro splicing of the 3' end of MHC transcripts we produced RNA from a transcription vector containing the region of the Drosophila muscle MHC gene beginning at the HindlII site in exon 17 and extending to the EcoRI site in exon 19. Fig. 1 diagrams the transcripts described in this communication. For all of the three-exon-containing transcripts, including those identified as wild type (WT), an internal deletion of 283 nts in exon 18 has been made. This leaves 150 nts at the 5' end and 67 nts at the 3' end of exon 18. We performed identical experiments (including gel isolation, PCR, and sequencing of spliced products) using transcripts containing all 500 nts of exon 18 with qualitatively identical results (data not shown). Interestingly, the decreased size of exon 18 significantly increases the efficiency of the splicing reaction without affecting the regulation, indicating that a large part of exon 18 is not required for proper regulation. Our initial analysis was performed on three-exoncontaining transcripts shown as MHC E17-18-19 WT in Fig. 1. RNA purified from splicing reactions at

129 E17-18-19 WT

various times after addition of the p r e - m R N A was analyzed (Fig. 2). Bands of the sizes expected for products and intermediates resulting from skip splicing of exon 17 to exon 19 are present in the + ATP reactions but missing in the zero time point and the 2 h, no ATP reactions. The products accumulate in a time-dependent manner and there is no evidence for spliced products or intermediates of the sizes expected for ligation of all three exons, even using various salt, magnesium, or ATP concentrations (data not shown). Our unpublished data confirmed the identities of the observed R N A species by the following criteria: MHC

4-

0.5 4-

1 4-

2 4"

I Time ATP

17-18-19 WT

368 MHC

I0 M

264

217

440

78

r-rr--1

17-18 WT

368

264

247 622

MHC

527

17-18 + Int 5

368

293

247

M H C 1 8 - 1 9 W3"

67 M H C 1 8 - 1 9 CAG

440

~ 8 1 67

~-] 440

404

78 • - UUU t o CAG c h a n g e

78

309 _ MHC 17-18-19 + Int5

368

293

217

440

368

264

217

440

78

368

293

Z17

440

78

368

264

217

78

293

217

78

-c171.1

78

[]

CAG

MHC 17-18-19 + l n t 5 CAG

MHC 17-XS-19

~

A Int 18

MHC 1 7 - 1 8 - 1 9 A Int 18 + lnt 5

~

G 368

Mac

17-1s-19

A Int 17

A l n t 1 7 CAG

I 17

I IS

368

217

368

217

I

440

L__._I 78

440

78

Fig. 1. Diagram of Drosophila melanogaster M H C 3' end p r e - m R N A s analyzed by in vitro splicing reactions. All of the three-exon-containing transcripts have a 283 nt deletion within exon 18. M H C E 1 7 - 1 8 19 A Int 17 and M H C E 1 7 - 1 8 - 1 9 A Int 18 are genomic-cDNA hybrids lacking introns 17 and 18, respectively. The sequence from the region immediately upstream of M H C exon 6 that was inserted at the exact intron 1 7 / e x o n 18 junction is shown in the + I n t 5 transcripts. T h e asterisk indicates that the terminal three nts of exon 18 were changed from U U U to C A G to create a consensus 5' splice site. N u m b e r s below the exons and introns indicate the size in nucleotides.

Fig. 2. Time course of splicing of M H C E 1 7 - 1 8 - 1 9 W T transcripts. Splicing reactions were performed using Kc cell nuclear extracts and in vitro synthesized 32p-labelled p r e - m R N A s diagrammed in Fig. 1 as M H C 1 7 - 1 8 - 1 9 WT. R N A purified from the reactions at the times shown (in h), that had been incubated in the presence ( + ) or absence ( - ) of ATP, was analyzed on a 5% polyacrylamide, 8 M urea gel. Lane M contains 32p end-labelled Mspl fragments of pBR322 as size markers.

(1) bands diagrammed as intermediate and product lariats were identified by their anomalous migration; their true molecular masses were confirmed by electrophoresis after gel isolation and debranching (Ruskin and Green, 1985); (2) the ligated product of splicing of exon 17 to exon 19 was identified by reverse transcription and PCR polymerization of the gel-isolated R N A using primers specific for exons 17 and 19. After determining that the single PCR product was the correct size for the ligated exons, the D N A was directly sequenced. This confirmed that the splice junctions used in vitro were identical to those used in vivo in embryonic and larval muscle. To determine the branchpoint(s) that mediate exon 17 to exon 19 splicing, R N A purified from the 2 h splicing reactions was subjected to primer extension branchpoint analysis (Rodriguez et al., 1984). Strong stops immediately 3' of adenosines at positions - 2 7 and - 3 3 (relative to the 3' splice site) were present in the + ATP but not in the - A T P reactions.

130 Both sites contain a 5 out of 5 nt match to the Drosophila consensus branchpoint sequence, C / o u A / G A C / u (Keller and Noon, 1985).

Kc cell nuclear extracts are unable to efficiently remove intron 17 To determine if exclusion of exon 18 from mature spliced products is due to the failure to remove intron 17, we performed splicing reactions on a two-exon transcript containing exons 17 and 18. A time course of the in vitro processing of M H C E17-18 W T (Fig. 1) is shown on the left side of Fig. 3. We observed A T P dependent lariats and extremely low levels of fully spliced products (identified by PCR) that accumulated late in the time course. We performed primer extension reactions on R N A purified from these time points as well as on R N A from a two h reaction ( + ATP) that had been treated with debranching enzyme. However, no primer extension stops indicative of branchpoints

could be detected, This is not surprising considering the low levels of intermediates observed in Fig. 3. The failure to efficiently utilize the 3' splice site sequence of exon 18 may be due to its very poor match to the Drosophila consensus P Y l I N C / T A G IG (Mount, 1982), since only 5 of the first 11 nts are pyrimidines. Additionally, the two possible branchpoints within the conserved distance of - 1 8 to - 4 0 nts from the 3' splice site (at positions - 2 5 and - 3 6 ) only match the consensus in 3 out of 5 nts. Unlike the 3' splice site of exon 18, the 3' splice site of exon 19 mediates efficient splicing to exon 17 (Fig. 2) and has very good pyrimidine content (10/11 nts = pyrimidines) in addition to having two perfect consensus branchpoint sequences at positions - 27 and - 33. We were interested in determining if improvement of the 3' splice site of exon 18 would permit efficient joining of exon 17 and 18 in vitro. Our unpublished results concerning in vitro splicing of transcripts from the 5' end of the M H C gene indicated that the branch-

E17-18 wild -type

l0 M

4-

E17-18

0.5

1

2

2

4-

4-

4-

-

I

I0 M

4-

+ Int 5

0.5

1

2

2

4-

4-

4-

=

I

Time ATP

-C-)

622 --

-[lr

]la ]

5 2 7 --

404 --

309 -

Fig. 3. Comparison of splicing of E17-18 transcripts containing wild-type or improved 3' splice sites. Splicing reaction time courses were performed on 32p-labelled pre-mRNAs (E17-18 WT and E17-18 + Int 5, diagrammed in Fig. 1) and the purified RNA was analyzed on a 5% polyacrylamide,8 M urea gel. Lanes labelled M contain end-labelled MspI fragments of pBR322.

131 point and 3' splice site of intron 5 mediated splicing of exon 5 to exon 6. Eight of eleven nucleotides preceding exon 6 are pyrimidines and there is a perfect branchpoint sequence beginning 26 nts upstream of the splice junction (see Fig. 1, + I n t 5 templates for the sequence). We thus tested the ability of the terminal 29 nts of intron 5 (containing the 3' splice site and branchpoint) to mediate splicing of exon 17 to exon 18 when placed at the exact intron 1 7 / e x o n 18 junction (Fig. 1, M H C 17-18 + Int 5). E17-18 + Int 5 transcripts were processed to generate mature products and intermediates of sizes expected for splicing of exon 17 to exon 18 (Fig. 3, right side). The identities of the lariat intermediate and lariat product were confirmed by a debranching assay performed on gel-isolated R N A (data not

shown). Reverse transcription and P C R analysis of the total R N A purified from each time point indicates that an A T P - d e p e n d e n t band of the correct size for ligated exon products appears in a time dependent manner (data not shown). Although we did not confirm the sequence of the splice junctions used in the ligated product, experiments discussed below confirm that intron 5 does mediate splicing of exon 17 to exon 18 using the correct 5' and 3' splice sites. We performed a primer extension analysis of R N A from all of the time points shown and on R N A from a 2 h + A T P reaction which had been debranched. A single primer extension stop was detected at position - 2 2 relative to the 3' splice site within the intron 5 insertion (data not shown). This sequence (underlined in Fig. 1) contains a 5 out of

E17-18-19 A Int 18 Wt

M1 D

E17-18-19 A Int 17

+ Int 5

+11-

Wt

,' ;e Ilo I

M1

+1

+ II.

m

CAG

I-

+1

I

-

C--3p

--

~

] Int 18

Int 17

ID]gmI]l

s22

-

1 17

1 18 I',91

-

1+'r

1 tel

527

404

ViT-1-~i

¸

~ii+~i!

~i~i!!~!~ii~!~!¸¸

~!i~

~'~' ~~ ~

i!~ii~i~ii!!ii~

!

~ i~

309

Fig. 4. Splicing of E17-18-19 A Int 18 and E17-18-19 /t Int 17 transcripts containing wild-type and mutant 3' or 5' splice sites. Pre-mRNAs containing all three terminal exons but only one intron (diagrammed in Fig. 1) were transcribed from hybrid constructs prepared by ligation of genomic and cDNA fragments and 2 h splicing reactions were performed. Lane M1 contains end-labelled MspI fragments of pBR322.

132 5 m a t c h to t h e Drosophila c o n s e n s u s b r a n c h p o i n t . N o stops w e r e d e t e c t e d in t h e i n t r o n 18 s e q u e n c e ups t r e a m f r o m t h e insertion. W e c o n c l u d e t h a t splicing o f exon 17 to exon 18 utilizes t h e i m p r o v e d b r a n c h p o i n t p r o v i d e d by the i n t r o n 5 insertion. In a similar series o f e x p e r i m e n t s , we d e t e r m i n e d if p r e - s p l i c i n g o f exon 18 to exon 19 p e r m i t t e d t h e j o i n i n g o f exon 17 to exon 18. T h e results f r o m splicing such g e n o m i c / c D N A hybrids a r e shown in Fig. 4. A l t h o u g h t h e r e is a w e a k i n t e r m e d i a t e b a n d (exon 17), t h e r e is no e v i d e n c e o f any spliced p r o d u c t s in t h e wild type A Int 18 reactions. A s in the t r u n c a t e d t r a n s c r i p t s d e s c r i b e d above, i m p r o v e m e n t o f t h e 3' splice site of exon 18 by i n s e r t i o n o f t h e i n t r o n 5 splice site a n d

branchpoint (+Int i n t r o n 17.

An improved 5' spfice site for exon 18 increases the efficiency of removal of intron 18 A splicing r e a c t i o n time c o u r s e i n d i c a t e s that wildtype t r a n s c r i p t s which lack exon 17 b u t c o n t a i n exon 18, i n t r o n 18 a n d exon 19 ( M H C E 1 8 - 1 9 W T in Fig. 1) a r e spliced at e x t r e m e l y low levels (Fig. 5). Since exon 19 has an excellent 3' splice site with m u l t i p l e c o n s e n sus b r a n c h p o i n t s we s u s p e c t e d t h a t the inefficient removal of i n t r o n 18 in t h e a b s e n c e o f any c o m p e t i n g u p s t r e a m 5' splice sites was d u e to the p o o r 5' splice

E18-19 WT

I M

E18-19

0

0.5

1

2

2

4-

4-

4.

4-

-

I

I M

5) results in efficient r e m o v a l o f

CAG

0

0.5

1

2

4-

4-

4-

4-

2 -

I

Time ATP

. C--~

622 527

- la~]----q 19

-

F

404

309

Fig. 5. Comparison of splicing of E18-19 transcripts containing wild-type or improved 5' splice sites. Time course splicing reactions were performed on E18-19 WT and E18-19 CAG transcripts (diagrammed in Fig. 1) and analyzed on a 6% acrylamide, 8 M urea gel. The terminal three nucleotides in exon 18 are changed from UUU to CAG in E18-19 CAG transcripts. The Kc cell nuclear extract used for these reactions differs from the one used for all other reactions discussed in this communication and contains a large amount of debranching activity that generates the linear intron intermediate. The ligated exon product is too small to detect on this gel but has been confirmed by PCR. Lane M contains end-labelled Mspl fragments of pBR322.

133 site of exon 18. We therefore used in vitro mutagenesis to change the 5' splice site of exon 18 from U U U IG U A A G U to the consensus CAG LG U A A G U (Mount, 1982). A splicing reaction time course on these mutant transcripts indicates that intron 18 removal was very efficient, as shown by the presence of the lariat product which accumulated in a time-dependent manner (Fig. 5). The ligated exons are too small to be detected on this gel, but the presence of this product has been verified by PCR using primers specific for exons 18 and 19 (data not shown). The nuclear extract used for this experiment contains high levels of debranching activity, resulting in a large amount of linearized intermediate lariat. This band as well as the lariat intermediate were identified by debranching assays performed on the gel-isolated bands. The lariat product was also debranched and migrated at approximately 440 nts relative to D N A size markers, close to the actual 446 nt size of intron 18 (debranching data not shown). When identical time course reactions were performed on these two transcripts (WT and CAG) using a different batch of Kc cell nuclear extract we found, to our surprise, that the wild-type transcripts also spliced efficiently (data not shown, but see Fig. 4, E 1 7 - 1 8 - 1 9 A Int 17 WT). We still observed a significant increase in spliced products with the 5' splice site mutation of exon 18, but the difference was not as striking as the example shown here. We suspect that these variations result from different levels or activities of factors that regulate 5' splice sites, such as S F 2 / A S F (Ge and Manley, 1990; Krainer et al., 1990a,b) or DSF (Harper and Manley, 1991), present in the different extracts (see Discussion). We identified the branchpoint utilized in these reactions using a 32p-labelled primer complementary to exon 19 and reverse transcription of unlabelled R N A purified from reaction time points and debranched material (data not shown). In contrast to the skip splicing of exon 17 to 19, only a single branchpoint at position - 2 7 was used in both the wild-type and mutant 5' splice site reactions. As indicated above this is a perfect consensus branchpoint sequence. We also tested whether the intron downstream of exon 18 can be efficiently removed in transcripts that contain exon 17 pre-joined to exon 18 (Fig. 4). The A Int 17 reactions clearly demonstrate that intron 18 can be spliced out, even in the wild-type transcripts (WT), as shown by the presence of the intron 18 lariat and the ligated exon products. As discussed above this indicates that the poor 5' splice site of exon 18 is functional in the absence of a competing upstream 5' splice site. However, the amounts of both products are substantially increased in the splice site mutant transcript (CAG), and the efficiency of this reaction can vary with the particular splicing extract used. The iden-

tity of the final spliced product was confirmed by PCR (data not shown). Inclusion of exon 18 in mature transcripts in vitro is dependent upon a constitutively utilized 3' splice site as well as a consensus 5' splice site; neither modification alone is sufficient to induce removal of both introns Since improvement of the 3' and 5' splice sites of exon 18 mediated increased removal of intron 17 or intron 18 respectively in single-intron-containing transcripts, we investigated the ability of these changes to mediate removal of both introns simultaneously in three-exon-containing transcripts (E17-18-19, Fig. 6A). In-vitro splicing reactions were performed for 2 h on transcripts containing wild type (WT) 3' and 5' splice sites, a 3' splice site from intron 5 ( + Int 5), a modified 5' splice site (CAG), or both the modified 5' splice site and the intron 5 insertion ( + Int 5 CAG). As previously shown in the time course in Fig. 2, the major product generated from the wild-type transcript is that resulting from splicing of exon 17 to exon 19 (skip splicing), as indicated by the large amount of lariat and ligated exon products. Insertion of the branchpoint and splice site of intron 5 in front of exon 18 results in a decrease in the amount of both products of the skip splicing reaction and the appearance of a weak lariat migrating just slower than the precursor R N A (Fig. 6A, + Int 5). Fully spliced product of the correct size for splicing of all three exons is also detected but at extremely low levels. It is interesting that ligation of all three exons is so inefficient in these transcripts containing an improved 3' splice site since the same nuclear extract that mediated wild-type splicing of exon 18 to exon 19 in two-exon-containing transcripts was used for this experiment. Thus the intron 5 insertion at the 3' splice site of exon 18 cannot efficiently out-compete the 3' splice site of exon 19 in the presence of the weak 5' splice site of exon 18. The improved 5' splice site of exon 18 also causes a decrease in the amounts of the skip splicing products and intermediates (Fig. 6A, E 1 7 - 1 8 - 1 9 CAG). In addition, lariat intermediate and product bands specific for intron 18 removal are now detected. These bands have been identified by their co-migration with lariat bands from E18-19 (Fig. 5) and E 1 7 - 1 8 - 1 9 A Int 17 (Fig. 4) splicing reactions which are necessarily specific for intron 18 removal since intron 17 is not present in these transcripts (data not shown). Again, fully spliced product including all three exons is not detected, presumably due to the inability to efficiently remove intron 17 even in the presence of a perfect 5' splice site for exon 18. The improvement of both the 3' and 5' splice sites of

134 exon 18 by the i n t r o n 5 i n s e r t i o n as well as t h e 5' splice site m u t a t i o n ( + Int 5 C A G ) results in p r o d u c t i o n o f large a m o u n t s o f an R N A species o f the c o r r e c t size for ligation o f all t h r e e exons. N o t e the large i n c r e a s e in the i n t r o n 17 l a r i a t p r o d u c t c o m p a r e d to t h e reaction in which only the 3' splice site is i m p r o v e d (comp a r e + Int 5 C A G with + Int 5 in Fig. 6A). Likewise, the i n t r o n 18 l a r i a t p r o d u c t is also i n c r e a s e d over t h a t seen with just t h e i m p r o v e d 5' splice site for exon 18 ( c o m p a r e + I n t 5 C A G with C A G in Fig. 6A).

T o c o n f i r m t h e i d e n t i t y o f the l i g a t e d exon p r o d u c t s , reverse transcription and PCR of RNA purified from t h e s e r e a c t i o n s was p e r f o r m e d using p r i m e r s specific to exon 17 a n d exon 19 ( d a t a n o t shown). T h e s e p r i m e r s s h o u l d d e t e c t p r o d u c t s resulting f r o m exon 17 to exon 19 splicing as well as exon 17, 18, 19 ligation. O n l y 20 P C R cycles o f 1 min at 94°C, 2 min at 50°C a n d 2 min at 72°C w e r e p e r f o r m e d in o r d e r to e n s u r e that t h e p o l y m e r i z a t i o n r e a c t i o n s w e r e t e r m i n a t e d while still in t h e e x p o n e n t i a l range. A s e x p e c t e d we d e t e c t e d P C R

E17-18-19

A Wt

M1

I m

+ Int 5

+11-

+ Int 5 CAG

+11-

+ll-

CAG

M2

M1

+1

l n t 18

Int 17 1196

-

1021

-

886

-

622

--

527

--

1+7 I J" 1191 117 I r8 I

F~7 I+~l 404

--

309

-

Fig. 6. In-vitro splicing of E17-18-19 transcripts containing wild-type or mutant 5' and 3' splice sites of exon 18. (A) Two hour splicing reactions were performed in the presence ( + ) or absence ( - ) of ATP on El7-18-19 WT, + Int 5, CAG, or + Int 5 CAG transcripts diagrammed in Fig. 1. RNA purified from these reactions was analyzed on a 5% acrylamide, 8 M urea gel. Lane MI contains end-labelled MspI fragments of pBR322 and lane M2 contains in vitro synthesized RNA size markers. (B) A time course of in vitro splicing of E17-18-19 Int 5 CAG transcripts demonstrates that the product of ligation of all three exons accumulates in a time dependent manner with no evidence of exon skipping. Lane M contains end-labelled MspI fragments of pBR322.

135

E17-18-19

E17-18-19

WT

I n t 5 CAG

2 4-

!1

2 -

0 4-

0.5 4-

1 4-

2 4-

2 -

I M

Time ATP

622

11+ 118119 I

527

404

r,,-q +/

YI(!

i+?+i~?

+i+!!,i i!!!!~+!(? iiii/ii

+~i+~ ~ ~!i ~

+~i~i~ ~ ~

+++

++~ ~+++i¸+++++++~+++ ¸+ +++++! ++¸¸+i¸++¸¸i

~ ~Y

~+i!~

~

~~i

i~=~

++++ iii++i++

309

Fig. 6 (continued).

products of the correct size for exon 17 to exon 19 splicing in the wild type, + Int 5, and CAG reactions but extremely low levels of this product in reactions containing both improved exon 18 splice sites ( + Int 5 CAG). Likewise, the double mutant reactions contained a product of the correct size for splicing together of all three exons but there was no evidence of this product in the other three reactions. To confirm that the correct splice junctions were utilized in the removal of both introns, the exon 17-18-19 spliced product from the + Int 5 CAG reaction was gel purified, reverse transcribed and amplified by PCR using

primers specific for exons 17 and 19; direct sequencing of the PCR products was performed (data not shown). By this means we determined that the 5' and 3' splice junction used in vivo in indirect flight muscle of the adult fly were also used in vitro. These experiments confirm that improvement of both exon 18 splice sites is necessary and sufficient for removal of both introns and for inhibition of skip splicing utilizing the Kc cell splicing machinery. To analyze the splicing pattern of the double mutant transcript further, we performed a splicing time course using E 1 7 - 1 8 - 1 9 + Int 5 CAG transcripts (Fig. 6B). It

136 is clear that the intron 18 lariat and ligated exon products resulting from ligation of all three exons accumulate in a time and ATP dependent manner. The intron 17 lariat product is not resolved from the precursor R N A on this gel. There appears to be a complete switch from skip splicing seen in wild-type transcripts to exon 18 inclusion, since products of the skip splicing reaction are totally eliminated in the double mutant.

Discussion

Kc cell extracts mimic the splicing pattern seen in vivo in embryonic and larval tissues The first aim of this study was to determine if Kc cell nuclear extracts splice the 3' end of Drosophila MHC transcripts in the embryonic/larval mode, the thorax-specific mode, or both. Our results indicate that Kc cell extracts, like larval muscles, skip exon 18 and splice exon 17 to exon 19. The skip splicing observed in Kc cell extracts is extremely efficient since there is no evidence of intermediates or products diagnostic of the alternative splicing pattern (ligation of all three exons). Splicing of exon 18 is also stringently regulated in vivo, where this exon is completely excluded in embryonic and larval muscles, but is always included in indirect flight muscles (Bernstein et al., 1986; Kazzaz and Rozek, 1989). Since exon 18 encodes an alternative C-terminus of the MHC protein, mis-expression of this exon might result in muscle containing ultrastructural or contractile defects. Some of the phenotypic differences between larval muscles and indirect flight muscles may be mediated by the alternative C-termini of the MHC molecules (indirect flight muscles have a more highly organized myofibrillar pattern than larval muscles and contract at greater frequency; see O'Donnell and Bernstein (1988) and references therein). Thus stringent control of exon 18 exclusion, as observed in the Kc cell extracts, is likely to be biologically relevant.

Intron 18 removal is activated by improvement of the 5' splice site of exon 18 The 5' splice site of exon 18 conforms exactly with the G U A A G U intron consensus sequence (Mount, 1982). However the terminal region of exon 18 differs from the consensus (UUU vs. C/AAG) at all three positions. This might lead to a poor ability to compete with the 5' splice site of exon 17 which is GCG [ G U A A G U (see Lear et al. (1990) for a discussion of 5' splice site competition). If this were the case, we would expect exon 18 to splice to exon 19 in the absence of the upstream 5' splice site as in MHC E18-19 and MHC E 1 7 - 1 8 - 1 9 A Int 17 transcripts.

Only a few of the Kc nuclear extracts we tested were able to efficiently remove intron 18 from these transcripts. This variability may reflect differences in the levels or activities of essential splicing factors such as ASF (also known as SF2) that activate use of proximal 5' splice sites (Ge and Manley, 1990; Krainer et al., 1990a,b) or DSF (Harper and Manley, 1991) which activates use of distal 5' splice sites. In any case, it is clear from the analysis of the wild-type transcripts that the 5' splice site of exon 17 has much greater affinity for the 3' splice site of exon 19 than does the 5' splice site of exon 18. A cooperative interaction between specific pairs of 5' and 3' splice sites has also been shown to regulate the alternative splicing of myosin light chain 1 / 3 (MLC 1/3) pre-mRNAs (Gallego and Nadal-Ginard, 1990). A major difference in the two systems is that MLC 1 / 3 alternative splicing is not tissue-specific, indicating that the affinity of the various splice sites is determined by the cis-acting elements alone whereas splicing of MHC transcripts requires a tissue-specific switch in affinities between 5' and 3' splice site of specific exon pairs, presumably mediated by tram-acting factors. Mutating the 5' splice site of exon 18 to improve its U1 snRNA binding potential enhances the ability of exon 18 to splice to exon 19, especially in extracts that were deficient in the ability to splice the wild-type transcripts (see Fig. 5). In pre-mRNAs containing the mutant 5' splice site in competition with the upstream 5' splice site ( E 1 7 - 1 8 - 1 9 CAG) intron 18 is removed fairly efficiently. However, this does not lead to removal of the upstream intron, likely because splicing factors present in Kc cell extracts are unable to interact efficiently with the 3' splice site of exon 18. This is in sharp contrast to results obtained for alternative splicing of rat preprotachykinin transcripts in vitro (Nasim et al., 1990; Kuo et al., 1991). Conversion of the 5' splice site of the internal (skipped) exon of preprotachykinin transcripts to consensus resulted in inclusion of the internal exon in mature mRNA, with complete removal of both introns and total inhibition of skip splicing. Differences in the two systems may reflect the need for more stringent regulation of both splice sites of exon 18 to ensure that exon 18 is always excluded in embryonic/larval tissue; in contrast, preprotachykinin transcripts are frequently spliced in both modes in the same tissue.

Role of the 3' splice site of exon 18 in exon skipping The efficient skip splicing of MHC exon 17 to exon 19 is aided by the excellent match of the 3' splice site and branchpoints of exon 19 with the Drosophila consensus sequences (Mount, 1982; Keller and Noon, 1985). The 3' splice site and branchpoint sequences immediately upstream of the skipped MHC exon 18

137 contain a poor match to the Drosophila consensus sequences (Bernstein et al., 1986; Rozek and Davidson, 1986). It is therefore not likely that sequences at the 3' splice site of exon 18 would mediate efficient binding of U2 snRNA to the branchpoint in the absence of tissue-specific activators. Similar suboptimal processing signals play a role in maintaining the correct ratio of avian sarcoma virus spliced and unspliced genomic R N A (Katz and Skalka, 1990; Fu et al., 1991). Interestingly, the intron preceding M H C exon 18 contains a potential branchpoint sequence at position - 6 1 relative to the 3' splice site that is followed by a polypyrimidine tract of 20 nts. A number of genes that have alternatively spliced transcripts (Helfman and Ricci, 1989; Smith and Nadal-Ginard, 1989; Smith et al., 1989b; Goux-Pelletan et al., 1990; Helfman et al., 1990; Guo et al., 1991) utilize branchpoints that are far removed from the 3' splice site. Unfortunately, we have not been able to determine the functional intron 17 branchpoint since exon 17 does not splice efficiently to exon 18 in the wild-type transcripts in vitro. We tested the possibility that skipping of exon 18 in vitro is due solely to competition between the poor 3' splice site of exon 18 and the strong 3' splice site of exon 19 for limited splicing factors. This model predicts that exon 17 should be spliced to exon 18 in transcripts terminated in exon 18 (MHC E17-E18) or in transcripts in which intron 18 has been removed (MHC E 1 7 - 1 8 - 1 9 zl Int 18). As shown in Figs. 3 and 4, splicing of exon 17 to 18 is very inefficient. This contrasts with the results seen for tropomyosin transcript splicing in H e L a cell extracts in which Helfman et al. (1988) found that ligation of an alternative exon of a tropomyosin transcript to the downstream exon promoted splicing of the alternative exon to an upstream exon. Although our results do not eliminate the possibility of competition between 3' splice sites in two-intron-containing transcripts, they clearly indicate that mechanisms other than competition must also play a role in exon 18 exclusion. The remaining models for exon 18 alternative splicing hinge upon the presence of positive (activator in adult thorax muscle) or negative (inhibitor in embryonic and larval muscle) splicing factors. Most instances of alternative splicing that have been well characterized to date involve negative regulation (for reviews also see McKeown, 1990; Maniatis, 1991; Sosnowski et al., 1989; Streuli and Saito, 1989; Inoue et al., 1990; Siebel and Rio, 1990; Chain et al., 1991; Tseng et al., 1991). It is possible that M H C exon 18 splice junctions are blocked by binding of factors present in Kc nuclear extracts (and presumably embryos and larvae) and that thorax-specific splicing factors remove this inhibition or that the inhibitors are not present or are inactive in thorax muscle. Another possibility is that the unusual 3' splice site

of exon 18 (poor polypyrimidine tract and terminal AG) cannot be efficiently recognized by the housekeeping splicing factors found in Kc cells. This hypothesis implies that thorax muscle must contain splicing factors that are capable of recognizing or activating this defective splice site. Deletion of this element should inhibit thorax-specific splicing in vivo if the tissue-specific factor directly interacts with the 3' splice site. However, MHC 3' end transcripts, expressed in vivo from a minigene construct containing a deletion of 12 nts upstream of the 3' splice of exon 18, were spliced in the correct tissue-specific manner in adult flies (Hess and Bernstein, 1991) indicating that the specific splice junction sequence is not absolutely required for recognition and inclusion of exon 18 in mature transcripts. Another explanation for exon 18 exclusion is that tissue-specific splicing factors may activate the 3' splice site of exon 18 via binding to adjacent or even distant sites within the primary transcript. This has been proposed for the activation of the female-specific 3' splice site of Drosophila doublesex (dsx) transcripts (Nagoshi and Baker, 1990; Hedley and Maniatis, 1991; Hoshijima et al., 1991; Ryner and Baker, 1991). Utilization of the dsx female-specific exon is dependent upon positive regulation mediated by tra and tra-2 binding greater than 200 nts downstream from the 3' splice site of the female-specific exon. The female-specific exon, like exon 18, is preceded by a purine tract, rather than a pyrimidine-rich sequence, that would not likely efficiently bind splicing factors such as U2AF (Ruskin et al., 1988; Zamore and Green, 1989, 1991) and polypyrimidine tract binding proteins (Garcia-Blanco et al., 1989; Wang and Pederson, 1990; Gil et al., 1991; Patton et al., 1991), which are known to mediate 3' splice site recognition. The authors suggest that tra-2 binding downstream may aid in the activation of this poor 3' splice site either by interaction with splicing factors bound at both ends of the exon or by alteration of secondary structure in such a manner as to make the 3' splice site more accessible to the splicing machinery. All of our data, including the splicing of exon 17 to 18 mediated by the intron 5 insertion, are consistent with a similar positive activation model for exon 18 inclusion. Since it is likely that the non-consensus splice sites that flank exon 18 would not be recognized by the constitutive splicing apparatus, we favor this model.

Recognition of the 3' splice site of exon 18 requires the presence of a constitutive splice junction, but this junction is not used efficiently when the downstream intron is present A better 3' splice site, although required, is not sufficient for efficient inclusion of exon 18 in mature transcripts in vitro. Splicing of MHC E 1 7 - 1 8 - 1 9 + Int

138 5 transcripts results predominantly in joining of exon 17 to exon 19, although at lower levels than in the wild-type transcripts. The most likely explanation for this is that the 3' splice site of exon 19 out-competes the modified 3' splice site of exon 18. Hess and Bernstein (1991) found similar results in vivo by P elementmediated transformation with an MHC minigene. Substitution of the 3' splice site of exon 18 by the same constitutive branchpoint and splice junction used in this study did not alter the normal larval pattern of exon 18 skipping. Whatever the mechanism, in vivo activation of splicing of exon 17 to exon 18 in the adult thorax would likely result in default removal of intron 18 since splicing of exon 18 to 19 occurs in the absence of competing upstream splice sites, although with varying efficiencies in different nuclear extracts. In conclusion we have determined that the failure to include exon 18 in mature transcripts in vitro is due to inability of the constitutive splicing machinery in Kc cell extracts to efficiently recognize the 3' splice site of exon 18, as well as the strong affinity between exon 17 and exon 19 splice junctions. In contrast to several other alternative splicing systems (Katz and Skalka, 1990; Nasim et al., 1990; Fu et al., 1991; Kuo et al., 1991) which splice transcripts in both alternative modes, the Drosophila Kc cell extracts always skip exon 18, and may therefore serve as a useful system for supplementation with muscle cell extracts in an attempt to identify trans-acting splicing factors. The in vitro system is also useful for screening mutant MHC transcripts prior to their analysis in vivo by the more time-consuming process of P element-mediated transformation.

Experimental Procedures Plasmid constructions pSP64/MHC 3' contains a 4.5 kb HindlII-HindlII fragment from the MHC gene which begins at the HindlII site in exon 17 and extends another 2 kb past the second poly (A) site (Bernstein et al., 1986). The HindlII-EcoRI, HindlII-Pst I, and Pst I-EcoRI fragments were gel isolated from this plasmid and inserted into the same sites in the pBS vector of Stratagene Inc. (La Jolla, CA) to create plasmids pBS/MHC HIII-RI from which E17-18-19 transcripts were generated, pBS/MHC HIII-Pst for generation of E17-18 transcripts and pBS/MHC Pst-RI for preparation of E1819 pre-mRNA. To create plasmid pKS/AcDNA, a HindlII-ClaI fragment from plasmid 90-pBS1A (subcloned from pCH1A (Hansen, 1986)) was subcloned into the pKS vector (Stratagene). pBS/MHC E17-1819 A Int 18 was created by inserting the PstI-EcoRI fragment from p K S / A c D N A into the same sites of pBS/MHC HIII-Pst. Likewise the HindlII-PstI frag-

ment from pKS/AcDNA was cloned into pBS/MHC Pst-RI digested with HindlII and PstI to generate pBS/MHC E17-18-19 A Int 17. Plasmid p K S / M H C 5' was prepared by inserting the SalI-XbaI fragment from the 5' end of the Drosophila muscle MHC gene (Wassenberg et al., 1987) into the same sites in the pKS vector. All intron 5 insertion constructs were prepared by insertion of the 29 nt BamHI-BgllI fragment from pKS/MHC 5' into the BgllI site at the junction of intron 17 and exon 18 in each construct. All of the constructs shown in Fig. 1 contain a deletion within exon 18 prepared by digestion with NsiI and PstI, treatment with T4 DNA polymerase to blunt the ends, and ligation. The 5' splice site mutation in exon 18 was generated using the Altered Sites in vitro mutagenesis system kit (Promega, Madison, WI) according to the manufacturer's instructions. Clones containing the desired deletions or mutations were screened by restriction enzyme analysis and the exact deletions determined by dideoxy sequencing using a Sequenase kit and the procedure provided by the manufacturer (United States Biochemical Corp., Cleveland, OH).

In vitro transcription and splicing Precursor RNAs were synthesized in 30/zl reactions containing 0.5-1.0/zg DNA template, 40 mM Tris-HC1 (pH 8.0), 8 mM MgC12, 50 mM NaC1, 2 mM spermidine, 30 mM dithiothreitol, 1 unit Inhibit-ACE (5'-3', Inc., West Chester, PA), 500 /zM each of ATP and CTP, 200/zM of unlabelled UTP and GTP, 2/zM [32p] UTP and GTP (20/zCi, 400 Ci/mMol), and 50 units of T3 or T7 RNA polymerase or 20 units of SP6 RNA polymerase. Reactions were performed for 75 min at room temperature followed by DNase digestion, two phenol-chloroform extractions and two ethanol precipitations. These conditions generated RNA with a specific activity of ~ 1 x 107 cpm//zg. Nuclear extracts (Dignam et al., 1983) were prepared from Kc Yale cells grown in serum-flee Echalier's D-22 medium (Whittaker M A Bioproducts, Walkersville, MD) to a density of 2.5 x 106 cells/ml. Typically ~ 3.6 x 109 cells yielded a total of 30 mg of protein with a concentration of 8-10 mg/ml. In vitro splicing (modified from Rio, 1988) was carried out in 25 /.d reactions containing 50-60% nuclear extract, 1 unit Inhibit-ACE, 3% polyvinyl alcohol, 6.0 mM creatine phosphate, 24 mM Hepes pH 7.6 (in addition to that provided by the nuclear extract), 1.4 mM MgC12, 3.0 mM ATP and ~ 10 ng of precursor RNA (50,000-100,000 cpm). After incubating at 22°C for the given times, the RNA was purified by proteinase K digestion at 30°C for 0.5 h, phenol-chloroform extraction and ethanol precipitation. Spliced products were analyzed on 4%-6% polyacrylamide (19:1 acrylamide/bisacrylamide), 8 M urea gels.

139

RNA analysis

Acknowledgments

R N A was gel isolated by excision of bands from 4 - 6 % polyacrylamide, 8 M urea gels and elution overnight in 0.5 M C H 3 C O 2 N H 4 , 10 m M (CH 3 CO2)2Mg • 4H20, 1 m M E D T A , 0,1% SDS at 22°C. Kc nuclear extracts were used to perform debranching assays (Ruskin and Green, 1985) on R N A purified from total splicing reactions or on gel-purified RNA. As seen in Results, some of these extracts contain debranching activity. Primer extension analyses (McKnight and Kingsbury, 1982) of total spliced products or gel-isolated R N A species were performed using 5' end-labelled (Maniatis et al., 1982) oligonucleotides with the sequence 5 ' G G T C G A A T C T T G G T G G G A A G G C C 3' which is complementary to exon 19 or an oligonucleotide with the sequence 5' C G A T G G T G A T G C C T C A G G C 3' which is complementary to exon 18. Reverse transcription sequencing (Bektesh et al., 1988) of gel-isolated products was performed using the above oligonucleotide complementary to exon 19. Reverse transcription for P C R analysis was p e r f o r m e d at 42°C for one h in 20 /zl reactions containing 50 m M Tris-HCl, p H 8.3, 6 m M MgC12, 40 m M KCI, 1 m M dGTP, 1 m M dATP, 1 m M TTP, 1 m M dCTP, 1 /xl Inhibit-ACE, 50 pmol of the negative strand oligonucleotide (shown above), a portion of the gel-isolated R N A or total R N A from splicing reactions, and 10-20 units of A M V reverse transcriptase (Life Sciences, St. Petersburg, FL). The reaction was then heated at 95°C for 10 min and the following components added: 8 / x l of 10 x P C R buffer (10 × = 100 m M Tris HC1, p H 8.3, 500 m M KCI, 0.1% gelatin), 0.8 /zl of 80 m M MgC12, 3.2/xg of G e n e 32 protein (Pharmacia, Piscataway, N J) or 1 unit of Perfect Match D N A Polymerase Enhancer (Stratagene), 50 pmol of the upstream oligonucleotide (Exon 17 specific = 5' G C G C C G A T C T G G C C G A G C A G 3' or Exon 18 specific = 5' G G C C G C A A G A G C G C G C T G C 3') and 1 unit of either Amplitaq D N A Polymerase (Cetus, Emeryville, CA) or 5 units of Replinase (Dupont, N E N Research Products, Boston, MA) in a total volume of 80/xl. After 20 to 30 cycles of 1 min at 94°C, 1 min at 45°C and 2 min at 65°C, the products were chloroform extracted and analyzed on 8% non-denaturing polyacrylamide gels. For some experiments P C R reactions were performed using 20-30 cycles of 1 min at 94°C, 2 rain at 50°C, and 2 min at 72°C. For sequencing of gel-isolated R N A we first confirmed that a single P C R product was produced and then unincorporated nucleotides and primers were removed by centrifugation through Centricon 30 microconcentrators (Amicon, Beverly, MA) according to the manufacturer's recommendations. The D N A was then directly sequenced using a Sequenase kit.

We are especially grateful to Norbert Hess for his encouragement and helpful discussions during the early stages of this project and to Kelleen Icuss for excellent technical assistance. We thank Don Rio for providing his splicing protocol prior to publication. We also wish to thank Bobbi Johnson, Gerald Rubin, Jim Donady, Barbara Hamkalo, Jacques Perrault, Ron Emeson and Geoff Rosenfeld for providing Kc, Schneider and H e L a cell lines, Michael Breindl for use of his tissue culture facilities, and William Stumph, Mike McKeown, Phil Singer and Richard Cripps for helpful criticism of the manuscript. This project was supported by National Institutes of Health research grant GM32443 and an Established Investigatorship to S.I.B. from the American H e a r t Association.

References Bandziulis, R.J., Swanson, M.S. and Dreyfuss, G. (1989) Genes Dev. 3, 431-437. Bektesh, S., Van Doren, K. and Hirsh, D. (1988) Genes Dev. 2, 1277-1283. Bernstein, S.I., Mogami, K., Donady, J.J. and Emerson, Jr., C.P. (1983) Nature 302, 393-397. Bernstein, S.I., Hansen, C.J., Becker, K.D., Wassenberg II, D.R., Roche, E,S., Donady, J.J. and Emerson, Jr., C.P. (1986) Mol. Cell. Biol. 6, 2511-2519. Bingham, P.M., Chou, T.-B., Mims, I. and Zachar, Z. (1988) Trends Genet. 4, 134-138. Breitbart, R.E., Andreadis, A. and Nadal-Ginard, B. (1987) Annu. Rev. Biochem. 56, 467-495. Breitbart, R.E. and Nadal-Ginard, B. (1987) Cell 49, 793-803. Chain, A.C., Zollman, S., Tseng, J.C. and Laski, F.A. (1991) Mol. Cell. Biol. 11, 1538-1546. Collier, V.L., Kronert, W.A., O'Donnell, P.T., Edwards, K.A. and Bernstein, S.I. (1990) Genes Dev. 4, 885-895. Dignam, J.D., Lebovitz, R.M. and Roeder, R.G. (1983) Nucleic Acids Res. 11, 1475-1488. Emerson, Jr., C.P. and Bernstein, S.I. (1987) Annu. Rev. Biochem. 56, 695-726. Fu, X.-D., Katz, R.A., Skalka, A.M. and Maniatis, T. (1991) Genes Dev. 5, 211-220. Gallego, M.E. and Nadal-Ginard, B. (1990) Mol. Cell. Biol. 10, 2133-2144. Garcia-Blanco, M.A., Jamison, S.F. and Sharp, P.A. (1989) Genes Dev. 3, 1874-1886. Ge, H. and Manley, J.L. (1990) Cell 62, 25-34. Ge, H., Zuo, P. and Manley, J.L. (1991) Cell 66, 373-382. George, E.L, Ober, M.B. and Emerson, Jr., C.P. (1989) Mol. Cell. Biol. 9, 2957-2974. Gil, A., Sharp, P.A., Jamison, S.F. and Garcia-Blanco, M.A. (1991) Genes Dev. 5, 1224-1236. Goux-Pelletan, M., Libri, D., d'Aubenton-Carafa, Y., Fiszman, M., Brody, E. and Marie, J. (1990) EMBO J. 9, 241-249. Guo, W., Mulligan, G.J., Wormsley, S. and Helfman, D.M. (1991) Genes Dev. 5, 2096-2107. Hansen, C. (1986) Thesis, San Diego State University, San Diego, CA.

140 Harper, J.E. and Manley, J.L. (1991) Mol. Cell. Biol. 11, 5945-5953. Hedley, M.L. and Maniatis, T. (1991) Cell 65, 579-586. Helfman, D.M., Ricci, W.M. and Finn, L.A. (1988) Genes Dev., 2, 1627-1638. Helfman, D.M. and Ricci, W.M. (1989) Nucleic Acids Res. 17, 5633-5650. Helfman, D.M., Roscigno, R.F., Mulligan, G.J., Finn, L.A. and Weber, K.S. (1990) Genes Dev. 4, 98-110. Hess, N., Kronert, W.A. and Bernstein, S.I. (1989) In L.H. Kedes and F.E. Stockdale (eds.), Cellular and Molecular Biology of Muscle Development. Liss, New York, pp. 621-631. Hess, N.K. and Bernstein, S.I. (1991) Dev. Biol., 146, 339-344. Hodgkin, J. (1989) Cell 56, 905-906. Hoshijima, K., Inoue, K., Higuchi, I., Sakamoto, H. and Shimura, Y. (1991) Science 252, 833-836. Inoue, K., Hoshijima, K., Sakamoto, H. and Shimura, Y. (1990) Nature 344, 461-463. Katz, R.A, and Skalka, A.M, (1990) Mol. Cell. Biol. 10, 696-704. Kazzaz, J.A. and Rozek, C.E. (1989) Dev. Biol. 133, 550-561. Keller, E.B. and Noon, W.A. (1985) Nucleic Acids Res. 13, 49714981. Krainer, A.R., Conway, G.C. and Kozak, D. (1990a) Genes Dev. 4, 1158-1171. Krainer, A.R., Conway, G.C. and Kozak, D. (1990b) Cell 62, 35-42. Krainer, A.R., Mayeda, A., Kozak, D. and Binns, G. (1991) Cell 66, 383-394. Kronert, W.A., Edwards, K.A., Roche, E.S., Wells, L. and Bernstein, S.I. (1991) EMBO J. 10, 2479-2488. Kuo, H., Nasim, F.H. and Grabowski, P.J. (1991) Science 251, 1045-1050. Lear, A.L., Eperon, L.P., Wheatley, I.M. and Eperon, I.C. (1990) J. Mol. Biol. 211, 103-115. Left, S.E., Rosenfeld, M.G. and Evans, R.M. (1986) Annu. Rev. Biochem. 55, 1091-1117. Maniatis, T., Fritsch, E.F. and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Maniatis, T. (1991) Science, 251, 33-34. Mattaj, I.W. (1989) Cell 57, 1-3.

McKeown, M. (1990) Genet. Eng. 12, 139-181. McKnight, S.L. and Kingsbury, R. (1982) Science, 217, 316-324. Miller, A. (1950) In M. Demerec (ed.), Biology of Drosophila. Wiley, New York, NY, pp. 420-534. Mount, S.M. (1982) Nucleic Acids Res. 10, 459-472. Nagoshi, R.N. and Baker, B.S. (1990) Genes Dev. 4, 89-97. Nasim, F.H., Spears, P.A., Hoffmann, H.M., Kuo, H.K. and Grabowski, P.J. (1990) Genes Dev. 4, 1172-1184. O'Donnell, P.T. and Bernstein, S.I. (1988) J. Cell. Biol. 107, 26012612. Patton, J.G., Mayer, S.A., Tempst, P. and Nadal-Ginard, B. (1991) Genes Dev. 5, 1237-1251. Rio, D.C. (1988) Proc. Natl. Acad. Sci. USA 85, 2904-2908. Rodriguez, J.R., Pikielny, C.W. and Rosbash, M. (1984) Cell 39, 603-610. Rozek, C.E. and Davidson, N. (1983) Cell 32, 23-34. Rozek, C.E. and Davidson, N. (1986) Proc. Natl. Acad. Sci. USA 83, 2128-2132. Ruskin, B. and Green, M.R. (1985) Science, 229, 135-140. Ruskin, B., Zamore, P.D. and Green, M.R. (1988) Cell 52, 207-219. Ryner, L.C. and Baker, B.S. (1991) Genes Dev. 5, 2071-2085. Siebel, C.W. and Rio, D.C. (1990) Science 248, 1200-1208. Smith, C.W.J. and Nadal-Ginard, B. (1989) Cell 56, 749-758. Smith, C.W.J., Patton, J.G. and Nadal-Ginard, B. (1989a) Annu. Rev. Genet. 23, 527-577. Smith, C.W.J., Porro, E.B., Patton, J.G. and Nadal-Ginard, B. (1989b) Nature 342, 243-247. Sosnowski, B.A., Belote, J.M. and McKeown, M. (1989) Cell 58, 449-459. Streuli, M. and Saito, H. (1989) EMBO J. 8, 787-796. Tseng, J.C., Zollman, S., Chain, A.C. and Laski, F.A. (1991) Mech. Dev. 35, 65-72. Wang, J. and Pederson, T. (1990) Nucleic Acids Res. 18, 5995-6001. Wassenberg 1I, D.R., Kronert,W.A., O'Donnell, P.T. and Bernstein, S.I. (1987) J. Biol. Chem. 262, 10741-10747. Zamore, P.D. and Green, M.R. (1989) Proc. Natl. Acad. Sci. USA 86, 9243-9247. Zamore, P.D. and Green, M.R. (1991) EMBO J. 10, 207-214.