U2 U6 base-pairing interaction (U2 nt 1-11 with U6 nt 87-97) occurs during the spliceosome cycle (7-9).

Transcription

1 Proc. Natl. Acad. Sci. USA Vol. 90, pp , August 1993 iochemistry A base-pairing interaction between U2 and small nuclear RNAs occurs in >150S complexes in HeLa cell extracts: Implications for the spliceosome assembly pathway (small nuclear ribonucleoproteins/glycerol gradients/ser/arg-rich proteins/psoralen crosslinking) DAVID A. WASSARMANt AND JOAN A. STEITZ* Department of Molecular iophysics and iochemistry, Howard Hughes Medical Institute, Yale University School of Medicine, 295 Congress Avenue, New Haven, CT Contributed by Joan A. Steitz, April 14, 1993 ASTRACT In mammalian cells, base pairing between the U2 and small nuclear RNAs is required during pre-rna splicing. We show by psoralen crosslinking of HeLa nuclear extract that U2 base pairing occurs within abundant ribonucleoprotein complexes that sediment at >150 S in glycerol gradients. All of the spliceosomal RNAs (Ul, U2, U4, US, and ) cosediment with these large complexes, suggesting that they may be related to small nuclear RNA-containing structures called speckles/coiled bodies or snurposomes, which have been visualized in mammalian or amphibian nuclei, respectively. In contrast to nuclear extract, S100 extract, which is splicing-defective and lacks the >150S complexes, does not contain base-paired U2. However, U2 base pairs form in S100 extract that has been made splicing-competent by supplementation with Ser/Arg-rich (SR) proteins, ATP, and an adenovirus splicing substrate. During splicing in supplemented S100 extract, U2 base pairing precedes the appearance of splicing intermediates and occurs initially in an ==60S spliceosome complex but also in progressively larger ( S) complexes. Possible functional relationships between the 60S spliceosome and the >150S complexes are discussed. Pre-mRNA splicing occurs within a ribonucleoprotein (RNP) complex, the spliceosome, composed of four small nuclear RNPs (snrnps), containing Ul, U2, U4+, and US small nuclear RNAs (snrnas), and numerous non-snrnp protein factors (for review, see refs. 1 and 2). Assembly of these components onto the pre-mrna to form an active spliceosome occurs in a specific order that is directed in part by base pairing of snrnas to conserved pre-mrna sequences at the splice sites and the intron branch site (for review, see refs. 1 and 2) and by extensive snrna-snrna base-pairing interactions between the U4 and snrnas and the U2 and snrnas (3-9). In addition to snrnps, a family of evolutionarily conserved Ser/Arg-rich proteins (called SR proteins) are required for pre-mrna splicing (10). Individual SR proteins have been shown to function as general splicing factors and can complement splicing-deficient HeLa S100 extract (10-13). In vitro studies have shown that mammalian spliceosome assembly can be divided into three major steps (1). (i) Formation of a commitment complex that contains the Ul snrnp and requires either a 5' splice site or a 5' splice site and a branch site. (ii) ATP-dependent binding of the U2 snrnp at the branch site. (iii) Assembly of the U5 and U4 snrnps as a preformed tri-snrnp complex. U4- base pairing is destabilized after spliceosome assembly but prior to the first cleavage step of the splicing reaction (3, 14, 15). However, it has not been established when the essential The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C solely to indicate this fact U2 base-pairing interaction (U2 nt 1-11 with nt 87-97) occurs during the spliceosome cycle (7-9). Glycerol gradient fractionation of in vitro splicing reaction mixtures has been used to determine approximate sedimentation coefficients for spliceosomes and various assembly intermediates: spliceosomes (60S), U4 U5 snrnp (25S), U4 snrnp (1SS), U2 snrnp (17S), U5 snrnp (20S), and Ul, U4, or snrnp (1OS) (ref. 16 and references therein). In vivo, spliceosomal snrnps and non-snrnp splicing factors are found in structures that are much larger than a single spliceosome. In mammalian somatic cell nuclei, spliceosomal snrnps are concentrated in structures called interchromatin granules (speckles) and coiled bodies, which belong to a class of interchromatin structures called nuclear bodies (ref. 17; for review, see ref. 18). In the germinal vesicle of frogs and salamanders, spliceosomal snrnps are not only on actively transcribed lampbrush loops but also in granules called snurposomes that are times larger than spliceosomes (19). Large multi-snrnp complexes have not been described in in vitro splicing extracts, but numerous nuclear transcripts have been shown to be released from mammalian cell nuclei in complexes that contain all of the spliceosomal snrnps (for review, see ref. 20). Sperling and coworkers (21, 22) have identified large nuclear RNPs that sediment as a discrete peak at -'200 S on sucrose gradients, and others (23) have described complexes that sediment heterogeneously between 150 and 350 S. We have identified in HeLa nuclear extract abundant >150S complexes that contain base-paired U2' and other spliceosomal RNAs. ase-paired U2 is not present in HeLa S100 extract but forms when the extract is incubated with SR proteins, ATP, and a splicing substrate. The interaction occurs prior to the first cleavage-ligation step of splicing, appearing initially and primarily in an --60S complex but also in progressively larger ( S) complexes. These results suggest that the >150S complexes are normal intermediates in the spliceosome cycle. They may function in preassembly of spliceosomes, in ordered splicing of multiintron transcripts, or in postsplicing recycling of snrnps. MATERIALS AND METHODS Extract Preparation. HeLa nuclear and S100 extracts were prepared according to Dignam et al. (24) except that buffer D was 50 mm KCl. Glycerol Gradient Fractionation. A 10% (vol/vol) glycerol solution (10% glycerol/50 mm KCl/1 mm MgCl2/20 mm Abbreviations: snrna, small nuclear RNA; snrnp, small nuclear ribonucleoprotein; SR protein, Ser/Arg-rich protein. tpresent address: Department of Molecular and Cell iology, Howard Hughes Medical Institute, University of California at erkeley, erkeley, CA *To whom reprint requests should be addressed.

2 7140 iochemistry: Wassarman and Steitz Hepes-KOH, ph 7.9) was layered onto an equal volume of a 50% glycerol solution, and the tubes were covered with Parafilm, tipped to a horizontal position for 2 hr at room temperature, and returned to a vertical position for 5-8 hr at 4 C. Gradients were loaded by replacing the top 400 Al with ,ul of sample. Reaction mixtures (300 j.l) contained either 150,ul of nuclear extract, 2.0 mm MgCl2, 0.5 mm ATP, and 20 mm creatine phosphate or 150,l4 of S100 extract, 2.0 mm MgCl2, 0.5 mm ATP, 20 mm creatine phosphate, 7.5,ul of SR proteins (see below), and 15 ul of adenovirus substrate (:20 fmol/,u4), containing the first intron of the major late transcript (25). Gradients were centrifuged at 15,000 rpm in an SW41 rotor (eckman) for 12 hr at 4 C, 525-,l4 fractions were removed from the top, and fraction 23 was used to remove pelleted material. Gradient Size Markers. y fractionating a rabbit reticulocyte lysate and assaying each fraction for rrna by A260, 80S ribosomes were shown to sediment in fraction 7 of a 10-50% gradient. y fractionating poliovirus (kindly provided by S. Jacobson and P. Sarnow, Yale University and University of Colorado Medical School) and probing each fraction for viral genomic sense RNA with a probe complementary to nt (26), 150S poliovirus particles (27) were shown to sediment in fractions 10 and 11. Northern lots. Crosslinked samples were fractionated on denaturing 6% polyacrylamide gels, and untreated samples were fractionated on 10%o gels. RNA was transferred by electroblotting to Zeta-Probe membrane (io-rad), and membranes were probed with [a-32p]utp-labeled full-length RNAs complementary to 7SK, 7SL, Ul, U2, U4, U5,, Ull, U12 (28-30), and 5S (kindly provided by. Peculis, Yale University) RNAs. To equalize signals from RNAs of different abundance, various amounts of each probe were used (2-8 x 106 cpm). lots were hybridized and washed according to Zeta-Probe manufacturer's procedures. Psoralen Crosslinking. Crosslinking was performed (31-33) for 20 min on ice with 4'-aminomethyl-4,5',8-trimethylpsoralen (HRI Associates, Emeryville, CA) at 20,ug/ml. SR Protein Complementation of S100 Extracts. A mixture of SR proteins (SRp3Oa and -b, -40, -55, and -70) from calf thymus was prepared according to Zahler et al. (10) and kindly provided by A. Zahler and M. Roth (Hutchinson Cancer Research Institute). Splicing experiments were performed in 20-,l reaction mixtures containing 10,ul of HeLa S100 extract, 2.0 mm MgCl2, 0.5 mm ATP, 20 mm creatine phosphate, the indicated amounts of SR proteins, and the standard adenovirus substrate (-20 fmol/,l). Reactions were incubated at 30 C for the times indicated. Splicing in HeLa nuclear extract was carried out in 25-,ul reaction mixtures containing 15 ul of extract, 2.0 mm MgCl2, 0.5 mm ATP, and 20 mm creatine phosphate. RESULTS AND DISCUSSION Identification of >150S Complexes in HeLa Nuclear Extract That Contain ase-paired U2-. To examine the sedimentation behavior of complexes containing base-paired U2 and snrnas, nuclear extract was incubated under splicing conditions without substrate, psoralen-crosslinked, and fractionated on a 10-30% glycerol gradient. Northern blots of RNA from gradient fractions revealed that crosslinked U2- was found exclusively in the gradient pellet, whereas crosslinked U4- not only was in the pellet but also resolved into two peaks, with S values expected for U4 (15S) and U4 U5 (25S) complexes (data not shown; ref. 16). We deduce that the U2- and U4- crosslinked species observed are the same as those previously characterized since they are of the same abundance, they migrate identically on denaturing polyacrylamide gels, and they are the only crosslinks to RNA observed in nuclear extract (31, 32, 34). Another U2- interaction has recently been identified in the yeast Saccharomyces cerevisiae, but it has not been detected by psoralen crosslinking in mammalian cell extracts (35). Fractionation of crosslinked nuclear extract was repeated on a 10-50% glycerol gradient, allowing reproducible detection of crosslinked U2- in fractions 12-18, peaking in fraction 15 (Fig. 1A). Similar results were obtained when nuclear extract was first fractionated and individual fractions were then crosslinked (data not shown). A small amount ('-25%) of crosslinked U4- cosedimented with crosslinked U2- in fractions 12-18, while the majority remained near the top of the gradient (Fig. 1A). A similar gradient profile was produced with unincubated nuclear extract, except that a minor amount (<5%) of crosslinked U2- was detected A _jt.ibi_.ia 4lPW ul 1 M Proc. Natl. Acad. Sci. USA 90 (1993) _ 105) S 80S * 4/ 6 U2/ 7SK 7SL U2 U'. ~~~~U 1 FIG. 1. Glycerol gradient fractionation of crosslinked and uncrosslinked HeLa nuclear extract. (A) Nuclear extract was incubated for 30 min under splicing conditions, psoralen-crosslinked, and fractionated on a 10-50%o glycerol gradient. RNA from each fraction was subjected to Northern blot analysis. The direction of fractionation is indicated by the arrow at the upper right (fraction 1 is the top of the gradient). Fractions 22 and 23 were inverted during loading of the gel. The blot was probed for U2,, and Ull snrnas, whose mobilities are indicated on the right along with crosslinked U2- (U2/) and U4- (). In a longer exposure, a small amount of crosslinked U2-U4- is present above U4- in lane + (data not shown). Similar blots probed individually for U2, U4, or identified the crosslinked species (data not shown). Lanes labeled - and + contain 1/60th of the total reaction mixture before and after crosslinking, respectively. M denotes 32P-labeled pr322 DNA digested with Msp I, with the fragment lengths in nucleotides indicated on the left. () Uncrosslinked nuclear extract was analyzed as in A. Lane T contains RNA from 1/60th of the amount of fractionated extract. 80S and 150S positions were determined by sedimentation in a parallel gradient of ribosomes and poliovirus, respectively. U2 U4

3 iochemistry: Wassarman and Steitz near the top of the gradient, peaking in fractions 5 and 6 (data not shown). Parallel gradients in which 80S ribosomes peaked in fraction 7 and 150S poliovirus particles sedimented in fractions 10 and 11 indicated that fractions 5 and 6 correspond to -60 S, whereas fractions are >150S (see 5S rrna in lanes 7 of Figs. 1 and 4). Comparisons with polyribosome peaks show that material in fraction 18 cosediments with polysomes containing four 80S ribosomes (data not shown). We conclude that the -60S peak probably corresponds to previously characterized 60S spliceosomes because these fractions are active for splicing when supplemented with micrococcal nuclease-treated nuclear extract (data not shown; ref. 36). To identify RNA species sedimenting in the >150S region, uncrosslinked nuclear extract was fractionated on a 10-50% glycerol gradient and the resulting Northern blot was probed for a number of small RNAs (Fig. 1). As expected, U2, U4, and peak in fractions containing the U2- and U4- crosslinks (Fig. 1, fractions 12-18). The remaining spliceosomal snrnas, Ul and U5, are also detected in these fractions. We estimate that =50% of each spliceosomal RNA in nuclear extract sediments at >150S, apparently contained in large complexes. Interestingly, a 5'-truncated form of Ul called U1* is largely absent from the complexes and is found primarily at the top of the gradient (fractions 1-7) (37). The small amounts (<5%) of 7SK, 7SL, Ul*, Ull, U12, and 5S RNA that copeak with the spliceosomal RNAs in fractions (Fig. 1 and data not shown) are probably due to nonspecific sticking, a characteristic of other large RNP complexes such as ribosomes. This is demonstrated in Fig. 1, lane 7, which appears to be overloaded because of sticking of these same RNAs to the 80S ribosome peak. Are the >150S Complexes Analogous to in Vivo Spliceosomal Structures? The >150S complexes are much larger than a single spliceosome and could conceivably represent higherorder spliceosome complexes containing multiple (4-10) copies of each snrnp (see below). Very large snrnp-rich structures called coiled bodies and snurposomes have been identified in mammalian cell nuclei and amphibian germinal vesicles, respectively, but their functions are not known (17-19). Coiled bodies may act in pre-mrna metabolism since snrnp localization in coiled bodies is disrupted when transcription is blocked; snurposomes have been proposed to function in the assembly and storage of snrnp complexes. The >150S complexes may be stable subunits of coiled bodies that survive the nuclear extract preparation protocol. In fact, mild sonication of mammalian nuclei releases a number of intron-containing pre-mrnas as S complexes containing splicing snrnps (refs. 21 and 23; for review, see ref. 20). We have not directly investigated whether pre-mrna is present in the >150S complexes, but several observations suggest that these complexes may be composed of spliceosomes assembled on endogenous pre-mrnas. (i) The lack of U1* RNA in the complexes suggests that functional Ul snrnps are necessary for assembly of the complexes; perhaps requiring base pairing of the Ul 5' end with pre-mrna 5' splice sites. (ii) The low level of crosslinked U4- relative to the amounts of U4 and RNA in the >150S complexes suggests that the majority of U4 and molecules in the complexes are not base paired, a conformation that occurs after spliceosome assembly but prior to splicing of a premrna. (iii) Data presented below (see Fig. 4) show that large ( S) complexes containing spliceosomal snrnps and base-paired U2 are built up progressively during splicing in S100 extract supplemented with SR proteins. The >150S Complexes Are Absent from HeLa S100 Extract. Previous studies showed that U2 and snrnas cannot be psoralen-crosslinked in HeLa S100 extract (32), whereas U4 and snrnas crosslink to a level comparable to that in Proc. Natl. Acad. Sci. USA 90 (1993) 7141 nuclear extract (see Fig. 4, compare lanes NE and S100). When S100 extract was incubated under splicing conditions without substrate and fractionated on a 10-50% glycerol gradient, all of the spliceosomal RNAs sedimented at the top of the gradient (data not shown; see Fig. 4). Unincubated S100 extract contains a small amount of base-paired U2 (Fig. 2A, lane 2) that sediments at =60 S; the same is true of unincubated nuclear extract (data not shown). The absence of the >150S complexes from S100 extract is probably due to their pelieting during the 100,000 x g centrifugation step in the preparation protocol (24). When nuclear extract is centrifuged at 100,000 x g, the >150S complexes are found exclusively in the pellet, and S100 extract that has not been centrifuged at 100,000 x g contains large amounts of the >150S complexes (data not shown). ase-paired U2 Is Generated by Making S100 Extract Splicing-Competent. Since U2 base pairing (7, 8) and SR proteins (10-13) are both required for splicing and are both absent from S100 extract, SR proteins may play a role in inducing the U2 base-pairing interaction. To test this idea, reaction mixtures containing S100 extract were supplemented with various concentrations of SR proteins, incubated under splicing conditions with or without the standard adenovirus substrate, and then psoralen-crosslinked. Northern blots probed for RNA reveal that U2 base pairing does occur in S100 extract supplemented with SR proteins (Fig. 2). In fact, there is a direct correlation between SR protein concentration and the amount of crosslinked U2- generated (Fig. 2A, lanes 3-16). At high concentrations, SR proteins alone stimulate U2 base pairing (Fig. 2A, lane (~d) A Ad z 0O SR (,l) r - o.o= rv o m +I--7I+-7Irl-- +I r7-7ilr-7i Ad o origin _r ~~~~~ 9 *1 S S U2/ -qp ' FIG. 2. SR proteins and a splicing substrate stimulate U2 base pairing in S100 extract. (A) S100 extract was incubated for 30 min under splicing conditions with the indicated amounts of SR proteins, either in the absence (-) or presence (+) of adenovirus substrate (Ad). Addition of 0.5 Al of prepared SR proteins (at 0.6 pg/ml) to 10,ul of S100 extract approximates the concentration of SR proteins in 10 /4 of nuclear extract. Reaction products were crosslinked and a Northern blot was probed for. Lane 1 contains RNA from 5 Al of crosslinked nuclear extract, and lane 2 contains RNA from 10 /4 of unincubated S100 extract. Positions of uncrosslinked and crosslinked RNA (U2/,, and *) are indicated between A and. U4-* (*) is a crosslink between fulllength U4 and a truncated form of (D.A.W., unpublished data). () S100 extract was incubated for 15 min under splicing conditions with the indicated amounts of adenovirus substrate at -20 fmol/ul. Reaction products were analyzed as in A.

4 , E v _ v 7142 iochemistry: Wassarman and Steitz Proc. Natl. Acad. Sci. USA 90 (1993) 15), while in the presence of splicing substrate, base-pairing nuclear extract. Intermediates and products are apparent at activity is enhanced >20 fold, even at low SR protein 15 min rather than the usual 30 min (compare lanes 2-6 with concentrations (Fig. 2A, lanes +). Splicing substrate alone 7-11), more closely resembling the time course of in vitro also stimulates U2 base pairing but to a lesser extent than splicing in yeast extracts (see references within ref. 2). This when combined with SR proteins (see Fig. 2, where lane 4 suggests that the SR protein-dependent step may be rate has the same amount of substrate as each of the + lanes in limiting for the mammalian splicing reaction in vitro. Fig. 2A). In contrast, the level of U4 base pairing does not U2 ase Pairing Is Initially Detected in an -60S Complex. Since crosslinked U2- in nuclear extract sediments appear to be enhanced by incubation with either SR proteins or splicing substrate (Fig. 2). exclusively at >150S, we suspected that the complexes Previous studies have shown that splicing efficiency in containing base-paired U2- generated in S100 extract S100 extract is directly correlated to SR protein concentration (10, 11, 13). Therefore, the increase in U2- base supplemented S100 extract was incubated with adenovirus would sediment similarly. However, when SR protein- pairing we observe at elevated SR protein concentrations is substrate, psoralen-crosslinked, and fractionated on a 10- probably the result of increased splicing activity. On the 50% glycerol gradient, crosslinked U2- sedimented instead in fractions 5-9, peaking in fraction 6, which corre- other hand, it is possible that SR proteins directly induce U2 base pairing. At least one SR protein, ASF/SF2, sponds to -60 S (Fig. 4A). The reaction in Fig. 4A was possesses RNA-binding and annealing activities, and several incubated for only 5 min, a time prior to the appearance of SR proteins contain RNA recognition motifs (11, 12, 38-40). splicing intermediates (Fig. 3); but a similar result was Moreover, a small amount of U2 base pairing is observed obtained for a reaction incubated for 15 min (data not shown), at a high SR protein concentration in the absence of substrate when both intermediates and products have been generated. (Fig. 2A, lane 15). Strikingly, after a 5- or 15-min incubation, small amounts U2 ase Pairing in S100 Extract Precedes the First Step of the snrnas and crosslinked U2- (seen on longer of the Splicing Reaction and Is ATP-Dependent. To determine exposures) sedimented in a periodic pattern (fractions 12, 15, when U2* base pairing occurs during the course of a 17, and 20) in the lower half of the gradient, overlapping the splicing reaction, S100 extract supplemented with SR proteins was incubated with adenovirus substrate and psoralen- (Fig. 4A and data not shown). These complexes were absent peak of the >150S complexes observed in nuclear extract crosslinked at various times. Northern blots probed for from samples prepared under identical reaction conditions, revealed that crosslinked U2- appears after 5 min, peaks but without ATP (Fig. 4). at 15 min, and then decreases in level over the rest of the time Thus, in an active splicing reaction mixture devoid of course (Fig. 3A, lanes 1-10). Splicing assays performed in the endogenous >150S complexes, =60S spliceosomes appear to same SR protein-supplemented S100 extract revealed that assemble progressively into polyspliceosomes. Conversely, intermediates and products are absent at 5 min, become the -200S large nuclear RNP complexes, which others have detectable at 15 min, and reach peak levels at 30 min (Fig. 3, derived from isolated nuclei, can be converted to 70S complexes by changing ionic conditions and to 30S complexes by lanes 2-6). Neither U2 base pairing (Fig. 3A, compare lanes 11 and 12) nor splicing intermediates (Fig. 3, compare mild RNase treatment (21, 22). Since the adenovirus substrate contains a single intron, the larger complexes gener- lanes 12 and 13) are detected in the absence of ATP. Similar experiments could not be performed during in vitro splicing ated in our reaction mixtures must involve interactions in nuclear extract because any increase in the amount of between spliceosomes assembled on separate pre-mrna base-paired U2' would be obscured by high levels of molecules. Yet, these assemblies may reflect the ability of endogenous base-paired U2. We conclude that U2 spliceosomes to make orderly contacts across exons in base pairing occurs early and then dissolves as splicing multiintron pre-mrnas, as has been suggested (41, 42). proceeds. The time course of U2- crosslinking, relative to SR-protein-supplemented S100 extract, which initially lacks the time course of splicing, is consistent with that previously endogenous >150S and spliceosome complexes, is an excellent system for studying de novo spliceosome assembly. Using observed for a U2--substrate double crosslink (32). Fig. 3 also demonstrates that splicing in S100 extract this system it may be possible to address a variety of questions. When does U2 base pairing occur relative to other complemented with SR proteins is significantly faster than in A inc(min) 60 Si1o NE S100 r+=- ATP r- rl-- r-+1- rw+i- Ad r:- --ATP ` ' inc (min) FIG. 3. Kinetics and ATP dependence of U2 origin QQ{, base pairing (A) and pre-mrna splicing () in S100 - extract complemented with SR proteins. (A) S100 extract, in the absence (-) or presence (+) of 1,ul of adenovirus substrate (Ad), was incubated under splicing conditions with 0.5,l4 of SR proteins (see Fig. 2A, lane 10) for the times indicated. Lanes 11 and 12 w U2/ - _ show reaction products incubated with (+) and without (-) ATP and creatine phosphate, respectively, r- 0 I I but from a different experiment than that in lanes Reaction products were analyzed as in Fig. 2. Positions of crosslinked U4- (), U2- (U2/), and U4-* (*) are indicated. () [a-32p]utp-labeled adenovirus substrate (lane 1) was spliced in S100 extract complemented with 0.5 j. of SR proteins (lanes 2-6) or in nuclear extract (lanes 7-11) for the times indicated. Lanes 12 and 13 show S100 reactions incubated with (+) and without (-) ATP and creatine phosphate, respectively. The positions of the splicing substrate, intermediates, and fl products are indicated between lanes 11 and 12 with exons as boxes, introns as lines, and introns lariats as loops.

5 A *S la to9 z _ "I"No t...:. Z CD m 80.e 0 as iochemistry: Wassarman and Steitz - z- Om Z S 80S..., U2/ FIG. 4. Glycerol gradient fractionation of S100 extract complemented with SR proteins. Splicing reaction mixtures containing S100 extract, SR proteins, adenovirus substrate, and MgCl2 were incubated either with (A) or without () ATP and creatine phosphate. After crosslinking and fractionation as in Fig. 1, the blots were probed for U2, U4, U5,, and 5S RNAs, whose positions are indicated on the right, along with crosslinked U2- (U2/) and U4- (). Lane NE shows RNA from 2.5 y4 of crosslinked nuclear extract, while lane S100 shows RNA from 1/60th of the sample loaded onto the gradient (corresponding to 2.5,ul of S100 extract). The periodic pattem of snrnas and of crosslinked U2- (seen in longer exposures) in the bottom half of the gradient was reproducibly obtained in several experiments. early steps in splicing, such as dissociation of U4 from U4-? Do SR proteins directly facilitate U2 base pairing? What are the signals that control assembly of the large spliceosomal complexes? Are the signals contained within the intron or contributed by a splicing factor? When do the large spliceosomal complexes function during the splicing cycle? Are they more prevalent when a multiintron substrate is used? Answers may reveal the relationship of the large complexes we have observed to in vivo spliceosomal structures. We thank A. Zahler and M. Roth for providing purified SR Proc. Natl. Acad. Sci. USA 90 (1993) 7143 proteins,. Peculis for constructing the 5S rrna probe, and S. Jacobson and P. Sarnow for providing materials and procedures for the poliovirus gradients. We also thank A. Zahler, M. Roth, J. Gall, K. Wassarman, and the C. Guthrie and J. Steitz laboratories for ideas and discussions that greatly improved this manuscript. This work was supported by the Howard Hughes Medical Institute and Grant GM26154 from the National Institutes of Health. 1. Green, M. R. (1991) Annu. Rev. Cell iol. 7, Guthrie, C. (1991) Science 253, row, D. & Guthrie, C. (1988) Nature (London) 334, Fabrizio, P. & Abelson, J. (1990) Science 250, Madhani, H. D., ordonne, R. & Guthrie, C. (1990) Genes Dev. 4, Vankan, P., McGuigan, C. & Mattaj, I. W. (1992) EMO J. 11, Wu, J. & Manley, J. L. (1991) Nature (London) 352, Datta,. & Weiner, A. M. (1991) Nature (London) 352, Wu, J. & Manley, J. L. (1992) Mol. Cell. iol. 12, Zahler, A. M., Lane, W. S., Stolk, J. A. & Roth, M.. (1992) Genes Dev. 6, Krainer, A. R., Conway, G. C. & Kozak, D. (1990) Genes Dev. 4, Fu, X.-D. & Maniatis, T. (1992) Science 256, Mayeda, A., Zahler, A. M., Krainer, A. R. & Roth, M.. (1992) Proc. Natl. Acad. Sci. USA 89, Lamond, A. I., Konarska, M. M., Grabowski, P. J. & Sharp, P. A. (1988) Proc. Natl. Acad. Sci. USA 85, Yean, S.-L. & Lin, R.-J. (1991) Mol. Cell. iol. 11, Konarska, M. M. & Sharp, P. A. (1988) Proc. Natl. Acad. Sci. USA 85, Carmo-Fonseca, M., Pepperkok, R., Carvalho, M. T. & Lamond, A. I. (1992) J. Cell iol. 117, rasch, K. & Ochs, R. L. (1992) Exp. Cell Res. 202, Wu, Z., Murphy, C., Callan, H. G. & Gall, J. G. (1991) J. Cell iol. 113, Sperling, R. & Sperling, J. (1990) The Eukaryotic Nucleus: Molecular iochemistry and Macromolecular Assemblies (Telford, Caldwell, NJ), Vol. 2, pp Sperling, R., Spann, P., Offen, D. & Sperling, J. (1986) Proc. Natl. Acad. Sci. USA 83, Spann, P., Feinerman, M., Sperling, J. & Sperling, R. (1989) Proc. Natl. Acad. Sci. USA 86, Wagatsuma, M., Obinata, M., Moroi, Y., Hanaoka, F. & Yamada, M. (1985) J. iochem. (Tokyo) 97, Dignam, J. D., Lebovitz, R. M. & Roeder, R. D. (1983) Nucleic Acids Res. 11, Solnick, D. (1985) Cell 42, Kitamura, N., Semler,. L., Rothberg, P. G., Larsen, G. R., Adler, C. J., Dorner, A. J., Emini, E. A., Hanecak, R., Lee, J. J., van der Werf, S., Anderson, C. W. & Wimmer, E. (1981) Nature (London) 291, Rueckert, R. R. (1990) Virology (Raven, New York), Vol. 2, pp Wassarman, D. A. & Steitz, J. A. (1991) Mol. Cell. iol. 11, Wassarman, K. M. & Steitz, J. A. (1992) Mol. Cell. iol. 12, lack, D. L. & Pinto, A. L. (1989) Mol. Cell. iol. 9, Hausner, T.-P., Giglio, L. M. & Weiner, A. M. (1990) Genes Dev. 4, Wassarman, D. A. & Steitz, J. A. (1992) Science 257, Wassarman, D. A. (1993) Mol. iol. Rep. 17, Rinke, J., Appel,., Digweed, M. & Luhrmann, R. (1985) J. Mol. iol. 185, Madhani, H. D. & Guthrie, C. (1992) Cell 71, Grabowski, P. J., Seiler, S. R. & Sharp, P. A. (1985) Cell 42, Lerner, M. R., oyle, J. A., Mount, S. M., Wolin, S. L. & Steitz, J. A. (1980) Nature (London) 283, Ge, H., Zuo, P. & Manley, J. L. (1990) Cell 66, Krainer, A. R., Mayeda, A., Kozak, D. & inns, G. (1991) Cell 66, Roth, M.., Zahler, A. M. & Stolk, J. A. (1991) J. Cell iol. 115, Robberson,. L., Cote, G. J. & erget, S. M. (1990) Mol. Cell. iol. 10, Grabowski, P. J., Nasim, F.-U. H., Kuo, H.-C. & urch, R. (1991) Mol. Cell. iol. 11,