B3 enhances splicing complex assembly on the 5 ؅ splice site of Bcl-x L . A , a splicing complex formation assay was performed on native acrylamide gels using uniformly labeled pre-mRNAs containing or lacking the B3 element. Transcripts were incubated at 30 °C in HeLa nuclear extracts, and aliquots were taken at different times (0, 15, and 30 min). Heparin was added 

Context in source publication

Context 1

... 50 ␮ l of activated protein A CL-4B beads (GE Healthcare) and by washing four times in a NET2 buffer (50 m M Tris-HCl pH 7.5, 150 m M NaCl, 0.05% Nonidet P-40, 0.5 m M dithiothreitol). All samples were resuspended in Laemmli buffer and separated by SDS-PAGE (10%). Gels were exposed 72 h on XAR film. Affinity Chromatography —The midAM RNA oligo was cou- pled to agarose-adipic acid beads (Sigma) according to the manufacturer’s recommendations. Twenty-five ␮ l of these beads were then incubated in 93.75 ␮ l of in vitro splicing mixture including the nuclear extract but without polyvinylalcohol for 10 min at 30 °C. To assess the role of U1 snRNP, the nuclear extract was incubated with a 2 Ј - O -methyl oligonucleotide complementary to the 5 Ј end of U1 snRNA. The beads were eluted twice with 200 ␮ l of buffer D then washed twice in 400 ␮ l of the same buffer. The elution and washing steps were repeated with increasing amounts of KCl (0.1, 0.25, 0.5, and 1.0 M ) followed by a final elution in Laemmli buffer. The eluted proteins were precipitated in trichloroacetic acid and separated by SDS-PAGE. The gel was silver-stained, and the bands of interest were cut out and destained. In-gel trypsin digestion and liquid chroma- tography-tandem mass spectrometry analysis was performed at the Genome Quebec Innovation Center at McGill University. Bcl-x L —We have shown previously that the deletion of a 86-nt region upstream of the 5 Ј splice site of Bcl-x L decreases the use of this splice site in vivo and in vitro (36). We constructed a variety of deletion mutants (Fig. 1, A and B ) to identify regions within B3 that are responsible for this activity in vitro and in vivo . Transcripts were produced (S2.13 and derivatives) and incubated in a HeLa nuclear extract for 2 h at 30 °C. Using RT-PCR to assess the production of the x S and x L splice isoforms, we confirmed that B3 is important for Bcl-x L usage in a HeLa extract since its removal shifts splicing to the Bcl-x S 5 Ј splice site (Fig. 1 C , compare lane 4 with lane 3 ). Removing the 3 Ј half of B3 ( ⌬ ML) had an effect that was almost as strong as ⌬ B3 (Fig. 1 C , compare lane 6 with lane 4 ). Deleting the upstream and downstream halves of ML ( ⌬ ML1 and ⌬ ML2, respectively) indicated that ML2 was the active portion that enforced splicing to the Bcl-x L 5 Ј splice site ( lanes 9 and 10 ). In contrast, removing the first half of B3 ( ⌬ AM) did not affect Bcl-x splicing (Fig. 1 C , lane 5 ). However, although removing the upstream portion of AM ( ⌬ AM1) had no effect, removing the downstream half ( ⌬ AM2) slightly decreased the relative level of Bcl-x ( lane 8 ). A deletion encompassing AM2 and the inactive ML1 region ( ⌬ midB3, lane 11 ) confirmed the positive contribution of AM2 to Bcl-x L splicing. Deleting both ML2 and AM2 (2x ⌬ ) almost completely eliminated splicing to the Bcl-x L 5 Ј splice site in vitro (Fig. 1 D , lane 5 ). The deletion of midAM did not decrease the relative production of Bcl-x L (Fig. 1 D , lane 6 ), whereas ⌬ midMID did ( lane 7 ), suggesting that the active portion in AM2 is at the 3 Ј end of this element. The most active portion of ML2 also appeared to be near its 3 Ј end since the magnitude of the effect of ⌬ midML was not as important as that observed with ⌬ ML2 (Fig. 1, D , lane 8 , and C , lane 10 , respectively). The impact of these deletions on Bcl-x splicing was verified in vivo by transfecting CMV promoter-driven minigenes in HeLa cells (Fig. 1 E ). The parent minigene X2.13 contains the same Bcl-x portion as the S2.13 transcript used in vitro . Although transcripts produced from the parent minigene are spliced preferentially to the Bcl-x s 5 Ј splice site (36), the deletion of B3 abrogates the production of Bcl-x L , an outcome also observed with ⌬ ML and ⌬ ML2 (Fig. 1 E , lanes 2 , 4 , and 8 ). The in vitro impact of all the other deletions was confirmed in vivo except for ⌬ AM1 and ⌬ midAM, which stimulated the use of Bcl-x L (Fig. 1 E , lanes 5 and 11 ), a situation not observed in vitro (Fig. 1 C , lane 7 , and D , lane 6 ). This result suggests the existence of a silencer element (see below). A silencer in the AM1-mi- dAM region could explain why deleting AM (which removes the AM2 enhancer) had no effect (Fig. 1 E , lane 3 ). Thus, three regions in B3 contribute to splicing control. Although we cannot differentiate between effects on Bcl-x L or Bcl-x S donor sites, the argument of proximity would suggest that ML2 and the downstream portion of AM2 stimulate splicing to the Bcl-x L 5 Ј splice site, whereas a sequence in midAM represses this event. To confirm the overall enhancer activity of B3 on the Bcl-x L 5 Ј splice site, we tested if B3 could stimulate complex formation on a simple pre-mRNA carrying the 5 Ј splice site of Bcl-x L . Because complex formation is more easily observed with a strong 3 Ј splice site, we produced hybrid pre-mRNAs carrying the 3 Ј splice site of the adenovirus major late transcript. As seen in Fig. 2, deletion of B3 eliminated the U2 snRNP-dependent assembly of splicing complex A. This result, therefore, indicates that B3 can stimulate early spliceosome assembly on the 5 Ј splice site of Bcl-x L . A Role for SRp30c in the Activity of B3 —We noted that AM2 and ML2 share the sequence CUUGGAU. However, this sequence does not appear essential because mutating the GG in the ML2 site had no effect on Bcl-x splicing in vitro (GG1 224 CC; Fig. 1 D , lane 9 ). Another feature of ML2 is that its downstream portion contains the sequence AGGAG, a sequence that we recently identified as an optimal binding site for SRp30c (13). Changing this AGGAG for AUUAG decreased splicing to the Bcl-x L site in vitro (Fig. 1 D , lane 10 ), consistent with a role for SRp30c in the activity of B3. inated the stimulation offered by the supplementation with SRp30c (2x ⌬ ; Fig. 4 A , lanes 9 –11 ). Notably, in this experiment, the double deletion (2x ⌬ ) promoted the use of a cryptic 5 Ј splice site (x M1 ) mapping in the midAM portion of B3 (see below). We also tested the addition of equivalent amounts of hTra2 ␤ , an SR-related protein often associated with splicing enhancer activity (41– 44). hTra2 ␤ did not significantly affect Bcl-x splicing in vitro , although it did stimulate the use of a distal 5 Ј splice site on an unrelated reporter pre-mRNA (Fig. 4 C ). We also tested the impact of exogenously expressing SRp30c in HeLa cells. SRp30c increased the production of the Bcl-x L RNA isoform derived from the X2.13 minigene (Fig. 5, A and B ). No improvement in Bcl-x L usage was detected when SRp30c was co-expressed with the variant lacking B3. The stable expression of a tagged SRp30c in HeLa and 293 cells was also associated with an increase in endogenous Bcl-x L levels (data not shown). Thus, SRp30c can stimulate the use of the Bcl-x L 5 Ј splice site in a B3-dependent manner both in vitro and in vivo . We also tested the activity of the SR protein ASF/SF2, which is 74% identical to SRp30c (47). Although ASF/SF2 improved splicing to the Bcl-x L site, consistent with previous results (34), this effect was independent of B3 (Fig. 5, C and D ). The Silencer Element Contains Cryptic 5 Ј Splice Sites —Our deletion analysis suggested the existence of a splicing silencer in the midAM portion of B3 since removing this region increased Bcl-x L usage in vivo (Fig. 1 E , lane 11 ). With the hope of identi- fying a factor responsible for this activity, we carried out affinity The current study links two observations made previously; specifically, that the 86-nt-long B3 element is required for the optimal use of the Bcl-x L 5 Ј splice site and that SRp30c can stimulate Bcl-x L splicing in vitro (36). We have shown here that SRp30c can modulate the activity of B3. Two elements in B3, AM2 and ML2, individually enhanced the relative use of the Bcl-x 5 Ј splice site, but the strongest effect on splicing was obtained when both ML2 and AM2 were deleted. Although we cannot rule out that B3 can repress the Bcl-x S 5 Ј splice site, we have shown that B3 behaves as an enhancer by helping to assemble splicing complexes on the 5 Ј splice site of Bcl-x L . Consistent with an enhancer function for ML2 and AM2, search engines designed to find putative enhancer motifs (RESCUE-ESE (45), PESX (46), ESR-Search (47)) identify such sequences in these elements (Fig. 7). The ML2 enhancer can explain the results of Taylor et al. (48) who used an RNA oligonucleotide complementary to an exonic portion upstream of the Bcl-x L 5 Ј splice site to shift splicing toward the Bcl-x S site in cells. Retrospectively, the impact of this oligonucleotide can be explained through obstruction of a portion of the ML2 enhancer (Fig. 7). Supplementing HeLa extracts with recombinant SRp30c stimulated Bcl-x L splicing, and the AM2 and ML2 elements were required for this effect. This observation was confirmed in vivo by showing that the SRp30c-mediated stimulation in the production of Bcl-x L was not observed when the minigene lacked B3. Recombinant SRp30c could bind to naked AM2 and ML2 RNAs. ML2 contains a sequence that fits the high affinity binding site for SRp30c (AGGA(G/C) Ref. 13). For AM2, it is possible that the two GGA and the GAG triplets contribute to the binding of SRp30c. Our UV cross-linking assays indicate that SRp30c can bind to the enhancer elements in the context of a nuclear extract. SRp30c binding to these sites may then help stabilize the binding of U1 snRNP to the downstream 5 Ј splice site of Bcl-x (Fig. 8). Our characterization of B3 also identified the upstream midAM silencer. However, SRp30c also stimulated Bcl-x L usage when the silencer element was mutated ( ⌬ midAM and weakX L ; data not shown), indicating that the stimulatory activity of SRp30c on the Bcl-x L 5 Ј splice site is probably direct and does not strictly rely on a possible repression of the upstream silencer. A direct stimulation by SRp30c is compatible with current models of action ...