Two different splicing events are promoted in response to SC35 overexpression

Except for the 3.0 and 1.3 kb SC35 transcripts [unspliced in the 3′ untranslated region (3′UTR) and polyadenylated at the proximal poly(A) site, respectively], the various SC35 mRNAs arise via a combination of two splicing events (Figure 3). The 2.1 and 1.7 kb species differ from the 2.0 and 1.6 kb transcripts by the inclusion of a 104 nt cassette exon delineated by the A2–D3 splice sites. In addition, the 2.1 and 2.0 kb species differ from the 1.7 and 1.6 kb transcripts by the retention of a 498 nt intron delineated by the D4–A4 splice sites. To evaluate the relative levels of these mRNAs prior to and following induction of SC35 expression, we performed semi- quantitative RT–PCR experiments. Southern blot analysis of RT–PCR products was performed with the H23, H24 and H121 oligonucleotide probes specific for different exon–exon junction sequences of the 3′UTR (Figure 3) and that allow us to discriminate between the various alternative SC35 mRNAs.

Analysis of exogenous SC35 transcripts from HeLa cells expressing coding (E11 and E23) or non-coding (F3 and F10) genomic sequences showed that inclusion of the 104 nt cassette exon and splicing of the 498 nt intron are concomitantly promoted in cells overexpressing the SC35 protein. The H24 probe revealed that the ratio of the expression levels of the 1.7 and 2.1 kb mRNAs, both containing the 104 nt cassette exon but differing by the splicing of the 498 nt intron, is significantly higher in E cells than in F cells (Figure 4A). Splicing of this intron appears to be related to the inclusion of the 104 nt cassette exon, since hybridization with the H23 probe (specific for mRNAs resulting from skipping of the cassette exon) indicated that the ratio of the 1.6 and 2.0 kb mRNA expression levels is only slightly increased upon induction of SC35 overexpression in E cells. Furthermore, both mRNAs appear to be 4- to 5-fold less abundant in E cells than in F cells, an observation that agrees with our northern blot analyses showing decreased expression of the major 2.0 kb mRNA in response to SC35 accumulation. The H121 probe showed that the ratio of the 1.7 and 1.6 kb mRNA expression levels in E cells is the reverse of that determined in F cells, confirming that inclusion of the cassette exon is stimulated in cells overexpressing SC35. Similar observations were obtained in HeLa cells overexpressing non-coding genomic sequences and transiently transfected with an SC35 constitutive expression vector (data not shown). We finally established that the expression pattern of endogenous SC35 mRNAs is affected similarly in HeLa cells expressing SC35 coding sequences (Figure 4B). Four days after Dox induction, the relative expression level of the 1.7 kb mRNA was clearly increased in cells stably transfected with coding cDNA (A) or genomic (E) SC35 sequences, as compared with that observed in uninduced cells or in cells expressing non-coding cDNA (C) or genomic (F) sequences. Altogether, northern blot and RT–PCR analyses demonstrate that the distribution of the various SC35 mRNAs is strongly affected in cells overexpressing SC35. In particular, the 1.7 kb mRNA, which differs from the 2.0 kb major species by both the retention of a 104 nt cassette exon and excision of a 498 nt intron, preferentially accumulates relative to other SC35 mRNAs.

Open in a separate window

Open in a separate window

Fig. 4. Southern blot analysis of RT–PCR products corresponding to exogenous and endogenous SC35 mRNAs expressed in HeLa cell clones. One-tenth of the PCR products corresponding to SC35 exogenous (A) or endogenous (B) mRNAs expressed in different HeLa cell clones were analyzed with the indicated probes. The structure and size of the transcripts corresponding to the PCR products are indicated and the positions of the PCR primers (arrowheads) and probes (short bold line) are shown. For each probe, the percentage of the hybridization signal corresponding to each PCR product was determined by PhosphorImager analysis for four independent clones of the same series and the mean values are indicated on the right. RT–PCR analysis of the GAPDH mRNA was performed to normalize for RNA amounts used in the RT step. (A) RT–PCR analysis of exogenous SC35 mRNAs from HeLa cell clones stably transfected with coding (E11 and E23) or non-coding (F3 and F10) SC35 genomic sequences and incubated for 48 h in the presence of Dox. (B) RT–PCR analysis of endogenous SC35 mRNAs expressed in HeLa cell clones stably transfected with coding cDNA (A), non-coding cDNA (C), coding genomic (E) or non-coding genomic (F) SC35 sequences prior to or 96 h after Dox induction.

SC35 specifically activates two splicing events in its own pre-mRNA in vitro

To eliminate the possibility of effects not directly mediated by SC35, we tested whether this splicing factor could recapitulate both the retention of the 104 nt cassette exon and the excision of the 498 nt intron in vitro. To this end, we generated three constructs (Figure 5): one containing the whole 498 nt intron flanked by its natural exons (wt-RI construct) and two others containing the 104 nt cassette exon. The first (wt-CE) corresponds to a wild-type 1249 bp fragment including the flanking whole introns and natural exons. The second (βglo-CE) contains a smaller portion (451 bp) of the SC35 gene including limited intronic sequences flanking the cassette exon, which was inserted in the middle of the intron separating exons 2 and 3 of the rabbit β globin gene. Consistent with the alternative splicing observed in vivo with both cassette exon and retained intron, these two elements are flanked by poor 5′ splice sites (only four successive bases matching with the consensus sequence) and suboptimal 3′ splice sites. In vitro assays were performed with standard HeLa nuclear extracts and the nature of the various splicing products was assessed by various criteria, such as abnormal electrophoretic behavior and debranching assays for lariat products, or RT–PCR analysis and sequencing of mRNA products. The results obtained with the different constructs showed that excision of the retained intron and inclusion of the cassette exon are highly and specifically activated by SC35 in vitro.

Open in a separate window

Fig. 5. Origin and schematic representation of the minigene constructs used in in vitro splicing analyses. The overall organization of the human SC35 gene is depicted and the different donor (D) and acceptor (A) splice sites are indicated. The β globin exon 2, exon 3 and intronic sequences represented in the βGlo-CE chimeric transcript are shown.

Due to weak splice sites, the wt-RI substrate was almost not spliced under standard splicing conditions (Figure 6, lane 2). However, when nuclear extracts were complemented with moderate amounts of SC35 (7–14 pmol), efficient splicing of the 498 nt intron (lanes 3 and 4) was observed. Moreover, a cryptic 5′ splice site (AG/GTAAAA), located 157 nt downstream of the bona fide 5′ splice site, was activated. These are specific for SC35, since complementation of the nuclear extract with ASF/SF2 (lanes 5 and 6) or 9G8 (lane 7) has no effect. Interestingly, complementation with SRp46 (lanes 8 and 9), expressed in human cells from an SC35-derived retrogene (Soret et al., 1998), also led to a partial splicing activation. To determine whether the splicing activation by SC35 was specific, we then analyzed splicing of the first intron of the major late adenovirus (MLA) transcription unit. As assessed by the accumulation of that intron (lane 11), splicing in the absence of added SR protein is quite efficient given the intron size (1021 nt), indicating that SR proteins were not limiting in our nuclear extracts. Complementation of these extracts with each individual SR protein, but especially with SC35 or ASF/SF2 (lanes 12 and 13), led to a splicing inhibition specific to the MLA intron, because such inhibition does not occur when introns of a similar size are spliced from the wt-CE and βglo-CE pre-mRNA (Figure 7). These results indicate that the SC35-dependent activation observed with the wt-RI pre-mRNA does not represent a general consequence of nuclear extract complementation by individual SR proteins.

Open in a separate window

Fig. 6. In vitro splicing of the retained intron (wt-RI) pre-mRNA. The wt-RI (left panel) and MLA (right panel) pre-mRNA substrates were subjected to in vitro splicing assays and the RNAs were analyzed by PAGE. Lanes 1 and 10, pre-mRNA starting material; lanes 2 and 11, splicing assays with the nuclear extract alone. The other assays were supplemented by individual SR proteins as indicated above the lanes. In the left panel, increasing amounts (7 and 14 pmol) of SC35 (lanes 3 and 4), ASF/SF2 (lanes 5 and 6) and SRp46 (lanes 8 and 9) or 10 pmol of 9G8 (lane 7) were added to the assays. In the right panel, the higher amount of each SR protein is added. Note that, for Figures ​Figures66 and ​and7,7, the use of three different nuclear extract preparations gave very similar results. The positions of precursors, intermediates and final products are indicated to the left of the autoradiograms.

Open in a separate window

Fig. 7. In vitro splicing of the SC35 cassette exon. The wt-CE (left panel) and the βGlo-CE (right panel) pre-mRNAs were spliced in the presence of nuclear extracts alone or supplemented with the individual SR proteins as in Figure 6. Lanes 1 and 6, pre-mRNA starting material; lanes 2 and 7, splicing assays with the nuclear extract alone. In the right panel, 7 and 14 pmol are used for SC35, ASF/SF2 and SRp46, and 10 pmol for 9G8. In the left panel, the higher amount of each SR protein was used. At the bottom of the right panel, two small zones, only including the 3- or 2-exon mRNA, are shown. The positions of the RNA products are indicated at the side of both panels.

Studies performed with the wild-type wt-CE pre-mRNA in HeLa nuclear extracts (Figure 7, lane 2) revealed that the splicing of the large 980 nt intron, including the cassette exon, is moderately efficient. This result was not surprising, since this whole region remains unspliced in vivo in the 2.5 and 3.0 kb SC35 mRNAs (Figure 3). Accordingly, the accumulation of the short 2-exon mRNA of 278 nt was weak. Very similar splicing patterns were observed when nuclear extracts were complemented with ASF/SF2 (lane 4) and SRp46 (lane 5), where the splicing efficiency in the presence of ASF/SF2 was slightly better. As already observed for the wt-RI substrate, complementation with SC35 (lane 3) resulted in a strong shift of splicing, revealed by the accumulation of a novel lariat intron of smaller size (the 598 nt intron 2) at the expense of the larger intron. An additional band with an apparent size of 540 nt, possibly corresponding to the free intron 1 of 278 nt, was also weakly visible. Concomitantly, the 382 nt mRNA, including the cassette exon, was detected, although at a weak level.

Splicing of the chimeric βglo-CE substrate in HeLa nuclear extract was more efficient than that of the wt-CE substrate, as shown by the strong accumulation of the 1044 nt intron containing the cassette exon and the detection of the 128 nt 2-exon mRNA (Figure 7, right panel, lane 7). However, the weak accumulation of the excised intron 1 and 2 showed poor recognition of the cassette exon. As previously, addition of ASF, 9G8 and SRp46 to the nuclear extracts (lanes 10–14) did not significantly modify this splicing pattern. In contrast, SC35 induced a dramatic shift of splicing, as revealed by the accumulation of the individual introns 1 and 2, of the splicing intermediates lacking intron 2 and of the 3-exon mRNA (lanes 8 and 9). Concomitantly, accumulation of the large 1044 nt intron and of the 2-exon mRNA was decreased. These results demonstrate that increasing concentrations of SC35 strongly stimulate the inclusion of the cassette exon and that the 451 nt region including the cassette exon is sufficient to promote this specific regulation.

Cell culture and transfection conditions

HeLa Tet-On cells (Clontech) were transfected by calcium phosphate co-precipitation according to standard protocols. Briefly, 4 × 10⁵ cells seeded in 60 mm dishes were transfected 24 h later with a mixture consisting of 5 µg of SC35 expression vectors, 2 µg of pNT-Hygro selection vector and 8 µg of carrier DNA (pBKS). Twenty-four hours after transfection, cells were incubated in fresh medium containing 150 µg/ml hygromycin (Sigma) and 100 µg/ml geneticin (Life Technologies). Resistant clones were isolated 15–20 days later and expanded for further analyses. Induction of SC35 expression was performed in the presence of 1 µg/ml Dox (Sigma) for various times. Before transient transfection, HeLa cells were incubated for 48 h in the presence of 1 µg/ml Dox to induce expression of SC35 sequences. Transfection was performed as described above in the presence of Dox.

Origin and derivation of probes

The EB650 probe detecting the different SC35 mRNAs was previously described (Sureau and Perbal, 1994). The HA probe (5′-GAGGCT AGCATAATCAGGAAC-3′) is specific for the tag sequence present at the 5′ end of exogenous SC35 RNAs. The SC35 probe specific for the 5′UTR of endogenous SC35 mRNAs corresponds to a 378 bp PCR-generated fragment. Oligonucleotides H23, H24 (Sureau and Perbal, 1994) and H121 (5′-CACAACTGCGCCTTTTCAAT-3′), specific for different exon–exon junctions, were used as probes for Southern blot analysis of RT–PCR products and northern blot analysis of SC35 mRNAs. The human GAPDH (Clontech) and β-actin-specific probes were used for normalization of northern blot hybridization signals. The GAPDH-specific probe (5′-CCAAAGTTGTCATGGATGACC-3′) was used for normalization of Southern blot hybridization signals.

Northern and western blot analysis

Total RNA and protein samples were purified with Tri-Reagent (Sigma) as recommended. Labeling of probes, northern blotting and hybridizations were performed according to standard procedures. Protein samples were subjected to SDS–PAGE and transferred onto a PVDF membrane (Immobilon P, Millipore). Detection was performed with an enhanced chemiluminescence (ECL) system (NEN Life Science). HA-tagged proteins were revealed with the 12CA5 monoclonal (Boehringer Mannheim) and peroxidase-conjugated antibodies (Jackson Immuno research Laboratories).

RT–PCR analysis

First strand cDNA was synthesized from 5 µg of DNase-treated total RNA with the Amersham cDNA synthesis kit. For analysis of cDNA corresponding to exogenous SC35 mRNAs, one-twentieth of the reaction was amplified by 20 PCR cycles with the HA (5′-GTTCCT GATTATGCTAGCCTC-3′) and the H74 (5′-AAGGCCCGGGTGACT GGAAGTGC-3′) primers. Amplification of cDNAs corresponding to endogenous SC35 messengers was performed by 22 PCR cycles under the same conditions except that H120 (5′-GGCCGCCACTCAGAGCT ATG-3′), specific for the SC35 5′UTR, was used instead of the HA primer. Amounts of cDNA subjected to PCR amplifications were normalized with primers specific for GAPDH sequences. PCR products amplified with Expand Long Template Taq DNA polymerase (Roche) were analyzed by Southern blot hybridization with oligonucleotidic probes. For NMD studies, amplification of EGFP-SC35 cDNAs was performed in the presence of 2 µCi of [α-³²P]dCTP by 22 or 25 PCR cycles with the EGFP-specific (5′-GACGAGCTGTACAAGTCC-3′) and H74 primers. Neomycin mRNA levels were used to normalize for transfection efficiency. Densitometric analyses were performed with a Molecular Dynamics PhosphorImager.

In vitro splicing assays

Transcription vectors used to generate pre-mRNA substrates were constructed as follows. A 765 bp fragment containing SC35 3′UTR sequences (positions 3209–3973 in DDBJ/EMBL/GenBank accession No. X75755) was PCR amplified and subcloned at the SacI and EcoRI sites in the pBKS vector (Stratagene) to generate the wt-RI construct. The wt-CE construct was obtained following subcloning of a 1240 bp PCR-generated fragment (positions 2125–3364) at the SacI and EcoRI sites in pBKS. The βGlo-CE construct was created from the BSSK HG2 vector, which contains the human β globin exon 2–intron 2–exon 3 region (provided by Dr R.Breathnach, INSERM, Nantes, France), by insertion of a 451 bp PCR-generated fragment (positions 2265–2715) at the HincII site in intron 2.

Capped ³²P-labeled pre-mRNA substrates were obtained by in vitro transcription with T7 or T3 RNA polymerase, and HeLa nuclear extracts and cytoplasmic S100 extracts were prepared as described (Bourgeois et al., 1999). In vitro splicing was performed in 25 µl assays containing 9 µl of nuclear extract plus 2 µl of S100 and the basic components for the splicing reaction (Bourgeois et al., 1999). The splicing reaction was carried out in the presence of 60 mM KCl and 3.2 mM MgCl₂, except for the MLA precursor (48 mM KCl and 1.6 mM MgCl₂), for 2 h at 30°C. Some assays were supplemented by purified recombinant SR proteins (7–14 pmol), as indicated. Splicing products were analyzed on 6% denaturing polyacrylamide gels and visualized by autoradiography.

RNA stability measurements

HeLa cells (3.5 × 10⁵) were incubated in the presence of actinomycin D (5 µg/ml) for 4–16 h. Total RNA was analyzed by northern blot hybridization with either the EB650 probe or the H23, H24 and H121 oligonucleotidic probes. To measure RNA stability in the absence of protein synthesis, similar experiments were performed after a 2 h cycloheximide (10 µg/ml) treatment.