SC35 autoregulates its expression by promoting splicing events that destabilize its mRNAs.

SC35 belongs to the family of SR proteins that regulate alternative splicing in a concentration-dependent manner in vitro and in vivo. We previously reported that SC35 is expressed through alternatively spliced mRNAs with differing 3' untranslated sequences and stabilities. Here, we show that overexpression of SC35 in HeLa cells results in a significant decrease of endogenous SC35 mRNA levels along with changes in the relative abundance of SC35 alternatively spliced mRNAs. Remarkably, SC35 leads to both an exon inclusion and an intron excision in the 3' untranslated region of its mRNAs. In vitro splicing experiments performed with recombinant SR proteins demonstrate that SC35, but not ASF/SF2 or 9G8, specifically activates these alternative splicing events. Interestingly, the resulting mRNA is very unstable and we present evidence that mRNA surveillance is likely to be involved in this instability. SC35 therefore constitutes the first example of a splicing factor that controls its own expression through activation of splicing events leading to expression of unstable mRNA.

Distinctive features of Drosophila alternative splicing factor RS domain: implication for specific phosphorylation, shuttling, and splicing activation.

The human splicing factor 2, also called human alternative splicing factor (hASF), is the prototype of the highly conserved SR protein family involved in constitutive and regulated splicing of metazoan mRNA precursors. Here we report that the Drosophila homologue of hASF (dASF) lacks eight repeating arginine-serine dipeptides at its carboxyl-terminal region (RS domain), previously shown to be important for both localization and splicing activity of hASF. While this difference has no effect on dASF localization, it impedes its capacity to shuttle between the nucleus and cytoplasm and abolishes its phosphorylation by SR protein kinase 1 (SRPK1). dASF also has an altered splicing activity. While being competent for the regulation of 5' alternative splice site choice and activation of specific splicing enhancers, dASF fails to complement S100-cytoplasmic splicing-deficient extracts. Moreover, targeted overexpression of dASF in transgenic flies leads to higher deleterious developmental defects than hASF overexpression, supporting the notion that the distinctive structural features at the RS domain between the two proteins are likely to be functionally relevant in vivo.

A novel family of plant splicing factors with a Zn knuckle motif: examination of RNA binding and splicing activities.

An important group of splicing factors involved in constitutive and alternative splicing contain an arginine/serine (RS)-rich domain. We have previously demonstrated the existence of such factors in plants and report now on a new family of splicing factors (termed the RSZ family) from Arabidopsis thaliana which additionally harbor a Zn knuckle motif similar to the human splicing factor 9G8. Although only around 20 kDa in size, members of this family possess a multi-domain structure. In addition to the N-terminal RNA recognition motif (RRM), a Zn finger motif of the CCHC-type is inserted in an RGG-rich region; all three motifs are known to contribute to RNA binding. The C-terminal domain has a characteristic repeated structure which is very arginine-rich and centered around an SP dipeptide. One member of this family, atRSZp22, has been shown to be a phosphoprotein with properties similar to SR proteins. Furthermore, atRSZp22 was able to complement efficiently splicing deficient mammalian S100 as well as h9G8-depleted extracts. RNA binding assays to selected RNA sequences indicate an RNA binding specificity similar to the human splicing factors 9G8 and SRp20. Taken together, these result show that atRSZp22 is a true plant splicing factor which combines structural and functional features of both h9G8 and hSRp20.

Characterization of SRp46, a novel human SR splicing factor encoded by a PR264/SC35 retropseudogene.

The highly conserved SR family contains a growing number of phosphoproteins acting as both essential and alternative splicing factors. In this study, we have cloned human genomic and cDNA sequences encoding a novel SR protein designated SRp46. Nucleotide sequence analyses have revealed that the SRp46 gene corresponds to an expressed PR264/SC35 retropseudogene. As a result of mutations and amplifications, the SRp46 protein significantly differs from the PR264/SC35 factor, mainly at the level of its RS domain. Northern and Western blot analyses have established that SRp46 sequences are expressed at different levels in several human cell lines and normal tissues, as well as in simian cells. In contrast, sequences homologous to SRp46 are not present in mice. In vitro splicing studies indicate that the human SRp46 recombinant protein functions as an essential splicing factor in complementing a HeLa cell S100 extract deficient in SR proteins. In addition, complementation analyses performed with beta-globin or adenovirus E1A transcripts and different splicing-deficient extracts have revealed that SRp46 does not display the same activity as PR264/SC35. These results demonstrate, for the first time, that an SR splicing factor, which represents a novel member of the SR family, is encoded by a functional retropseudogene.

The SR splicing factors ASF/SF2 and SC35 have antagonistic effects on intronic enhancer-dependent splicing of the beta-tropomyosin alternative exon 6A.

Exons 6A and 6B of the chicken beta-tropomyosin gene are mutually exclusive and selected in a tissue-specific manner. Exon 6A is present in non-muscle and smooth muscle cells, while exon 6B is present in skeletal muscle cells. In this study we have investigated the mechanism underlying exon 6A recognition in non-muscle cells. Previous reports have identified a pyrimidine-rich intronic enhancer sequence (S4) downstream of exon 6A as essential for exon 6A 5'-splice site recognition. We show here that preincubation of HeLa cell extracts with an excess of RNA containing this sequence specifically inhibits exon 6A recognition by the splicing machinery. Splicing inhibition by an excess of this RNA can be rescued by addition of the SR protein ASF/SF2, but not by the SR proteins SC35 or 9G8. ASF/SF2 stimulates exon 6A splicing through specific interaction with the enhancer sequence. Surprisingly, SC35 behaves as an inhibitor of exon 6A splicing, since addition to HeLa nuclear extracts of increasing amounts of the SC35 protein completely abolish the stimulatory effect of ASF/SF2 on exon 6A splicing. We conclude that exon 6A recognition in vitro depends on the ratio of the ASF/SF2 to SC35 SR proteins. Taken together our results suggest that variations in the level or activity of these proteins could contribute to the tissue-specific choice of beta-tropomyosin exon 6A. In support of this we show that SR proteins isolated from skeletal muscle tissues are less efficient for exon 6A stimulation than SR proteins isolated from HeLa cells.

Cloning of human 2H9 heterogeneous nuclear ribonucleoproteins. Relation with splicing and early heat shock-induced splicing arrest.

Using antibody 2H9 from our heterogeneous nuclear ribonucleoproteins (anti-hnRNP) monoclonal antibody library, we previously showed in HeLa cells that a 35-37-kDa protein doublet switches from the hnRNP complexes to the nuclear matrix following a 10-min heat shock at 45 degrees C (1 Lutz, Y., Jacob, M., and Fuchs, J. P. (1988) Exp. Cell Res. 175, 109-124). cDNA cloning and sequencing revealed an hnRNP protein (2H9) which is a new member of the hnRNP F, H/H' family. Protein 2H9 displays two consensus sequence-type RNA binding domains (CS-RBD) showing 80-90% homology with two of the three CS-RBDs of hnRNP F and H/H'. Another common feature is the presence of two glycine/tyrosine-rich auxiliary domains located at the C terminus and between the two CS-RBDs. At the functional level we show that specific anti-2H9 peptide antibodies can directly inhibit an in vitro splicing system. Moreover, the 2H9 protein doublet is no more present in nuclear extracts from such briefly stressed cells, which interestingly correlates with the inability of these extracts to catalyze in vitro splicing reactions. Taken together, our data suggest that these proteins are involved in the splicing process and also participate in early heat shock-induced splicing arrest by transiently leaving the hnRNP complexes. These 2H9 proteins, which are encoded by a single gene located on human chromosome 10, were also found to be associated with nuclear bodies in situ.

The human hnRNP-M proteins: structure and relation with early heat shock-induced splicing arrest and chromosome mapping.

With anti-hnRNP monoclonal antibody 6D12 we previously showed in HeLa cells that as early as 10 min after the onset of a heat shock at 45 degrees C, a 72.5-74 kDa antigen doublet leaves the hnRNPs and strongly associates with the nuclear matrix, the effect being reversed after a 6 h recovery at 37 degrees C. cDNA cloning and sequencing enabled us to identify these antigens as hnRNP-M proteins and further to show that the correct sequence differs by an 11 amino acid stretch from the originally published sequence. We also show that monoclonal antibodies raised against synthetic hnRNP-M peptides can directly inhibit in vitro splicing. Furthermore, stressing cells at 45 degrees C for 10 min is sufficient to abolish the splicing capacity of subsequently prepared nuclear extracts which, interestingly, do not contain the hnRNP-M proteins any more. Taken together, our data suggest that these proteins are involved in splicing as well as in early stress-induced splicing arrest. Further in situ hybridization assays located the hnRNP-M encoding gene on human chromosome 19.

Titration of serine/arginine (SR) splicing factors during adenoviral infection modulates E1A pre-mRNA alternative splicing.

Alternative splicing of the adenovirus-2 E1A pre-mRNA involves the use of three 5' splice sites and is modulated during infection because the 13S mRNA and 9S mRNA reactions are predominant during the early and late periods, respectively. We had previously reproduced in vitro the 13S to 9S modulation with nuclear extracts isolated from infected HeLa cells and shown that high molecular weight viral RNAs are involved in this modulation, most likely by sequestering or titrating general splicing factors. To further test this hypothesis, we titrated splicing factors from an uninfected nuclear extract using competitor RNA or by progressive inactivation of splicing factors with monoclonal antibodies. We found that the 13S to 9S modulation occurs when titrating only with certain RNAs (essentially adenoviral RNAs), and also by progressively inactivating the 9G8 SR splicing factor. The demonstration that late nuclear extracts contain levels of active SR splicing factors limiting for the 13S reaction has been made by complementation experiments. We show that late nuclear extracts do not complement SR factor-deficient extracts, whereas late extracts treated with micrococcal nuclease complement them. Furthermore, complementation of late nuclear extracts with each of the three 30-35-kDa SR factors (9G8, SC35, and SF2/ASF) restores an efficient 13S mRNA reaction. Thus, our results provide evidence that the 13S to 9S modulation is triggered through a titration of SR factors required for the 13S mRNA reaction by major late transcripts that accumulate in nuclei late in infection.

The gene encoding human splicing factor 9G8. Structure, chromosomal localization, and expression of alternatively processed transcripts.

The 9G8 factor is a 30-kDa member of the SR splicing factor family. We report here the isolation and characterization of the human 9G8 gene. This gene spans 7745 nucleotides and consists of 8 exons and 7 introns within the coding sequence, thus contrasting with the organization of the SC35/PR264 or RBP1 SR genes. We have located the human 9G8 gene in the p22-21 region of chromosome 2. The 5'-flanking region is GC-rich and contains basal promoter sequences and potential regulatory elements. Transfection experiments show that the 400-base pair flanking sequence has a promoter activity. Northern blot analysis of poly(A)+ RNA isolated from human fetal tissues has allowed us to identify five different species, generated by alternative splicing of intron 3, which may be retained or excised as a shorter version, as well as the use of two polyadenylation sites. We also show that the different isoforms are differentially expressed in the fetal tissues. The persistence of sequences between exon 3 and 4 results in the synthesis of a 9G8 protein lacking the SR domain which is expected to be inactive in constitutive splicing. Thus, our results raise the possibility that alternative splicing of intron 3 provides a mechanism for modulation of the 9G8 function.

Characterization and cloning of the human splicing factor 9G8: a novel 35 kDa factor of the serine/arginine protein family.

By adopting a monoclonal antibody approach, we have identified a novel splicing factor of 35 kDa which we have termed 9G8. The isolation and characterization of cDNA clones indicate that 9G8 is a novel member of the serine/arginine (SR) splicing factor family because it includes an N-terminal RNA binding domain (RBD) and a C-terminal SR domain. The RNA binding domain of 9G8 is highly homologous to those of the SRp20 and RBP1 factors (79-71% identity), but the homology is less pronounced in the cases of SF2/ASF and SC35/PR264 (45-37% identity). Compared with the other SR splicing factors, 9G8 presents some specific sequence features because it contains an RRSRSXSX consensus sequence repeated six times in the SR domain, and a CCHC motif in its median region, similar to the zinc knuckle found in the SLU7 splicing factor in yeast. Complete immunodepletion of 9G8 from a nuclear extract, which is accompanied by a substantial depletion of other SR factors, results in a loss of splicing activity. We show that a recombinant 9G8 protein, expressed using a baculovirus vector and excluding other SR factors, rescues the splicing activity of a 9G8-depleted nuclear extract and an S100 cytoplasmic fraction. This indicates that 9G8 plays a crucial role in splicing, similar to that of the other SR splicing factors. This similarity was confirmed by the fact that purified human SC35 also rescues the 9G8-depleted extract. The identification of the 9G8 factor enlarges the essential family of SR splicing factors, whose members have also been proposed to play key roles in alternative splicing.

Contrasted cis-acting effects of downstream 5' splice sites on the splicing of a retained intron: the adenoviral E1A pre-mRNA model.

The adenoviral E1A pre-mRNA contains an upstream intron (the 216 nucleotide intron) which is spliced only weakly both in vivo and in vitro. We have chosen the E1A transcript as a model to analyse, in vitro, the role of downstream cis-elements involved in the alternative splicing of this retained intron. By using a series of constructs containing specific deletions, mutations and/or truncations, we show that the 13S 5' splice site, positioned 259 nucleotides downstream of the 216 nucleotide intron, is the main cis-element which activates the splicing of this intron. Our results establish the importance of a downstream 5' splice site for the activation of the 3' splice site, which is known to be suboptimal within this retained intron. Unexpectedly, the 12S 5' splice site, although positioned at an ideal distance (121 nucleotides) from the upstream intron, does not exhibit such a cis-acting effect. In contrast, its improvement to a consensus sequence may even result in a slight negative cis-acting effect in the presence of the 13S 5' splice site, which is the first observation of such a feature. We have shown that this unexpected behaviour is due, at least partly, to the unusual characteristics of the wild-type upstream intron, which requires a hairpin structure between the branch sites and the 3' splice site to reduce the operational distance between these two sites. Possible mechanisms involved in the contrasted cis-acting effects of the 13S and 12S 5' splice sites are discussed.

Modulation of alternative splicing of adenoviral E1A transcripts: factors involved in the early-to-late transition.

The E1A pre-mRNA of adenovirus is spliced into three mRNA species (13S, 12S, and 9S mRNAs) by the use of three alternative 5'-splice sites. The 13S and 9S mRNAs predominate during the early and late periods of infection, respectively. With HeLa nuclear extracts isolated in early and late periods of infection, we were able to reproduce a 13S-9S modulation that resembles that occurring in infected cells. An in vitro analysis of the cis-acting parameters involved in the 13S-9S switch indicates that the 13S mRNA splicing inhibition is one of the first events of the late period and leads to the subsequent stimulation of the 9S mRNA reaction. The new abilities of the late nuclear extract for the 9S mRNA reaction were also confirmed by analyzing splicing of a major late transcript containing leaders 1 and 2 separated by the wild-type intervening sequence (IVS) of 1021 nucleotides. Complementation experiments show that the trans-acting factor(s) are micrococcal nuclease sensitive. They were partially characterized by induction experiments, and we show that the primary factors responsible for the 13S-9S modulation in vitro are viral RNAs of high molecular weight that accumulate late in infection. We postulate that the splicing modulation of E1A pre-mRNA results from an indirect mode of action for these viral RNAs, based on a sequestration of common splicing factors that are not present in vast excess in HeLa cells.

Differential block of U small nuclear ribonucleoprotein particle interactions during in vitro splicing of adenovirus E1A transcripts containing abnormally short introns.

We have studied the consequences of decreasing the donor site-branch site distance on splicing factor-splice site interactions by analyzing alternative splicing of adenovirus E1A pre-mRNAs in vitro. We show that the proximal 13S donor site has a cis-inhibiting effect on the 9S and 12S mRNA reactions when it is brought too close to the common branch site, suggesting that the factor interactions in the common 3' part of the intron are impaired by the U1 small nuclear ribonucleoprotein particle (snRNP) binding to the displaced 13S donor site. Further analysis of the interactions was carried out by studying complex assembly and the accessibility to micrococcal nuclease digestion of 5'-truncated E1A substrates containing only splice sites for the 13S mRNA reaction. A deletion which brings the donor site- branch site distance to 49 nucleotides, which is just below the minimal functional distance, results in a complete block of the U4-U5-U6 snRNP binding, whereas a deletion 15 nucleotides larger results in a severe inhibition of the formation of the U2 snRNP-containing complexes. Sequence accessibility analyses performed by using the last mini-intron-containing transcript demonstrate that the interactions of U2 snRNP with the branch site are strongly impaired whereas the initial bindings of U1 snRNP to the donor site and of specific factors to the 3' splice site are not significantly modified. Our results strongly suggest that the interaction of U1 snRNP with the donor site of a mini-intron is stable enough in vitro to affect the succession of events leading to U2 snRNP binding with the branch site.

The 216-nucleotide intron of the E1A pre-mRNA contains a hairpin structure that permits utilization of unusually distant branch acceptors.

A recently characterized 216-nucleotide intron-splicing reaction occurs within the adenovirus E1A pre-mRNA through the use of three branch acceptor sites, located at 59, 55, and 51 nucleotides from the 3' splice site. To investigate the role of the cis-acting sequence elements in the selection of such unusually distant branch sites, transcripts differing in sequence downstream of the branch sites were analyzed for in vitro splicing. Initial results suggested that secondary structure could be involved in the use of distant branch sites. The involvement of a hairpin structure, including a nine-G C-base-pair stem, was supported by the results of site-directed mutagenesis analyses. Mutations that destroyed or weakened this hairpin resulted in an inefficient splicing reaction. In contrast, complementary mutation or deletion of two bulges, which involved a restoration or reinforcement of the hairpin, resulted in a reactivation or improvement of the splicing efficiency, respectively. Therefore, we conclude that the hairpin structure shortens the operational distance between the 3' splice site and the branch acceptors and brings the branch sites into the branch-permissive window, 18 to 40 nucleotides upstream of the 3' splice site. Our results confirm the importance of the constraint of distance for the splicing reaction and show that this constraint may be overcome by means of a stable hairpin formation.

A 5' splice-region G----C mutation in exon 1 of the human beta-globin gene inhibits pre-mRNA splicing: a mechanism for beta+-thalassemia.

We have characterized a Mediterranean beta-thalassemia allele containing a sequence change at codon 30 that alters both beta-globin pre-mRNA splicing and the structure of the hemoglobin product. Presumably, this G----C transversion at position -1 of intron 1 reduces severely the utilization of the normal 5' splice site since the level of the Arg----Thr mutant hemoglobin (designated hemoglobin Kairouan) found in the erythrocytes of the patient is very low (2% of total hemoglobin). Since no natural mutations of the guanine located at position -1 of the CAG/GTAAGT consensus sequence had been isolated previously, we investigated the role of this nucleotide in the constitution of an active 5' splice site by studying the splicing of the pre-mRNA in cell-free extracts. We demonstrate that correct splicing of the mutant pre-mRNA is 98% inhibited. Our results provide further insights into the mechanisms of pre-mRNA maturation by revealing that the last residue of the exon plays a role at least equivalent to that of the intron residue at position +5.

In vitro splicing of adenovirus E1A transcripts: characterization of novel reactions and of multiple branch points abnormally far from the 3' splice site.

During the analysis of the in vitro alternative splicing of the natural E1A transcript of adenovirus, other minor reactions were detected (Schmitt et al., 1987, Cell 50, 31-39). We report here their characterization. The first reaction concerns the excision of a 216 nucleotide intron delineated by the 9S 5' splice site and a 3' splice site 216 nucleotides downstream. It can occur on the premRNA transcript and the 13S and 12S mRNA species. Strikingly, the reaction uses one of 3 branch points located 51, 55 or 59 residues upstream of the 3' splice site, a distance which is unusually long since all the branch points mapped up to now are located between 18-37 nucleotides of the 3' splice site. The dramatic accumulation of the corresponding lariat intermediates, likely related to this long spacing indicates that the second splicing step is relatively unefficient. The second kind of reaction analysed is a cryptic splicing which uses a 3' splice site generated by the junction of the 13S mRNA exons, and leads to the formation of psi 12S and psi 9S mRNAs. In vitro, this reaction occurs only from a 13S mRNA transcript, and not from the 13S mRNA newly formed in the splicing assay, consistent with what has been observed in vivo. Thus, both the well known alternative and the minor reactions occurring in vivo from E1A premRNA and mRNAs are detected in vitro, implying that most of the alternative splicing machinery is reconstituted in the in vitro system.

Alternative splicing of E1A transcripts of adenovirus requires appropriate ionic conditions in vitro.

We have developed an in vitro splicing system using a HeLa cell nuclear extract that is highly active for the alternative splicing of the natural E1A transcripts. The efficiency of using the three alternative 5' splice sites is strongly dependent on the ionic conditions in the reaction, and the simultaneous production of the 13S, 12S, and 9S mRNA species is observed only at appropriate salt concentrations. All the intermediate and final splicing products have been extensively characterized and it has been demonstrated that the same major branch site is used for all the alternative reactions. The ratio of 13S to 9S mRNAs formed is close to that observed in vivo early in infection, suggesting that most of the mechanisms giving rise to alternative splicing are preserved in vitro.

The different intron 2 species excised in vivo from the E2A premRNA of adenovirus-2: an approach to analyse alternative splicing.

In the early period of cellular infection by adenovirus 2, the E2A region gives rise to 2 major mRNA species of 2.0 and 2.3 kilobases, formed by alternative excisions of intron 2 (Gattoni et al., 1986, J. Mol. Biol. 187, 379-307). We have analysed the excision pathways of this intron. Two major intron species of 626 and 337 nucleotides, generated by the use of 2 consensus 3' splicing sites and a minor intron species of 520 nucleotides, generated by the use of another weaker 3' splicing site, are identified, the 3 species sharing a common 5' splicing site. They are detected predominantly in the lariat form. For the 2 major species we analyzed, the branched nucleotides are localized at consensus branching sequences, 26 or 25 nucleotides upstream from the 3' terminal AG. Our results confirm that the first reactions of cleavage at the 5' end of introns and branching occur in vivo as described in in vitro systems. The second predominant form of intron 2 is the linear segment, whereas the nicked lariat form which is very minor, might not be a genuine product of in vivo splicing. All intron 2 molecules show practically intact 5' and 3' terminal sequences, indicating that they are well protected against nuclease attack throughout their life. Therefore, these results indicate that the primary reaction following the excision of the lariat intron is debranching. In addition, the existence of a potential 5' splicing site contiguous to the major internal 3' splicing site raised the possibility of an elimination of the major 626 nucleotide intron in 2 cycles of excision. However, we demonstrate that intron 2 is systematically excised by a one cycle process, which is likely to represent the general rule for the production of correctly spliced mRNA.

Splicing of the E2A premessenger RNA of adenovirus serotype 2. Multiple pathways in spite of excision of the entire large intron.

During the early period of infection, the 4.9 kb (kb = 10(3) bases) E2A premRNA of adenovirus serotype 2 is matured mainly into a 2.0 kb mRNA by excision of introns of 2233 and 626 nucleotides. In order to define all the possible steps of splicing occurring in vivo, we characterized splicing intermediates present after a limiting treatment of cells with cycloheximide. Three complementary methods of analysis were used: RNA transfer analysis, S1 nuclease mapping and complementary DNA-RNase assay. Our principal conclusions concerning the poly(A)+ species are as follows. The RNA intermediate family detected is more complex than expected, since two major RNA intermediates of 4.6 kb and 4.3 kb, two minor intermediates of 2.9 kb and 2.6 kb, and a 2.3 kb RNA, which represents a minor alternative mRNA form, are revealed. Despite its large size and the presence of multiple internal donor and acceptor signals, intron 1 is exclusively excised as a whole. Intron 2 is either primarily excised as a whole, removing the standard 626-nucleotide sequence, or a smaller sequence of 337 nucleotides is removed, generating the 2.3 kb alternative mRNA. Kinetics of the ligation reaction demonstrate that the minimal time for excision of intron 2 is no more than two minutes, indicating a high level of co-ordination of the multiple individual reactions occurring during excision of an intron. Besides the major pathway for E2A premRNA splicing, namely the excision first of intron 2, followed by the excision of intron 1 after a lag time of five minutes, a minor pathway (used with a frequency of 10%) can be detected where the order of intron excision was inverted. With the alternative variant of excision of intron 2, at least three different pathways are therefore used to mature the E2A premRNA. RNA intermediates resulting from the cleavage at the 5' end of introns and branching can be detected by S1 mapping experiments, but their low accumulative level (1% relatively to the initial premRNA) precluded their direction by RNA transfer experiments and their complete characterization.

High specificity of the cDNA-RNase assay to detect accurate splicing in vitro.

e previously developed a splicing assay (Keohavong et al., 1982) that we designated as a cDNA-RNase assay to analyze the ligation reaction between exons of premessenger RNA during in vivo or in vitro splicing. It was important to determine the specificity of this splicing assay, since the accuracy of in vitro splicing must always be demonstrated clearly. To do this, we constructed DNA probes derived from adenovirus E1A cDNA carrying deletions or insertions of 2-6 bases. After hybridizing them to the wild-type mRNA, the ability of single-strand-specific RNases to detect small mismatches of the RNA-DNA hybrids was examined. The demonstration that an imprecision in the splicing reaction of as little as 2 nucleotides can be detected with an efficiency of 99% indicates the high specificity of the splicing assay and its usefulness for the verification of accurate splicing in in vitro systems.