Logo of molcellbPermissionsJournals.ASM.orgJournalMCB ArticleJournal InfoAuthorsReviewers
Mol Cell Biol. 2001 Mar; 21(6): 1986–1996.

Switch in 3′ Splice Site Recognition between Exon Definition and Splicing Catalysis Is Important for Sex-lethal Autoregulation


Maintenance of female sexual identity in Drosophila melanogaster involves an autoregulatory loop in which the protein Sex-lethal (SXL) promotes skipping of exon 3 from its own pre-mRNA. We have used transient transfection of Drosophila Schneider cells to analyze the role of exon 3 splice sites in regulation. Our results indicate that exon 3 repression requires competition between the 5′ splice sites of exons 2 and 3 but is independent of their relative strength. Two 3′ splice site AG's precede exon 3. We report here that, while the distal site plays a critical role in defining the exon, the proximal site is preferentially used for the actual splicing reaction, arguing for a switch in 3′ splice site recognition between exon definition and splicing catalysis. Remarkably, the presence of the two 3′ splice sites is important for the efficient regulation by SXL, suggesting that SXL interferes with molecular events occurring between initial splice site communication across the exon and the splice site pairing that leads to intron removal.

Alternative splicing is a versatile mechanism of gene expression regulation that can turn on or off genes or generate different protein isoforms with distinct biological properties (for recent reviews, see references 12 and 32). Despite the prevalence of alternative splicing in higher eukaryotes, relatively little is known about the underlying molecular mechanisms of regulation. Genes that participate in the Drosophila melanogaster sex determination cascade have become good model systems for understanding splicing control because genetic data have defined both regulatory factors and their target pre-mRNAs (reviewed in reference 46).

Sex-lethal (SXL), an RNA- binding protein with preference for U-rich sequences, promotes female-specific patterns of splicing on at least three transcripts: (i) its own pre-mRNA, where SXL promotes exon 3 skipping (6); (ii) transformer pre-mRNA, where SXL promotes a switch between alternative 3′ splice sites (9); and (iii) male-specific-lethal 2, where SXL promotes retention of an intron at the 5′ untranslated region (4, 28, 62). The biological consequences of these alternative splicing events are dramatic and control multiple aspects of sexual determination in the fruitfly. Sxl and tra transcripts spliced in the female-specific mode encode SXL and TRA proteins, while transcripts that follow the alternative splicing pathway can only encode truncated polypeptides. Expression of full-length TRA controls somatic sexual differentiation and sexual behavior, while expression of SXL maintains the female differentiation state throughout the life of the fly. Retention of the msl-2 intron allows SXL to act as a translational repressor and to inhibit MSL-2 protein expression, thereby turning off dosage compensation in female flies (5, 20, 29).

The mechanism by which SXL controls tra splicing has been investigated in vivo and in vitro. Two 3′ splice sites are present in intron 1. The proximal site, used in both males and females (hence the name non-sex-specific) contains a high-affinity binding site for SXL at the polypyrimidine (Py) tract. The distal site is used in a female-specific fashion. Evidence from experiments in transgenic flies and transient transfections of cells in culture, as well as in vitro biochemical analysis, indicate that SXL represses the use of the non-sex-specific site (27, 51, 54). In vivo and in vitro results are consistent with a model in which SXL prevents the binding of the splicing factor U2AF to the Py tract of the non-sex-specific site, thereby diverting U2AF and splicing to the female-specific site (21, 54). Blockage of U2AF binding is also important for regulation of msl-2 splicing in vitro (34).

Several lines of evidence suggest a different mechanism for Sxl autoregulation. First, although the Py tract associated with one of the 3′ splice sites preceding exon 3 in D. melanogaster contains a relatively long stretch of uridines and is a potential binding site for SXL (25), its mutation does not abolish regulation (26, 41), in contrast to tra (27, 51). Second, multiple U-rich sequences, relatively distant from the 5′ and 3′ splice sites, contribute to exon 3 skipping (26, 41), and cooperative binding of SXL to these sequences, mediated through an amino-terminal glycine and an asparagine-rich domain, is important for regulation (55). Third, ectopic expression in male transgenic flies of a chimeric protein in which the splicing activation domain of U2AF was fused to SXL RNA-binding domains results in disruption of tra regulation but not of Sxl regulation (21). Because this chimeric protein promotes the splicing of pre-mRNAs containing SXL binding sites at the Py tract (54), as U2AF does, these data argue that antagonizing U2AF activity is insufficient to explain SXL-mediated exon skipping.

Results using transgenic flies suggest that a key regulatory step in Sxl autoregulation is the inhibition of exon 3 5′ splice site (26). Inhibition of 5′ splice site recognition can result in exon skipping because of defects in exon definition. Exon definition is the process by which early recognition of splice sites in relatively short exons embedded in long introns occurs through molecular interactions involving factors bound to the 3′ splice site preceding the exon and the downstream 5′ splice site (reviewed in reference 7). It is believed that the factors involved in this process are the same as those involved in early communication between splice sites across introns: U1 snRNP bound to the 5′ splice site, U2AF65 bound to the Py tract, and bridging activities, for which members of the SR family of factors are strong candidates. How the communication between splice sites across exons is switched to communication across introns for selection of the actual splicing partners is currently unknown.

Here we report a role for the 3′ splice site AG in exon definition that can be physically separated from its role in splicing catalysis, arguing for a switch in splice site recognition in the transition between these two processes. Since the presence of two 3′ splice sites appears to be important for regulation of exon skipping by SXL, this observation opens the possibility that the presence of SXL interferes with molecular events leading to such a switch.


Plasmids and mutagenesis.

Plasmids copia-SxlTE234, pBShsp-cat, and pBShsp-SxlF1cDNA were described previously (41).

Deletion or substitution mutants in the Sxl minigene including exons 2 to 4 were obtained from plasmid copia-SxlTE234 using the procedure described by L. O. F. Penalva and J. Valcárcel (Techical tips online [http://tto.trends.com]). All mutations were confirmed by sequencing. The sequence changes introduced were as follows: 5′ splice site mutants, as indicated in Fig. Fig.2B;2B; 5′ss ex2Mut, the sequence of exon 2 5′ splice site (in DNA form AG/GTAAA, the slash indicating the exon-intron boundary) was changed into CAATGAA; 3′ splice site mutants, as indicated in Fig. Fig.3B3B and and5A;5A; ΔInt2, deletion of intron 2 sequences between positions 12 and 2919 (the sequence between exon 2 5′ss and the distal 3′ splice site of exon 3 is indicated in Fig. Fig.3D);3D); and ΔInt2MutAG, the sequence AG CCCAGAAAGAAGCAG, corresponding to the distal 3′ splice site AG and the 5′ end of exon 3 was replaced by (CA)8.

FIG. 2
Role of the relative strength of 5′ splice sites. (A) Alignment of the 5′ splice site regions of exons 2 and 3 of Sxl genes from D. melanogaster (mel), D. subobscura (sub), and D. virilis (vir). (B) Sequences of the 5′ splice site ...
FIG. 3
Role of exon 3 alternative 3′ splice sites. (A) Alignment of exon 3 3′ splice site region of Sxl genes from D. melanogaster (mel), D. subobscura (sub), and D. virilis (vir). The proximal and distal AG's are separated by a space from the ...
FIG. 5
Switch in 3′ splice site recognition is important for SXL regulation. (A) Sequences of exon 3 proximal 3′ splice site mutants. Exon 3 is represented as a gray box; the positions of the proximal and distal 3′ splice site, as well ...

Transfection and RNA isolation.

Transfections were performed using Lipofectin (Gibco) according to the manufacturer's recommendations. Typically, 3 μg of each plasmid (pBSHS-TE234 or mutant derivatives and either pBShsp-Sxl-CF1 or pBShsp-cat) were used to transfect Schneider cell cultures at 80% density in 60-cm2 plates. At 30 h after transfection, cells were harvested, washed once with cold 1 × phosphate-buffered saline, and then lysed in 0.5% NP40, 10 mM Tris-HCl (pH 8.5), 1.5 mM MgCl2, and 150 mM NaCl. Nuclei were spun down for 5 min at 800 × g, and the supernatant (cytoplasmic fraction) was transferred to a new tube, digested for 20 min at 65°C with proteinase K (0.2 mg/ml) in 1% Sodium dodecyl sulfate – 100 mM Tris (pH 7.5)–12.5 mM EDTA–150 mM NaCl, extracted twice with phenol-chloroform and once with chloroform, and precipitated with isopropanol; the pellet was then washed with 75% ethanol and resuspended in H2O. The purified RNA was quantified by determining the absorbance at 265 nm.

Preparation of total RNA from Drosophila adult flies.

Total RNA from adult flies was prepared as described previously (11).


A total of 15 μg of total RNA purified from transfected cells was treated twice with 10 U of RNase-free DNase (Roche) for 1 h at 37°C. RNA was purified after each DNase digestion by extraction with phenol-chloroform and ethanol precipitation. Reverse transcription (RT) was carried out using Expand Reverse Transcriptase (Roche) according to the manufacturer's instructions. PCR was performed as described by Sakamoto et al. (41): 2 min at 94°C, followed by 25 cycles of 1 min at 94°C, 1 min at 60°C, and 1 min at 72°C, and then a final 5 min at 72°C, using primers PT1 and PT2, designed by Sakamoto et al. (41) to provide sequence tags for the transfected minigenes. PCR products were analyzed in 1.8% agarose gels. For the analysis of intron retention in clone 5′ss ex2Mut (Fig. (Fig.2D),2D), a primer corresponding to exon 3 sequences (CACTGACTCTTAAGATAGTATGTAG) replaced PT1 in the PCR. For the analyses shown in Fig. Fig.5,5, primer PT1 was replaced by primer ex2-14 (TCAAGTCAACTGCAACTCACC).

The RT-PCRs were quantitative under these conditions (see Fig. Fig.11 and data not shown). All of the data presented were obtained using these conditions, and results that required further amplification (for example, because of low transfection efficiency or low RNA yields) were disregarded.

FIG. 1
Minigene structure, conserved sequences, and validation of the assay. (A) Schematic representation of the Sxl minigene present in plasmid copia-SxlTE234. Boxes represent exons 2 to 4 of Sxl mRNAs, thick lines represent introns, and patterns of splicing ...

Analysis of exon 3 3′ splice site usage.

A total of 15 μg of total RNA from adult males (see Fig. Fig.4)4) or from transfected cells (see Fig. Fig.6)6) was analyzed by RT-PCR using the following primers: GGTTGCTTTGCGTTACAAAAAC (antisense, exon 3) and CCCCCATATGGCTACAACAA (sense, exon 2). PCR reactions were amplified for 25 cycles of 1 min at 94°C, 1 min at 55°C, and 30 s at 72°C, followed by a final extension of 10 min at 72°C. The products of amplification were analyzed by electrophoresis on 6% polyacrylamide non denaturing gels.

FIG. 4
Relative use of exon 3 3′ splice sites in Drosophila male flies. RT-PCR was performed using primers corresponding to exons 2 and 3 that should generate products of amplification differing by 18 bp, depending on whether the proximal or the distal ...
FIG. 6
Use of AGs in proximal 3′ splice site mutants. RNA isolated from transfected cells was analyzed by RT-PCR as in Fig. Fig.4.4. The transfected reporter plasmid is indicated at the top. Cotransfection with control plasmid (−) or ...


To address the role of competing splice sites in Sxl autoregulation, Drosophila Schneider cells were transiently transfected with a minigene containing the Sxl genomic region between exons 2 and 4 (Fig. (Fig.1A)1A) under the copia transposable element promoter or with mutants of this construct. RNA was isolated from the transfected cultures and analyzed by RT-PCR as described previously (41) using oligodeoxynucleotides corresponding to transcribed vector sequences (PT1 and PT2 in Fig. Fig.1A),1A), thus allowing specific detection of transcripts derived from the Sxl minigene and not endogenous Sxl transcripts.

To verify the sensitivity of our detection procedure to changes in the relative accumulation of alternatively spliced transcripts, we performed the experiment shown in Fig. Fig.1B.1B. Cotransfection of the plasmid containing the Sxl minigene (copia-SxlTE234) with the vector pBShsp-cat resulted in accumulation of transcripts that included exon 3 (lane 1), as is the case in male flies. Cotransfection of copia-SxlTE234 with an expression plasmid containing SXL cDNA under a heat shock promoter (pBShsp-SxlF1cDNA) resulted in the accumulation of transcripts that skipped exon 3 (lane 2), as in female flies. When these two types of RNAs were mixed in equal amounts, the products of RT and PCR amplification accumulated in similar proportion (lane 3). When the RNAs were mixed at 1:3 or 3:1 ratios, the relative abundance of the amplification products changed proportionally (lanes 4 and 5). Figure Figure1C1C shows a comparison between the ratios of input RNAs and the ratios between the signals of the amplified DNA products quantified by SYBR Green (Molecular Probes) staining and phosphorimager analysis. The data indicate that our assay system allows the accurate detection of threefold (and probably smaller) changes in the relative accumulation of alternatively spliced products. They also indicate that there are no significant differences in the relative amplification efficiencies of the two products.

Regulation of 5′ splice site usage.

Competing sites of different strengths often establish default patterns of alternative splicing that are then subject to modulation by regulatory factors (for a review, see reference 12). It has been proposed that blockage of the exon 3 5′ splice site is the key regulatory step in SXL-induced skipping of exon 3 (26). Both exon 2 and exon 3 5′ splice sites are conserved among three Drosophila species (melanogaster, suboscura, and virilis) which have diverged for ~60 million years (10, 38). The 5′ splice site of exon 3 is predictably stronger in the three species, particularly because of the presence of a guanidine at intronic position 5 which is conserved in 85% of higher eukaryotic 5′ splice sites (22) but absent from the 5′ splice site of exon 2. This configuration of splice site strengths could reflect that proper default and/or regulated Sxl splicing requires a strong 5′ splice site in exon 3, a weak 5′ splice site in exon 2, or both. To test these hypotheses, a series of mutant minigenes were prepared (Fig. (Fig.2B)2B) in which (i) both exons contained identical 5′ splice sites, either that of exon 2 [mutant 5′ss(2-2)] or that of exon 3 [mutant 5′ss(3-3)]; (ii) the relative position of the two 5′ splice sites was swapped [mutant 5′ss(3-2)]; and (iii) a strong consensus 5′ splice site was placed in exon 2, in competition with either the natural 5′ splice site of exon 3 [mutant 5′ss(CS-3)] or with the exon 2 5′ splice site placed in exon 3 [mutant 5′ss(CS-2)]. Figure Figure2C2C shows that the accumulation of alternatively spliced products was very similar in all cases, both in the absence or in the presence of SXL. These data argue against a role for the precise identity and/or strength of the competing 5′ splice sites in either establishing the default splicing pathway or in achieving regulation by SXL.

We noticed that, in this particular series of experiments, an additional product of amplification accumulated in the absence of SXL. This product was purified and cloned. Sequence analysis revealed that it corresponds to the use of a cryptic 3′ splice site 24 nucleotides upstream from the natural 3′ splice site associated with exon 4. This cryptic site is spliced to exon 3 5′ splice site in the absence of SXL and apparently is not spliced to exon 2 in the presence of SXL.

To test whether inhibition of exon 3 5′ splice site could occur in the absence of a competing 5′ splice site, exon 2 5′ splice site was inactivated by mutation (5′ss ex2Mut). The results shown in Fig. Fig.2D2D indicate that in the absence of a functional 5′ splice site in exon 2, the blockage of the exon 3 5′ splice site was significantly less dramatic than for wild-type RNA, despite the fact that all the sequences required for SXL function are present in the transcript. Although we cannot rule out that SXL still affects the kinetics of splicing but shows little effect on the accumulation of final products, the result argues that the effects of SXL are reduced in the absence of a functional competing 5′ splice site. This conclusion is also consistent with previous results analyzing minigenes in which both exon 2 and intron 2 were deleted (41).

Regulation of 3′ splice site usage.

A feature conserved among Sxl genes is the presence of two 3′ splice sites preceding exon 3, of which the distal site is associated with a more extensive Py tract (Fig. (Fig.3A).3A). Sequencing of cDNA clones from D. melanogaster male flies indicated that both sites are used (6). To test the possible roles of the alternative sites, a series of mutant minigenes were generated (Fig. (Fig.3B).3B). Deletion of the distal site, including both the Py tract and the 3′ splice site AG (3′ssdΔ1), resulted in significant accumulation of transcripts in which exon 3 was skipped in the absence of SXL (Fig. (Fig.3C,3C, lane 3). This result suggests that the distal 3′ splice site, with its long and U-rich Py tract, plays an important role in exon 3 definition.

Because several AGs follow the distal 3′ splice site and because it has been reported that downstream AGs can be activated upon mutation of natural 3′ splice sites (50), we reasoned that in order to determine the effect of mutations at the distal AG and to prevent activation of the downstream AGs, these needed to be either deleted (3′ssdΔ2) or mutated (3′ssdMut), together with the natural AG. Remarkably, deletion (3′ssdΔ2) or substitution (3′ssdMut) of the 3′ splice site AG, without introducing changes at the Py tract, resulted in even stronger defects in exon definition (lanes 5 and 9, respectively). The downstream AG-rich stretch had, by itself, no effect in either default or regulated splicing (3′ssdΔ4, lanes 11 and 12), indicating that the loss of exon definition in mutants 3′ssdΔ2 and 3′ssdMut could be attributed exclusively to mutation of the distal 3′ splice site AG. Accordingly, mutation of the distal 3′ splice site AG alone to CA also resulted in efficient exon skipping even in the absence of SXL (3′ssdCA, lanes 13 and 14), indicating that downstream AGs could not be used for exon 3 definition.

The effects of deletion of the Py tract alone (3′ssdΔ3, lane 7) or of deletion of both signals (3′ssdΔ1, lane 3) appear less dramatic than the effects of deletion or mutation of the 3′ splice site alone. This difference may be related to the fact that cryptic splice sites become activated in these mutants: the amplification products corresponding to inclusion of exon 3 in these mutants appear as a doublet (lanes 3 and 7); cloning and sequencing of the products revealed that both the proximal 3′ splice site and, unexpectedly, a cryptic 3′ splice site located 20 nucleotides upstream from the proximal 3′ splice site in intron 2 (indicated by an arrow in Fig. Fig.3B)3B) were utilized. Sequencing also revealed an additional minor source of heterogeneity in the use of a cryptic 3′ splice site 24 nucleotides upstream of exon 4, as observed in Fig. Fig.22C.

We conclude that deletion of the Py tract associated with the downstream 3′ splice site of exon 3 causes, in the majority of transcripts, either skipping of the exon or activation of a cryptic 3′ splice site upstream. Deletion of the AG dinucleotide associated with the distal 3′ splice site causes efficient exon skipping. Taken together, the results argue that sequences associated with the distal 3′ splice site play a critical role in the proper definition of exon 3. Efficient exon 3 inclusion in the absence of SXL is essential for the viability of male flies (46) (see also Discussion).

Previous work identified transcripts that used either the proximal or the distal 3′ splice sites associated with Sxl exon 3 (6). The strong effects of mutation of the distal 3′ splice site in exon definition, however, suggest that the majority of the splicing events between exons 2 and 3 implicate recognition of the distal site. This would also be consistent with the predicted relative strength of their Py tracts.

To analyze the relative use of the distal and proximal sites in RNAs isolated from the transfected cells, RT-PCR analysis was performed using oligonucleotide primers that allowed the amplification of the sequences at the junction between exons 2 and 3. The products of amplification were cloned, and 20 independent clones were sequenced. Surprisingly, 70% of the clones sequenced corresponded to cDNAs in which the proximal (and predictably weaker) 3′ splice site was used. Similar estimates were obtained by electrophoretic analysis of the amplification products (see below).

To further validate the analysis, the same RT-PCR approach was used to analyze the use of exon 3 alternative 3′ splice sites in RNAs purified from male flies. The products of amplification were fractionated on a polyacrylamide gel that allows separation of the alternative products, that differ by 18 bp. The results (Fig. (Fig.4)4) show that the majority of the transcripts were spliced to the proximal site (compare lanes 1 and 3). This result was also confirmed by cloning the amplification products and sequencing 25 independent clones. As was the case for the minigene used in transient-transfection assays, more than 70% of the transcripts were spliced to the proximal 3′ splice site. We conclude that while the distal 3′ splice site of exon 3 has an important role in exon definition, the majority of the actual splicing events take place using the proximal 3′ splice site.

To further substantiate a role for the 3′ AG in exon definition, uncoupled from its function in splicing catalysis, we prepared a minigene in which most of intron 2 was deleted such that splicing between exons 2 and 3 could not take place. The deletion also removed the proximal 3′ splice site as well as part of the Py tract associated with the distal site (Fig. (Fig.3D).3D). In this construct (ΔInt2) the 5′ splice sites of exons 2 and 3 compete for the only available 3′ splice site of exon 4. In the absence of SXL, exon 3 was spliced to exon 4 (Fig. (Fig.3D,3D, lane 1). The AG corresponding to exon 3 distal 3′ splice site, as well as other AGs immediately downstream, were mutated in construct ΔInt2MutAG (Fig. (Fig.3D).3D). A significant fraction of splicing took place between exons 2 and 4 in these transcripts (Fig. (Fig.3D,3D, lane 2). This result indicates that the distal AG plays a critical role in the recognition and activation of exon 3 5′ splice site despite the fact that this AG cannot participate in splicing reactions, and the associated Py tract has been significantly shortened (from eight to three U's).

Role of the proximal 3′ splice site.

The results presented above indicate that the distal 3′ splice site is essential for exon definition, while the proximal site is preferentially used in catalysis. The analysis of mutants at the proximal 3′ splice site was therefore particularly intriguing. Different concentrations of SXL-encoding plasmid were cotransfected together with the reporter to quantitatively assess differences in response to the SXL protein, which accumulated in proportion to the amount of DNA transfected (data not shown). Mutation of the AG associated with the proximal 3′ splice site to CA (Fig. (Fig.5A,5A, A,3′sspCA)3′sspCA) did not have any effect in exon 3 definition (Fig. (Fig.5B,5B, lane 5). Cloning and sequencing of the amplification product, as well as RT-PCR analyses (Fig. (Fig.6,6, lanes 3 and 4), indicated that the distal 3′ splice site AG was utilized, indicating that use of the proximal site is not strictly required for exon 3 inclusion. Strikingly, however, exon skipping in the presence of SXL was strongly reduced (Fig. (Fig.5B,5B, compare lanes 2 to 4 with lanes 6 to 8): 100-fold-higher concentrations of SXL-encoding plasmid were required to achieve 50% exon skipping. These differences are likely to have physiological relevance because even twofold variations in SXL expression can have important developmental consequences (46). The dramatic effects of these minimal mutations are in sharp contrast with the modest effects that individual (or even multiple) deletion of other regulatory sequences (e.g., SXL-binding sites) have in SXL function (26, 41). Taken together, the data suggest that dual recognition of the 3′ splice sites of exon 3 is critical to allow efficient regulation by SXL.

One possible caveat of this result was that the mutation not only inactivated the proximal 3′ splice site but also affected a putative branch point associated with the distal site. Although the precise location of the branch point is not known, it is conceivable that the sequence changes introduced in mutant 3′sspCA could have improved the strength of the distal 3′ splice site branch point, somehow antagonizing the repressive effect of SXL. To address this issue, two new mutants were prepared in which (i) the proximal AG was mutated to GG, which should be neutral regarding the creation of cryptic branch points, and (ii) the proximal AG and adjacent sequences were deleted (Fig. (Fig.5A,5A, A,3′sspΔ),3′sspΔ), thus disrupting a branch point associated with the distal 3′ splice site in that region. Analysis of these mutants indicated that neither the AG-to-GG substitution (data not shown) nor deletion of the AG affected exon 3 inclusion (Fig. (Fig.5B,5B, lane 9), but they significantly reduced the effect of SXL in exon skipping (Fig. (Fig.5B,5B, compare lanes 2 to 4 and lanes 10 to 12), as was the case with the AG-to-CA mutation. RT-PCR analysis showed that the distal AG was used in the 3′sspΔ1 mutant as well (Fig. (Fig.6,6, lane 5), confirming that the distal AG can functionally undergo catalysis in the absence of a proximal AG. Taken together, the results of Fig. Fig.5B5B suggest that mutation or deletion of the proximal 3′ splice site compromises SXL regulation and argue that the occurrence of a switch in 3′ splice site recognition between exon definition and splicing catalysis allows SXL to efficiently interfere with exon 3 inclusion.

The switch in 3′ splice site recognition could take place by very different mechanisms, depending on whether or not the Py tract associated with the proximal site is functionally recognized by early splicing factors such as U2AF65/35 (see Discussion). To address this issue, a mutant was generated in which all the uridines in the Py tract associated with the proximal 3′ splice site were mutated to cytidines (Fig. (Fig.5A).5A). Such mutations are predicted to result in significant decreases in the affinity of U2AF65 binding (47, 48, 61). The results of Fig. Fig.5C5C indicate that the mutation was rather neutral regarding both exon definition and regulation by SXL, suggesting that the strength of the Py tract associated with the proximal site is not important for 3′ splice site activation or regulation and arguing that the switch in 3′ splice site recognition involves a single recognition event at the Py tract and possibly a single branch point recognition event.


We report in this manuscript that two 3′ splice site AGs associated with Sxl exon 3 have very different roles in alternative splicing of this exon. The distal AG is critical for proper exon inclusion, presumably through interactions with factors bound to the 5′ splice site (e.g., U1 snRNP) across the exon (7). The proximal AG, in contrast, is preferentially used for catalysis, and its mutation, rather than affecting exon definition, compromises efficient regulation by SXL.

Two separable steps in 3′ splice site recognition.

Analyses of protection to RNase H-mediated degradation indicated that the 3′ splice site AG is recognized twice during the splicing reaction in higher eukaryotes, first at early steps of spliceosome assembly and then again at the time of the second transterification reaction leading to exon ligation (45). Several lines of evidence argue that the two steps are, at least to some degree, independent. First, pre-mRNAs containing 3′ splice sites with strong Py tracts can undergo the first transterification reaction in vitro in the absence of a 3′ splice site AG (39), as is the general case for budding yeast pre-mRNAs (36, 40). Second, 3′ exon ligation can be accomplished in trans by exposing a spliceosome that has carried out the first step to an exogenously added 3′ exon (2), persuasively demonstrating that catalysis does not necessarily require early AG recognition within the same molecule. Our results show that the particular configuration of 3′ splice sites in Sxl exon 3 allows the physical separation of the two steps on two different AGs, and they also argue for a critical function for the 3′ splice site AG sequence in exon definition.

The first AG recognition step is important for definition of the 3′ splice site region in cooperation with the Py tract and involves the two subunits of U2AF. U2AF65 binds to the Py tract (48, 59, 60), and U2AF35 binds to the AG region (34, 58, 63). Recognition of the AG by U2AF35 is particularly critical for the splicing of pre-mRNAs containing a weak 3′ splice site, which are therefore classified as AG-dependent substrates (39, 58; reviewed in reference 35). Interestingly, although the Py tract associated with the distal site of Sxl exon 3 is relatively strong (containing eight uridines in a row), the adjacent AG plays a critical role in exon 3 definition. The distal AG is important for the use of exon 3 5′ splice site even when the sequence cannot undergo splicing reactions and the associated Py tract has been significantly shortened (Fig. (Fig.3D).3D). It is possible that the stabilization of U2AF65 binding afforded by U2AF35 is critical for exon definition or that interaction of U2AF35 with the distal AG promotes other steps in spliceosome assembly (23, 57). Remarkably, the effects of deletion of the distal 3′ splice site AG are more dramatic than the effects of deleting only the Py tract or even of deleting both signals. One possible explanation for these observations is that recognition of the Py tract by U2AF65 initiates a cascade of events leading to exon definition and that this cascade relies on recognition of the 3′ splice site AG by U2AF35. If this interaction does not occur, abortive exon definition ensues. In the absence of the Py tract, however, U2AF is directed to 3′ splice sites upstream and exon definition occurs through those signals, albeit less efficiently. Efficient exon 3 inclusion in the absence of SXL is a critical feature of the Sxl genetic switch: even modest levels of exon skipping in males could produce enough SXL protein to initiate the autoregulatory loop, promoting the accumulation of more transcripts lacking exon 3 and therefore the accumulation of more SXL protein, ultimately leading to lethality (reviewed in reference 46).

Despite the important role of the distal AG in exon definition, catalysis occurs primarily using the proximal AG (Fig. (Fig.44 and and6).6). One possible explanation for these results could be that the distal AG is not a functional splice site but rather acts as an exonic splicing enhancer sequence (ESE) of the proximal site. Indeed, the sequence encompassing the distal AG and sequences immediately downstream is purine-rich, and ESEs made of purine repeats have been shown to facilitate the use of upstream weak 3′ splice sites (56; reviewed in reference 8). Several results make this scenario very unlikely and argue that the distal AG is being recognized as a functional 3′ splice site for purposes of exon definition. First, all known ESEs are composed of more than two functionally relevant nucleotides and, as in the case of purine-rich enhancers, their activity is proportional to the number of short sequence repeats (52). In contrast, only deletion or mutation of the distal AG (and not of all the AGs downstream) has an effect in exon 3 definition. Second, the distal AG is preceded by a functional Py tract (Fig. (Fig.3C),3C), whereas the proximal site is not (Fig. (Fig.5C).5C). Third, the distal Py tract and AG can function as a 3′ splice site region when the proximal AG is mutated (Fig. (Fig.55 and and6).6). Fourth, the distal AG has functions in splice site activation even when the proximal site has been deleted and the distal site itself cannot undergo splicing (Fig. (Fig.33D).

Mechanisms of 3′ splice site switch.

Two molecular mechanisms could account for the switch in 3′ splice site recognition. First, early factors such as U2AF65/35 could reassemble on the upstream 3′ Py tract and AG or be simultaneously bound to both sites and be differentially efficient for establishing interactions leading to exon definition (distal > proximal) or leading to intron removal (proximal > distal). The absence of effects of mutations at the Py tract upstream from the proximal AG (Fig. (Fig.5C),5C), however, suggest that the proximal AG is not activated from an upstream Py tract following the conventional mechanism of 3′ splice site activation.

An alternative scenario is that recognition of the distal site results in branch point definition, from which the AG used for catalysis will be specified. How the 3′ AG is determined at the time of exon ligation is not completely understood, but work on substrates containing long Py tracts distant from the 3′ splice sites, as well as trans-splicing assays in vitro, suggested the existence of a 5′-to-3′ linear scanning mechanism that would identify the first AG from the Py tract-branch point region (13, 49, 50). Competition and local context effects within a window of sequence from the branch point also contribute to defining the AG undergoing catalysis (13, 14, 33, 37, 50, 53).

In Sxl exon 3, early recognition of the distal site allows the use of the proximal one for catalysis. Other examples of such dual use have been reported previously. First, a mutation in the human β-globin gene, associated with a form of β-thalassemia, generates a 3′ splice site AG between the branchpoint and the normal 3′ splice site, leading to aberrant splicing and the synthesis of a nonfunctional globin mRNA. Surprisingly, experimental inactivation of the wild-type AG in the mutant RNA affected the efficiency with which the upstream cryptic splice site was used in vitro (30, 61). A second example was found in the early transcriptional unit of polyomavirus, which produces three tumor antigens (StAg, MTAg, and LTAg) encoded by alternatively spliced transcripts. It was found that the use of a StAg 3′ splice site required the presence of a MTAg 3′ splice site located 14 nucleotides downstream (19). These studies led to the conclusion that selection of the proximal site could occur after branch formation, possibly through a scanning mechanism, and therefore that the 3′ splice site requirements for 5′ cleavage and lariat formation can be uncoupled from those required for 3′ cleavage and ligation. Furthermore, observations made using plant introns in which 3′ splice sites sequestered within stems of hairpin structures could be activated by “helper” downstream 3′ splice sites were consistent with the possibility that 3′ splice sites can be recognized at least twice during spliceosome assembly prior to catalysis (31).

Switch in 3′ splice site recognition and SXL function.

Our most intriguing result was the apparent requirement of a switch in 3′ splice AG recognition for efficient SXL function. An alternative explanation would be that the proximal AG is important for SXL binding. The binding specificity of SXL, however, is mainly dictated by U-rich sequences (24, 42, 47, 48), and although Sakashita and Sakamoto reported a preference for adjacent AG's, these were located 3′, not 5′, from U-rich stretches (42). In addition, while mutation of the proximal AG strongly inhibits SXL function (Fig. (Fig.5B),5B), drastic changes in a single SXL binding site (particularly in the U8 Py-tract associated with the distal site) do not have substantial effects in SXL regulation (26, 41).

How can the switch in AG recognition facilitate SXL function? The second AG recognition step has been associated with the second step of the splicing reaction (15, 53). To accomplish its role in promoting exon skipping, however, SXL needs to act even before the first step of catalysis, either by forcing branch point formation in the vicinity of exon 4 rather than exon 3 or by affecting splice site pairing across the introns. Our observations therefore suggest that SXL discriminates whether the proximal or the distal site will be used even before the 2′-5′ branch is formed. The first implication of this hypothesis is that recognition of different AGs results in exon definition complexes of different composition and/or different properties for establishing partnerships with neighboring 5′ splice sites. The second implication is that the different complexes are differentially sensitive to SXL function. It is conceivable that splitting the two steps of 3′ splice site recognition between two physically different AGs opens a window of opportunity for SXL repression. Binding of U2 snRNP to the branch point, for example, could be less stable when the proximal site is recognized. This lower stability could have little effect in splice site activation in the absence of SXL but be very detrimental in its presence.

SXL could interfere with the assembly of splicing factors or somehow inactivate the factors bound. Genetic results have shown that mutant alleles of the gene sans-fille (snf), which encodes a protein component of both U1 and U2 snRNPs, show deficient Sxl autoregulation when combined with particular Sxl mutants (1, 18, 43, 44). It is conceivable that interactions between SXL and SNF (16, 17) compromise splice site activation.

If the precise structure of the 3′ splice sites and the apparent switch in 3′ splice site recognition is important for SXL function, why are the cryptic 3′ splice sites that are activated by mutation of the natural sites properly regulated, both in transient tranfections (Fig. (Fig.5)5) and in transgenic flies (26)? One possibility is that the cryptic sites also have a similar configuration of nearby competing 3′ splice sites. Indeed, analysis of 3′ splice site utilization indicated that both the cryptic site and the proximal site (now in a distal configuration relative to the cryptic site) are used (Fig. (Fig.3C3C and data not shown).

A plausible reason for maintaining the duality in 3′ splice site recognition is that other 3′ splice sites regulatable by SXL would not be strong enough to prevent even small levels of exon skipping in the absence of SXL. As mentioned above, efficient inclusion of exon 3 is critical for male survival (46). The strong signals required for full exon definition would then demand a mechanism like the switch in 3′ splice site recognition mentioned above to allow efficient regulation by SXL. Thus, the configuration of 3′ splice sites preceding exon 3 could have evolved to accomplish both stringent exon inclusion in the absence of regulator and an opportunity for the regulator to interfere.


We thank Hiroshi Sakamoto, Kunio Inoue, and Yoshiro Shimura for the kind gift of plasmids and protocols; Douglas Black, Peter Nielsen, and Chris Smith for discussions; Witold Filipowicz for pointing out related results in plants; an anonymous reviewer for pointing out a similar example of 3′ splice site recognition in polyoma early transcripts; and Iain Mattaj and Bertrand Séraphin for suggestions on the manuscript.

L.O.F.P. was supported by an EMBO postdoctoral fellowship. M.J.L. was supported by Fundación Ramón Areces (Spain) and Marie Curie-European Union postdoctoral fellowships.

L.O.F.P. and M.J.L. contributed equally to this study.


1. Albrecht E B, Salz H K. The Drosophila sex determination gene snf is utilized for the establishment of the female-specific splicing pattern of Sex-lethal. Genetics. 1993;134:801–807. [PMC free article] [PubMed]
2. Anderson K, Moore M J. Bimolecular exon ligation by the human spliceosome. Science. 1997;276:1712–1716. [PubMed]
3. Anderson K, Moore M J. Biomolecular exon ligation by the human spliceosome bypasses early 3′ splice site AG recognition and requires NTP hydrolysis. RNA. 2000;6:16–25. [PMC free article] [PubMed]
4. Bashaw G J, Baker B S. The msl-2 dosage compensation gene of Drosophila encodes a putative DNA-binding protein whose expression is sex specifically regulated by Sex-lethal. Development. 1995;121:3245–3258. [PubMed]
5. Bashaw G J, Baker B S. The regulation of the Drosophila msl-2 gene reveals a function for Sex-lethal in translational control. Cell. 1997;89:789–798. [PubMed]
6. Bell L R, Horabin J I, Schedl P, Cline T W. Positive autoregulation of sex-lethal by alternative splicing maintains the female determined state in Drosophila. Cell. 1991;65:229–239. [PubMed]
7. Berget S M. Exon recognition in vertebrate splicing J. Biol Chem. 1995;270:2411–2414. [PubMed]
8. Blencowe B J. Exonic splicing enhancers: mechanism of action, diversity and role in human genetic diseases. Trends Biochem Sci. 2000;25:106–110. [PubMed]
9. Boggs R T, Gregor P, Idriss S, Belote J M, McKeown M. Regulation of sexual differentiation in D. melanogaster via alternative splicing of RNA from the transformer gene. Cell. 1987;50:739–747. [PubMed]
10. Bopp D, Calhoun G, Horabin J I, Samuels M, Schedl P. Sex-specific control of Sex-lethal is a conserved mechanism for sex determination in the genus Drosophila. Development. 1996;122:971–982. [PubMed]
11. Campuzano S, Carramolino L, Cabrera C V, Ruiz-Gomez M, Villares R, Boronat A, Modolell J. Molecular genetics of the achaete-scute gene complex of D. melanogaster. Cell. 1985;40:327–338. [PubMed]
12. Chabot B. Directing alternative splicing: cast and scenarios. Trends Genet. 1996;12:472–478. [PubMed]
13. Chen S, Anderson K, Moore M J. Evidence for a linear search in bimolecular 3′ splice site AG selection. Proc Natl Acad Sci USA. 2000;97:593–598. [PMC free article] [PubMed]
14. Chiara M D, Palandjian L, Kramer R F, Reed R. Evidence that U5 snRNP recognizes the 3′ splice site for catalytic step II in mammals. EMBO J. 1997;16:4746–4759. [PMC free article] [PubMed]
15. Chua K, Reed R. The RNA splicing factor hSlu7 is required for correct 3′ splice site choice. Nature. 1999;402:207–210. [PubMed]
16. Cline T W, Rudner D Z, Barbash D A, Bell M, Vutien R. Functioning of the Drosophila integral U1/U2 protein Snf independent of U1 and U2 small nuclear ribonucleoprotein particles is revealed by snf(+) gene dose effects. Proc Natl Acad Sci USA. 1999;96:14451–14458. [PMC free article] [PubMed]
17. Deshpande G, Samuels M E, Schedl P D. Sex-lethal interacts with splicing factors in vitro and in vivo. Mol Cell Biol. 1996;16:5036–5047. [PMC free article] [PubMed]
18. Flickinger T W, Salz H K. The Drosophila sex determination gene snf encodes a nuclear protein with sequence and functional similarity to the mammalian U1A snRNP protein. Genes Dev. 1994;8:914–925. [PubMed]
19. Ge H, Noble J, Colgan J, Manley J L. Polyoma virus small tumor antigen pre-mRNA splicing requires cooperation between two 3′ splice sites. Proc Natl Acad Sci USA. 1990;87:3338–3342. [PMC free article] [PubMed]
20. Gebauer F, Merendino L, Hentze M W, Valcárcel J. The Drosophila splicing regulator Sex-lethal directly inhibits translation of male-specific-lethal 2 mRNA. RNA. 1998;4:142–150. [PMC free article] [PubMed]
21. Granadino B, Penalva L O F, Green M R, Valcárcel J, Sánchez L. Distinct mechanisms of splicing regulation in vivo by the Drosophila protein Sex-lethal. Proc Natl Acad Sci USA. 1997;94:7343–7348. [PMC free article] [PubMed]
22. Green M R. Pre-mRNA splicing. Annu Rev Genet. 1986;20:671–708. [PubMed]
23. Guth S, Martinez C, Gaur R K, Valcarcel J. Evidence for substrate-specific requirement of the splicing factor U2AF(35) and for its function after polypyrimidine tract recognition by U2AF(65) Mol Cell Biol. 1999;19:8263–8271. [PMC free article] [PubMed]
24. Handa N, Nureki O, Kurimoto K, Kim I, Sakamoto H, Shimura Y, Muto Y, Yokoyama S. Structural basis for recognition of the tra mRNA precursor by the Sex-lethal protein. Nature. 1999;398:579–585. [PubMed]
25. Horabin J I, Schedl P. Regulated splicing of the Drosophila sex-lethal male exon involves a blockage mechanism. Mol Cell Biol. 1993;13:1408–1414. [PMC free article] [PubMed]
26. Horabin J I, Schedl P. Sex-lethal autoregulation requires multiple cis-acting elements upstream and downstream of the male exon and appears to depend largely on controlling the use of the male exon 5′ splice site. Mol Cell Biol. 1993;13:7734–7746. [PMC free article] [PubMed]
27. Inoue K, Hoshijima K, Sakamoto H, Shimura Y. Binding of the Drosophila Sex-lethal gene product to the alternative splice site of transformer primary transcript. Nature. 1990;344:461–463. [PubMed]
28. Kelley R L, Solovyeva I, Lyman L M, Richman R, Solovyev V, Kuroda M I. Expression of msl-2 causes assembly of dosage compensation regulators on the X chromosomes and female lethality in Drosophila. Cell. 1995;81:867–877. [PubMed]
29. Kelley R L, Wang J, Bell L, Kuroda M I. Sex lethal controls dosage compensation in Drosophila by a non-splicing mechanism. Nature. 1997;387:195–199. [PubMed]
30. Krainer A R, Reed R, Maniatis T. Mechanisms of human β-globin pre-mRNA splicing. In: Berg P, editor. The Robert A. Welch Foundation conferences on chemical research. XXIX. Genetic chemistry: the molecular basis of heredity. Houston, Tex: Welch Foundation; 1985. pp. 353–382.
31. Liu H-X, Goodall G J, Kole R, Filipowicz W. Effects of secondary structure on pre-mRNA splicing: hairpins sequestering the 5′ but not the 3′ splice site inhibit intron processing in Nicotiana plumbaginifolia. EMBO J. 1995;14:377–388. [PMC free article] [PubMed]
32. López A J. Alternative splicing of pre-mRNA: developmental consequences and mechanisms of regulation. Annu Rev Genet. 1998;32:279–305. [PubMed]
33. Luukkonen B G M, Séraphin B. The role of branchpoint-3′ splice site spacing and interaction between intron terminal nucleotides in splice site selection in Saccharomyces cerevisiae. EMBO J. 1997;16:779–792. [PMC free article] [PubMed]
34. Merendino L, Guth S, Bilbao D, Martinez C, Valcarcel J. Inhibition of msl-2 splicing by Sex-lethal reveals interaction between U2AF35 and the 3′ splice site AG. Nature. 1999;402:838–841. [PubMed]
35. Moore M J. Intron recognition comes of AGe Nat. Struct Biol. 2000;7:14–16. [PubMed]
36. Parker R, Patterson B. Architecture of fungal introns: implications for spliceosome assembly. In: Inouye M, Dudock B, editors. Molecular biology of RNA: new perspectives. New York, N.Y: Academic Press; 1987. pp. 133–150.
37. Patterson B, Guthrie C. A U-rich tract enhances usage of an alternative 3′ splice site in yeast. Cell. 1991;64:181–187. [PubMed]
38. Penalva L O F, Sakamoto H, Navarro-Sabate A, Sakashita E, Granadino B, Segarra C, Sanchez L. Regulation of the gene Sex-lethal: a comparative analysis of Drosophila melanogaster and Drosophila subobscura. Genetics. 1996;144:1653–1664. [PMC free article] [PubMed]
39. Reed R. The organization of 3′ splice sequences in mammalian introns. Genes Dev. 1989;3:2113–2123. [PubMed]
40. Rymond B C, Rosbash M. Cleavage of 5′ splice site and lariat formation are independent of 3′ splice site in yeast mRNA splicing. Nature. 1985;317:735–737. [PubMed]
41. Sakamoto H, Inoue K, Higuchi I, Ono Y, Shimura Y. Control of Drosophila Sex-lethal pre-mRNA splicing by its own female-specific product. Nucleic Acids Res. 1992;20:5533–5540. [PMC free article] [PubMed]
42. Sakashita E, Sakamoto H. Protein-RNA and protein-protein interactions of the Drosophila Sex-lethal mediated by its RNA-binding domains. J Biochem. 1996;120:1028–1033. [PubMed]
43. Salz H K. The genetic analysis of snf. a Drosophila sex determination gene required for activation of Sex-lethal in both the germline and the soma. Genetics. 1992;130:547–554. [PMC free article] [PubMed]
44. Salz H K, Flickinger T W. Both loss-of-function and gain-of-function mutations in snf define a role for snRNP proteins in regulating Sex-lethal pre-mRNA splicing in Drosophila development. Genetics. 1996;144:95–108. [PMC free article] [PubMed]
45. Sawa H, Shimura Y. Alterations of RNase H sensitivity of the 3′ splice site region during the in vitro splicing reaction. Nucleic Acids Res. 1991;19:3953–3958. [PMC free article] [PubMed]
46. Schütt C, Nöthiger R. Structure, function and evolution of sex-determining systems in dipteran insects. Development. 2000;127:667–677. [PubMed]
47. Singh R, Banerjee H, Green M R. Differential recognition of the polypyrimidine-tract by the general splicing factor U2AF65 and the splicing repressor Sex-lethal. RNA. 2000;6:901–911. [PMC free article] [PubMed]
48. Singh R, Valcárcel J, Green M R. Distinct binding specificities and functions of higher eukaryotic polypyrimidine tract-binding proteins. Science. 1995;268:1173–1176. [PubMed]
49. Smith C W J, Porro E B, Patton J G, Nadal-Ginard B. Scanning from an independently specified branch point defines the 3′ splice site of mammalian introns. Nature. 1989;342:243–247. [PubMed]
50. Smith C W J, Chu T T, Nadal-Ginard B. Scanning and competition between AGs are involved in 3′ splice site selection in mammalian introns. Mol Cell Biol. 1993;13:4939–4952. [PMC free article] [PubMed]
51. Sosnowski B A, Belote J M, McKeown M. Sex-specific alternative splicing of RNA from the transformer gene results from sequence-dependent splice site blockage. Cell. 1989;58:449–459. [PubMed]
52. Tanaka K, Watakabe A, Shimura Y. Polypurine sequences within a downstream exon function as a splicing enhancer. Mol Cell Biol. 1994;14:1347–1354. [PMC free article] [PubMed]
53. Umen J G, Guthrie C. Prp16p, Slu7p, and Prp8p interact with the 3′ splice site in two distinct stages during the second catalytic step of pre-mRNA splicing. RNA. 1995;1:584–597. [PMC free article] [PubMed]
54. Valcárcel J, Singh R, Zamore P D, Green M R. The protein Sex-lethal antagonizes the splicing factor U2AF to regulate alternative splicing of transformer pre-mRNA. Nature. 1993;362:171–175. [PubMed]
55. Wang J, Bell L R. The Sex-lethal amino terminus mediates cooperative interactions in RNA binding and is essential for splicing regulation. Genes Dev. 1994;8:2072–2085. [PubMed]
56. Watakabe A, Tanaka K, Shimura Y. The role of exon sequences in splice site selection. Genes Dev. 1993;7:407–418. [PubMed]
57. Wu J Y, Maniatis T. Specific interactions between proteins implicated in splice site selection and regulated alternative splicing. Cell. 1993;17:1061–1070. [PubMed]
58. Wu S, Romfo C M, Nilsen T W, Green M R. Functional recognition of the 3′ splice site AG by the splicing factor U2AF35. Nature. 1999;402:832–835. [PubMed]
59. Zamore P D, Green M R. Biochemical characterization of U2 snRNP auxiliary factor: an essential pre-mRNA splicing factor with a novel intranuclear distribution. EMBO J. 1991;10:207–214. [PMC free article] [PubMed]
60. Zamore P D, Patton J G, Green M R. Cloning and domain structure of the mammalian splicing factor U2AF. Nature. 1992;13:609–614. [PubMed]
61. Zhuang Y, Weiner A M. The conserved dinucleotide AG of the 3′ splice site may be recognized twice during in vitro splicing of mammalian mRNA precursors. Gene. 1990;90:263–269. [PubMed]
62. Zhou S, Yang Y, Scott M J, Pannuti A, Fehr K C, Eisen A, Koonin E V, Fouts D L, Wrightsman R, Manning J E, Lucchesi J. Male-specific lethal 2, a dosage compensation gene of Drosophila, undergoes sex-specific regulation and encodes a protein with a RING finger and a metallothionein-like cysteine cluster. EMBO J. 1995;14:2884–2895. [PMC free article] [PubMed]
63. Zorio D A, Blumenthal T. Both subunits of U2AF recognize the 3′ splice site in Caenorhabditis elegans. Nature. 1999;402:835–838. [PubMed]

Articles from Molecular and Cellular Biology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...