INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The temporal, developmental, and tissue-specific regulation of alternative pre-mRNA splicing is an important feature of gene control employed by metazoan cells (reviewed in reference 22). Yet, how alternative splice site choice is regulated is in most cases still unknown; specific trans-acting factors and corresponding cis-acting sequence elements have been identified only for a few pre-mRNAs. In a growing number of systems, members of the SR protein family (24) seem to be involved in regulation of alternative splicing, often by enhancing recognition of suboptimal splice sites through binding to exonic splicing enhancers (for reviews, see references 3, 14, and 21).

Adenovirus gene expression is to a large extent regulated at the level of alternative pre-mRNA splicing (reviewed in reference 6). We are using the adenovirus major late region 1 (L1) as a model pre-mRNA to study the mechanisms controlling alternative splice site usage in adenovirus-infected cells. In the L1 unit, a common 5' splice site can be joined to two alternative 3' splice sites, resulting in the formation of the so-called 52,55K (proximal 3' splice site) and IIIa (distal 3' splice site) mRNAs. Early during virus infection, the 52,55K 3' splice site is used exclusively, whereas the IIIa splice site becomes the preferred site late during virus infection (reviewed in reference 6).

We have previously shown that IIIa splicing is negatively regulated by hyperphosphorylated SR proteins that bind to a 49-nucleotide-long intronic repressor element, the 3RE, located immediately upstream of the IIIa branch site (8). SR protein binding to the 3RE results in inhibition of IIIa splicing by preventing U2 snRNP recruitment to the spliceosome. In late virus-infected cells, the inhibitory effect of the 3RE on IIIa splicing is alleviated by a virus-induced dephosphorylation of SR proteins, rendering them nonfunctional as repressor proteins of IIIa splicing (9).

Importantly, in our previous experiments we did not address whether SR protein dephosphorylation is sufficient to fully explain the enhanced IIIa splicing phenotype observed late during virus infection. It is noteworthy that the IIIa 3' splice site contains a short atypical pyrimidine tract that binds U2AF⁶⁵ inefficiently in vitro (17). This observation led us to consider the possibility that inactivation of the repressive activity of SR proteins on IIIa splicing might not be sufficient to explain the high IIIa splicing activity observed in late virus-infected cells. We therefore set out to search for additional IIIa sequence elements that contribute to the enhanced splicing phenotype of the IIIa pre-mRNA in nuclear extract prepared from adenovirus-infected cells (Ad-NE). For these experiments we made use of the observation that -globin splicing is repressed in Ad-NE (17). Here we show that replacing the -globin branch site/polypyrimidine tract with the branch site and atypical pyrimidine tract from the IIIa pre-mRNA is sufficient to convert -globin from a transcript that is repressed in splicing to a pre-mRNA that is activated in Ad-NE. We refer to this sequence element as the IIIa virus infection-dependent splicing enhancer element (3VDE), since it functions as a splicing enhancer only in the context of the virus-infected extract.

Collectively, our results show that the 3RE and the 3VDE are the critical elements controlling IIIa splicing in Ad-NE, with the 3VDE making the most significant contribution. We further show that enhanced splicing of the IIIa pre-mRNA in Ad-NE is not mimicked by an increase in U2AF⁶⁵ interaction with the IIIa 3' splice site. This observation suggests that IIIa splicing may operate by a novel mechanism that does not require efficient U2AF recruitment to the IIIa 3' splice site.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids and transcript synthesis. Plasmids IIIa, glob (the rabbit -globin first intron [5]), glob (3RE, 3VDE), glob (3RE), glob (3VDE), and IIIa (-3RE) have previously been described (8). Plasmids glob (IIIa-bp), glob (IIIa-py), IIIa (-3VDE), and IIIa (-3RE, 3VDE) were constructed by PCR cloning using appropriately positioned restriction endonuclease cleavage sites and designed primers. Plasmids IIIa-1G and IIIa-2G were reconstructed using synthetic double-stranded oligonucleotides. The nucleotide sequence of 3VDE is AGUACUAAGC_GGUGAUGUUUCUGAUCAG, and the corresponding -globin sequence is GUGCUGAC_UUCUCUCCCCUGGGCUGUUUUCAUUUUCUCAG. The branch sites are shown in bold, and the underlines show the break points used to construct hybrid transcripts glob (IIIa-bp) and glob (IIIa-py), respectively. All plasmid sequences were verified by DNA sequencing. Plasmid maps and sequences are available on request or at http://www.bmc.uu.se/IMIM/GA.html.

In vitro splicing reactions. Nuclear extract preparation from uninfected (HeLa-NE) or late adenovirus-infected HeLa spinner cells was as previously described (10, 16). Splicing reaction mixtures were incubated at 30°C for 90 to 160 min in a total volume of 25 µl containing 5 to 25 fmol of transcript, 40% nuclear extract, 2.6% polyvinyl alcohol, 12% glycerol, 12 mM HEPES (pH 7.9), 60 mM KCl, 2 mM ATP, 20 mM creatine phosphate, 0.3 mM dithiothreitol, and 2.5 mM MgCl₂. The optimal MgCl₂ concentration for IIIa splicing is 2.5 mM (O. Mühlemann, unpublished observation), and the splicing efficiency is two- to threefold lower in 3.2 mM MgCl₂ (the standard concentration used in most laboratories). Splicing of -globin is unaffected by MgCl₂ concentrations between 2 and 5 mM (unpublished observation). Following incubation, RNA was analyzed on denaturing 8% polyacrylamide gels. Dried gels were subjected to PhosphorImager quantification as previously described (10, 18). All splicing reactions were performed multiple times with at least three different batches of HeLa-NE and Ad-NE, and average values and standard deviations are shown.

Prespliceosome formation. Standard splicing reactions were set up as described above except that polyvinyl alcohol was omitted and the amount of ³²P-labeled RNA was doubled. From the reactions, incubated at 30°C, aliquots were removed at different time points, mixed with heparin (final concentration, 0.5 µg/µl) and resolved on a 4% (84:1 acrylamide/diacrylylpiperazine) native polyacrylamide gel, which was cast in a buffer containing 50 mM Tris-glycine and 5% glycerol. In the running buffer, glycerol was omitted.

U2 snRNA depletion. Oligonucleotide-directed RNase H cleavage of U2 snRNP in Ad-NE was done exactly as described elsewhere (13). An aliquot of the cleaved extract was used to verify, by Northern blotting, that the RNase H treatment effectively destroyed the designated U snRNA, the remaining of the extract was used for splicing assay. Oligonucleotide E15 (2) directed against the 5' end of U2 snRNA was used in the depletion reaction. Mock depletion of Ad-NE was performed in the absence of oligonucleotide.

Western blot. One, 3, and 10 µl of HeLa-NE (9 mg/ml) and Ad-NE (9 mg/ml) were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on a 12% gel under reducing conditions and transferred to a nitrocellulose membrane using a semidry transfer apparatus. Filters were treated as previously described (18). U2AF⁶⁵ was detected using monoclonal antibody mc3 (4) and visualized by chemiluminescence according to the protocol of the manufacturer (Amersham).

UV cross-linking. Approximately 100 fmol of ³²P-labeled transcript IIIa (-3RE) (contains the IIIa pyrimidine tract) and IIIa (-3RE, -3VDE) (contains the -globin polypyrimidine tract) were incubated in HeLa-NE and Ad-NE under splicing conditions for 15 min at 30°C and then cross-linked by UV irradiation (output, 1,200 µW/cm²; distance, 1 cm) for 15 min on ice. RNA was digested with 10 µg of RNase A (Pharmacia) at 37°C for 60 min. U2AF⁶⁵ was immunoprecipitated with protein A-Sepharose (Pharmacia)-bound monoclonal antibody mc3 (4) for 1 h at room temperature. The precipitate was washed three times and resolved by SDS-PAGE on a 12% gel under reducing conditions. Labeled proteins were visualized by autoradiography.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

A viral splicing enhancer activates IIIa splicing in Ad-NE. In most of our previous work we have used transcripts with an artificial exonic splicing enhancer attached to the 3' end of the IIIa transcript (12). Thus, appending the strong adenovirus major late first leader 5' splice site (we refer to this as a U1 enhancer [23]) to the IIIa second exon results in a more than 50-fold stimulation of IIIa splicing in HeLa-NE (Fig. 1, compare lanes 1 and 3). This observation has been instrumental in our previous work, since IIIa transcripts without the U1 enhancer show very little, if any, splicing activity in most HeLa-NE preparations, even under conditions optimized for IIIa splicing (Fig. 1, lane 3, and data not shown). Since the U1 enhancer increases the basal level of IIIa splicing in HeLa-NE, splicing of the IIIa-U1 transcript in Ad-NE results in only a modest (2.5-fold) activation (Fig. 1, lanes 1 and 2; reference 16). In contrast, the IIIa transcript is spliced efficiently in Ad-NE irrespective of the presence or absence of the U1 enhancer (Fig. 1, compare lanes 2 and 4). This result suggests that a virus infection-specific splicing enhancer that is nonfunctional in HeLa-NE activates IIIa pre-mRNA splicing in Ad-NE. To obtain a more sensitive assay, all subsequent experiments were done with splicing substrates lacking the U1 enhancer.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Enhancement of IIIa splicing in Ad-NE. IIIa transcripts, either containing an artificial second exon enhancer (IIIa-U1) or not (IIIa), were spliced in HeLa-NE or Ad-NE. Positions of pre-mRNAs and splicing products are indicated to the left for IIIa-U1 and to the right for IIIa. Fold activation of splicing in Ad-NE relative to HeLa-NE is shown for each transcript at the bottom.

Identification of the minimal IIIa sequence element conferring enhanced splicing in Ad-NE. To identify the IIIa sequence element(s) required for activated splicing in Ad-NE, we replaced short sequences in the rabbit -globin pre-mRNA with the corresponding sequence from IIIa (Fig. 2A). The idea was to localize the minimal IIIa element conferring an enhanced splicing phenotype to -globin in Ad-NE.

View larger version (34K): [in this window] [in a new window]   FIG. 2.   Mapping of the IIIa virus infection-dependent splicing enhancer, conferring an enhanced splicing phenotype in Ad-NE. (A) Schematic representation of the IIIa and -globin (glob) hybrid pre-mRNAs used to map the 3VDE. Sequences originating from IIIa are shaded, the 3VDE is in black, and the 3RE is shown as a striped box. -Globin sequences are shown as open boxes (exons) and thin lines (introns). Positions of the IIIa and -globin branch sites are marked ( and *, respectively). (B) The indicated pre-mRNAs were spliced in HeLa-NE or Ad-NE, and products were resolved by gel electrophoresis and visualized by autoradiography. Positions of pre-mRNA and splicing products are shown for IIIa (shaded boxes) and -globin and hybrid transcripts (open boxes). (C) PhosphorImager quantitation of splicing efficiencies of hybrid transcripts in HeLa-NE (open bars) and Ad-NE (shaded bars). All experiments were done multiple times using at least three different batches of HeLa-NE and Ad-NE. Average splicing efficiencies and standard deviations are shown.

View larger version (59K): [in this window] [in a new window]   FIG. 3.   The IIIa pyrimidine tract is critical for the enhancer activity of the 3VDE. (A) Schematic representation of the IIIa and -globin (glob) hybrid transcripts used. (B) The indicated pre-mRNAs were incubated under splicing conditions in HeLa-NE or Ad-NE, and products were resolved by gel electrophoresis and visualized by autoradiography. (C) PhosphorImager quantitation of splicing efficiencies of the hybrid transcripts in HeLa-NE (open bars) and Ad-NE (shaded bars). All experiments were done multiple times using at least three different batches of HeLa-NE and Ad-NE. Average splicing efficiencies and standard deviations are shown. (D) The IIIa pyrimidine tract is sufficient to enhance A-complex formation in Ad-NE. The indicated transcripts were incubated under splicing conditions in HeLa-NE or Ad-NE. Aliquots were withdrawn at the indicated time points and analyzed for spliceosomal complex formation by native gel electrophoresis.

Mutational analysis of the 3VDE. To further demonstrate that the IIIa pyrimidine tract is indeed the critical enhancer element activating IIIa splicing in Ad-NE, we crippled the IIIa pyrimidine tract by mutating selected pyrimidines within 3VDE (Fig. 4). Such mutations have previously been analyzed (17). However, in that study IIIa transcripts activated by an artificial U1 enhancer were used. Thus, questioning the significance of the results obtained for splicing under conditions were the 3VDE functions as the splicing enhancer. We therefore reinvestigated the effect of IIIa pyrimidine tract mutations on IIIa splicing in the absence of the artificial U1 enhancer. As shown in Fig. 4, substituting the UC pair at positions -9 and -8 with GG completely abolished IIIa splicing in Ad-NE. Mutating residue -10 from U to G caused a 10-fold reduction in IIIa splicing in Ad-NE. Collectively these results underscore our conclusion that the IIIa pyrimidine tract is a critical element required for high IIIa splicing in Ad-NE. Note that the low level of wild-type IIIa splicing in HeLa-NE was reduced to nondetectable levels on transcripts IIIa-1G and IIIa-2G (Fig. 4), making it impossible to quantitate the effect of the pyrimidine mutations on IIIa splicing in HeLa-NE.

View larger version (29K): [in this window] [in a new window]   FIG. 4.   Mutational analysis of the 3VDE showing that the IIIa pyrimidine tract is the enhancer element. (A) Sequence of the 3VDE. The IIIa pyrimidine tract is shown highlighted (transcript IIIa), and changes in mutants IIIa-1G and IIIa-2G are indicated. Asterisks indicate positions of the two alternative IIIa branch sites (11). (B) The indicated pre-mRNAs were incubated under splicing conditions in HeLa-NE or Ad-NE, and products were resolved by gel electrophoresis and visualized by autoradiography. (C) PhosphorImager quantitation of splicing efficiencies of IIIa transcripts with mutated pyrimidine tracts in HeLa-NE (open bars) and Ad-NE (shaded bars). All experiments were done multiple times using at least three different batches of HeLa-NE and Ad-NE. Average splicing efficiencies and standard deviations are shown.

U2 snRNP is essential for IIIa pre-mRNA splicing in Ad-NE. The observation that the IIIa pyrimidine tract did not function together with the -globin branch site in generating a spliced mRNA raised the question of whether U2 snRNP was necessary for 3VDE function. Hypothetically, the 3VDE may promote splicing in Ad-NE by a U2 snRNP-independent mechanism. To test this possibility, we functionally inactivated U2 snRNP in Ad-NE by oligonucleotide-directed RNase H cleavage of the 5' end of U2 snRNA (12). As shown in Fig. 5, incubation of Ad-NE with increasing amounts of an U2-specific oligonucleotide, during RNase H treatment, abolished IIIa splicing (lanes 3 and 4), whereas mock treatment did not adversely affect IIIa splicing (lane 2). This result demonstrates that U2 snRNP is, indeed, required for IIIa splicing in Ad-NE.

View larger version (37K): [in this window] [in a new window]   FIG. 5.   Oligonucleotide-directed RNase H cleavage of the 5' end of U2 snRNA showing that a functional U2 snRNP is required for IIIa splicing in Ad-NE. Ad-NE untreated or treated with or without (Mock) increasing amounts of oligonucleotide E15 (2) were incubated under splicing conditions, and products were resolved by gel electrophoresis and visualized by autoradiography.

The 3VDE is required for activated IIIa splicing in Ad-NE. Next, we asked whether the 3RE and the 3VDE also are essential for enhanced IIIa pre-mRNA splicing in Ad-NE. In these experiments, we replaced IIIa sequence elements with the corresponding sequences from -globin. As shown in Fig. 6, IIIa (-3RE) was spliced in Ad-NE almost as efficiently as the wild-type IIIa transcript (lanes 2 and 4). However, since removal of the 3RE alleviates repression of IIIa splicing by SR proteins (9), the basal level of IIIa (-3RE) splicing in HeLa-NE is increased compared to the wild-type IIIa transcript (compare lanes 1 and 3), which results in a significant decrease in the fold activation in Ad-NE. This finding supports our previous conclusion that viral inhibition of SR protein activity makes an important contribution to IIIa 3' splice site activation, by suppressing IIIa splicing in HeLa-NE. The results also show that the 3RE is not the most significant element controlling IIIa splicing in Ad-NE. Interestingly, the IIIa (-3VDE) transcript was spliced almost as efficiently as -globin in HeLa-NE (lanes 5 and 9). Furthermore, the IIIa (-3VDE) transcript was only moderately activated in Ad-NE (lane 6). This residual activation can be attributed to the reduced repressive activity of SR proteins in Ad-NE (9). Accordingly, the double mutant IIIa (-3RE, -3VDE) showed an even higher basal splicing activity in HeLa-NE, but its splicing was slightly repressed in Ad-NE (lanes 7 and 8). Thus, replacing the 3RE and the 3VDE in the IIIa pre-mRNA with the corresponding sequences from -globin was sufficient to convert IIIa from a pre-mRNA that is enhanced in Ad-NE compared to HeLa-NE, to a transcript which, similar to -globin, is repressed in Ad-NE. Collectively, our results demonstrate that the 3VDE is the major element controlling IIIa 3' splice site activity, with the 3RE making a smaller but important contribution.

View larger version (37K): [in this window] [in a new window]   FIG. 6.   The 3VDE functions as a Janus element: it abrogates IIIa splicing in HeLa-NE and enhances IIIa splicing in Ad-NE. (A) Schematic representation of the IIIa and -globin hybrid transcripts used. (B) The indicated pre-mRNAs were incubated in HeLa-NE or Ad-NE, and splicing products were resolved by gel electrophoresis and visualized by autoradiography. (C) PhosphorImager quantitation of splicing efficiencies of the hybrid transcripts in HeLa-NE (open bars) and Ad-NE (shaded bars). All experiments were done multiple times using at least three different batches of HeLa-NE and Ad-NE. Average splicing efficiencies and standard deviations are shown.

U2AF pre-mRNA interaction is reduced in Ad-NE. U2AF is an essential splicing factor required for processing of prototypical pre-mRNAs (19). U2AF binds to the polypyrimidine tract at the 3' splice site and aids in the recruitment of U2 snRNP to the branch site (20). We have previously reported an inverse correlation between recombinant U2AF⁶⁵ binding to a 3' splice site and the splicing efficiency in Ad-NE (17). Thus, splicing of pre-mRNAs that bind U2AF⁶⁵ efficiently is repressed in Ad-NE, whereas pre-mRNAs with atypical pyrimidine tracts, such as IIIa, which bind U2AF⁶⁵ inefficiently, are enhanced in Ad-NE. As shown above (Fig. 3), the IIIa pyrimidine tract appears to play a major role in controlling IIIa 3' splice site activation. Since the steady-state amount of U2AF⁶⁵ does not change during virus infection (Fig. 7A; reference 10), we conclude that the inhibition of splicing of prototypical pre-mRNAs in Ad-NE does not result from reduced amounts of U2AF⁶⁵ in late virus-infected cells.

View larger version (39K): [in this window] [in a new window]   FIG. 7.   U2AF65 binding to pyrimidine tracts in Ad-NE. (A) Western blot analysis of the steady-state amounts of U2AF65 in HeLa-NE (lanes 1 to 3) and Ad-NE (lanes 4 to 6). The blot was probed with monoclonal antibody mc3 (4), which is specific for U2AF65. (B) 32P-labeled transcripts IIIa (-3RE) (lanes 1 and 3) or IIIa (-3RE, -3VDE) (lanes 2 and 4) were incubated in HeLa-NE or Ad-NE. Proteins bound to the RNAs were UV cross-linked, digested with RNase A, and immunoprecipitated with mc3. The immunoprecipitate was separated on an SDS-12% polyacrylamide gel, and labeled proteins were visualized by autoradiography.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Viruses typically inhibit host cell gene expression to gain full access to the biosynthetic machinery of the cell. Thus, many viruses inhibit host cell RNA processing and RNA transport (7). Since RNA splicing is a prerequisite for nuclear-to-cytoplasmic export of most cellular mRNAs, a virus-induced suppression of host cell RNA splicing may be an important regulatory mechanism by which viruses inhibit host cell gene expression. Some viruses, such as herpes simplex virus, vaccinia virus, and most RNA viruses, contain genes essentially lacking introns. Such viruses could potentially completely shut off host cell RNA splicing without a significant impact on virus-specific gene expression. Most DNA viruses, like adenovirus, still depend on a functional splicing machinery for expression of viral genes. Thus, all adenovirus genes except pIX (1) contain introns. Therefore, adenovirus, instead of abolishing RNA splicing in infected cells, appears to redirect the specificity of the splicing machinery late during the infectious cycle, such that splicing of generic pre-mRNAs is reduced and splicing of certain viral mRNAs is enhanced.

We have previously shown that adenovirus reduces the functional activity of the classical SR proteins through a virus-induced dephosphorylation (9). However, inactivation of the SR family of splicing factors poses a significant problem. It is well established that SR proteins are essential for generic pre-mRNA splicing (for reviews, see references 3, 14, and 21). Why are late-specific adenovirus pre-mRNAs spliced more efficiently under conditions where the functional activity of SR proteins is reduced? The results presented here appears to provide an important piece to the puzzle. We show that the adenovirus IIIa branch site/polypyrimidine tract functions as a virus infection-dependent splicing enhancer, the 3VDE. This sequence element is essential for regulated IIIa pre-mRNA splicing and sufficient to convert -globin from a transcript that is repressed to a pre-mRNA that, similar to the IIIa pre-mRNA, is enhanced in Ad-NE (Fig. 2, compare lanes 7 and 8 with lanes 9 and 10). Interestingly, the 3VDE is inhibitory for splicing in HeLa-NE, probably because the 3VDE contains a weak pyrimidine tract which does not efficiently bind the general splicing factor U2AF (Fig. 7B). Thus, replacing the 3VDE with the branch site/polypyrimidine tract from -globin increases basal IIIa splicing dramatically in HeLa-NE and thereby essentially abolishes its activation in Ad-NE (Fig. 6). Similarly, transfer of the 3VDE to -globin drastically inhibits -globin splicing in HeLa-NE (Fig. 2, compare lanes 9 and 7) and converts -globin to a pre-mRNA that now is enhanced in Ad-NE (Fig. 2, compare lanes 7 and 8). Collectively, these results show that the 3VDE functions as a Janus element, it inhibits splicing in HeLa-NE and functions as a splicing activator element in Ad-NE.

Although the 3VDE functions as the primary element causing elevated IIIa mRNA splicing in Ad-NE, our results also show that the previously characterized 3RE (8) makes a smaller but important contribution to the tight control of IIIa pre-mRNA splicing. Thus, transfer of both the 3RE and 3VDE to -globin inhibits globin splicing more effectively than each element separately (Fig. 2B). Similarly, removal of both elements from the IIIa pre-mRNA increases IIIa splicing in HeLa-NE more than removal of each element individually (Fig. 6). Our results show that the 3RE and the 3VDE are the critical viral elements controlling the splicing phenotype of a pre-mRNA in Ad-NE. However, our results also indicate that the 3RE and 3VDE have evolved to function efficiently in combination with other auxiliary splicing signals in the IIIa pre-mRNA. This is illustrated by transcript glob (3VDE), which is activated in Ad-NE, although only to about 50 to 70% of the splicing efficiency the IIIa pre-mRNA (Fig. 2 and 3). More remarkably, the glob (3RE, 3VDE) transcript regains only approximately 10% of the splicing efficiency of the wild-type IIIa pre-mRNA in Ad-NE (Fig. 2). However, the important point is that this transcript is spliced more efficiently in Ad-NE compared to HeLa-NE (Fig. 2). The identity of these auxiliary signals is currently under investigation.

The surprising finding that transcripts IIIa (-3RE, -3VDE) and IIIa (-3VDE) were spliced more efficiently in Ad-NE compared to the wild-type IIIa transcript raises the question whether inadvertent expression of the IIIa protein may be negative for virus multiplication. The IIIa protein is characterized as a structural component of the viral capsid. So why is IIIa splicing subjected to such a tight control during virus infection? Experiments are in progress to determine whether unregulated IIIa protein expression has negative effects on virus multiplication.

The observation that a virus infection-dependent splicing enhancer controls IIIa pre-mRNA splicing may have a wider significance. Other viruses may have evolved similar strategies to be able to efficiently produce viral mRNAs under conditions where host cell gene expression is limited through viral inactivation of key cellular splicing factors. In late adenovirus-infected cells, the activity of the SR family of splicing factors is severely reduced by a virus-induced dephosphorylation (9). Since hyperphosphorylated SR proteins are essential for generic pre-mRNA splicing, such a posttranslational modification would be expected to reduce host cell pre-mRNA splicing. Our results further suggest that the 3VDE may provide a mechanism by which adenovirus can sustain an efficient splicing of the IIIa pre-mRNA, even under conditions of limiting concentrations of functional SR proteins.

This hypothesis makes several predictions that currently are under investigation. For example, other adenovirus pre-mRNAs that show an enhanced splicing late during infection should have evolved similar infection-dependent splicing enhancers. As shown in Fig. 5, U2 snRNP is required for IIIa splicing in Ad-NE. Thus, the enhanced IIIa splicing observed in Ad-NE may result from a viral factor and/or an alternative cellular factor that substitute for the general splicing factor U2AF to promote U2 snRNP recruitment to the IIIa 3' splice site. As shown in Fig. 7B, the increase in IIIa splicing in Ad-NE is not accompanied by an increase in U2AF⁶⁵ binding to the IIIa 3' splice site, suggesting that the 3VDE may function as a splicing enhancer in the absence of efficient U2AF⁶⁵ recruitment. Interestingly, the steady-state amounts of U2AF⁶⁵ are identical in HeLa-NE and Ad-NE (Fig. 7A), yet the RNA binding capacity to the -globin 3' splice site is slightly reduced in Ad-NE (Fig. 7B). This result suggests that RNA binding by U2AF⁶⁵ is reduced by a virus-induced posttranslational modification. Based on our previous work (9), we would expect this modification to be a virus-induced dephosphorylation. However, it remains to be shown that U2AF⁶⁵ is a phosphoprotein.

Collectively, our data suggest that the 3VDE may operate through an alternative, potentially U2AF-independent mechanism. Although such a model is attractive, preliminary experiments suggest a more complex regulation, since the IIIa pre-mRNA is not spliced in U2AF-depleted Ad-NE (unpublished observation). Thus, a significance of U2AF for IIIa 3' splice site activity cannot be excluded. Potentially, other viral or cellular splicing factor(s), essential for IIIa splicing, also are lost from the nuclear extract during U2AF depletion. Alternatively, the results may suggest that U2AF is still required for IIIa splicing in Ad-NE, although it does not make direct contact with the RNA. Perhaps it becomes tethered to the 3VDE by interacting with another factor that makes the direct contact with the IIIa pyrimidine tract. Experiments are in progress to resolve questions of this type.

Irrespective of the mechanistic details of how the 3VDE functions, the results presented here are novel and important, because they demonstrate the existence of a regulatory element that has evolved to function as a splicing enhancer only in the context of a virus infection. We propose that other mammalian viruses have evolved similar virus infection-dependent splicing enhancers to take control of the RNA biosynthetic machinery in the infected cell.