Viruses have evolved a number of mechanisms to modulate cellular splicing. Several mammalian DNA viruses inhibit cellular RNA splicing by inducing dephosphorylation of SR splicing cofactors, which also leads to changes in viral splicing patterns (21
). In this report we demonstrate that EBV SM protein specifically directs alternative SS selection, suggesting that viruses may also regulate cellular gene expression at the level of alternative splicing.
The probability of alternative splicing is increased when the size of introns exceeds 200 to 250 nucleotides, possibly due to a transition from intron definition to exon definition as the mechanism for specifying SSs (7
). An additional factor increasing the likelihood of alternative splicing is the presence of suboptimal SSs, which are more likely to be dependent on the binding of splicing factors to exonic or intronic splicing enhancers that facilitate or stabilize spliceosome formation (17
). The final exons of the human STAT1 gene fit both of these criteria, with an intron of almost 900 nucleotides and a 5′ SS of only moderate strength as predicted by 5′ SS strength modeling algorithms (39
). As might be expected, there is a clustering of predicted exonic splicing enhancers in exon 23. Elimination of spliceosome formation by mutation of the constitutive 5′ SS did not result in efficient utilization of any downstream 5′ SS, consistent with their predicted low SS strength. The ability of SM to activate the alternative α′ 5′ SS suggests that it acts as a splicing factor by binding to one or more sites on the STAT1 RNA. Deletion of the distal portion of exon 23 inhibited SM activity as did overexpression of ASF/SF2, which binds to the same region. Moreover, ASF/SF2 competed with SM for binding to STAT1 RNA, further indicating that SM binds to the same region of STAT1 RNA and that such binding is important for function.
However, it is likely that SM interacts with multiple sites on the RNA, as it also bound to intronic sequences distal to the alternative α′ 5′ SS, and splicing enhancers are generally close to the SS, where bound splicing factors may facilitate spliceosome formation. Nevertheless, it is interesting that binding of ASF/SF2 may inhibit use of the downstream alternative 5′ SS, either by altering the local composition of factors associated with splicing complexes along the STAT1 RNA or by preventing multimeric binding of SM along the STAT1 RNA in a manner analogous to hnRNPs. It should be noted that although RNase-sensitive transfer of radiolabel from RNA to SM occurred after UV cross-linking, suggesting that SM directly contacts RNA, formation of SM-RNA complexes may require cellular proteins as these assays were performed with cell lysate rather than purified SM protein.
The production of two isoforms of STAT1 expression by alternative cleavage and polyadenylation has been well described, and several lines of evidence indicate that the ratio of STAT1β to STAT1α modulates the cellular response to viral and mycobacterial infection, IFN exposure, and apoptotic stress (1
). We have previously shown that EBV SM leads to increased total STAT1 levels, resulting in increased expression of several genes stimulated by type I IFNs (IFN-α and IFN-β) (31
). However, SM also increases the relative amount of the β form of the RNA, increasing the STAT1β/STAT1α RNA ratio (31
), which has been shown to inhibit the ability of the cell to respond to IFN-γ (1
). Demonstration of the specific effects of SM on the cellular IFN response will therefore require a comparative analysis of genes that are primarily responsive to IFN-γ but not to IFN-α/β, as there is considerable cross talk between type I and type II IFN signaling (11
). Thus, most IFN-γ-responsive genes (ISGs) are also upregulated by increases in STAT1β via the STAT1/STAT2/IRF-9 complex in which STAT1β is active. Thus, the predicted effect of SM (which increases both STAT1 α and β levels) on ISGs is that there will be a change in the relative levels of ISGs depending on how responsive they are to STAT1β, rather than an overall decrease in IFN-γ-responsive gene expression. The alternative SS described in this report generates an mRNA that is predicted to undergo either NMD or translation into the β form of STAT1 protein. The alternative processing induced by SM is thus reminiscent of recently described “poison cassette” exons present in SR protein genes themselves, which, when included as a result of alternative splicing, result in mRNAs that undergo NMD (22
). The extent to which SM-induced, alternatively spliced STAT1 mRNAs undergo NMD as opposed to translation remains to be determined.
The ability of a viral protein to influence cellular splicing and interact with cellular splicing factors by competing for binding to RNA or spliceosome components increases the potential for combinatorial diversity among splicing factors that are thought to regulate alternative splicing of greater than 70% of human genes (20
). The presence of an alternative SS which is used only in the presence of a viral protein raises the possibility of yet another level of posttranscriptional regulation of cellular gene expression by infecting viruses and increases the potential complexity of a “splicing code.”