Previous studies have shown that increasing the pyrimidine content of the PPT of HIV-1 hnRNP A/B-dependent repressed 3′ splice sites reduces ESS silencing activity (8
). These results suggested that reduced binding of the essential splicing factor U2AF65 to the PPT may be the rate-limiting step in hnRNP A/B-dependent ESS inhibition (3
). Our studies have directly tested this hypothesis and showed that, although 5′ splice site selection was not inhibited by interference with U2AF65 binding to the PPT of either 3′ splice site A2 or A3, 3′ splice site selection was inhibited. Negative regulation of alternative splicing by interference with U2AF65 binding is a common feature of numerous alternative splicing pathways. The Drosophila melanogaster
sex-specific splicing regulatory factor Sxl competes with U2AF65 for binding to a region containing the PPT of the male-specific default 3′ splice site within tra
). The splicing of msl-2
pre-mRNA is also inhibited by competition between Sxl and U2AF65 for the PPT of the repressed 3′ splice site (35
). The human PPT binding protein represses splicing of an adenovirus-derived in vitro splicing substrate by competing with U2AF65 for binding to the precursor RNA (44
). In each of the above cases, inhibition of U2AF65 binding to the PPT results from the binding of a negative regulatory factor to an adjacent or overlapping region of the PPT within the intron of the pre-mRNA. In contrast, we have shown that inhibition of U2AF binding to the PPT was caused by the activity of an exonic negative regulatory element positioned downstream of the 3′ splice site.
How does the ESS element exert its effects on the binding of U2AF65 from a position approximately 75 nt downstream of the PPT? Our previous data have suggested that splicing repression by ESSV does not occur as a result of a simple binding of one trans
-acting factor to one cis
element, since an N-terminal-truncated hnRNP A1 protein containing the RNA-binding domain (UP1) does not restore ESSV activity to hnRNP A/B-depleted HNE (8
) (Fig. ). Similar results have been reported for inhibition of splicing by HIV-1 ESS2 and ESS3, the ESS in the K-SAM alternative exon within the human fibroblast growth factor receptor 2 pre-mRNA, and the ESS in the cytoskeletal 4.1R protein (12
). These results suggest that the binding of hnRNP A/B proteins to an ESS via the RNA binding domain is not sufficient to inhibit splicing and that the C-terminal region of hnRNP A1, which may mediate multimerization of the protein, is also necessary for splicing inhibition (13
). In addition, we have shown above that ESSV activity apparently proceeds in a unidirectional manner. Within the context of substrate HS1-ESSVB (Fig. ), ESSV repressed splicing only to 3′ splice site A3 approximately 75 nt upstream and not to the 3′ splice sites (A4c, A4a, A4b, and A5) located in a region approximately 75 to 125 nt downstream. Together, these results are consistent with a recently described model proposing that ESSs serve as high-affinity binding sites for hnRNP A/B proteins and as a nucleation points for multimerization of hnRNP A/B proteins to regions upstream of the ESS (51
). In further support of this model, footprinting analyses of tat
exon 3 RNA, which contains an HIV-1 hnRNP A/B-dependent ESS (ESS3), have shown that purified hnRNP A1 protects several regions within the RNA upstream of the silencer (15
). This multimerization of hnRNP A/B proteins may not be sufficient to explain inhibition of U2AF65 binding since we were unable to show inhibition of binding of purified HisU2AF to ESSVB3′ by addition of excess purified hnRNP A1 in the absence of HNE (data not shown). This result suggests that there may be an additional factor(s) necessary to transmit inhibition of U2AF65 binding from hnRNP A/B proteins bound at the ESS. Together, these results are consistent with the model shown in Fig. , where inhibition of U2AF65 binding to an upstream PPT requires a higher-order RNA-protein structure relying on protein-protein interactions between hnRNP A/B proteins and possibly other proteins.
FIG. 7. Working model of the ATP-independent complexes formed in vitro. (A) Assembly of splicing components under conditions of repression. hnRNP A/B proteins (circles) bound to ESSV (gray box) nucleate binding of additional hnRNP A/B proteins to upstream cis (more ...)
Note that the HIV-1 3′ splice site A3 used throughout this study contains a short PPT interspersed with purines, and therefore splicing of the intron is predicted to be AG dependent (39
). The binding of U2AF65 to short PPTs, which are characteristic of AG-dependent introns, is stimulated by the binding of U2AF35 to a region that includes the 3′ splice junction (26
). Therefore, it is possible that ESSV indirectly inhibits the binding of U2AF65 to the A3 PPT by interfering with stabilization mediated by the binding of U2AF35 to the 3′ splice junction. It is also possible that ESSV interferes with the binding of the branch point binding protein (mBBP), which is also a component of E complexes and which acts to stabilize the binding of U2AF65 (7
). Further characterization of the ATP-independent, heparin-sensitive complex observed in the E complex EMSAs presented here will be necessary clarify these issues.
We do not yet know if the other hnRNP A/B-dependent ESS elements inhibit U2AF65 binding by the same mechanism as does ESSV. Based on the previously described shared characteristics of the hnRNP A/B-dependent silencers, this would be an expected result. However, Tange et al. (47
) have reported that the hnRNP A/B proteins in HNE supplemented with the SR protein SF2/ASF do not inhibit U2AF65 binding to the PPT of HIV 3′ splice site A7. In its native context A7 is negatively regulated by ESS3 and an intronic splicing enhancer. The binding of U2 snRNP to the branch point sequence in this case appears to be inhibited (47
). Thus, it is possible that, in different sequence contexts, different rate-limiting steps of spliceosome assembly may be affected by hnRNP A/B-dependent ESS elements.
We have shown that ESSV did not affect the rate of formation of an ATP-independent, heparin-sensitive complex that requires intact U1 snRNP. Our results indicate that this complex, containing U1 snRNP, presumably bound at the 5′ splice site, is stable, even though splicing of the substrate is almost completely inhibited by ESSV. It is not yet clear whether this complex is a functional precursor to the E complex or is an aberrant splicing complex. As evidence for the former, we have found that the 5′ splice site is capable of pairing with nonrepressed alternative 3′ splice sites despite the presence of an ESSV-repressed 3′ splice site in the substrate. However, we cannot exclude the possibility that the selected alternative 3′ splice sites influence the activity of the 5′ splice site.
Our results suggest that U1 snRNP is stably associated with 5′ splice site D1 independent of splicing, as intact U1 snRNP is required for formation of a stable complex that forms on a splicing substrate which is almost completely inhibited from splicing by ESSV. The tethering of U1 snRNP to HIV-1 5′ splice sites independent of splicing may serve several functions during virus replication. Hybridization of U1 snRNA to 5′ splice site D4 has been shown to be necessary for stabilization of both env
mRNA and unspliced RNA, suggesting that U1 snRNP bound to 5′ splice site D4 has a function independent of splicing (29
). In addition, it has been proposed that the tethering of U1 snRNP to HIV-1 5′ splice site D1 is primarily responsible for prevention of premature cleavage and polyadenylation at a 3′ processing signal located approximately 70 nt downstream of the cap site in the primary viral transcript (5
). According to this model, synthesis of all viral transcripts would require the binding of U1 snRNP to 5′ splice site D1 independent of splicing, because approximately 50% of the viral precursor RNA remains unspliced in an infected cell. A splicing-independent function of U1 snRNP in the context of HIV-1 infection remains to be established.