INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Avian leukosis viruses (ALVs) have a single copy of an ~100-nucleotide (nt) sequence termed the direct repeat element within their 3' untranslated region upstream of the 3' long terminal repeat. Nondefective avian sarcoma viruses (ASVs) contain two such direct repeat elements which have approximately 80% sequence homology. The downstream element is in the 3' untranslated region, and the upstream element is between the env and src genes approximately 100 nt upstream of the src 3' splice site. The direct repeat elements are composed of two subelements (dr1 and dr2) (Fig. 1) (31). Precise deletion of both ASV dr1 elements results in a defective virus phenotype characterized by reduced cytoplasmic levels of unspliced RNA, increased turnover of unspliced viral RNA, and very low virus particle production (27). Deletion mutants with single dr1 elements, either the downstream dr1 (DDR) or upstream dr1 (UDR), are replication competent but exhibit delayed-replication phenotypes (26). Delayed phenotypes have also been reported for single dr1 constructs lacking the src gene (24). When the dr1 elements are inserted into human immunodeficiency virus (HIV)-based reporter constructs lacking the Rev-binding site (RRE), they facilitate Rev-independent transport and expression of unspliced RNA (24, 26, 34). Thus, they have the properties of constitutive transport elements (CTEs). Such posttranscriptional elements are present in the RNA transcripts of members of the type D retrovirus and hepadnavirus families (6, 7, 13, 36). In the case of the type D simian retrovirus Mason-Pfizer monkey virus, several factors have been identified that bind to their CTEs and facilitate unspliced viral RNA transport (10, 29, 30). Although nuclear proteins specifically binding to the ASV and ALV dr1 elements have not yet been reported, such factors are likely to mediate the CTE activity.

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Schematic representations of the pJD100-C infectious RSV proviral construct, riboprobe template, and sequence alignment. (A) Diagrams of the wild-type RSV proviral DNA, showing sites of interest marked by nucleotide numbers, viral unspliced and spliced RNA species, and antisense riboprobe template used to analyze viral RNA by RNase protection assays. pMap21BS spans the env 3' splice site (nt 5042 to 5258) and the src 3' splice site (nt 6983 to 7330), with a heterologous spacer sequence between the RSV sequences. The sizes of the fragments protected by each RNA species are indicated below the riboprobe map. 5'ss, 5' splice donor; env 3'ss, env 3' splice acceptor; cryp 5'ss, cryptic 5' splice site; src 3'ss, src 3' splice acceptor. (B) Sequence comparison of the PrC RSV UDR and the DDR. Mutations of the UDR and DDR used in this study are indicated. Asterisks indicate lack of corresponding bases. Dashes represent the identity with the wild-type sequence. Nucleotides in boldface type indicate identity between the UDR and DDR sequences.

In addition to the effects on unspliced RNA accumulation and turnover, cells infected with ASV mutants with deleted dr1 elements are characterized by an increased steady-state level of src mRNA and a decrease in spliced env mRNA (27). We and others have previously shown that the region between the env and src genes contains an RNA element or elements that act to inhibit splicing at the src 3' splice site (1, 21). Mutagenesis of a 23-nt sequence immediately upstream of the UDR results in an approximately twofold selective increase in src splicing (1). We termed this sequence the suppressor of src splicing (SSS). The effect of dr1 deletions on src mRNA levels suggests that these elements might also play a role in inhibiting src splicing.

We wished to obtain evidence as to whether the same or different factors are responsible for the effects of the UDR on unspliced RNA transport and/or stability and on src RNA splicing. To this end, we compared a panel of UDR point mutants for their effects on CTE activity and RNA splicing in one- and two-dr1 element virus constructs and found a direct correlation between the mutations and these two activities. The results also indicated that both dr1 elements function in the context of ASV to selectively inhibit splicing at the src 3' splice site and increase the level of unspliced RNA. This may explain the increase in virus fitness resulting from the presence of two dr1 elements in ASV. Our experiments further suggest that in the wild-type virus the dr1 elements act together with the SSS element to inhibit src splicing.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids. The chloramphenicol acetyltransferase (CAT) reporter plasmid pCMV138 and the -galactosidase expression plasmid pCMV110 were obtained from Thomas Hope (The Salk Institute, La Jolla, Calif.). Enzymes used for cloning were obtained from New England Biolabs, Inc. (Beverly, Mass.), and Roche Molecular Biochemicals (Indianapolis, Ind.). PCR primers used in this study are listed in Table 1. Nucleotide numbers correspond to the sequence of the Prague C (PrC) strain of Rous sarcoma virus (RSV) (25). Plasmids pUDR(+), pDDR(+), and pDDR() have been previously described (26). To facilitate construction of other CAT constructs, the XhoI site in the 3' exon of pUDR(+) was removed by partial digestion with XhoI, the cohesive ends were blunted with T4 DNA polymerase, and blunt-end ligation was performed. To generate the other CAT constructs, PCR products containing the PrC wild-type UDR and UDR mutations were synthesized using an infectious PrC RSV plasmid pATV-8 as the template (14). The antisense primers were A51, A64, A65, A66, A67, A68, A69, A70, A71, A72, A73, A81, A82, A83, and A84. The sense primer was S43. The PCR products were cleaved with ClaI and XhoI, and the ClaI/XhoI fragments (nt 6864 to 6983) were inserted into pUDR(+), which was also cut with ClaI and XhoI. Thus, the CAT clones contain both the UDR (nt 6897 to 6989) and flanking sequences at both ends (nt 6864 to 7037). To generate DDR CAT clone pSW19C, a PCR product spanning nt 8811 to 8909 was produced from template pJD100 using primers S57 and CMS A1. This product was cleaved with AccI and ClaI and cloned into the unique ClaI site of pCMV138. The same strategy was used to create DDR mutants pSW21C and pSW22C using sense primer S62 and antisense primers 2B4 or 2B5 for production of PCR products.

Cell culture and DNA transfection. Secondary chicken embryo fibroblasts (CEF) were cultured in SGM (medium 199 [Bethesda Research Laboratories, Inc., Gaithersburg, Md.] supplemented with 10% [vol/vol] tryptose phosphate broth and 5% [vol/vol] calf serum). CEF were transfected with 7.5 µg of proviral DNA per 60-mm-diameter plate by the DEAE-dextran procedure as previously described (22) or 6.5 µg of CAT clone DNA plus 2.5 µg of pCMV110 per 35-mm-diameter well by the calcium phosphate coprecipitation procedure essentially as previously described (33). Cells were passaged every 3 days.

Enzyme assays. CAT assays were performed using CEF extracts harvested 48 h posttransfection (18). The amount of cell extract analyzed was adjusted based upon LacZ activity produced by the cotransfected control plasmid, pCMV110. Typically, a 5-µl aliquot of the 400 µl of cell extract obtained from a 35-mm-diameter well was used. Data shown are the average of a minimum of four independent transfections. The reverse transcriptase (RTase) assays were carried out as described previously (35). Briefly, after transfecting CEF with proviral DNA, culture medium was analyzed for RTase activity at various time points posttransfection. The amount of [-³²P]dTTP incorporated into the product is proportional to the virions released to the medium.

RNA analysis. Medium from infected cells was changed 6 h prior to RNA harvest (5). In RNA stability experiments, parallel cell cultures were treated with medium containing dactinomycin at a concentration of 1 µg/ml (Sigma Biochemicals, St. Louis, Mo.). At various times after this, total cellular RNA was harvested using TRI REAGENT (Molecular Research Center, Inc., Cincinnati, Ohio). Cytoplasmic and nuclear RNAs were isolated as previously described (27a). Briefly, cells were treated with 0.5% NP-40 and 0.5% sodium deoxycholate followed by Dounce homogenization. The separation of nuclear and cytoplasmic fractions was monitored by phase-contrast microscopic observation. The purified nuclei appeared to be free of cytoplasmic tags. RNase protection assays were carried out essentially as described previously (5). Total cellular RNA (5 to 20 µg) and in vitro-transcribed pMap21BS riboprobe (6 × 10⁶ cpm) were used for each assay. RNA and riboprobe were hybridized at 57°C in 90% formamide for 14 to 16 h. Digestion with T₁ RNase (Roche) was carried out at room temperature for 15 min. Samples were analyzed on 5% polyacrylamide gels containing 7 M urea. Relative molar ratios of viral RNA species were calculated based on radioactivity measurement with a beta imager (Packard InstantImager) and normalized for the number of uridine residues in the different protected RNA bands.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

CTE activity and virus replication are correlated in single-dr1 constructs with UDR point mutations. We first created point mutations across the UDR element as shown in Fig. 1B. We selected the region between nt 6934 and 6982 for mutagenesis because this is the most highly conserved region of dr1 elements in different ASV and ALV strains. Furthermore, preliminary results indicated that the mutated region was within an 80-nt-minimum dr1 sequence sufficient to support approximately 50% of the virus replication exhibited by virus with the entire dr1 element and that the most 5'-proximal 20 nt of this sequence was relatively insensitive to mutagenesis (S. Winistorfer and C. M. Stoltzfus, unpublished data). The UDR point mutations were first tested for their ability to replace Rev and the HIV type 1 (HIV-1) RRE in a CAT reporter assay. This assay utilizes a reporter construct (pCMV138) in which the cat gene is placed between the 5' and 3' splice sites of an intron within the HIV-1 env gene (Fig. 2A). CAT expression, which peaks at approximately 48 h posttransfection, is therefore dependent on the accumulation and expression of cytoplasmic unspliced RNA. The effects of the same mutations in the context of the single upstream dr1 virus construct DDR-C were also tested. DDR-C has a delayed replication phenotype compared to the wild-type virus, with peak RTase levels occurring at day 9 posttransfection (26). The assay depends on transient transfection of virus DNA constructs followed by spread of infectious virus to surrounding cells. Mutants 25C, 31C, 32C, and 33C were most strongly affected in the CAT assay for CTE activity (Fig. 2A). The corresponding virus mutants WG25, WG31, WG32, and WG33 were also the most affected in virus replication as determined by medium RTase activities (Fig. 2B). On the other hand, mutations 28C, 37C, and 29C had relatively small effects on UDR CTE activity. The corresponding mutations also had only minor effects on virus replication. The remaining mutations had intermediate effects on CTE activity (24C, 26C, 27C, 81C, 30C, 34C, 35C, and 36C). This was also true for replication of the corresponding virus mutants. These results indicated that the effects of dr1 mutations on CTE activities as determined by the CAT reporter assays were directly correlated with effects of the same mutations on virus replication.

View larger version (51K): [in this window] [in a new window]   FIG. 2.   Effects of UDR mutations on CTE activity correlate with effects on virus production. (A) UDR (nt 6864 to 7037) was cloned into the intron region of pCMV138 in both the sense and antisense orientation [pUDR-C(+) and UDR(), respectively]. UDR mutations shown in Fig. 1B were also cloned into pCMV138. Values shown are based on at least four independent transfections. Transfected cells were harvested at 48 h posttransfection, corresponding to peak CAT levels. (B) RTase activities of the indicated UDR single-dr1 virus mutants on day 9 posttransfection compared to DDR-C. At this time, cells were 50 to 100% infected as determined by morphological transformation, and the values represented peak levels of RTase. The values shown are the average of three independent experiments. Standard deviations are indicated by error bars.

Selective increase in src splicing and increased turnover rate of unspliced RNA are characteristic of UDR point mutants strongly affecting CTE activity. We previously showed that deletion of both dr1 elements resulted in decreased steady-state levels of unspliced and env mRNA and increased levels of spliced src mRNA (27). This suggested that the dr1 elements, in addition to their effects on CTE activity, might act to inhibit splicing. We determined if UDR point mutations in single-dr1 constructs had similar effects. Figure 3 shows the results of RNase protection assays of total RNA isolated from cells infected with selected viral mutants. The wild-type single-dr1 virus (DDR-C) showed an elevated level of src mRNA and a decreased unspliced RNA level compared to the two-dr1 wild-type parental virus (compare to pJD100-C [see Fig. 7A]). This difference will be discussed below. WG30, with a mutation which had moderate effects on CTE activity (Fig. 2A) and virus replication (Fig. 2B), was not significantly different from DDR-C. In contrast, viruses with mutations, shown in Fig. 2 to strongly affect CTE activity and virus replication (WG25, WG33, and the double mutant WG2533), had significantly elevated molar ratios of spliced src mRNA and reduced molar ratios of unspliced RNA compared to DDR-C. In addition, protected bands corresponding to a double spliced mRNA were significantly elevated in these mutants. This mRNA species arises by splicing from a cryptic 5' splice site within the env gene to the src 3' splice site (Fig. 1A). This multiply spliced RNA has previously been detected in CEF transfected with an ASV mutant in which the polypyrimidine tract upstream of the src 3' splice site was improved (35). It was also detected in nonpermissive mammalian cells transfected with ASV DNA (4). In both of these cases, the use of the cryptic 5' splice site was correlated with a selective increase in splicing at the src 3' splice site.

View larger version (50K): [in this window] [in a new window]   FIG. 3.   Selective increase in src mRNA in cells transfected with single dr1 virus clones containing UDR point mutants. (A) Representative RNase protection assays of total cellular RNA (10 µg) for viral RNA species. The locations of the protected bands are indicated. (B) Percentages of different viral RNA species in infected cells were determined by measurements of radioactivity as discussed in Materials and Methods. Values are the average of three independent experiments. Asterisks indicate values significantly different (P < 0.05) from the parental single dr1 virus (DDR-C). Standard deviations are indicated by error bars.

View larger version (103K): [in this window] [in a new window]   FIG. 4.   Rapid initial turnover rate of unspliced RNA of single-dr1 UDR mutant viruses. (A) RNase protection assays of total cellular RNA harvested at various time points after addition of dactinomycin (1 µg/ml). (B) The amounts of radioactivity remaining at the indicated time points after addition of drug were measured and the amounts of viral RNA remaining versus time after the drug addition were determined. The values shown are averages of three independent experiments. Standard deviations are indicated by error bars.

View larger version (58K): [in this window] [in a new window]   FIG. 5.   The unstable fraction of UDR mutant unspliced RNA is located in the nucleus. DDR-C- and WG33-infected cells were treated as in Fig. 4 with dactinomycin, and fractionation of the nucleus and cytoplasm was carried out at the indicated time points after the drug addition. Shown are the protected bands for the unspliced RNA species in the nuclear and cytoplasmic fractions. Percent of remaining RNA are given for two independent experiments (Exp1 and Exp2).

UDR mutations in the context of two-dr1 wild-type virus constructs result in a delayed virus replication phenotype. The UDR point mutations selectively affected src splicing and nuclear unspliced RNA turnover in the context of a single-dr1 virus. This confirmed and extended previous results based on deletions of the dr1 elements (24, 27). The effects of these same mutations in the context of two-dr1 virus constructs were studied. We previously showed that the UDR deletion mutant containing only the DDR exhibited a delayed-replication phenotype compared to wild-type virus (26). To determine if this resulted from loss of UDR function or was a secondary effect of the deletion, we measured the replication kinetics of selected UDR point mutants shown in Fig. 1B in the context of two-dr1 virus constructs (Fig. 6). Some of the viral mutants (WG25bd, WG31bd, and WG33bd) exhibited markedly delayed phenotypes. The delay was increased for the double mutant WG2533bd that combines the two single WG25bd and WG33bd mutations. The delay seen for WG2533 was identical to that seen for the single-DDR virus construct UDR (data not shown). In the case of the triple mutant WG872533bd, which has an SSS mutation in addition to the dr1 mutations, there was a further delay compared to WG2533bd. Only slight replication delays were observed for mutants WG24bd, WG29bd, and WG30bd. Thus, the results in Fig. 6 indicated that the effects of the dr1 mutations on wild-type replication kinetics directly correlated with the effects of these same mutations on CTE assays and replication in the single-dr1 context (compare data in Fig. 6 and Fig. 2).

View larger version (25K): [in this window] [in a new window]   FIG. 6.   UDR mutations in the context of two-dr1 virus constructs cause a delayed virus replication phenotype. All of the proviral constructs used contain a wild-type DDR with a mutated UDR. After transfection of CEF, culture media were harvested at indicated time points and analyzed for RTase activity. At least three independent experiments were carried out for each mutant. Representative data are shown.

Selective increase in src mRNA splicing of UDR point mutants in two-dr1 wild-type virus constructs. To determine if the replication delays were correlated with changes in RNA splicing or RNA turnover, we performed RNase protection analyses to measure molar ratios of spliced and unspliced viral mRNA species in total RNA isolated from mutant and wild-type infected cells (Fig. 7A). In mutants with a delayed phenotype (WG25bd, WG31bd, WG32bd, WG33bd, WG2533bd, and WG872533bd) there were small but significant increases in the percentages of spliced src mRNA. Corresponding decreases in percentages of unspliced RNA were also observed. There was also a significant increase in src splicing when WG872533bd was compared to WG2533bd, suggesting that in the context of the wild-type virus both the dr1 and SSS act together to inhibit src splicing. The percentages of RNA spliced at the env 3' splice site on the other hand were not significantly different from each other.

View larger version (68K): [in this window] [in a new window]   FIG. 7.   UDR mutations in two-dr1 virus constructs cause a selective increase in src mRNA splicing but do not significantly affect RNA stability. (A) Relative molar ratios of viral RNA species in cells infected with mutant viruses based on RNase protection assay of total cellular RNA harvested on day 5 posttransfection. The values shown are the average of data from at least three independent experiments. Asterisks indicate values significantly different (P < 0.05) from those for wild-type pJD100-C. (B) Half-lives of RNA species of mutant viruses were determined as described in the legend to Fig. 4. Values shown are averages of three independent experiments. Standard deviations are indicated by error bars.

DDR mutations in the context of two-dr1 wild-type virus constructs also result in a delayed virus replication phenotype. We pointed out when discussing the data of Fig. 3 that the DDR deletion mutant DDR-C exhibited an increase in src splicing compared to the wild-type virus (compare Fig. 3B and 7A). This result suggested that the DDR as well as the UDR might act to inhibit splicing at the src 3' splice site. Thus, we determined if DDR point mutations had similar effects on replication of two-dr1 ASV constructs. Preliminary data indicated that point mutations within the DDR, like the UDR, had a range of effects on CTE activity (data not shown). For the experiments shown here we selected the three DDR point mutations shown in Fig. 1B to test mutants with small effects and those with strong effects. As shown in Fig. 8A and B, mutants pSW19C and pSW19, respectively, affected CTE activity and virus replication only slightly compared to wild-type pJD100. On the other hand, mutants SW21C and SW22C and the corresponding virus mutants (SW21 and SW22) had greater effects on CTE activity and virus replication. These three mutations were then tested in the context of two-dr1 virus constructs. Figure 9A shows that the replication kinetics of SW19bd were not significantly different from those of the wild-type pJD100-C control. On the other hand, the replication kinetics of SW21bd and SW22bd were markedly delayed compared to those of the wild-type virus. The results indicated that, similar to the UDR mutants, the magnitude of the replication delay was directly related to the effect of the DDR mutation on CTE activity and on virus replication in the single-dr1 context.

View larger version (31K): [in this window] [in a new window]   FIG. 8.   Effects of DDR mutations on CTE activity correlate with effects on viral production of single-dr1 DDR viruses. (A) DDR mutations were cloned into pCMV138 as described in the legend to Fig. 2. Locations of mutations are shown in Fig. 1B. The values shown are the average of six independent experiments. (B) RTase activity of indicated single-dr1 viruses with DDR mutations at different times after transfection. At least three independent experiments were carried out for each mutant. Representative data are shown.

View larger version (37K): [in this window] [in a new window]   FIG. 9.   DDR mutations in two-dr1 virus constructs resulted in a delayed virus replication phenotype as described in the legend to Fig. 6 (A) and selective increases in src mRNA splicing as described in Fig. 7 (B). The results are based on three experiments, and in panel B, values significantly different from those for the wild type (P < 0.05) are labeled with asterisks. Standard deviations are indicated by error bars.

Selective increase in src mRNA splicing of DDR point mutants in two-dr1 wild-type virus constructs. DDR mutants exhibiting delayed phenotypes also exhibited a selective increase in the level of spliced src mRNA (Fig. 9B). Mutants with the most effect on replication (SW21bd and SW22bd) showed the greatest increase in the level of spliced src mRNA. This was concomitant with a relative decrease in unspliced RNA level. The half-lives of the RNA species as determined by actinomycin D chase experiments were not significantly different from those of the wild type (data not shown). Since changes in stabilities could not account for the changes in the levels of the different viral RNA species, we concluded that mutations of the DDR in the context of two-dr1 viruses, like the UDR mutations, selectively affect splicing at the src 3' splice site.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Our results have indicated that point mutations of both the UDR and DDR elements selectively elevate splicing of ASV RNA at the src 3' splice site. Point mutations having the most effect on CTE activity and virus replication in single-dr1-containing virus constructs also have the most effect on the level of src splicing in the double-dr1 constructs. This suggests that the same putative dr1-binding factor or factors necessary to facilitate cytoplasmic unspliced-RNA accumulation in the single-dr1 virus constructs are also necessary for src splicing inhibition in the two-dr1 constructs. In the context of the two-dr1 wild-type virus, these same mutations result in a delayed virus replication phenotype. This suggests that the replication delay is caused by the concomitant reduction in the level of unspliced viral RNA to serve as mRNA and genomic RNA. Our previous results have indicated that mutations that optimize the polypyrimidine tract of the src 3' splice site result in an oversplicing virus phenotype. This causes a replication delay and selection of revertants in which inefficient splicing is restored (35). Similar results were also reported earlier for ASV env 3' splice site branch point mutants (8, 15). Thus, there appears to be strong selective pressure for ASV to maintain inefficient splicing, and even small changes in unspliced-RNA levels appear to have significant effects on the kinetics of virus replication.

There are several possible models to explain the effect of the dr1 mutants on src splicing in the two-dr1 virus context. First, there may be increased retention of unspliced RNA precursors in the nucleus because of inefficient transport to the cytoplasm. In wild-type virus constructs, two-dr1 elements may be needed for maximum efficiency of unspliced RNA export. Nuclear retention may increase the exposure of RNA precursors to the cellular splicing machinery. Second, the putative binding of a factor or factors to dr1 elements may directly inhibit formation of functional spliceosomes at the src 3' splice site. The latter model requires that the dr1 act on splicing from both an intron location approximately 100 nt upstream and at an exon location 1.8 kb downstream from the src 3' splice site. It is possible that the src gene RNA could loop out, allowing factors binding downstream of it to associate with spliceosome components by protein-protein interactions. In dr1 mutant-infected cells there was an increase in splicing at the src 3' splice site but not a corresponding increase in splicing at the env 3' splice site. This would appear to favor the latter hypothesis. However, we cannot rule out a model in which nuclear retention results in selective splicing at the src 3' splice site. If this model were true, we might expect to see differences in the unspliced viral RNA nuclear-to-cytoplasmic ratios when cells infected with two-dr1 constructs bearing mutations in a single dr1 are compared to wild-type. However, we have not found detectable difference when these ratios were determined by cell fractionation experiments (W. Guo and C. M. Stoltzfus, unpublished data).

Effects of dr1 deletions on stability of unspliced RNA have previously been reported (24, 27). We have extended these studies to show that viral unspliced RNA in cells infected with virus constructs containing a single mutated dr1 rapidly turns over in the nucleus. In contrast to the inhibitory effects on src splicing, which requires both dr1 elements, our data indicated that a single wild-type dr1 is sufficient to stabilize unspliced RNA in the nucleus. This effect on unspliced RNA stability may be a consequence of failure to transport unspliced RNA efficiently. Alternatively, the instability of unspliced RNA in the nucleus may reduce the amount of mutant unspliced RNA available to be transported to the cytoplasm.

We have previously reported that deletion or point mutagenesis of the 23-nt SSS region immediately upstream of the UDR results in elevated src splicing (1). McNally and Beemon showed that a fragment containing both the SSS and UDR can inhibit splicing of a heterologous cellular c-myc gene when the fragment was placed within the intron of the gene (21). Amendt et al. showed that splicing of src RNA substrates in HeLa cell nuclear extracts was inhibited by addition of CEF nuclear extract. This specific inhibition occurred with substrates containing both the SSS and UDR sequences upstream of the src 3' splice site (1). We showed above that in the context of the two-dr1 virus constructs the SSS and UDR both contribute to the total reduction in src splicing. This suggests that each of the elements function separately and may bind to a different inhibitory factor or factors.

Our results suggest that factors binding to CTEs of a simple retrovirus RNA affect its splicing. Previous data have suggested that Rev posttranscriptional factors of lentiviruses, in addition to their role in transport of viral RNA, may bind to the RRE and directly influence the splicing of viral RNAs. Kjems et al. showed that addition of HIV-1 Rev or the basic RNA-binding domain subfragment of Rev to in vitro splicing reactions specifically inhibits splicing of substrates containing the HIV RRE and results in the formation of nonfunctional spliceosome complexes (16, 17). Other data obtained by transfection of HIV-1 constructs containing mutations in splice sites are consistent with the interaction of HIV-1 Rev with the spliceosomal machinery (11, 19). Equine infectious anemia virus (EIAV) RNA undergoes alternative splicing to either four-exon or three-exon multiply spliced RNAs. Expression of EIAV Rev is required for the exon skipping (20). This may occur because of overlap between splicing enhancers and the EIAV RRE (3, 9). Alternative splicing may therefore be regulated by competition of cellular SR proteins and EIAV Rev for binding at the enhancer site (3). Further understanding of the mechanism by which splicing at the ASV src 3' splice site is inhibited by the UDR will require isolation of the putative binding factors and reproduction of the phenomenon in an in vitro splicing system.

Our results using upstream dr1 mutants in the context of one- and two-dr1 viruses containing the src gene have shown an excellent correlation between ASV replication kinetics and accumulation of unspliced RNA as determined by the CTE assay and by analysis of RNA from mutant-infected cells. This correlation is in good agreement with previous studies of Ogert and Beemon who mutated the downstream dr1 element of a src deleted Prague ASV construct with a single-dr1 element in the 3' UTR (23). It has previously been shown that ASV dr1 mutant unspliced RNA does not accumulate in the nucleus, as might be expected if the dr1 element was indeed a CTE (24, 27). Our results given above have shown that this can be explained by a rapid turnover of mutant unspliced RNA in the nucleus.

However, some of the dr1 mutant unspliced RNA continues to be transported to the cytoplasm. Our previous data have indicated that an additional effect of dr1 deletions or point mutations is a rate of particle assembly that is lower than expected based on the amount of unspliced viral RNA and viral Gag precursor present in the cytoplasm (27). We have hypothesized that, because of reduced levels of cytoplasmic unspliced gag-pol mRNA due to nuclear events, Gag protein levels may drop below a threshold necessary for efficient particle assembly. Alternatively, a reduced rate of particle assembly may result from an additional function of the dr1 element targeting unspliced RNA to particular sites in the cytoplasm favorable for Gag translation leading to virus assembly (26, 27). Results from other laboratories using different src-deleted ASV constructs with dr1 mutations have indicated that there are modest reductions in cytoplasmic unspliced RNA levels, but there appears to be little or no change in virus particle production. Instead, packaging of viral RNA within these particles appears to be reduced significantly when the dr1 element is mutated (2, 28). Further studies will be necessary to resolve these discrepancies and to determine which of the above dr1 defects are most critical for virus replication.