A Premature Termination Codon in Either Exon of Minute Virus of Mice P4 Promoter-generated Pre-mRNA Can Inhibit Nuclear Splicing of the Intervening Intron in an Open Reading Frame-dependent Manner*

How premature translation termination codons (PTCs) mediate effects on nuclear RNA processing is unclear. Here we show that a PTC at nucleotide (nt) 385 in the NS1/2 shared exon of P4-generated pre-mRNAs of the autonomous parvovirus minute virus of mice caused a decrease in the accumulated levels of doubly spliced R2 relative to singly spliced R1, although the total accumulated levels of R1 plus R2 remained the same. The effect of this PTC was evident within nuclear RNA, was mediated by a PTC and not a missense transversion mutation at this position, and could be suppressed by improvement of the large intron splice sites and by mutation of the AUG that initiated translation of R1 and R2. In contrast to the PTC at nt 385, the reading frame-dependent effect of the PTC at nt 2018 depended neither on the initiating AUG nor the normal termination codon for NS2; however, it could be suppressed by a single nucleotide deletion mutation in the upstream NS1/2 common exon that shifted the 2018 PTC out of the NS2 open reading frame. This suggested that there was recognition and communication of reading frame between exons on a pre-mRNA in the nucleus prior to or concomitant with splicing.

How premature translation termination codons (PTCs) mediate effects on nuclear RNA processing is unclear. Here we show that a PTC at nucleotide (nt) 385 in the NS1/2 shared exon of P4-generated pre-mRNAs of the autonomous parvovirus minute virus of mice caused a decrease in the accumulated levels of doubly spliced R2 relative to singly spliced R1, although the total accumulated levels of R1 plus R2 remained the same. The effect of this PTC was evident within nuclear RNA, was mediated by a PTC and not a missense transversion mutation at this position, and could be suppressed by improvement of the large intron splice sites and by mutation of the AUG that initiated translation of R1 and R2. In contrast to the PTC at nt 385, the reading frame-dependent effect of the PTC at nt 2018 depended neither on the initiating AUG nor the normal termination codon for NS2; however, it could be suppressed by a single nucleotide deletion mutation in the upstream NS1/2 common exon that shifted the 2018 PTC out of the NS2 open reading frame. This suggested that there was recognition and communication of reading frame between exons on a pre-mRNA in the nucleus prior to or concomitant with splicing.
Premature termination codons (PTCs) 1 have been shown to result in decreased levels of PTC-containing mRNAs in many organisms, and there is an increasing appreciation of the effect of PTCs on RNA levels in mammalian cells. In some of these cases, PTCs have been shown to affect nuclear-associated mRNA abundance by a process termed nonsense-mediated decay, which has been suggested to degrade fully spliced mRNAs in the nucleus, possibly during mRNA export (reviewed in Refs.1 and 2).
PTCs have also been implicated in altering nuclear RNA processing events other than decay in mammalian cells, which results in both intron retention and exon skipping, suggesting that PTCs may influence splice site selection (1). In two cases, exon skipping of the 66-nucleotide exon 51 of fibrillin FBN1 RNA (3) and exon skipping and intron retention for the P4generated pre-mRNAs that generate the nonstructural proteins of the autonomous parvovirus minute virus of mice (MVM) (4,5), the effects of PTCs on RNA processing have been shown to be reading frame-dependent.
MVM is an autonomous parvovirus that is organized into two overlapping transcription units that produce three major classes of RNA (6 -8) (see Fig. 1). Transcripts R1 and R2 are generated form a promoter (P4) at map unit 4 and encode the viral nonstructural proteins NS1 and NS2, respectively, whereas the R3 transcripts are generated from a promoter at map unit 38 (P38) and encode the viral capsid proteins (6,9). Both NS1 and NS2 play essential roles in viral replication and cytotoxicity (10), and so maintenance of their relative steady state levels, which is controlled at least in part by alternative splicing, is critical to the viral life cycle (Refs. 11-13; reviewed in Ref. 14). All MVM mRNAs generated during infection or following transfection are very stable (15), and no viral proteins are known to participate in the alternative splicing of MVM pre-mRNAs (reviewed in Ref. 14).
There are two types of introns in MVM P4-generated transcripts (14) (see Fig. 1). An overlapping downstream small intron, which undergoes an unusual pattern of overlapping alternative splicing using two donors (D1 and D2) and two acceptors (A1 and A2) (16), is located between nt 2280 and 2399 and is common to both P4-and P38-generated transcripts. An upstream large intron, located between nt 514 and 1989, is additionally excised from a subset of P4-generated pre-mRNAs to generate R2 mRNA. This upstream intron utilizes a nonconsensus donor at nt 514 and has a weak polypyrimidine tract at its 3Ј splice site (13).
The NS2-specific exon is a 290-nt alternatively spliced exon that is translated in two open reading frames (ORFs). In singly spliced R1, this region utilizes ORF3 to encode NS1; in doubly spliced R2, this exon utilizes ORF2 to encode NS2 (Fig. 1). We have previously shown that PTCs in the NS2-specific exon caused a decrease in the accumulated levels of R2 relative to R1, although the total accumulated levels of R1 plus R2 remained the same. This decrease was a consequence of the artificially introduced translation termination signal acting in cis rather than the absence of a functional viral gene product and was shown to be evident in nuclear RNA and independent of RNA stability, suggesting an effect on nuclear splicing (4,5). Although perhaps not directly comparable with PTCs that cause genetic diseases in higher organisms, analyses of PTCs introduced into viral genes such as MVM have proven to be informative models for the effects of PTCs on RNA accumulation (4,5).
Efficient inclusion of the NS2-specific exon as an internal exon in vivo and consequent excision of the upstream large intron from P4-generated pre-mRNA to generate R2 require an internally redundant, bipartite exon splicing enhancer (ESE) comprised of 5Ј and 3Ј elements within the NS2-specific exon (17). The function of this ESE is sensitive to the presence of PTCs (5). A nonsense but not a missense mutation in the NS2 open reading frame at nt 2018 within the 5Ј element of the ESE affected definition of the NS2-specific exon by interfering with the ability of the bipartite ESE to strengthen interactions at the upstream large intron polypyrimidine tract. A PTC at nt 2018 alone resulted in retention of the upstream intron without a net decrease in the total accumulated P4-generated product.
When the PTC at nt 2018 was combined with a mutation in the 3Ј element of the bipartite enhancer, the NS2-specific exon was skipped. These effects were independent of RNA stability and were shown to be reading frame-dependent; single nucleotide deletions in front of the 2018 mutation, which removed the PTC from the NS2 open reading frame, fully suppressed its effect (5).
Although the NS1/NS2 shared exon is a 5Ј terminal exon, PTCs at nt 385 within this exon have also been shown to result in a decrease in the accumulated level of doubly spliced R2 relative to singly spliced R1, in a manner independent of the stability of R2 or R1 (4). In this report we show that the effect of the PTC at nt 385 was evident within nuclear RNA, was dependent on a PTC and not a missense transversion mutation at this position, and could be suppressed by improvement of the large intron splice sites. The effects of the 385 PTC and a PTC at nt 2018 were at least partially additive, suggesting that they operated by at least partially independent mechanisms. This was further underscored by the observation that the effect of the PTC at nt 385 depended upon the AUG that initiated translation of R1 and R2, whereas the reading frame-dependent effect of the PTC at nt 2018 depended neither on the initiating AUG nor on the normal termination codon for NS2. The effect of the PTC at nt 2018 could, however, be suppressed by a single nucleotide deletion mutation in the upstream NS1/2 common exon that shifted the 2018 PTC out of the NS2 ORF, suggesting that there was communication of reading frames between exons. Our observations are most consistent with a model in which reading frame can be recognized on a pre-mRNA molecule in the nucleus, prior to or concomitant with splicing.

EXPERIMENTAL PROCEDURES
Mutant Construction-Construction of p385UTT and p2018 TAA has been previously described (4).
The mutants p385 CAA, p385 TAA , and pSTOP(Ϫ) were constructed by m13-based oligonucleotide mutagenesis as described previously (4). Mutant oligonucleotides were homologous to the viral DNA except at the nucleotides that were to be altered. pSTOP(Ϫ) is a deletion of two nucleotides at 2380 and 2381. The changes made in p385 CAA and p385 TAA are shown in Fig. 2A. pAUG(X) was constructed by cutting the wild type (WT) clone of MVM at the NcoI site at nt 260 and chewing back with mung bean nuclease to create a deletion between nt 261 and 265 (which deletes the initiating AUG at nt 260). To generate pAUG(XXX) and pAUG(XXX)2018 TAA , mutations of the Kozak consensus sequences (18) at nt 474 and 504 were initially generated by m13based oligonucleotide mutagenesis (the 474 mutation changes the WT sequence 5Ј-GACATG-3Ј to 5Ј-TACATT-3Ј; the 504 mutation changes the WT sequence 5Ј-GAAATG-3Ј to 5Ј-CCAATT-3Ј), and then the m13 fragment bearing these mutations was subcloned into pAUG(X) and pAUG(X)2018 TAA , respectively. All the final clones of the above mutants were sequenced to confirm that only the desired mutations were introduced, and the absence of the P38 promoter product R3 for these mutants confirms that no NS1 was made.
Transfection and RNA Isolation-Murine A92L cells, the normal tissue culture host for MVM(p), were grown as described previously (6) and transfected with wild type and mutant MVM plasmids, using either DEAE-Dextran (4) or LipofectAMINE-Plus reagent (as described in the LipofectAMINE-Plus reagent kit manufactured by Life Technologies, Inc.). RNA was typically isolated 48 h post-transfection, after lysis in guanidinium thiocyanate, by centrifugation through CsCl exactly as described previously (15). RNase protection assays RNase protection assays were performed as described previously (15), using an [␣-32 P]UTP-labeled, SP6-generated antisense MVM RNA probe from MVM nucleotides 1854 -2378 (probe B). This probe extends from before the acceptor site of the large intron to within the small intron common to all MVM RNAs and distinguishes P4-RNA species using the large intron acceptor and either of the alternative small intron donors (designated M for the major splice donor (D1) at nt 2280 and m for the minor splice donor (D2) at nt 2317 (15)). Thus probe B can distinguish between unspliced, minor, and major forms of both R1 and R2. For analysis of RNA produced after transfection with p2018 TAA, p1T2018 TAA, p2T2018 TAA , p4T2018 TAA , p385 TAA ϩ2018 TAA , p2T385UTT, p4T385UTT, pCSD2018 TAA , pAUG(X)2018 TAA , pAUG(XXX)2018 TAA , p2018 TAA D1(Ϫ)A2(Ϫ), and p2018 TAA STOP(Ϫ), antisense RNase protection probes homologous to the 2018 and polypyrimidine tract mutations being analyzed were constructed and used. Because in all cases the relative ratio of bands within an individual lane was the important result, rigorous attempts were not made to equilibrate the amounts of specific RNA between lanes. This would be expected to vary from sample to sample depending upon transfection efficiencies and does not effect the comparison of relative ratios within a sample. RNase protection products were analyzed on a Betagen-scanning phosphorimage analyzer, and molar ratios of MVM RNA were determined by standardization to the number of uridines in each protected fragment. Nuclear RNA was isolated after the pelleting of nuclei twice through sucrose exactly as described previously (4). The nuclear RNA fractions were determined to be Ͼ95% pure as monitored by the relative absence of mature rRNA and the presence of rRNA precursors in these preparations by ethidium bromide staining following gel electorophoresis (data not shown), also as described (4).

RESULTS
A Nonsense but Not a Missense Mutation at nt 385 in the NS1/NS2 Common Exon of P4-generated Pre-mRNAs Resulted in a Reduced Accumulated Level of R2 Relative to R1, Likely Because of Effects on Excision of the Downstream Large Intron-Introduction of an ochre (TAA) PTC at nt 385 in the NS1/NS2 common exon of P4-generated pre-mRNA, either by point mutagenesis or by insertion of a 15-nt linker with overlapping PTCs in all three reading frames, led to a significant inversion in the accumulated levels of doubly spliced R2 relative to singly spliced R1 in total RNA, compared with that generated by wild type, although the total P4-generated product (R1ϩR2) was unchanged (Fig. 2, A and B). This result was similar to that previously seen when a PTC was introduced at nt 2018 in the internal NS2-specific exon of P4-generated pre-mRNA (4, 5) but of somewhat greater magnitude. A missense transversion (CAG) at nt 385 showed no such effect (Fig.  2, A and B). When the PTCs at nt 385 and 2018 were combined, an even greater decrease in the accumulated level of R2 relative to R1 was seen (Fig. 2, A and C). That the effects of the two mutations were at least somewhat additive suggested that the mechanisms behind their effects were at least partially independent, a possibility that is further supported by the observations described below. The effect of the PTC at nt 385 was also evident in pure preparations of nuclear RNA, suggesting that the effect of this PTC was manifest within the nucleus (Fig. 2D), as was previously reported for the PTC at 2018 (5).
Subtle improvements of the large intron splice sites could suppress the effect of the PTC at nt 385, consistent with the suggestion that this effect was nuclear. As shown in Fig. 3 (A  and B), improvement of either the large intron 5Ј splice site to consensus or the large intron 3Ј splice site polypyrimidine tract by as few as two additional pyrimidines, restored the accumulated levels of R2 relative to R1 to near wild type levels. Improvement of the large intron 5Ј splice site to consensus also suppressed the phenotype of the PTC at 2018 (Fig. 3, A and C); we have previously reported that improvements of the 3Ј polypyrimidine tract also suppresses the effect of the PTC at nt 2018 (5). Improvement of the large intron 3Ј splice site polypyrimidine tract by four pyrimidines resulted in splicing of the P4 product almost exclusively to R2 (Fig. 3B, sixth lane).
That the effects of the PTCs can be overcome by improvements of the large intron splice sites, which do not appear in the final spliced R2 mRNA (Fig. 3), further suggests that the effects of the PTCs are independent of effects on RNA stability, as has been shown previously (4,5). Taken together, these results suggested that PTCs in either exon of R2 can interfere with the nuclear excision of the upstream large intron from P4-generated pre-mRNAs.

The Effect of the PTC at nt 385, but Not That of the PTC at 2018, Could Be Suppressed by Mutations in the Initiating
AUG-The accumulated level of R2 relative to R1 was affected by a PTC but not by a missense transversion at nt 385, and we have previously shown that the effect of a PTC at nt 2018 was reading frame-dependent, i.e. it was suppressed by a frameshift mutation in front of nt 2018 (5). Therefore, we chose to ask whether the effects of the PTCs were linked to translation in the cytoplasm by determining whether their effect was dependent on the initiating AUG that begins the reading frame for NS1 and NS2. Although such mutations may conceivably have other effects on RNA processing, similar mutations have been used to suppress the effects of PTCs in the triosephoshate isomerase gene RNA in order to link the PTC effects to cytoplasmic translation (19). As can be seen in Fig. 4, destruction by point mutagenesis of the initiating AUG at nt 260 suppressed the effect of the PTC at nt 385 but not the effect of the represent forms of R1, R2, and R3, respectively, that are unspliced in the small intron region and are smaller than the corresponding bands generated by the wild type and 2018 mutant because the wild type protection probe B used for these experiments was not entirely homologous in the small intron region of these mutants.
PTC at nt 2018. Translation reinitiation has also been shown to abrogate nonsense-mediated decay of the triosephoshate isomerase gene RNA (19). There are also two internal AUG codons in frame in the NS1/NS2 shared exon (at nt 474 and 504), and the effect of the 2018 PTC was also resistant to mutation of all three AUGs. Interestingly, the triple AUG mutations were less effective than the nt 260 mutation alone in suppressing the effect of the PTC at nt 385 (Fig. 4). This may have been because nt 474 and 504 lie close to the large intron donor at nt 514 and because mutations at these sites may have interfered with an exonic signal required for splicing of the large intron. That there was a reduction in the accumulated levels of R2 relative to R1 in RNA generated by a mutant in which the three AUGs were destroyed but no PTC had been introduced (pAUG(XXX)) is consistent with this possibility (Fig. 4). Thus, nonsense codon-mediated reduction of the accumulated levels of R2 relative to R1 caused by the PTC at nt 385 but not at nt 2018 was dependent upon the reading-frame initiating AUG at nt 260.
The Effect of the PTC at nt 2018 Is Not Dependent upon the Normal Stop Codon in R2 but Rather upon Communication between the Two Exons-The effect of the PTC at nt 2018 has previously been shown to be reading frame-dependent; a single nucleotide mutation at nt 2011 that shifted the R2 reading frame into the open NS1 reading frame prior to the 2018 PTC could suppress its effect (5). That the 2018 PTC could not be suppressed by destruction of the initiating AUG codon suggested that the R2 reading frame must be marked from another point. As shown in Fig. 5 (A and B), this marker is not the normal termination codon for the major isoform of NS2. (The major isoform of NS2 is encoded by an R2 molecule that uses the small intron pair D1/A1 and is found in approximately 75% of the MVM mRNAs.) When the NS2 major isoform TAA codon was mutated so that the NS2 open reading frame was fused in-frame into the capsid gene VP1 open reading frame that terminated at nt 4552, the effect of the 2018 PTC on the accumulated levels of R2 relative to R1 remained unchanged (p2018 TAA STOP(Ϫ)). The 2018 PTC was, however, suppressed by a single nucleotide deletion at nt 505 in the upstream NS1/2 common exon which shifts the PTC into the NS1 (ORF) in the final R2 spliced product (Fig. 5, A and B). This result suggested that there was communication of reading frame between the two exons before final splicing of the intervening intron.
Studies with the T-cell receptor gene have suggested that a spliceable downstream intron was required for some effects of PTCs on RNA processing (20). This was not the case, however, for the PTC at MVM nt 2018. Mutations that disabled both the outside small intron donor and the acceptor (D1 and A2; Fig. 1) prevent excision of the small intron (probably because the remaining internal D2-A2 pair that is separated by only 59 nt is too close to be spliced together (14,21)); however, the upstream intron is still excised at wild type efficiency in this mutant (Fig. 5, A and C). When a PTC was introduced at nt 2018 in this small intron-disabling mutant, the full effect of the PTC on the accumulated levels of R2 relative to R1 was still evident (p2018 TAA D1(Ϫ)A2(Ϫ) ; Fig. 5, A and C). Thus the 2018 PTC did not require a prior splicing event downstream for its effect. DISCUSSION A PTC transversion mutation but not a missense transversion mutation at nucleotide 385 in the 5Ј-terminal NS1/2 common exon of P4-generated pre-mRNAs of the autonomous parvovirus MVM resulted in an increased level of the singly spliced R1 mRNA relative to doubly spliced R2 mRNA in the nucleus of transfected cells. This effect could be suppressed by improvement of the splice signals of the large intron, suggest-ing that the effect on MVM RNA processing was at the level of excision of the large intron. Mutation of the initiating AUG at nt 260 also suppressed the effect of the 385 PTC, suggesting a link to the establishment of the reading frame of this 5Ј-terminal exon. These results help place the MVM P4-generated pre-mRNAs in a growing list of examples in which PTCs can either directly or indirectly affect nuclear RNA processing in an open reading frame-dependent manner (1,22).
We have previously shown that a PTC at nt 2018 in the downstream internal NS2-specific exon of MVM P4-generated pre-mRNA also inhibited excision of this large intron, likely by virtue of interfering with the function of an ESE within the NS2 specific exon (5). In that case, the effect of the PTC on splicing was shown to be dependent upon an intact open reading frame. Here we show further that the effect of the PTC at nt 2018, although reading frame-dependent, was not affected by alterations of either the protein translation initiating AUG or terminating TAA signals, which have been shown to be important for PTC effects in other systems (1,19,20,22). Surprisingly, the effect could be suppressed by a frameshift mutation in the 5Ј terminal NS1/2 common exon of R2 message, suggesting that the effect of the 2018 PTC was dependent upon communication of reading frame between the two exons in the nucleus, prior to the completion of the splicing process.
PTCs have been shown to have significant effects on RNA processing associated with the nucleus of mammalian cells (1,20). In some cases PTCs have been shown to trigger nucleusassociated nonsense-mediated decay; in other cases PTCs have been shown to affect other nuclear processes that lead to altered steady state RNA levels (1). We have recently shown that the PTC at nt 2018 likely affects splicing of MVM P4-generated pre-mRNA by interfering with the nuclear function of an ESE within the internal NS2-specific exon of MVM RNA (5). These observations are most consistent with "nuclear scanning"-type models of reading frame recognition (1,23). Because our results suggest that recognition of the PTC at nt 2018 is not influenced by the initiating AUG and yet can be suppressed by a frameshift mutation in an upstream exon, such a model would have to accommodate recognition of frame between the upstream NS1/2 common exon and the NS2-specific exon prior to splicing. On the other hand, because the PTC at nt 385 is suppressible by mutation of the initiating AUG, it is formally possible that the effect of this PTC could be linked to cytoplasmic translation. This seems unlikely, however, because the large intron begins only 129 nt downstream of the PTC at nt 385. The observation that the effects of the two PTC mutations together are additive suggests that they may function somewhat differently.
Although the mechanisms of PTC action in the nucleus are not well understood, their effect has provoked a growing appreciation that the recognition of reading frame in the nucleus of mammalian cells can influence RNA processing events (1,3). How such recognition can occur is not known; however, there have recently been some interesting observations that suggest that an appropriate apparatus may be in place. Ribosomal proteins and RNAs, elongation factor subunits eIF2A (20) and eIF4E (24), and aminoacyl tRNAs have recently been detected in the nucleus of mammalian cells (25). The U5 snRNP, which binds to exon sequences adjacent to 5Ј and 3Ј splice sites and has a role in juxtaposing exons together during splicing (26,27), contains a 116-kDa protein that is both essential for splicing and closely related to the ribosomal translocase EF-2 (28). In addition, a 200 S ribonuclear protein particle that contains pre-mRNA and splicing factors but is much larger than the 60 S splicesome identified in extracts competent for splicing in vitro has been identified in mammalian cells (29). Our previous observations (4,5) and those in this manuscript taken together provide strong evidence for the existence of such an exonscanning reading frame recognition mechanism in the nucleus of mammalian cells that can interfere with the processing of PTC-containing RNAs.