INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Sendai virus (SeV), a prototype paramyxovirus, contains a nonsegmented negative-strand RNA genome. This genome RNA of 15,384 nucleotides (nt) is found in a helical nucleocapsid core assembled with a predicted 2,564 copies of the N protein (i.e., if there is but one N subunit per 6 nt, the rule of six [2a]), to which ca. 300 copies of the P (phosphoprotein) and ca. 50 copies of the L (large) protein are bound. Paramyxovirus genomes serve as templates for both monocistronic mRNAs and a full-length complementary antigenome strand, the intermediate in genome replication. The nucleocapsid core can carry out mRNA synthesis in vitro, and when it is provided with unassembled N protein, genome replication also takes place. The synthesis of both genomes and antigenomes is coupled to their assembly, and they are consequently found only as nucleocapsids assembled with N protein (reviewed in reference 22).

Six mRNAs (in the order N, P, M, F, HN, and L) are transcribed from the N RNA genome by the P-L polymerase. All of these viral mRNAs except the P-gene mRNA express a single primary translation product from a single open reading frame (ORF). The paramyxovirus P gene mRNAs, in contrast, generally contain alternate ORFs that overlap the N terminus as well as the middle region of the P-protein ORF, and they express several proteins. For SeV, the C-protein ORF overlaps the N-terminal region of the P ORF and is accessed via ribosomal choice during translational initiation (8) (Fig. 1a). The highly conserved, cysteine-rich V ORF which overlaps the middle of the P ORF, on the other hand, is accessed by a mechanism involving transcriptional choice, i.e., cotranscriptional mRNA editing (3, 29-31).

View larger version (39K): [in this window] [in a new window]   FIG. 1.   (a) Schematic representation of the paramyxovirus P-gene mRNAs. The mRNAs are indicated as horizontal lines, and the ORFs are represented as boxes. For each virus group, the upper line shows the mRNA which is an exact copy of the gene, and the beginning of the ORF box indicates the ribosomal start codon. When more than one ORF box is attached to the line, the ORFs are accessed by alternate initiation codons. The boxes below indicate alternate downstream ORFs which are fused into place by G insertions in the mRNAs. The positions of the insertions are shown by the dotted vertical lines. The three possible ORFs are indicated by differences in shading. (b) Comparison of paramyxovirus editing sites. The sequences are written as positive-strand RNA, 5' to 3', and are grouped into the three genera of the Paramyxovirinae. Spaces have been introduced to emphasize the different elements of the sequence. The short G run which is expanded on mRNA editing is shown on the right, together with the pattern of G insertions which occurs for each group (dotted brackets). Note that the A run which precedes the G run is the only part of this cis-acting sequence which is strictly conserved according to genera. Also note that the second A residue upstream of the rubulavirus G run is replaced by a G (highlighted with rectangle), which presumably accounts for why rubulaviruses insert a minimum of two G residues when stuttering begins (19, 31). The shaded boxes indicate sequence conservations. When the highlighted AC is changed to UU as in the other respiroviruses, SeV edits its mRNA with multiple G insertions, like PIV3.

The subfamily Paramyxovirinae is organized in three genera: respiroviruses (formerly the Paramyxovirus genus), including SeV and bovine parainfluenza virus type 3 (bPIV3); morbilliviruses (e.g., measles and distemper viruses); and rubulaviruses (e.g., mumps virus and simian virus 5). Most of these viral P genes contain an A_nG_n purine run at the start of the internal, overlapping V ORF (Fig. 1b). mRNAs with expanded G runs are transcribed from these genes in addition to those which are faithful copies of their templates, and the number of G insertions which occur for each virus group mirrors their requirements to switch between the in-frame and out-of-frame ORFs (reviewed in reference 19). For the morbilliviruses and SeV, which require a +1 frameshift to access the V ORF from the genome-encoded P ORF, a single G is added as the predominant insertional event (Fig. 1a). For the rubulaviruses, which require a +2 frameshift to access the remainder of the P ORF from the genome-encoded V ORF, two G's are added at high frequency when insertions occur. For bPIV3, where both V and another ORF (called D) overlap the middle of the genome-encoded P ORF, one to six G's are added at roughly equal frequency, so that mRNAs encoding all three overlapping ORFs are expressed. Although the available evidence suggests that these Gs are added cotranscriptionally, the term "mRNA editing" has nevertheless been retained to describe these events. RNA editing is defined here as a process in which nucleotide insertion, deletion, or base substitution produces an RNA whose sequence (and informational capacity) differs from that of its template, other than by splicing and by 5' and 3' end formation (1, 4).

Paramyxovirus mRNAs are made in the cytoplasm, and these viruses consequently must fend for themselves in all aspects of mRNA synthesis. All negative-strand virus RNA polymerases (RNAPs) which polyadenylate their mRNAs are thought to do so by stuttering on a short run of template U residues (4 to 7 nt long), and it was this observation that first suggested that the G insertions would similarly occur by pseudo-templated transcription (3, 17, 29). Paramyxovirus mRNA editing is thought to take place as follows (see Fig. 7a): (i) the viral polymerase is postulated to pause before the end of the template C run (nt 1051 to 1053); (ii) the nascent chain, whose 3' end is base paired to the template, slips backward by one (SeV and morbilliviruses) or two (rubulaviruses) template positions (Fig. 1); and (iii) as a result, one or two of the template C residues are copied a second time when transcription resumes processively. In the realignment of nascent mRNA and template, U:G (but not A:C) pairs are permitted, and in analogy to ribosomal frameshifting, the region where alternate base pairing occurs after realignment is called the slippery sequence (2, 16, 33). For most viruses, this cycle of slippage and pseudo-templated synthesis occurs only once, but for bPIV3 it is postulated to occur repeatedly, generating a range of multiple G insertions. The presence of a counting mechanism has therefore been invoked to explain these different patterns of G insertions (26).

The replacement of the SeV editing region with that of bPIV3 in a SeV minigenome leads to mRNAs with G insertions whose distribution resembles those found in bPIV3 infections (18). The counting mechanism is thus apparently controlled in large part by a cis-acting sequence. Here we report experiments which show that these controlling sequences lie immediately upstream of the slippery A₆G₃ purine run and propose a model to account for the counting mechanism.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of minigenomes. Construction of the internal deletion SeV minigenome (SH22) is described in reference 15. Briefly, pSH22 is based on pSP65, with the minigenome inserted directly downstream of the T7 promoter and carrying the hepatitis delta virus genomic ribozyme directly after the minigenome. Each minigenome contains 423 nt of the 5' trailer/L-gene region of SeV, a 54-nt polylinker region into which the 103-nt SeV editing cassette has been inserted between EcoRI and XbaI sites, and 146 nt of the 3' end of the SeV negative-strand genome including the leader/N-gene region.

Minigenome expression and direct limited primer extension of the mini-mRNAs. The various minigenomes and the SeV N, P, and L genes were expressed in A549 cells basically as described elsewhere (6, 18). The cells were grown as monolayers in 9-cm-diameter dishes and infected at 2 to 5 PFU/cell with a vaccinia virus expressing T7 RNA polymerase (vTF7-3 [11]). At 1 h postinfection (hpi), the medium was replaced with a transfection mix composed of 20 µl of home-made transfectACE, pGEM-L (1 µg), pGEM-hPIV1-N (2.5 µg), pGEM-^HA P30 (2.5 µg), pGEM-mini-genome (5 µg), and minimal essential medium to 1 ml. After 2 h at 33°C, an extra 6 ml of minimal essential medium was added and the cells were incubated for 40 h before harvesting. After removal of medium, the cells were solubilized and scrapped into 150 mM NaCl-50 mM Tris (pH 7.4)-10 mM EDTA-0.6% Nonidet P-40. Nuclei were removed by pelleting at 12,000 × g for 5 min. To recover mRNA and viral nucleocapsids (6), the cytoplasmic extracts were centrifuged in a step gradient composed of a 5.7 M CsCl cushion, a 40% CsCl solution, and a 20% CsCl solution at 35,000 rpm (overnight, in an SW55 rotor). RNAs from either viral nucleocapsids (found at the 40%-20% interface) or mRNA pellets were analyzed by limited primer extension using Moloney murine leukemia virus reverse transcriptase (RT; Gibco-BRL) and the ³²P-labeled SeV-edit primer (5' GATGTGTTCTCTCCTATG) in a reaction mixture containing ddATP to terminate the extension upstream of the purine run as described in reference 26. The products were separated on a 10% sequencing gel.

Generation of recombinant SeV constructs (rSeV) containing mutations upstream of or at the editing site. The recovery system was as described previously (12, 13). Briefly, one 9-cm-diameter petri dish of A549 cells was infected with 2 to 3 PFU of vaccinia virus TF7-3 per cell and transfected 1 h later with 1.5 µg of pGEM-L, 5 µg of pGEM-N, 5 µg of pGEM-^HAP (which does not express the C proteins), 15 µg of FL3-1, and 5 µg of one of the pN/P^Xho/M shuttle vectors. All mutations were introduced simultaneously in the N/P^Xho/M shuttle vectors as well as the pGEM-^HAP to avoid loss of the mutation by recombination. After 24 h, cytosine arabinoside (100 µg/ml) was added to inhibit vaccinia virus replication; 24 h after that, the cells were scraped into their medium and directly injected into the allantoic cavities of two to three 10-day-old embryonated chicken eggs. Three days later, the allantoic fluids were harvested and reinjected into eggs undiluted. For further passages, the viruses were diluted 1/500 before injection.

Analysis of the mRNAs by direct limited primer extension or on the RT-PCR product. To examine the length of the purine run, mRNA pelleted through a 5.7 M CsCl cushion was used directly for limited primer extension with primer SeV-edit as described (26). Alternatively, an RT reaction was carried out with the oligonucleotide PEcoP, and a 1/10 aliquot was used for PCR with 50 pmol of each primer (PEcoP and PEag [5' CCAGCCAACGGCCGCCC]) in 10 mM Tris (pH 8.3)-50 mM KCl-2 mM MgCl₂-200 µM deoxynucleoside triphosphate (dNTP). PCR was carried out in 50 µl with 1 U Taq polymerase in a GeneAmp PCR system 9600 as follows: denaturation 94° for 20 s, elongation at 72°C for 30 s, and annealing at 45° for 20 s, for 18 cycles.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Minigenome mRNA editing in transfected cells. (i) The minimum G run is three. We previously used a synthetic minigenome in a cell transfection assay to examine the role of cis-acting sequences on the editing phenotype (18). The viral functions required to replicate and transcribe the minigenome (N, P, and L) are expressed from T7-promoted pGEM plasmids in this system, and the T7 RNA polymerase is provided by coinfection with a recombinant vaccinia virus (vTF7 [Fig. 2a]). The minigenome T7 transcript, a negative strand with exact viral ends, is first assembled with viral N protein, and these negative-strand nucleocapsids are transcribed and replicated by the SeV N, P, and L proteins (Fig. 2a). mRNA (CsCl pellet RNA) is then prepared from these cells, and the presence of G insertions is determined by limited primer extension. As mentioned above, when the minigenomes were engineered to contain the mRNA editing region of either bPIV3 or SeV, the resulting mRNA was found to be edited accordingly (18).

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Minigenome mRNA editing in transfected cells. (a) SeV minigenome replication and transcription was reconstituted in vaccinia virus vTF7-infected cells via the transfection of pGEM plasmids which express the PIV1 N protein (sphere), the P protein with a 10-amino-acid deletion including the editing site (square), and the L protein (oval), as well as the minigenome (see text). T7 refers to T7 RNA polymerase expressed from vTF7-3. The T7 minigenome transcript is assembled with N protein and is replicated and transcribed by the SeV and hPIV1 proteins. The ensuing mRNAs (bottom line) are examined for G insertions by limited primer extension. (b) Limited primer extension analysis of mRNAs and antigenomes from cells transfected with minigenomes containing A6G1-5 editing sites (indicated above the lanes). Cytoplasmic extracts of the transfected cells were prepared at 24 hpi, and their SeV RNAs were separated on CsCl density gradients into fractions containing the mRNAs (pellet) and genomes/antigenomes (banded material) (Materials and Methods). A 5'-32P-labeled primer was then extended across the editing site with RT in the presence of ddATP to limit the extension (schematized above). The intense lower band in each mRNA lane indicates the uninserted mRNAs. The fraction of one-G-insertion mRNAs was determined by densitometry and is shown below the mRNA panel. The results of a parallel transfection of the A6G4 construct in which pGEM-L was withheld is shown on the left, and the absence of bands here indicates that the other signals observed are dependent on the SeV polymerase. An editing-inactive construct, A4GAG3, was included as another negative control.

(ii) The 2 nt upstream of the purine run are a key determinant of the editing phenotype. When 18 nt upstream of the SeV A₆G₃ run were replaced with the corresponding sequence of bPIV3 and the G run was increased from three to four (as in bPIV3), a significant fraction of the resulting mRNAs contained multiple G insertions, i.e., displayed a PIV3 phenotype (18). We therefore constructed minigenomes Swap2, -5, and -8, in which 2, 5, and 8 nt upstream of the A₆G₃ run were replaced with the corresponding sequence of bPIV3, to more precisely define the cis-acting sequence responsible for this altered phenotype (Fig. 3). When as few as 2 nt of bPIV3 were placed upstream of the purine run, the minigenome mRNA had a G insertion pattern reminiscent of that found in natural bPIV3 infections. Bands corresponding to two to five G insertions were now detected, and the overall fraction of edited mRNA rose from 15 to >40% (Fig. 3 and 4b). The altered editing phenotype was not influenced by the number of G's in the purine run, as Swap8 constructed in the A₆G₄ background behaved similarly to Swap8 in the A₆G₃ background (Fig. 3) as did Swap5 and Swap2 (data not shown). To control that all minigenomes were of hexamer length and that the G insertions had occurred during mRNA synthesis, the CsCl-banded antigenomes were also directly examined. Only a single strong band at the nonedited position was detected (Fig. 3). Thus, the altered insertion pattern of the Swap constructs was due not to the extra G in the polypurine run (5' A₆G₄ versus A₆G₃) but rather to the differences in the upstream sequence.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   Minigenome mRNA editing in cells transfected with Swap2, -5, and -8. mRNA editing patterns of minigenome constructs in which 2, 5, and 8 nt of the SeV sequence were replaced with that of bPIV3 (Swap2, -5, and -8, highlighted by gray boxes at the top) were examined as described for Fig. 2. The Swap series was also examined in the A6G4 background, and only the result for Swap8 is shown in the last lane.

View larger version (37K): [in this window] [in a new window]   FIG. 4.   mRNA editing in cells infected with rSeV-Swap2, -5, and -8. Parallel A549 cell cultures were infected with 10 PFU/cell of rSeV-Swap2, -5, or -8 per cell. Other cultures were infected with either bPIV3, SeV, or an rSeV, for reference, or mock infected. CsCl pellet RNA from the various cultures was prepared at 24 hpi, and additions to the purine run were examined by limited primer extension (a). The positions of bands representing the uninserted (0) and inserted (+1, etc.) mRNAs are indicated on the sides. As negative controls, primers were also extended on the Swap8 and SeV DNAs used to generate the rSeV. The results were quantified by densitometry and are shown as histograms in panel b. The overall fraction of the mRNA which contain insertions (edited) and that with insertions of >1 G are indicated for each panel. The insertion patterns of each minigenome and rSeV were examined in at least three separate transfections or infections. One typical result is shown.

mRNA editing in rSeV infections. The vTF7-infected/plasmid-transfected cell system, although highly artificial, has the advantage that the translational consequences of the mRNA editing do not feed back on the editing process itself, as all viral proteins (N, P, and L) are provided via T7 RNA polymerase. mRNA editing can thus be studied in isolation. However, unlike the case for natural infections, the transfected cells constitutively produce large amounts of viral proteins whose stoichiometric balance is not subject to viral regulation, and this may also affect editing. Moreover, in our modified transfection system, the N^hPIV1 and ^HAP30 genes are used in place of N^SeV and ^HAP, which reduces the editing frequency. To study mRNA editing in natural virus infections, we constructed rSeV containing mutations at the editing site (Materials and Methods). The amino acids of the P sequence altered by the editing mutations (Fig. 5) appear to lie in a noncritical region of the protein (7), as all of these rSeV were prepared as readily as the wt control. A549 cells infected with these viruses all accumulated similar amounts of N and M protein (at 24 hpi) as judged by immunoblotting, and similar levels of genomes and antigenomes as judged by primer extension, as the wt control (data not shown).

View larger version (23K): [in this window] [in a new window]   FIG. 5.   Amino acid substitutions in the mutant virus P proteins. The nucleotide sequence of the P-gene P ORF (positions 1034 to 1054) is shown in groups of three, and their encoded amino acids are indicated below in one-letter code. The nucleotide substitutions of the various mutants are indicated by lowercase letters in bold, and the corresponding amino acid changes are underlined and in bold.

View larger version (27K): [in this window] [in a new window]   FIG. 6.   mRNA editing in rSeV with AC, AU, or UU upstream of A5G3 purine runs. Parallel A549 cell cultures were infected at 10 PFU/cell with the various rSeV whose editing regions are outlined above, and additions to the purine run of their P mRNAs were examined by limited primer extension. The results were quantified by densitometry and are shown as histograms in panel b, along with that of rSeV-Swap2 from Fig. 4. The insertion pattern of each rSeV was examined in at least three separate infections. One typical result is shown.

Alteration of the A run and mRNA editing. Morbilliviruses, which, like SeV, edit their P mRNAs by the insertion of predominantly a single G residue, are a more closely related group of viruses than the respiroviruses. Morbilliviruses also contain a more strongly conserved sequence at the editing site (5' YCC AUU A₅G₃) than the respiroviruses, where only 5' ANY A₆ is conserved (Fig. 1b). As morbilliviruses contain a run of only five adenosines, we were interested in the editing phenotype of a SeV which contained the morbillivirus cis-acting sequence. This was accomplished by converting the sequence 5' A₆G₃C to A₅G₃CC (Fig. 6) so that (i) hexamer genome length is maintained and (ii) the hexamer phase of the first G of the G run moves upstream by one position, which is in fact to that conserved in morbilliviruses (20). The concomitant amino acid substitutions in the P protein are shown in Fig. 5. Again, these viruses were prepared readily and grew to similar levels in eggs as wt virus (data not shown).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

An interesting feature of paramyxovirus mRNA editing is that each particular virus inserts G residues in a pattern matched to the organization of its P gene ORFs (Fig. 1a). Furthermore, the SeV transcription complex can be induced to edit its P mRNA like that of h/bPIV3 (all respiroviruses), i.e., to switch its editing phenotype, by simply substituting the bPIV3 editing region for that of SeV. Our results have highlighted the importance of the sequence immediately upstream of the A_nG_n slippery sequence for this phenotype switch, rather than the number of G's in the G run. A minimum of three G's were found to be required for SeV editing to occur, and increasing the G run to four or five was tolerated and did not affect the editing phenotype. We have not examined the effects of further extending the G run. However, a recombinant measles virus engineered to contain three extra G's at the editing site (5' AUU A₅G₆) was found to insert two to four G's with increased frequency (28).

One unexpected finding of this work is that replacing the SeV slippery sequence (A₆G₃) with that of measles virus (A₅G₃) inactivates the editing process, unlike the analogous h/bPIV3 (A₆G_4-5) swap. It is possible that SeV stuttering is adversely affected by a run of only five adenylates and/or by the displacement of the hexamer phase of the G run with respect to the N-protein subunit upstream by 1 position (to shorten the A run) (20). However, neither of these possible requirements appears to be critical for stuttering, as replacement of the upstream AC with the conserved UU restores the G insertions. Most unexpectedly, significant amounts of mRNAs with multiple G insertions were generated during this restored editing. This unexpected switch to the PIV3 phenotype highlights how little we know of how the cis-acting sequence determines the distribution of G insertions. It also suggests that there may well be differences in how the various viral polymerases interact with the cis-acting sequence and/or that we have as yet identified only one element of the cis-acting sequence, that which has been conserved. Despite our limited information, the decision of the viral polymerase to stutter or not to stutter, and the extent of the stuttering, can be considered in terms of a competitive kinetic model, as this does not depend on detailed structural information of the transcription elongation complex.

A competitive kinetic model for paramyxovirus polymerase stuttering. Cellular RNAPs respond to intrinsic signals in the template DNA and nascent RNA which divert a fraction of the transcription complexes from the path of rapid chain elongation, e.g., to pause or to terminate the chain (23). These processes are among the best-studied examples of transcriptional choice and thus the most useful model for us. By analogy to Escherichia coli RNAP (32), the paramyxovirus elongation complex has two choices at any template position I: it can extend the nascent chain by one nucleotide to form a transcript that is I + 1 residues in length, or it can be induced by features of the template or nascent chain sequence to form a processively unstable complex (Fig. 7). At this point (the first branchpoint; see below), there is a significant probability that the active site together with its nascent chain are repositioned one residue upstream of the template C which has just been copied (shown as C1052 in Fig. 7). This template C is then copied a second time when nucleotide addition recurs, resulting in a pseudo-templated G insertion.

View larger version (35K): [in this window] [in a new window]   FIG. 7.   A stuttering model for paramyxovirus mRNA editing. (a) The putative RNA-RNA hybrid between the polyprimidine tract of the negative-strand genome (top strand, written 3' to 5') and the polypurine run of the nascent mRNA chain (bottom strand, written 5' to 3') when the active site of the transcription complex has just incorporated G1052 (top left). At the editing site (shown as C1052, highlighted with a gray box), the transcription complex has the choice of realigning its nascent RNA chain upstream on the template before the next nucleotide is added. Here, the minimum requirement for stuttering, that the realigned hybrid be nearly as stable as its predecessor, has been met because non-Watson-Crick U:G pairs (highlighted in gray) do not disrupt the helical stack. If the rate constant for pseudo-templated addition of the G residue (kpseudo) is greater than that for realignment, Kstutter will be mostly determined by krealign. Having stuttered once at template C1052, the transcription complex is back to where it started and has the same choice. When the upstream AC is replaced with UU, however, krestutter is increased relative to kstutter (see text). Escape from this process requires the active site to move on to a position where realignment of the nascent chain upstream does not occur. (b) A model for the thermodynamic reaction pathway of the transcription elongation complex at the editing site. Barriers heights are given as G*, the free energy of activation. The black dot represents the transcription elongation complex. Note that the activation energy for extending the chain from I to I + 1 (G°forward) is shown as being similar for strictly templated (reaction coordinate 1052) and pseudo-templated (reaction coordinate 1051) nucleotide addition. When AC is found upstream of the purine run, the ground state of the transcription elongation complex after realignment (reaction coordinate 1051) will be determined mostly by the stability of the hybrid and is therefore higher than that of its predecessor due to the presence of the U:G pair. When UU rather than AC is found at this site, the ground state of the realigned complex is more stable than its predecessor, presumably due to additional interactions between the exit channel of the polymerase and the upstream nucleotides. This leads to a greater fraction of the mRNAs containing pseudo-templated additions and to a greater number of insertions when stuttering takes place.