Identification of a Mutation in Editing of Defective Newcastle Disease Virus Recombinants That Modulates P-Gene mRNA Editing and Restores Virus Replication and Pathogenicity in Chicken Embryos

Abstract

Editing of P-gene mRNA of Newcastle disease virus (NDV) enables the formation of two additional proteins (V and W) by inserting one or two nontemplated G residues at a conserved editing site (5′-AAAAAGGG). The V protein of NDV plays an important role in virus replication and is also a virulence factor presumably due to its ability to counteract the antiviral effects of interferon. A recombinant virus possessing a nucleotide substitution within the A-stretch (5′-AAgAAGGG) produced 20-fold-less V protein and, in consequence, was impaired in replication capacity and completely attenuated in pathogenicity for chicken embryos. However, in a total of seven serial passages, restoration of replication and pathogenic capacity in 9- to 11-day-old chicken embryos was noticed. Determining the sequence around the editing site of the virus at passage 7 revealed a C-to-U mutation at the second nucleotide immediately upstream of the 5′-A₅ stretch (5′-GuUAAgAAGGG). The V mRNA increased from an undetectable level at passage 5 to ca. 1 and 5% at passages 6 and 7, respectively. In addition, similar defects in another mutant possessing a different substitution mutation (5′-AAAcAGGG) were restored in an identical manner within a total of seven serial passages. Introduction of the above C-to-U mutation into the parent virus (5′-GuUAAAAAGGG) altered the frequency of P, V, and W mRNAs from 68, 28, and 4% to 15, 44, and 41%, respectively, demonstrating that the U at this position is a key determinant in modulating P-gene mRNA editing. The results indicate that this second-site mutation is required to compensate for the drop in edited mRNAs and consequently to restore the replication capacity, as well as the pathogenic potential, of editing-defective NDV recombinants.

Newcastle disease virus (NDV) belongs to the new genus Avulavirus within the family Paramyxoviridae. NDV is the causative agent of one of the most devastating diseases of poultry worldwide. The negative-strand RNA virus genome of NDV encodes six genes (in the order 3′-NP-P-M-F-HN-L-5′). Except for the P gene mRNA, all express one major viral structural protein from a single open reading frame (ORF). The P gene mRNAs of the subfamily Paramyxovirinae express several proteins by the use of alternative reading frames and by a process of mRNA editing. The mRNA editing, which involves insertion of pseudotemplated G residue(s) occurs by a polymerase stuttering mechanism (10, 11, 24). Like in other members of paramyxovirinae, the P gene mRNA of NDV is edited by inserting one or two G residues into a G run within a conserved editing locus (5′-AAAAAGGG) (16, 21). As a result, three P gene derived mRNA species are produced which encode for the P, V, and W ORFs (unedited, with +1 and +2 frameshifts, respectively). The V and W proteins are colinear to the N-terminal half of the P protein but have different carboxy-terminal parts.

The majority of the accessory proteins, including V and/or C proteins encoded by the P genes of different members of the subfamily Paramyxovirinae have been shown to be dispensable for virus replication in cell cultures (4, 5, 12, 13, 14, 20, 23). Interestingly, however, mutants lacking V or C accessory proteins were strongly attenuated in pathogenicity in vivo (5, 12, 13). The mechanism of the in vivo attenuation was demonstrated to involve the interferon system in which accessory proteins, particularly V or C proteins, inhibit interferon signaling or suppress interferon production (reviewed in references 7 and 8). Additional roles of paramyxovirus accessory proteins in viral RNA transcription and synthesis, as well as in virus assembly and propagation, were also recently uncovered, demonstrating functional versatility of these proteins (1, 9, 22).

The dual function of the V protein of NDV in virus replication, as well as in virus virulence, was shown by generating mutants lacking the entire V and W proteins or only the unique C-terminal part of the V protein or by constructing a mutant expressing low levels of V protein (16). The C terminus of the V protein, which is highly rich in cysteines, is of particular interest since it is conserved in all three genera of paramyxovirinae, with the exception of human parainfluenza virus types 1 and 3 (6, 15). NDV mutants lacking the entire V and W proteins or only the C-terminal portion were severely impaired in virus pathogenicity and virus propagation both in vitro and in vivo, indicating that the C terminus of V protein is responsible for this incompetence. Very recently, the NDV V protein was demonstrated to have an interferon-antagonist activity, and this activity was shown to be located in the C terminus of the V protein (17). Therefore, the mechanism by which V protein plays a role in virulence is presumably due to its ability to antagonize the antiviral effects of interferon and its exact role in virus replication remains to be established.

As opposed to the mutations that completely abolished the function of NDV V protein, a single A-to-G change within the A-run only reduced the editing frequency and thus lowered the level of additional proteins generated by RNA editing (16). The mutant NDV-P1 expressing low-level V protein grew in 9- to 11-day-old embryonated eggs to a titer of 6.7 log₁₀ 50% embryo infectious dose (EID₅₀)/ml at passage level 5. In passages 6 and 7 the titers of the mutant in embryonated eggs have increased to 7.1 and 9.3, respectively. The latter titer is identical to the titer of the parent virus. This dramatic increase in replication efficiency was also accompanied with increase in pathogenicity for 9- to 11-day-old embryonated specific-pathogen-free (SPF) chicken eggs. Compared to the parent virus, the mutant at passage level 5 was attenuated by as much as 10⁶-fold for young embryos. The degree of attenuation was, however, dramatically decreased in subsequent passages and the mutant was almost as pathogenic as the parent virus at passage level 7 for 9- to 11-day-old embryonated SPF chicken eggs.

Here we describe a compensatory mutation, which is responsible for the restoration of the replication efficiency and pathogenicity in 9- to 11-day-old embryonated SPF chicken eggs. Moreover, introducing the mutation into the backbone of the parent virus provided direct evidence for the role of the second-site mutation in modulating the editing frequency of NDV. A mechanism for how NDV restores RNA editing to ensure the expression of P-gene derived proteins for efficient NDV replication and virulence is discussed.

MATERIALS AND METHODS

Construction of editing-site mutants.

The plasmid pflNDV, expressing the full-length antigenome RNA of the lentogenic ND vaccine strain, Clone-30 (19), was used to introduce mutations. Construction and generation of the mutant NDV-P1, possessing an A-to-G replacement at position 5′-A3 (5′-AAgAAGGG) was described previously (16). Another mutant containing a replacement at position 5′-A4 in the A-stretch of the editing site was constructed by performing PCR with the template pflNDV with forward primer 4 (5′-GCTCCTCGCGGCTCAGACTCG-3′, nucleotides 151 to 171) and reverse primers A4 (5′-CCATGGGCCCTGTTTAGCATTGGACG-3′, nucleotides 2269 to 2294). In order to introduce the identified second-site mutation into the backbone of the parent Clone-30, PCR was performed with the same forward primer 4 and a reverse primer T2 (5′-CCATGGGCCCTTTTTAACATTGGACG-3′). PCR products were then digested with AatII/ApaI and cloned into the same sites of pflNDV. The region newly introduced into each clone was sequenced to rule out PCR-introduced errors. The resultant full-length clone with a single nucleotide substitution at position 5′-A4 and the clone possessing the identified second-site mutation were named NDV-PC4 and NDV-T2, respectively (Fig. 1).

FIG. 1.

Recombinant NDV constructs. A schematic representation of the NDV gene order is shown in the negative-strand genomic RNA. The nucleotide sequences around the editing site (positions 2274 to 2291) are presented in a positive sense (5′ to 3′). The introduced nucleotide modifications (lowercase letters) and amino acid substitutions (one-letter code) in the P proteins of mutant viruses are shown in boldface.

Recovery of recombinant viruses.

Transfections and virus recoveries were done as described preiviously (16). Approximately 1.5 × 10⁶ BSR-T7/5 cells stably expressing phage T7 RNA polymerase (3) were grown overnight in 3.2-cm-diameter dishes. Cells were transfected with plasmid mixtures containing 5 μg of pCite-NP, 2.5 μg of pCite-P, 2.5 μg of pCite-L, and 10 μg of one of the full-length clones by using a mammalian transfection kit (CaPO₄ transfection protocol; Stratagene). Three days after transfection, supernatant was harvested and inoculated into the allantoic cavity of 9- to 11-day-old embryonated SPF chicken eggs. After 3 to 4 days of incubation, the presence of virus in the allantoic fluid was determined by rapid plate hemagglutination (HA) test (2) with chicken erythrocytes. The infectious titer, expressed as the EID₅₀ was calculated by using the method of Reed and Muench (18).

Determining sequences around the P gene mRNA editing site.

Total RNA was isolated from BSR-T7/5 cells infected with serially passaged recombinant viruses by using the RNeasy kit (Qiagen). Reverse transcription by avian myeloblastosis virus reverse transcriptase on 1 μg of total RNA was primed with NDV P gene-specific oligonucleotide P#13 (5′-CCACCCAGGCCACAGACGAAG-3′, nucleotides 2176 to 2196) or oligo(dT) primer to amplify only mRNAs. DNA amplification was then performed with NDV P gene-specific oligonucleotides P#13 and P#17 (5′-ATGAATTCAGCTGTTGGA-3′, nucleotides 2671 to 2688). The PCR products were analyzed on a 1% agarose gel and used directly for sequencing or cloned by using Invitrogen's TOPO TA cloning kit. Sequences of ca. 300 bp flanking the editing site were determined from independent colonies and examined for the presence of any mutation around the conserved editing site.

Serial passaging of recombinant viruses in embryonated eggs.

Recombinant viruses were independently passaged serially in 9- to 11-day-old embryonated SPF chicken eggs. The titer of the progeny obtained after each passage was determined by inoculating 10-fold serial dilutions into the allantoic cavity of embryonated eggs. The eggs were observed daily and NDV-specific mortality was scored during the incubation period of 7 days. The HA test was carried out, and the infectious titer, expressed as the EID₅₀, and the mortality (expressed as 50% embryo-lethal dose [ELD₅₀]) was calculated by using the method of Reed and Muench (18).

RESULTS

Serial passage of NDV-P1 yielded progeny viruses that exhibit enhanced replication and restored pathogenicity.

Recently, we have shown that NDV mutants lacking the complete V and W proteins or only the cysteine-rich carboxy-terminal part of the V protein were severely impaired in cell culture growth and completely unable to propagate in 9- to 11-day-old embryonated eggs. In contrast, a mutant NDV-P1 possessing a single nucleotide substitution in the 5′-A₅G₃ sequence (5′-AAGAAGGG, Fig. 1) grew in eggs, albeit to very low infectious titers. The modification in NDV-P1 led to expression of V protein at a 20-fold-lower level and extreme attenuation for the chicken embryos. The findings indicated that the V protein of NDV plays an important role in virus replication, as well as in virus virulence (16). The infectious titers of NDV-P1 after the fifth and sixth egg passages were 6.7 and 7.1 log₁₀ EID₅₀/ml, respectively. One further passage yielded a titer of 9.3 log₁₀ EID₅₀/ml, a value that is identical to the titer obtained for the parent virus (Fig. 2).

FIG. 2.

Infectious titers of recombinant NDVs. Embryonated chicken eggs were inoculated at a dose of 4 log₁₀ EID₅₀ per egg and incubated for 4 days. Tenfold serial dilutions of the harvested allantoic fluid samples were prepared and inoculated into 9-to 11-day-old embryonated eggs (10 eggs/dilution). After 3 to 4 days of incubation, an HA test was carried out, and the calculated infectious titers in the log₁₀ EID₅₀/ml are shown on the top of the bars.

To determine whether there is also a concomitant increase in virus virulence, 10-fold serial dilutions of passages 5, 6, and 7 of NDV-P1 were inoculated into 11-day-old embryonated SPF chicken eggs and incubated for 1 week. Consistent with the previous observations, NDV-P1 at passage level 5 was not lethal for embryos even with a dose as high as 6 log₁₀ EID₅₀. Passage 6 of NDV-P1 showed some embryo mortality and passage 7 was nearly as virulent as the parent virus (Fig. 3). The ELD₅₀ subsequently dropped from >6.7 log₁₀ EID₅₀ at passage 5 to 4.7 and 1.1 log₁₀ EID₅₀ at passages 6 and 7, respectively. For comparison, the ELD₅₀ of the parent virus is 0.3 log₁₀ EID₅₀. Chicken embryos inoculated with the parent virus, rNDV, already started to die at 3 days postinoculation, at doses higher than 4 log₁₀ EID₅₀/ml (Fig. 3). The phenotypic changes of NDV-P1 between passages 5 and 7 initially suggested reversion of the introduced mutation at the editing site. We therefore sequenced several independent clones from passage 7 to control the presence or absence of the introduced mutation at the 5′-A₅ stretch. In contrast to our expectation, all of the clones maintained the introduced A-G substitution. The stability of the introduced mutation suggested the occurrence of a second-site compensatory mutation.

FIG. 3.

Pathogenicity of serially passaged NDV-P1 in SPF chicken embryos. Embryonated eggs were inoculated with the parent virus at passage 3 (rNDV) or the mutant NDV-P1 at passage 5 (P1-5), passage 6 (P1-6), or passage 7 (P1-7) and then incubated for 7 days or until the embryos had died. NDV-P1, which was completely safe for 11-day-old chicken embryos at passage 5, became nearly as virulent as the parent virus at passage 7.

Localization of a second-site compensatory mutation.

In order to localize the mutation responsible for enhancing the replication capacity and restoring pathogenicity of NDV-P1, we sequenced several independent clones derived from passages 5, 6, and 7 (Table 1). Approximately 300 bp flanking the editing-site sequence were examined for the occurrence of any deviation from the NDV-P1 virus sequence. All clones derived from passage 5 contained no alterations in the sequenced region. For passage 6, 4 of 102 clones contained a C-to-U subtitution (5′-GuUAAgAAGGG), which is the second nucleotide immediately upstream of the 5′-A₅G₃ editing-site sequence. Surprisingly, the proportions of independent clones possessing this mutation at passage 7 dramatically increased to 88% (Table 1), demonstrating the growth advantage of NDV-P1 progeny viruses possessing this compensatory mutation. Sequencing of PCR products derived from genomic RNA showed that the second-site mutation is incorporated during genome replication at position 2278 of NDV genome. This mutation is referred as C-U₂₂₇₈ here. These results indicate that a total of seven serial passages in NDV-P1 enabled the virus to introduce a compensatory mutation that overcame the dramatically decreased replication capacity and pathogenicity in chicken embryos.

TABLE 1.

Occurrence of a second-site compensatory C-U₂₂₇₈ mutation in editing-site mutants at various passages

Mutant virus	Passagea	Total no. of coloniesb	No. of colonies positive for:			% C-U mutation
			P	V	W
rNDV	3	116	79	32	5	0
NDV-P1	5	56	56	0	0	0
NDV-P1	6	102	101	1	0	4
NDV-P1	7	96	90	5	1	88
NDV-PC4	3	29	29	0	0	0
NDV-PC4	7	28	27	1	0	89
NDV-T2	2	27	4	12	11	100c

^aNumbers of serial passages in 9- to 11-day-old embryonated SPF chicken eggs.

^bTotal number of independent colonies sequenced to determine the occurrence of a mutation.

^cThe C-U₂₂₇₈ mutation was artificially introduced.

The second-site compensatory mutation C-U₂₂₇₈ partially restores P-gene mRNA editing frequency.

In earlier studies we have demonstrated the presence of a link between the level of V-protein expression and the efficiency of virus replication, as well as virus pathogenicity in chicken embryos (16). To determine whether the exponential increase in the proportion of clones possessing the second-site compensatory mutation during serial passages of NDV-P1 is accompanied by increased editing frequency, the proportions of P, V, and W ORFs were determined from the above-sequenced plasmid clones. As shown in Fig. 4 and Table 1, the amount of edited mRNAs increased from an undetectable level at passage 5 to 1 and 6% at passages 6 and 7, respectively. The total amount of edited mRNAs of 6% at passage 7 is more than fivefold-lower than that of the parent virus. However, both V and W mRNAs were detected in passage 7 at a proportion comparable to that of the parent virus, demonstrating restoration of P-gene mRNA editing, albeit at a low level.

FIG. 4.

Pattern of P-gene mRNA editing in cells infected with serially passaged NDV-P1. The sequences around the editing site from independent clones (Table 1) derived from rNDV or NDV-P1 at passage 5 (P1-5), passage 6 (P1-6), or passage 7 (P1-7) were determined. The percentages of unedited mRNAs representing P ORF (P) or edited mRNAs with insertion of one G residue coding for V ORF (V) or with insertions of two G residues coding for W ORF (W) are shown on the top of the bars.

Similar defects in another editing-site mutant were also restored by the C-U₂₂₇₈ compensatory mutation.

To determine the involvement of the second-site C-U₂₂₇₈ mutation in restoring defects caused by a different substitution mutation, we constructed another mutant with a single base substitution at position 5′-A4 (5′-AAAcAGGG, Fig. 1). The mutant NDV-PC4 was recovered as previously described (16) and serially passaged in 9- to 11-day-old embryonated SPF chicken eggs for a total of seven passages. The infectious titers and embryo mortalities at passages 3 and 7 were then determined. As shown in Fig. 5A, the mutation in NDV-PC4 suppressed the efficiency of replication by a 100-fold at passage 3, which was restored to parent virus level within four further passages. Consistent with the observation made for NDV-P1, NDV-PC4 at passage 3 was completely safe for 9- to 11-day-old embryonated eggs even at a dose as high as 6 log₁₀ EID₅₀ (Fig. 5B). This phenotype was dramatically altered within the subsequent four passages, and the mutant is nearly as pathogenic as the parent virus at passage 7. The ELD₅₀ dropped from >6 log₁₀ EID₅₀ at passage 3 to 0.2 log₁₀ EID₅₀ at passage 7.

FIG. 5.

(A) Infectious titers of recombinant NDV-PC4 at passages 3 and 7. Titers were determined as described under the legend for Fig. 2. (B) Pathogenicity of NDV-PC4 at passage 3 (PC4-3) and passage 7 (PC4-7) in SPF chicken embryos. Mortality was determined as described in the legend for Fig. 3. (C) Editing frequency of NDV-PC4 at passages 3 (PC4-3) and passage 7 (PC4-7). mRNA editing was determined from independent clones (Table 1) as described under the legend for Fig. 4.

To determine whether the restoration in replication capacity and pathogenic potential of NDV-PC4 mutant was also accompanied by the second-site C-U₂₂₇₈ mutation and increased editing frequency, the region around the editing site was amplified and cloned. The sequence around the editing site was then determined from independent clones derived from passages 3 and 7. The results show that the introduced point mutation in the 5′-A₅ stretch has been retained. As expected, all sequenced clones derived from passage 3 do not contain the C-U₂₂₇₈ mutation. In contrast, 89% of the clones derived from passage 7 possess the second-site compensatory mutation (Table 1). The mutation in NDV-PC4 suppressed the editing frequency to undetectable level at passage 3, and the patterns of editing at passage 7 were 96, 4, and 0% for P, V, and W mRNAs, respectively (Fig. 5C). The absence of W mRNA in passage 7 of NDV-PC4 may be explained by the relatively small number of total colonies analyzed compared to that of NDV-P1 (Table 1). Taken together, the results demonstrate that seven serial passages in embryonated chicken eggs enabled editing-defective NDV mutants to introduce the compensatory mutation that partially overcame the decrease in editing frequency and restored virus replication and virulence.

The second-site compensatory mutation C-U₂₂₇₈ modulates P-gene mRNA editing.

Although the C-U₂₂₇₈ mutation clearly plays an important role in partially restoring the editing frequency of an editing-defective mutant, it was of interest to determine the influence of this mutation in the backbone of the parent virus. We therefore introduced this mutation upstream of the 5′-A₅G₃ sequence of the parent virus (5′-GuUAAAAAGGG). The modified full-length cDNA clone, NDV-T2, together with three support plasmids expressing NDV NP, P, and L proteins, was transfected into BSR-T7/5 cells, and virus recovery was done as described in Materials and Methods. HA was detected in eggs already after the first passage. After one further passage, a titer of 9.4 log₁₀ EID₅₀/ml was obtained, which is very similar to that of the parent virus (9.3 log₁₀ EID₅₀/ml). In addition, the ELD₅₀ of the recombinant NDV-T2 was 0.2 log₁₀ EID₅₀, a value again comparable to that of the parent virus (0.3 log₁₀ EID₅₀). We then determined the editing frequency by sequencing independent clones derived from the second passage. Surprisingly, this alteration gave rise to a virus that produces only 15% unedited P mRNA, which is 4.5-fold less than that produced by the parent virus. The rest 85% of the mRNAs were edited, constituting 44% V and 41% W mRNAs (Fig. 6 and Table 1). In the parent virus backbone, the C-U₂₂₇₈ mutation predominantly affected W mRNA by increasing the insertion of two G's by 10-fold, whereas V mRNA increased by a moderate 1.5-fold. Therefore, this mutation is a key determinant in modulating NDV P-gene mRNA editing by influencing the overall frequency and pattern of mRNA editing.

FIG. 6.

Comparison of putative mRNA-template hybrids of rNDV and NDV-T2 during the editing process. The template negative-strand genomes (top strands) were written 3′ to 5′, and the mRNA chains (bottom strands) were written 5′to 3′. The editing frequency of both viruses for P, V, and W mRNAs are shown in percentage. The only difference between the two viruses at position 2278 (C-U₂₂₇₈) was shown in boldface lowercase. The unique putative base pairings are underlined. As much as 85% of P-gene derived mRNAs in NDV-T2 are edited, whereas only 32% of the mRNAs in the parent virus represent V and W mRNAs.

DISCUSSION

The V protein of NDV plays a crucial role both in virus replication and pathogenicity. Earlier studies demonstrated that NDV mutants lacking both V and W proteins or lacking only the unique cysteine-rich C-terminal region of the V protein produced 5,000-fold-less infectious progeny in cell culture than that produced by the parent virus. Surprisingly, the mutants were completely unable to propagate in 9- to 11-day-old embryonated chicken eggs (16). These findings demonstrated that the V protein of NDV is indispensable for virus replication, although P-gene-derived accessory proteins of Paramyxoviridae were categorized as nonessential gene products (4, 5, 12, 13, 23, 20). Another mutant, NDV-P1, possessing a single-nucleotide substitution in the 5′-A₅ run was, in contrast, viable in embryonated eggs. This mutation resulted in downregulation of V protein expression instead of complete abrogation (16). NDV-P1 was dramatically attenuated for chicken embryos with an ELD₅₀ of >6.7 log₁₀ EID₅₀ compared to 0.3 log₁₀ EID₅₀ for the parent virus. The involvement of V protein in virulence is presumably associated with its ability to counteract the antiviral effects of interferon, which is encoded by the carboxy terminus of V (17). The studies described in this report show how editing-defective NDVs restore P-gene mRNA editing and hence secure expression of V and presumably W protein to overcome the selective pressure placed on them for better replication in embryonated eggs.

The mutants NDV-P1 and NDV-PC4 possess single-nucleotide substitutions at the 5′-A₃ and -A₄ positions, respectively, in the 5′-A₅G₃ editing-site sequence. The effects of the mutations in NDV-P1 and NDV-PC4 at low passage levels were very dramatic resulting in an at least 100-fold titer decrease compared to that of the parent virus. In addition, both mutants were completely nonpathogenic for 9- to 11-day-old chicken embryos even at high doses. All of the sequenced independent clones derived from low passages of both mutants represented unedited P mRNA, demonstrating that the edited V and W mRNAs are below the detection limit. Within a total of seven passages, the mutants showed restored efficiency of replication and increased pathogenicity for chicken embryos. The proportion of edited mRNAs augmented from an undetectable level (0%) at low passages to 6 and 4% for NDV-P1 and NDV-PC4, respectively, at passage 7. In NDV-P1 passage 7, although not in NDV-PC4, both V and W mRNAs were detected, demonstrating restoration of P-gene mRNA editing, albeit at a low level. The possibility that W protein may play a role in restoring virus replication and virulence could not be ruled out from these experiments. However, in previous experiments, recombinants either lacking V and W proteins or lacking only the C terminus of V protein exhibited similar defect in replication and virulence (16). This finding strongly suggested that, V rather than W is responsible for restoring virus virulence and presumably virus replication.

The striking result of the studies described in this report is the identification of a second-site, compensatory mutation capable of modulating mRNA editing. The proportion of the progeny virions of NDV-P1 and NDV-PC4 that incorporated the C-U₂₂₇₈ mutation increased from 0% at low passages to 86 and 89%, respectively, at passage 7. The occurrence of an identical second-site mutation in both recombinants strongly indicates that the C-U₂₂₇₈ substitution is essential to compensate for the loss of function caused by disruption of the A-stretch. Introduction of the C-U₂₂₇₈ mutation into the backbone of the parent virus provided a direct evidence for the key role of this second-site mutation in modulating the editing process. The mutant NDV-T2 edited 85% of its P-gene derived mRNAs as opposed to only 32% for the parent virus. This mutation not only raised the total amounts of edited mRNAs but also considerably altered the editing pattern of P, V, and W mRNAs from 68, 28, and 4% to 15, 44, and 41%, respectively. Although P proteins of negative-strand RNA viruses are generally known to be required for efficient virus replication, it is surprising that NDV-T2 is able to replicate to the level of the parent virus despite a 4.5-fold-lower level of P mRNA. The fact that P, V, and W proteins differ only at their C terminus, as well as the finding showing the involvement of V protein in virus replication, suggest that V and/or W proteins might at least partially compensate for the functions of P protein. To provide further insight for the requirement of P protein in NDV replication, we designed a full-length cDNA that should produce only edited mRNAs but not P mRNA. Despite repeated recovery attempts, we failed to rescue an infectious virus (data not shown), suggesting that P protein is not entirely dispensable for NDV replication or completely replaceable by V and W proteins.

Similar to NDV, ca. 30% of Sendai virus (SeV) mRNAs are edited by mostly inserting a single G residue. The effects of introducing alterations into the sequence around the SeV P-gene mRNA editing site (5′-AACA₆G₃) have been extensively studied (10, 11). In a recombinant SeV in which the 5′-A-run was shortened from six to five nucleotides, no mRNA editing could be detected. However, substitutions of the upstream nucleotides of the 5′-AACA₅G₃ sequence from AAC to AAU or AUU restored editing activity to 24 or 44%, respectively. In the latter case, significant fractions of the mRNAs contained two to six G insertions. Hausmann et al. (11) also provided experimental evidence that the insertions of the nontemplated G residues occur by a stuttering process. Furthermore, approximately six nucleotides upstream of the editing site, particularly the positions more proximal to the 5′-A₆G₃ sequence, were shown to determine the number of times the polymerase stutters before normal transcription resumes (10). Our data are thus consistent with these findings, showing that the nucleotides immediately upstream of the editing site are key determinants in modulating the frequency and pattern of insertion of pseudotemplated G residues. Moreover, based on sequence comparison around the editing site of different paramyxoviruses, the NDV and SeV are the only known paramyxoviruses that do not have a U residue at the second nucleotide upstream of the 5′-A₅G₃ or 5′-A₆G₃ sequences, respectively. As shown for NDV-T2, the presence of U at position 2278 increased the level of edited mRNAs and altered the pattern of editing. It would be of interest to determine whether other members of the subfamily Paramyxovirinae produce high percentage of edited transcripts in a pattern similar to that of NDV-T2.

The only difference between NDV-T2 and the parent virus is the formation of unique base pairings in the upstream sequence. As shown in Fig. 6, the unique upstream C:U/A:U pairing in NDV-T2 is more efficient than the corresponding C:C/G:U pairing in rNDV in promoting one G insertion. For inserting two G's, the upstream 3′-ACA template strand of NDV-T2 would require to base pair with 5′-UUA, whereas the corresponding pairing in the parent rNDV would be between 3′-ACG and 5′-CUA. The result is generation of 41% W mRNA (with two G insertion) in NDV-T2 as opposed to only 4% W mRNA in the parent virus (Fig. 6). It appears that the nature and most likely the stability of the mRNA-template hybrid in the most upstream 5′-A₅G₃ sequence of NDV is a key determinant in controlling how many times the polymerase stutters and when to resume normal transcription after an editing event. However, the editing process is less efficient in the absence of an intact A-stretch at the editing site, suggesting that mutations within the uridine-rich region of the negative-strand genome may induce secondary structure formation that prevent or weaken specific interaction with the transcription-editing complex. Therefore, nucleotide insertion and editing-site recognition might be two distinct processes that cooperatively regulate P-gene mRNA editing. Further studies should reveal the exact mechanism involved in the editing of paramyxoviruses.