Mutations at Alternative 5′ Splice Sites of M1 mRNA Negatively Affect Influenza A Virus Viability and Growth Rate

ABSTRACT Different amino acid sequences of influenza virus proteins contribute to different viral phenotypes. However, the diversity of the sequences and its impact on noncoding regions or splice sites have not been intensively studied. This study focuses on the sequences at alternative 5′ splice sites on M1 mRNA. Six different mutations at the splice sites were introduced, and viral growth characteristics for those mutants generated by reverse genetics with 12 plasmids were examined, for which G12C (the G-to-C mutation at the first nucleotide of the intron for the mRNA3 5′ splice site), C51G (at the 3′ end of the exon of the M2 mRNA 5′ splice site), and G146C (for the first nucleotide of the intron for mRNA4) are lethal mutations. On the other hand, mutants with the mutation G11C (at the 3′ end of exon of the mRNA3 5′ splice site), G52C (for the first nucleotide of the intron for M2 mRNA), or G145A (at the 3′ end of the exon of mRNA4) were rescued, although they had significantly attenuated growth rates. Notably, these mutations did not change any amino acids in M1 or M2 proteins. The levels of precursor (M1 mRNA) and spliced products (M2 mRNA, mRNA3, and mRNA4) from the recombinant mutant virus-infected cells were further analyzed. The production levels of mRNA3 in cells infected with G11C, G52C, and G145A mutant viruses were reduced in comparison with that in wild-type recombinant virus-infected ones. More M2 mRNA was produced in G11C mutant virus-infected cells than in wild-type-virus-infected cells, and there was little M2 mRNA and none at all in G145A and G52C mutant virus-infected ones, respectively. Results obtained here suggest that introducing these mutations into the alternative 5′ splice sites disturbed M1 mRNA splicing, which may attenuate viral growth rates.

Influenza A virus is a pathogen of humans, birds, and other mammals for respiratory tract infections. Pandemic influenza A virus infection causes high morbidity and mortality and has major social and economic impacts in the world. The virus is a member of the family Orthomyxoviridae and contains eight segmented, negative-stranded genomic RNAs. These eight segments together encode viral RNA polymerase complex members PB2, PB1, and PA; glycoprotein hemagglutinin (HA); nucleoprotein (NP); neuraminidase (NA); matrix protein (M1); ion channel protein (M2); nonstructural proteins NS1 and NS2; and an alternatively translated protein, PB1-F2 (9). The M1 matrix protein is a major structural component of the virus particle and forms a layer beneath the lipid cellderived envelope. Inside the virion and in infected cells during the late stages of virus replication, the M1 protein associates with viral ribonucleoproteins (vRNPs) (3,4,39) and plays an important role in influenza A virus budding (19,20,28). Expression of M1 protein in mammalian cells results in budding of virus-like particles (16). The protein contains a specific amino acid sequence whose function resembles the function of the late domain of retrovirus matrix proteins regarding virus budding, which evidences the role of M1 in virus budding (20). The mutation of certain residues in M1 protein markedly influences the morphology of virus particles (6,8,12).
In addition to encoding the M1 protein, the M gene of influenza A virus also encodes the ion channel transmembrane protein M2 through alternative splicing. The M2 ion channel protein is abundantly expressed in the plasma membrane of virus-infected cells but is significantly underrepresented in virions because of very limited molecules that are incorporated into virus particles (5,22,43,44). M2 protein is likely needed for efficient vRNP uncoating during viral entry (17), and M2 mutant influenza viruses are extremely attenuated (10,40). Other functions of M2 include virus assembly and budding (32).
During cellular mRNA maturation, introns are removed precisely and flanking exons are ligated. Alternative splicing of precursor mRNAs is one of the most important mechanisms for introducing protein diversity in eukaryotes. Influenza A virus M1 mRNA is colinear with viral RNA, whereas M2 mRNA is encoded by an alternative spliced transcript (23,24,37,41). M1 mRNA contains three alternative 5Ј splice sites (5Ј ss): a 5Ј ss at position 12, which produces mRNA3; a 5Ј ss at position 52, which produces M2 mRNA; and a 5Ј ss at position 146, which produces mRNA4 (1,23,24,38). Although mRNA3 and M2 mRNA are seen in all influenza A virus strains, mRNA4 exists only in some strains, such as A/WSN/33 (38). The sequences of the mRNA3 5Ј ss CAG/GUAGAU (the slash without accompanying parentheses around the two relevant nucleotides indicates the exon/intron boundary) and the mRNA4 5Ј ss GAG/GUUCUC resemble that of the consensus 5Ј ss AAG/GUAAGU closely. The M2 mRNA 5Ј ss sequence AAC/GUA(U/C)GU, however, does not fit well with the consensus because of a C rather than a G at the 3Ј end of the 5Ј exon. Alternative 5Ј splicing of influenza A virus M1 mRNA is controlled by viral polymerase and cellular splicing factors (36,37). Early during infection, the distal 5Ј ss is used to produce mRNA3. At a relatively late stage in infection in cells, newly synthesized polymerases bind to the virus mRNA 5Ј end that encompasses the first 11 or 12 nucleotides of the 5Ј terminus, thereby blocking the mRNA3 5Ј ss located at position 12 of the M1 mRNA (36,37). Moreover, the cellular SF2/ASF splicing factors interact with the 3Ј exon of the M1 mRNA and enhance activation of the M2 mRNA 5Ј ss (36).
Since only M1 mRNA and M2 mRNA encode two functional viral M1 and M2 proteins, which are important for viral growth, several interesting questions arise. (i) Why does the mRNA3 5Ј ss exist and why has it been preserved during evolution? (ii) Why is the M2 mRNA 5Ј ss weaker than mRNA3 as a signal for splicing? (iii) As the mRNA4 5Ј ss exists only in certain strains of influenza A viruses, how does it affect the viability for these strains? Experimental results obtained from a number of earlier studies may provide some answers (10,36,38). The present study introduced a series of mutations at alternative 5Ј ss of M1 mRNA. By analyzing viral viability and growth rates of the recombinant viruses generated by reverse genetics with 12 plasmids, we determined whether these mutations were lethal or whether they would reduce viral growth rates. The findings may provide a clue as to why specific alternative 5Ј ss signatures in influenza virus genomes are evolutionarily preserved.

MATERIALS AND METHODS
Cell culture. Madin-Darby canine kidney (MDCK) cells and 293T human embryonic kidney cells were maintained in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% fetal bovine serum (Gibco). All cells were maintained at 37°C under conditions of 5% CO 2 .
Construction of a mutant with a mutation at an alternative 5 ss. Mutation constructs were introduced into pPOLI-M-RT by using a QuikChange kit (Stratagene). Sequence analysis confirmed that only the specifically introduced mutations were present in the plasmids.
Plasmid-based reverse genetics. Mutant viruses (A/WSN/33 strain) were generated using the 12-plasmid-based reverse genetics system described by Fodor et al. (15). Plasmids were kindly provided by George G. Brownlee. For virus rescue, 10 6 293T cells were cotransfected with four protein expression plasmids (pcDNA-PB2, pcDNA-PB1, pcDNA-PA, and pcDNA-NP) and eight viral RNA (vRNA) transcription plasmids (pPOLI-PB2-RT, pPOLI-PB1-RT, pPOLI-PA-RT, pPOLI-HA-RT, pPOLI-NP-RT, pPOLI-NA-RT, pPOLI-M-RT, and pPOLI-NS-RT) by using Lipofectamine 2000 (Invitrogen). After 24 h, the transfection medium was removed from the cells and replaced with fresh medium containing 0.5% fetal bovine serum, penicillin, and streptomycin. The transfected cells were maintained for 2 to 3 days after transfection. The medium from transfected cells was collected daily and assayed for the presence of the influenza virus by attempting to create plaques with a 0.5-ml aliquot on MDCK cells by using standard methods. The remaining medium was transferred into 25-cm 2 flasks containing subconfluent MDCK cells for amplification of any rescued virus. The rescued virus showed a specific property characteristic of the influenza A/WSN/33 virus (i.e., it formed plaques on MDCK cells in the absence of trypsin). The plaques formed by the rescued virus were comparable in size to those formed by an authentic wild-type influenza A/WSN/33 virus grown on the same MDCK cells.
Sequencing of recombinant viruses. M, NP, and NS vRNA from transfected or infected supernatants was reverse transcribed by using primers 5Ј-AGTAGAA ACAAGGTAGTTTTT-3Ј (corresponding to nucleotides [nt] 1007 to 1027 of M cRNA), 5Ј-AGTAGAAACAAGGGTATTTTT-3Ј (corresponding to nt 1545 to 1565 of NP cRNA), and 5Ј-AGTAGAAACAAGGGTGTTTT-3Ј (corresponding to nt 871 to 890 of NS cRNA) and SuperScript II reverse transcriptase (Invitrogen). The reverse-transcribed cDNA was amplified by PCR by using specific primers 5Ј-AGCAAAAGCAGGTAGATATT-3Ј (corresponding to nt 1 to 20 of M cRNA), 5Ј-AGCAAAAGCAGGGTAGATAA-3Ј (corresponding to nt 1 to 21 of NP cRNA), and 5Ј-AGCAAAAGCAGGGTGACAAA-3Ј (corresponding to nt 1 to 20 of NS cRNA) and the KOD-plus kit (Toyobo). PCR products were then purified with a gel extraction kit (Qiagen) and sequenced. Sequencing of the 5Ј and 3Ј ends of the M vRNA was performed by rapid amplification of cDNA ends. For the sequencing of the 3Ј end of M vRNA, purified M vRNA was first transcribed by using primer 5Ј-AGCAAAAGCAGGTAGATATTGAAAGV N-3Ј (the underlined sequence is complementary to nt 1003 to 1027 of the M vRNA). The reverse-transcribed cDNA products were amplified by using primer 5Ј-GGATGGGGGCTGTGACCACTGAAGTGGC-3Ј (complementary to nt 575 to 602 of the M vRNA) and the Super SMART PCR cDNA synthesis kit (Clontech) by following the manufacturer's instructions. For the sequencing of the 5Ј end of M vRNA, purified M vRNA was transcribed by using primer 5Ј-AAGCAGTGGTATCAACGCAGAGTACAGCAAAAGCAVN-3Ј (the underlined sequence is complementary to nt 1018 to 1027 of the M vRNA). The reverse-transcribed products were amplified by using primer 5Ј-GCCACTTCA GTGGTCACAGCCCCCATCC-3Ј (corresponding to nt 575 to 602 of the M vRNA) and Super SMART PCR cDNA synthesis kit (Clontech) by following the manufacturer's instructions. PCR products were then purified with a gel extraction kit (Qiagen) and sequenced.
Plaque assay. Confluent MDCK cells in 35-mm dishes were washed with phosphate-buffered saline, and serial dilutions of the virus were adsorbed onto cells for 1 h at 37°C. Unadsorbed virus was removed by washing with phosphatebuffered saline, and cells were overlaid with 3 ml of overlay Dulbecco's modified Eagle's medium supplemented with 3% agarose. After incubation for 72 h at 35°C, cells were fixed with 10% formalin for 1 h. Following formalin removal, cells were stained with crystal violet and plaques were visualized. Visible plaques were counted, and concentrations of PFU/ml were determined. The plaque numbers and sizes were obtained from at least three independent experiments.
Western blotting. Transfected cells were lysed in lysis buffer (0.6 M KCl, 50 mM Tris-HCl [pH 7.5], 0.5% Triton X-100) at 72 h posttransfection. Cell lysates were analyzed by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to a polyvinylidene difluoride membrane (Amersham Biosciences). The transfer membrane was first blocked for 1 h at room temperature with Tris-buffered saline-Tween (TBS-T) containing 5% skim milk, followed by either overnight incubation with 14C2 mouse anti-M2 immunoglobulin G (IgG) monoclonal antibody (Ab; Affinity Bioreagents) (43) (diluted 1:1,000) or 1D6 anti-M2 cytoplasmic tail monoclonal Ab (diluted 1:100) (kindly provided by Robert A. Lamb) at 4°C or incubation with anti-M1 IgG monoclonal Ab (diluted 1:500) (Biodesign International) for 2 h at room temperature. The membrane was washed three times with TBS-T and then incubated for 1 h at room temperature with a 1:2,000 dilution of horseradish peroxidase-conjugated antimouse IgG Ab. Signals were detected by using the Western immunoblot ECL detection system (Amersham Biosciences) and exposed to X-ray film (Kodak).
RNA extraction and reverse transcription-PCR (RT-PCR). Total RNA from transfected 293T cells transfected via reverse genetics with 12 plasmids or virusinfected MDCK cells were collected using an RNeasy mini kit (Qiagen) following the manufacturer's instructions. The cDNA was made using a SuperScript II reverse transcription kit (Invitrogen) by following the manufacturer's instructions. The specific primers for viral M mRNAs and M vRNA were described previously by Cheung et al. (10). Primers for mRNA4 were sense primer 5Ј-G AACACCGATCTTGAGGCCTAT-3Ј (the underlined sequence corresponds to nt 130 to 145 of the M cRNA; the italicized sequence corresponds to nt 740 to 744 of M cRNA) and antisense primer 5Ј-CTGTTCCTTTCGATATTCTTC-3Ј (corresponding to nt 95 to 75 of the M vRNA). Primers for NP vRNA were described previously by Liang et al. (26). Amplified products were further analyzed by 2% agarose gel electrophoresis.
Real-time quantitative PCR. The specific primers used for M1 mRNA, M2 mRNA, and mRNA3 were described previously by Cheung et al. (10). The specific primers for mRNA4 are described in the previous paragraph. Values were normalized with the ␤-actin mRNA level. The primers for ␤-actin mRNA were sense primer 5Ј-GCTCGTCGTCGACAACGGCTC-3Ј and antisense primer 5Ј-CAAACATGATCCTGGGTCATCTTCTC-3Ј. The PCR amplification yielded a product of 352 bp. The cDNA was amplified using Sybr green real-time PCR master mix (Bio-Rad), 5 mM of each primer and 0.5 l of the cDNA product in a total volume of 50 l. Forty cycles of PCR (one cycle consists of 10 min at 95°C, 15 s at 95°C, and 1 min at 60°C) were performed using the Bio-Rad iQ5 system. A melting curve analysis was performed to verify the specificity of the products; the relative values were calculated using the ⌬⌬CT method. Each experiment was performed in triplicate.
Primer extension assay. Primer extension reactions were performed by using the primer extension system-AMV reverse transcriptase kit (Promega) by following the manufacturer's instructions. Briefly, 5 g total RNA was mixed with M vRNA-specific 32 P-labeled primer and positive-sense RNA-specific 32 P-la-

10874
CHIANG ET AL. J. VIROL. beled primer described by Mullin et al. (27). Primer extensions were performed at 42°C for 90 min. Transcription productions were denatured at 90°C for 10 min, separated on 6% polyacrylamide gels containing 7 M urea in Tris-borate-EDTA buffer, and detected by autoradiography.

RESULTS
Genomic signatures for mRNA3 5 ss are required for M gene synthesis. The mRNA3 5Ј ss is located at nt 11 and 12 of M1 mRNA, which can produce mRNA3 with only a coding potential of nine amino acids (Fig. 1). The sequences AG/GU of mRNA3 5Ј ss in all influenza A viruses were analyzed and were highly conserved without variation. To determine the impact of the conserved mRNA3 5Ј ss on influenza A viruses, mutations at this splice site were introduced and recombinant viruses were generated by using reverse genetics with 12 plasmids. Nucleotide changes were made at 3Ј end of the 5Ј exon (position 11) or intron donor site (position 12) of the M1 mRNA ( Fig. 2A). The plasmid carrying mutated M vRNA and the remaining seven genomic vRNA expression plasmids were cotransfected with PB1, PB2, PA, and NP protein expression plasmids into 293T cells. The G-to-C substitution was made at position 11 (G11C) such that the mRNA3 5Ј ss sequence AC/GU did not fit the consensus splicing sequence AG/GU (wild type). The resulting mutant (the G11C mutant) was rescued, although it had a reduced growth rate ( Fig. 2A; see Fig.  7A). The G-to-C substitution was also made at position 12 (G12C) to knock out the splicing at the mRNA3 5Ј ss. The G12C mutant could not be rescued ( Fig. 2A). These results indicate that mutations at the alternative 5Ј splice junctions for mRNA3 affect the production of the influenza A virus. Neither the G11C mutation nor the G12C mutation alters any amino acid, because positions 11 and 12 are not within the regions coding for M1 and M2 proteins (Table 1).
This study further investigated M1 mRNA splicing in the 293T cells transfected with the indicated 12 plasmids. The production of the spliced products and expression levels of the M1 and M2 proteins were examined. The M1 mRNA and all spliced mRNAs, including M2 mRNA, mRNA3, and mRNA4, were detected in the wild type and in the G11C mutant ( When the sequence at the mRNA3 5Ј ss was changed, change in the promoter sequence in 3Ј end of M vRNA was also introduced accordingly. It has been reported that positions 11 and 12 are within the promoter region of the 3Ј end of vRNA and the 5Ј end of cRNA (Fig. 3A). We hypothesized that G11C and G12C would decrease M vRNA synthesis and reduce the expression of M1 mRNA, which consequently affect the production of M1 and M2 proteins. To test this hypothesis, primer extension and RT-PCR were conducted to detect M gene synthesis. Significant decreases in the M vRNA, M1 mRNA, and M cRNA levels were seen for the G11C mutant Neither M1 mRNA nor M cRNA was detected in the G12C mutant (Fig. 3B, lane 4). However, due to the superior sensitivity of the RT-PCR assay, the weak band for M1 mRNA remained detectable in the G12C mutant (Fig. 2B, lane 4). The level of NP vRNA, the control, did not change in the G11C or G12C mutant (Fig. 3C, right panel, lanes 6 to 8). These results, therefore, suggest that the G-to-C substitutions at positions 11 and 12 of the M gene affect promoter activity during M gene synthesis.
The conserved 3Ј and 5Ј terminal nucleotides on vRNA form a corkscrew secondary structure through partially complemen-FIG. 1. Structures of M1 mRNA in influenza A virus and its three alternative spliced mRNA products, namely, mRNA3, M2 mRNA, and mRNA4, of influenza virus A/WSN/33. Solid squares at the 5Ј end represent the 5Ј cap from host cells. Gray, hatched, and white rectangles represent coding regions. Bold lines at the 5Ј and 3Ј ends of the mRNAs represent noncoding regions. The N terminus of the M2 protein and the segment of mRNA4 protein corresponding to the coding region each overlap that of M1 protein by 9 residues and 40 residues, respectively. The C terminus of the mRNA4 protein segment corresponding to the coding region overlaps that of M1 protein by 14 residues. Italic letters represent the introns of the splice site sequences. aa, amino acids. tary base-pairing, and this is required for the promoter activity (7,13,14). The cRNA promoter is complementary to the vRNA promoter and structures a corkscrew base-pairing between the 5Ј nt 10Ј to 12Ј and the 3Ј nt 11 to 13 (Fig. 4A) (2). Two mRNA3 5Ј ss mutations at position 11 (G11C) and 12 (G12C) of M1 mRNA, as seen here, inevitably disrupted the potential base-pairing in M cRNA promoter (Fig. 4A). To avoid the structure disruption, two compensatory mutations, G11C-CM12 and G12C-CM13, at positions 12 and 13 in the 3Ј end of M cRNA of G11C and G12C, were made to restructure the base-pairing of the promoter regions (Fig. 4A). While the newly made G11C-CM12 mutant still presented a growth rate as low as that of the G11C mutant, the new G12C-CM13 mutant was successfully rescued, in contrast to the lethal G12C mutant (Fig. 4A). Significant levels of M vRNA, M1 mRNA, and M cRNA were restored in G12C-CM13 (Fig. 4B, lane 4). G11C-CM12, on the other hand, did not recover the M RNA synthesis (Fig. 4B, lane 3). We further examined the splicing by real-time RT-PCR and found that M1 mRNA level of G11C-CM12 was decreased (Fig. 4B, lane 3), followed by the reduction of spliced M2 mRNA, mRNA3, and mRNA4 (Fig. 4C,  lanes 3, 7, 11, and 15). In terms of protein synthesis levels, M1 and M2 of G11C-CM12 were also decreased (Fig. 4D, lane 2).
The G12C-CM13 mutant, on the other hand, produced levels of M1 mRNA, M2 mRNA, and mRNA4 (Fig. 4C, lanes 4, 8,  and 16), but not mRNA3 (Fig. 4C, lane 12), that were comparable to those of the wild type. Expression of the M1 and M2 proteins were barely affected by this mutation (Fig. 4D, lane 3).   These results indicate that conserved genomic signatures for mRNA3 5Ј ss among all influenza A viruses are required for the formation of the structure of promoter region, which is essential for M gene synthesis.
A weak splice site for M2 mRNA is essential for efficient influenza A virus growth. After the importance of mRNA3 5Ј ss was determined, the impact of the M2 mRNA 5Ј ss on the influenza A virus was studied. The C-to-G substitution at the 3Ј end of the 5Ј exon (position 51) of M2 mRNA (C51G) was made to increase the strength of the 5Ј ss. The resulting mutant (C51G mutant) could not be rescued (Fig. 5A). The M1 mRNA level was reduced moderately in 12-plasmid-transfected cells in which the C51G mutation was made in the M gene (Fig. 5B, lane 3). The amount of M2 mRNA in C51G was proteins from M2 mRNA 5Ј ss mutants at 72 h after transfection of 293T cells with 12 plasmids via reverse genetics. The M1 and M2 proteins were each detected by mouse anti-M1 Ab and 1D6 mouse anti-M2 cytoplasmic tail Ab, respectively. Anti-M2 1D6 Ab (against C terminus of M2 protein) was used for Western blotting to reduce the interference of Ab affinity caused by amino acid substitution near the N terminus of the M2 protein.
VOL. 82, 2008 M1 mRNA 5Ј SPLICE SITES AND VIRUS VIABILITY twofold higher than that in the wild type (Fig. 5B, lane 7). However, mRNA3 and mRNA4 splicing levels were significantly decreased in the C51G mutant (Fig. 5B, lanes 11 and 15) in comparison with those of the wild type (lanes 10 and 14). Production levels of M2 protein in the wild type and the C51G mutant were comparable (Fig. 5C, lanes 1 and 2). However, the expression level of the M1 protein was markedly reduced in the C51G mutant (Fig. 5C, lane 2). The mutation at position 51 caused the ninth amino acid to change from Thr to Arg in both M1 and M2 proteins (Fig. 5A). Whether or not this change would jeopardize the viability of recombinant virus was also examined. One A50C C51G mutant, which corresponds to a T9R mutation, and an A50G mutant, which corresponds to a T9A mutation, were made and tested. While the former recombinant virus was lethal, the latter was rescued, yet it had a reduced growth rate (Fig. 5A). It seems that Thr at position 9 is essential for virus survival with an adequate growth rate. This may explain why the M2 mRNA 5Ј ss exists in such a sequence combination (AAC/G, a weak 5Ј ss) so that it can encode Thr9.
On the other hand, the G-to-C substitution was made at position 52 (the first nucleotide of the intron) to knock out the splicing for the production of M2 mRNA. The resulting recombinant virus was rescued but had a growth rate that was lower (by approximately 2 logs) than that of wild type ( Fig. 5A; see Fig. 7A). As expected, M2 mRNA splicing was impaired by the mutation of G at the donor site of the M2 mRNA intron (Fig. 5B, lane 8). The G52C mutant did not express the M2 protein (Fig. 5C, lane 3) in the 12-plasmid-transfected cells. These experimental results have also been obtained by Cheung et al. (10). mRNA4 5 ss is important for the growth of A/WSN/33. After investigating mRNA3 and M2 mRNA 5Ј ss, we further turned our attention to the downstream mRNA4 5Ј ss. Unlike the highly conserved sequences in mRNA3 and M2 mRNA 5Ј ss, the mRNA4 5Ј ss sequences differed among influenza A viruses (38). All of the influenza A virus M gene sequences (up to 19 January 2008) were analyzed. Twenty-two strains out of 6,192 strains contained the potential mRNA4 5Ј ss (Table 2). These strains belong to different subtypes and were isolated from various hosts, including humans, birds, and swine. Many of these strains listed in Table 2 To determine whether the mRNA4 5Ј ss plays a role in viral growth, we introduced two mutations at this 5Ј ss. The G-to-A substitution at position 145 (the 3Ј end of the 5Ј exon) was made to weaken the 5Ј ss. The resulting mutant (the G145A mutant) was rescued, although it had a low growth rate ( Fig. 6A; see Fig. 7A). When G was replaced by C at position 146 (intron donor site) to knock out the production of mRNA4, the recombinant virus (G146C) was not rescued (Fig.  6A). The same results were also obtained for A/Puerto Rico/ 8/34 (data not shown).
Together these data suggest that the existence of mRNA4 5Ј ss in certain strains of influenza A viruses, such as A/WSN/33 and A/Puerto Rico/8/34, may affect viral growth. In addition, the mRNA4 levels were strongly reduced when A was substi-tuted for G at position 145 and C was substituted for G at position 146 (Fig. 6B, lanes 15 and 16). The M1 and M2 protein expression levels in G145A and G146C mutant virustransfected cells were lower than those in wild-type-transfected cells (Fig. 6C). In order to know whether other mutations at position 146 would lead to lethal phenotypes, we made two additional mutants of A/WSN/33 M1 mRNA. The G146A mutant (corresponding to an amino acid change from V to I) was able to be rescued, yet it had a low growth rate. G146U (resulting in an amino acid change from V to F), on the other hand, was a lethal mutation (Fig. 6A). The results indicate that Val at position 41 of M1 protein is important for virus survival and efficient growth.
Attenuated growth characteristics of influenza A viruses with mutations at alternative 5 ss of M1 mRNA. Table 1 summarizes the effects of mutations at the alternative 5Ј ss for mRNA3, M2 mRNA, and mRNA4 on production of progeny influenza A viruses. Notably, G12C, C51G, and G146C are lethal mutations, whereas mutants with G11C, G52C, and G145A can be rescued. The stabilities of the introduced mutations were examined by sequencing of the full length of the M gene segment. These recombinant viruses were found to be stable for at least 10 passages and were not reverted to the wild type. We also did not observe any nucleotide substitution. Sequences of other gene segments, including NP and NS genes, that may associate with M gene were also examined and found no mutations (data not shown). This study further examined the growth properties of rescued recombinant viruses in MDCK cells. Cells were infected with recombinant viruses at a multiplicity of infection of 0.001; the viruses yielded in the culture supernatant at various times were determined by plaque assay. The results demonstrate that G11C, G52C, and G145A mutant viruses grew significantly more slowly than the wildtype virus did (Fig. 7A). The maximum titers of G11C, G145A, and G52C viruses were approximately 1 to 2 logs lower than that of the wild-type recombinant virus (Fig. 7A). These results demonstrate that variations in M1 mRNA 5Ј ss attenuated influenza A virus growth. Notably, these mutations did not change any amino acid sequence in the encoded virus proteins (Table 1). The plaque diameter of the G52C recombinant virus after 72 h postinfection was approximately 0.5 mm, four times smaller than that of the wild-type, G11C, and G145A recombinant viruses (approximately 2 mm in diameter at 72 h postinfection) (Fig. 7B). These data are consistent with the findings that M2 protein expression is not required for the growth of the influenza A virus in cell cultures and that influenza A virus can undergo multiple cycles of replication without M2 ion channel activity (10,40).

Disturbance of M splicing and protein expression in the recombinant mutant virus-infected cells.
After demonstrating that growth of G11C, G52C, and G145A mutant recombinant viruses was attenuated by mutations at alternative 5Ј ss of M1 mRNA, this study then examined the splicing of M1 mRNA and expression levels of M1 and M2 proteins in G11C, G52C, and G145A mutant virus-infected cells. Quantitative real-time RT-PCR using specific primers for each spliced mRNA was performed. The amount of mRNA3 in wild-type infected cells at 12 h postinfection was approximately 10-fold higher than that at 24 h (7.73% versus 0.77%; Student's t test, P ϭ 0.0003) and approximately 8-fold higher than that at 36 h (7.73% versus 1%; Student's t test, P ϭ 0.0004) postinfection (Fig. 8A). In contrast, as time increased, the amount of M2 mRNA in wild-type virus increased. When the mRNA3 5Ј ss was weakened (via the G-to-C mutation at position 11), splicing efficiency of the producing M2 mRNA increased approximately 1.3-fold at 24 h (23.09% versus 17.49%; Student's t test, P ϭ 0.0011) and approximately 1.3-fold at 36 h (32.99% versus 25.5%; Student's t test, P ϭ 0.0029) postinfection in comparison to the efficiency of wild-type virus. The M2 mRNA was not produced in the G52C mutant (the M2 mRNA 5Ј ss knockout mutation). On the other hand, when the mRNA4 5Ј ss was weakened, the production of M2 mRNA also decreased approximately 2.3-fold at 24 h (17.49% versus 7.66%; Student's t test, P ϭ 0.016) and approximately 1.7-fold at 36 h (25.5% versus 15.27%; Student's t test, P ϭ 0.036) postinfection in comparison to wild-type M2 mRNA production.
Expression levels of the M1 and M2 proteins at 12, 24, and 36 h after infection of MDCK cells were determined. As time increased, the levels of M1 and M2 in wild-type virus-and mutant-infected cells increased (Fig. 8B). The expression levels of M1 protein in G52C and G145A mutants (Fig. 8B, lanes 8 and 9 and lanes 11 and 12, respectively) were lower than those in wild-type virus (lanes 2 and 3) at 24 and 36 h after infection of MDCK cells. Although the G11C mutant did not change the expression of M1 protein levels, the expression of M2 protein was reduced slightly (Fig. 8B, lanes 5 and  6). As expected, no M2 protein was detected in the G52 mutant virus-infected cells (Fig. 8B, lanes 8 and 9). Furthermore, expression of M2 proteins was significantly decreased in the G145A mutant (lanes 11 and 12). Taken together, the results suggest that mutations in the alternative 5Ј ss of M1 mRNA affect the production of M2 mRNA and the protein it encodes.

DISCUSSION
The influenza virus harbors an enormous genomic diversity because of gene reassortment or accumulation of point muta- tions. Many studies have revealed that sequence variations in the coding region change virus phenotypes. However, the impact of sequence variation in noncoding regions or splice sites has not been extensively studied. This work clearly demonstrated that the mutations at alternative 5Ј ss of M1 mRNA were lethal or attenuated the growth rate for influenza A virus. These findings may explain why these splicing signatures must be preserved throughout viral evolution.
Why do additional splice sites in the M gene of influenza A viruses exist despite the fact that only M1 mRNA and M2 mRNA encode two functional viral M1 and M2 proteins? Specifically, we asked three questions in the beginning of this study. (i) Why does the mRNA3 5Ј ss exist and why has it been preserved during evolution? (ii) Why is M2 mRNA 5Ј ss a weaker signal than mRNA3 is for splicing? (iii) As mRNA4 5Ј ss exists only in certain strains of influenza A viruses, how does it affect the viability of these strains? By introducing mutations into the splice site junctions and examining M RNA synthesis, splicing, and protein expression in detail, we revealed clues to answer these questions. We firstly observed that the mRNA3 5Ј ss has been preserved because the splice site signature G/G at positions 11 and 12 of M1 mRNA corresponds to CC at the 3Ј end of vRNA, which is important to form a promoter structure needed for efficient viral RNA synthesis. Secondly, our mutation analysis shows that changing the M2 mRNA 5Ј ss into a stronger one would inevitably prevent the M2 protein residue at position 9 from being coding as Thr, which would adversely affect the virus viability and growth rate. This explains why the M2 mRNA 5Ј ss has to stay in a weaker form than mRNA3 does. Lastly, for mRNA4 5Ј ss, we noticed that the signature G as the first intron nucleotide for mRNA4 is responsible for encoding Val at position 41 of M1 protein, and this might be important for virus survival and efficient growth. This could explain why the mRNA4 5Ј ss is there. There is, however, another novel reason for the existence of the mRNA4 5Ј ss. We found that the mutation at position 145 (the last nucleotide of the 5Ј exon for mRNA4) did not change any amino acid in M1 or M2 protein but did attenuate the viral growth rate (Fig. 7A). This attenuation might have been a result from the disturbance of M mRNA splicing (Fig. 8).
It has been proposed that the mRNA3 5Ј ss is spliced early in infection to ensure that the M2 protein is expressed when needed (36,37). Splicing of the M2 mRNA 5Ј ss occurs late in infection, when alternative splicing occurring at mRNA3 5Ј ss is blocked by the binding of the viral polymerase complex at the first 11 or 12 nucleotides of the 5Ј end of M1 mRNA (37). Two mRNA3 5Ј ss mutants, with mutation G11C (which weakens the mRNA3 5Ј ss) or G12C (which knocks out the mRNA3 5Ј ss), were generated to affect the production of mRNA3. A decrease in mRNA3 and an increase in M2 mRNA in these mutants were expected. Increased M2 mRNA levels in G11C recombinant virus-infected cells were observed at 24 h and 36 h postinfection (Fig. 8A). Splicing of the mRNA3 knockout mutation (G12C) could not be validated, because precursor M1 mRNA was barely detectable (Fig. 2B, lane 4). The undetectable level of M1 mRNA was possibly due to the impaired M gene synthesis ( Fig. 3B and C). Impaired synthesis was likely because the sequence signatures for mRNA3 5Ј ss are also important signatures for the M vRNA promoter. The terminal 13 and 12 nucleotides of the 5Ј and 3Ј ends, respectively, are highly conserved in eight influenza A virus RNA segments. The first 12 to 14 nucleotides at the 3Ј end of the vRNAs and the first 11 to 13 nucleotides at the 3Ј end of the cRNA comprise the core promoter region (25, 29-31, 33, 34, 42). A conventional chloramphenicol acetyltransferase reporter assay system demonstrated that disruption of base pairs at position 11 or 12 in the 5Ј strand of cRNA significantly reduced the vRNA level, indicating that base-pairing between the 5Ј and 3Ј ends of the cRNA promoter is essential for viral replication (11,21). The mRNA3 5Ј ss is located at nt 11 and 12 of M1 mRNA, corresponding to nt 11 and 12 of the M cRNA promoter region (Fig. 3A). Therefore, the G11C and G12C mutants were expected to disrupt the base-pairing of the cRNA promoter. Experimental findings demonstrated that the G11C and G12C mutants reduced the M vRNA level in 12-plasmidtransfected cells (Fig. 3B and C). These results are consistent with findings obtained by Crow et al. (11) and Kim et al. (21). In addition to the role of the mRNA3 5Ј ss in modulating the use of the alternative 5Ј ss of M1 mRNA in infected cells (37), the fact that G at position 12 is within the vRNA promoter region definitely accounts for why the mRNA3 5Ј ss sequence is highly conserved in all influenza A viruses.
The M2 mRNA 5Ј ss exists for M2 protein production. The M2 protein is essential for vRNP uncoating during viral entry (17). This study demonstrates that the mutant with the G52C mutation (which knocks out the M2 mRNA 5Ј ss) and lacking the M2 protein grew slowly in MDCK cells (Fig. 7A), a result that is in agreement with the findings obtained by Cheung et al. (10) and Watanabe et al. (40), who showed that M2 protein expression in cell culture was not required for the growth of influenza A virus. Interestingly, the mutant with the C51G mutation (which enhances the weak M2 mRNA 5Ј ss) with an increased amount of produced M2 protein was not active in viral growth (Fig, 5A). Unlike the abundant M2 protein in the C51G mutant, M1 protein was expressed only in a fairly small amount (Fig. 5C). Strong M2 mRNA 5Ј ss may end up with extremely efficient splicing of M1 mRNA, resulting in only a tiny amount of precursor left (Fig. 5B). Without sufficient M1 protein, the virus may not assemble very effectively. Additionally, the C51G mutation changed the amino acid sequence from Thr to Arg at the ninth position of M1 and M2 proteins ( Table 1). Previous reports indicate failure to generate a series of recombinant viruses with mutations at the dinucleotide GU of the 5Ј exon of M2 mRNA 5Ј ss, because the 9th or 10th amino acid of the M1 protein involved in lipid membrane binding in the process of virus replication were altered by those mutations (10,35). The ninth amino acid substitution in the M1 protein affected rescue of the C51G mutant, a conclusion in agreement with the result obtained by Cheung et al. (10), suggesting that the ninth amino acid Thr is essential for production of influenza A virus.
A/Puerto Rico/8/34 (H1N1) is another influenza A virus that also contains mRNA4 5Ј ss. We have introduced G145A and G146C mutations into A/Puerto Rico/8/34 through an eightplasmid-based reverse genetics system (18). In a result similar to the results obtained from WSN, the G145A mutant was rescued but had a low growth rate, and G146C was a lethal mutation. This study also demonstrated that a mutation corresponding to a change of Val to Leu (G146C) at position 41 of M1 mRNA was a lethal mutation for the A/WSN/33 influenza A virus (Table 1). Amino acid changes at positions 41, 95, and 218 in the M1 protein of A/WSN/33 would result in the filamentous phenotype (12). Assembly of the G146C mutant is affected by the amino acid change at position 41. The knockout mutation in mRNA4 5Ј ss (G146C) is a nonsynonymous mutation. Therefore, it is difficult to assess the importance of mRNA4 5Ј ss with knockout mutation experiments. However, when the last nucleotide of the 5Ј exon at position 145 was changed from G to A, this synonymous mutation caused the recombinant virus to have a growth rate lower than that of wild-type virus (Fig. 7A), suggesting that the splicing site signature is important for the A/WSN/33 virus. The functions of the resulting amino acids in the M1 protein remain unclear. Table 2 lists some key prototype viruses that have been passaged multiple times and are high-yield viruses, and the existence of mRNA4 5Ј ss could be one of the reasons promoting such high-yield characteristics. Therefore, it is necessary to check whether mRNA4 can be translated into protein in virusinfected cells. If it was translated, it would yield a 54-residue polypeptide that has 40 and 14 amino acids that are identical to those of the N and C termini of the M1 protein, respectively (Fig. 1). We have indeed performed the test and did not detect this 54-residue peptide in infected cells by using M1 Ab. We have generated polyclonal Abs against those 20 residues at the C terminus of the putative M4 protein; however, those Abs could not clearly detect any bands potentially representing M4 (by molecular weight) either. Therefore, it remains unclear whether mRNA4 can be translated to a functional protein in infected cells.
Importantly, this study demonstrated growth rates for G11C, G52C, and G145A mutant viruses that were lower than that of the wild-type virus (Fig. 7A). Notably, these nucleotide mutations did not alter any amino acid sequence in the M1 and M2 proteins. A disturbance in M1 mRNA splicing in those mutant recombinant virus-infected cells was observed (Fig. 8A). When the upstream mRNA3 5Ј ss is weak, M2 mRNA 5Ј ss can be used more efficiently to produce more M2 mRNA; however, when the downstream mRNA4 5Ј ss is weakened, the utilization of M2 mRNA 5Ј ss also decreased, resulting in a level of M2 mRNA lower than that of wild-type virus (Fig. 8A). The phenomenon is likely due to different interferences caused by an adjacent 5Ј ss located in either the upstream or downstream region. The protein expression levels for both M1 and M2 in those virus-infected cells were decreased (Fig. 8B). It has been reported that low expression levels of M1 and M2 proteins in infected cells significantly reduce the influenza A virus titer (5), which is consistent with the findings obtained in this study.
In summary, several mutations were introduced into recombinant viruses. Viral viability and growth characteristics were analyzed. These mutations either caused viral lethality or reduced the viral growth rate. These results suggest that the alternative 5Ј ss signatures in the M1 mRNA of the influenza A virus are important for viral viability and an efficient growth rate and provide a clue as to why those splice sites exist.