Multiple Gene Segments Control the Temperature Sensitivity and Attenuation Phenotypes of ca B/Ann Arbor/1/66

Abstract

Cold-adapted (ca) B/Ann Arbor/1/66 is the influenza B virus strain master donor virus for FluMist, a live, attenuated, influenza virus vaccine licensed in 2003 in the United States. Each FluMist vaccine strain contains six gene segments of the master donor virus; these master donor gene segments control the vaccine's replication and attenuation. These gene segments also express characteristic biological traits in model systems. Unlike most virulent wild-type (wt) influenza B viruses, ca B/Ann Arbor/1/66 is temperature sensitive (ts) at 37°C and attenuated (att) in the ferret model. In order to define the minimal genetic components of these phenotypes, the amino acid sequences of the internal genes of ca B/Ann Arbor/1/66 were aligned to those of other influenza B viruses. These analyses revealed eight unique amino acids in three proteins: two in the polymerase subunit PA, two in the M1 matrix protein, and four in the nucleoprotein (NP). Using reverse genetics, these eight wt amino acids were engineered into a plasmid-derived recombinant of ca B/Ann Arbor/1/66, and these changes reverted both the ts and the att phenotypes. A detailed mutational analysis revealed that a combination of two sites in NP (A114 and H410) and one in PA (M431) controlled expression of ts, whereas these same changes plus two additional residues in M1 (Q159 and V183) controlled the att phenotype. Transferring this genetic signature to the divergent wt B/Yamanashi/166/98 strain conferred both the ts and the att phenotypes on the recombinant, demonstrating that this small, complex, genetic signature encoded the essential elements for these traits.

Influenza virus can be isolated from avian and mammalian species, and new strains emerge each year from this gene pool (29). These new strains have the potential to create epidemics that cause approximately 40,000 deaths annually in the United States or pandemics that can result in tens of millions of deaths. Influenza B virus has been isolated from both humans and seals, and the primary mechanisms driving emergence of new type B strains are reassortment among human isolates and antigenic drift (23). Influenza B virus causes excess morbidity and mortality and has been at least partially responsible for 16 epidemics during the past 70 years (31).

Vaccination is the main method of control for influenza epidemics. FluMist, a live, attenuated, trivalent, influenza vaccine has been shown to be safe, well tolerated, and effective in controlled clinical studies in both children and adults and was recently licensed for the prevention of influenza A and B (for a review, see reference 3). FluMist contains three different 6:2 genetic reassortants, each containing the hemagglutinin (HA) and neuraminidase (NA) gene segments of a currently circulating wild-type (wt) strain (i.e., H3N2, H1N1, and B) combined with the six internal gene segments (PB2, PB1, PA, NP, M, and NS) of the master donor virus (MDV) derived from either ca A/Ann Arbor/6/60 (MDV-A) or ca B/Ann Arbor/1/66 (MDV-B).

The properties that attenuate the vaccine and control its replication in humans, thereby conferring its safety and contributing to its effectiveness, are inherited from the six internal gene segments of the MDVs. MDV-B was produced by serial passage of the parental wt B/Ann Arbor/1/66 isolate at successively lower temperatures in eggs and primary chicken kidney (PCK) cells, which resulted in a vaccine strain that differed genetically from the parent isolate. These genetic differences resulted in the phenotypic expression of three characteristic markers that differentiate MDV-B from wt B/Ann Arbor/1/66. These characteristic phenotypes for MDV-B are as follows: (i) cold adaptation (ca), efficient growth at 25°C; (ii) temperature sensitivity (ts), restricted replication at 37°C; and (iii) attenuation (att), restricted replication in the lower respiratory tracts of ferrets (21). In addition, these phenotypes are genetically stable; multiple plaque isolates of an MDV-B derived vaccine retained the ca, ts, and att phenotypes after prolonged replication in the lower respiratory tract of immunosuppressed hamsters (28).

There is limited information describing which gene segments are necessary and sufficient for expression of the characteristic phenotypes for MDV-B. By using classical reassortment techniques, it was reported that the MDV-B PA gene segment could transfer the ts and att phenotypes to the heterologous B/Hong Kong/1732/76 strain (5, 6). Mutation of methionine 431 to isoleucine in this reassortant reverted the ts phenotype (i.e., non-ts) of the virus, demonstrating the critical nature of M431. However, introduction of the PA I431 mutation into MDV-B did not revert the ts phenotype (5), suggesting that other loci on the same and/or other MDV-B gene segments contributed to the ts phenotype.

The types of genetic analyses described above using classical reassortment techniques and heterologous strains are time-consuming. Interpretation of these data are confounded by the fact that gene segments from different strains generally differ from each other by many other sequence differences in addition to the individual residues being targeted for study. The generation of infectious influenza B virus from cloned cDNA, however, facilitates the genetic analysis, since mutations in all eight gene segments can be specifically introduced by recombinant DNA methods. No selection system is needed to obtain the appropriate recombinant virus, and interpretation of the data is not confounded by other strain differences. Previously, we reported the generation of wt B/Yamanashi/166/98 by transfection of eight plasmids by using the RNA pol I/pol II bidirectional transcription system (8, 9). Here, we generated MDV-B entirely from plasmids. This approach allowed us to generate an isogenic non-ts, non-att recombinant virus that differed by eight amino acids on three gene segments (PA, NP, and M) from MDV-B. The minimal genetic loci responsible for expression of the ts and att phenotypes were identified by constructing specific recombinant viruses with changes in a subset of these eight differences. Introducing a six-amino-acid subset of these differences into a divergent wt strain transferred the ts and att phenotypes, confirming their role in the expression of these phenotypes.

MATERIALS AND METHODS

Cells and virus isolates.

ca B/Ann Arbor/1/66 (MDV-B) was originally obtained from H. F. Maassab at the University of Michigan. MDCK cells were originally obtained from the European Collection of Cell Cultures. Primary chick kidney (PCK) cells were derived in house by trypsinization of the primary tissue and suspension in MEM (Earle's) medium containing 5% fetal calf serum. PCK cells were seeded in 96-well cell culture plates for 48 h in order to prepare monolayers with >90% confluency.

Cloning of plasmids.

Viral RNA was isolated and reverse transcription-PCR was performed as described previously with modifications (8). Briefly, the PCR fragments were digested with BsmBI (or BsaI for the NP segment) and inserted into the BsmBI sites of pAD3000 in a two- or three-fragment ligation reaction. Two to four clones of each plasmid were sequenced and compared to the consensus sequence of MDV-B. Plasmid sequences were corrected, when necessary, by routine cloning or utilizing the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) such that the coding capacity was identical to that of MDV-B. The resultant plasmids were designated pAB121-PB1, pAB122-PB2, pAB123-PA, pAB124-HA, pAB125-NP, pAB126-NA, pAB127-M, and pAB128-NS. The PB1 and HA plasmids, pAB121-PB1 and pAB124-HA, each had two silent nucleotide changes and pAB128-NS had one silent nucleotide change compared to the consensus sequence (PB1, A924G; C1701T; HA, T150C; and T153C NS, A416G). These nucleotide changes did not result in amino acid alterations and were retained to facilitate identification of the recombinant viruses.

Construction of plasmids with wt nucleotide substitutions in the PA, NP, PB2, and M gene segments were constructed by QuickChange mutagenesis of pAB123-PA, pAB125-NP, pAB122-PB2, and pAB127-M plasmids or by amplification of the target region with primers containing the desired mutation and subsequent ligation into the plasmid. The resulting plasmids were sequenced to ensure that the cDNA did not contain unwanted mutations.

The sequence of template DNA was determined by using Rhodamine or dRhodamine dye-terminator cycle sequencing ready reaction kits with AmpliTaq DNA polymerase FS (Perkin-Elmer Applied Biosystems, Inc., Foster City, CA). Samples were separated by electrophoresis and analyzed on PE/ABI model 373, model 373 Stretch, or model 377 DNA sequencers.

Temperature sensitivity assays in PCK cells and MDCK cells.

For plaque assays, virus dilutions were incubated for 30 to 60 min at 33°C on confluent MDCK monolayers. The cells were overlaid with 0.8% agarose and incubated at 33°C or 37°C in 5% CO₂ cell culture incubators. Three days after infection the cells were stained with 0.1% crystal violet solution or immunostained with chicken anti-MDV-B antiserum, and the number of plaques was determined.

The ts assay was also performed by titration of the virus samples at 33 and 37°C on 96-well plates of PCK cells, and the median infectivity was determined [50% tissue culture infective dose(s) (TCID₅₀)] by examining the cytopathic effect. Serial 10-fold dilutions of the virus samples were prepared in 96-well blocks. The diluted virus samples were then transferred to the washed PCK monolayer in the 96-well plates. At each dilution of the virus sample, replicates of six wells were used for infection with the diluted virus, and the titer of each virus sample was determined in two to four replicates. Each assay included uninfected control cells and a ts control virus. In order to determine the ts phenotype of the virus samples, the plates were incubated for 6 days at 33 and 37°C in 5% CO₂ cell culture incubators, and the virus titers were calculated by the Karber method and reported as the mean (n = 4) log₁₀ TCID₅₀/ml ± the standard deviation. The difference in virus titer at 33 and 37°C was used to determine the ts phenotype. A virus was classified as temperature sensitive (ts) if its titer at 37°C was at least 2 logs (100-fold) lower than its titer at 33°C. The results shown in the figures are derived from at least two replicates.

Attenuation assay in ferrets.

Recombinant viruses obtained after transfection were passaged one time in embryonated chicken eggs to produce a virus stock. Nine-week-old ferrets (Triple F Farms, Syre, PA) were lightly anesthetized with isoflurane and inoculated intranasally by droplets using 0.5 ml per nostril of virus. Three days after infection ferrets were euthanized, and their lungs and nasal turbinates were harvested. A 10% lung tissue suspension was prepared by using OptiMEM I, serially diluted, and 0.1 ml was injected into 10- or 11-day-old embryonated chicken eggs, followed by incubation for 3 days at 33°C. The allantoic fluids were harvested, the presence of virus in the lung tissues was detected by the hemagglutination assay, and the titer was calculated as 50% egg infectious dose per gram of tissue (log₁₀ EID₅₀/g ± the standard deviation). Virus replication in nasal turbinates was determined by plaque assay and expressed as log₁₀ PFU per gram of tissue (log₁₀ PFU/g).

RESULTS

Generation of MDV-B from eight plasmids.

To generate MDV-B entirely from cloned cDNA, either cocultured 293T-MDCK or COS7-MDCK cells were transfected with pAB121-PB1, pAB122-PB2, pAB123-PA, pAB124-HA, pAB125-NP, pAB126-NA, pAB127-M, and pAB128-NS, representing each of the eight gene segments. At 7 days after transfection, supernatants of transfected cells were titrated by plaque assay on MDCK at 33°C. The virus titers in the supernatants of cocultured 293T-MDCK and COS7-MDCK cells were 5.0 × 10⁶ and 7.6 × 10⁶ PFU/ml, respectively. The regions encompassing the five silent nucleotide substitutions were sequenced, confirming that the virus in the supernatant of the infected cells was indeed derived from the plasmids. These results demonstrated that plasmid derived, recombinant MDV-B, designated R1 (rMDV-B), was efficiently recovered after transfection of the eight plasmids.

In order to demonstrate the fidelity of plasmid rescue R1 (rMDV-B) was evaluated for the characteristic ts phenotype by measuring the TCID₅₀ of the virus on PCK cells at two different temperatures. The differences in virus titer between 33 and 37°C for the parental MDV-B and R1 (rMDV-B) were 3.4 and 3.7 log₁₀, respectively (Table 1). Thus, R1 (rMDV-B) expressed the ts phenotype that was indistinguishable from the parental, nonrecombinant MDV-B.

TABLE 1.

Temperature sensitivity phenotype for parental, nonrecombinant MDV-B and R1 (rMDV-B) generated from eight plasmids^a

Virus	Mean log10 TCID50/ml ± SD		Difference (log10)	Phenotype
	33°C	37°C
ca B/Ann Arbor/01/66	8.5 ± 0.1	5.1 ± 0.1	3.4	ts
R1 (rMDV-B)	7.0 ± 0.1	3.3 ± 0.1	3.7	ts

^aPCK cells were infected with the parent virus MDV-B and the plasmid-derived recombinant virus R1 (MDV-B). The virus titer was determined at two different temperatures. A difference of 2.0 log₁₀ or greater indicates expression of the ts phenotype.

Eight amino acid loci in the internal gene segments are unique to MDV-B.

As the first step in elucidating the molecular changes responsible for the characteristic biological properties of MDV-B, including the ts and att phenotypes, the nucleotide and the deduced amino acid sequences of the internal genes of MDV-B were compared to sequences of wt influenza B viruses deposited in the Los Alamos influenza database using the BLAST search algorithm (1, 22). This analysis identified eight amino acids unique to MDV-B that were not present in any other influenza B strain (Table 2); these changes were located in the PA (two changes), NP (4) and M1 (2) open reading frames. In addition, one change in PB2 was observed in only one other wt influenza B virus. Sequence analysis of wt5, a virus derived from the original wt B/Ann Arbor/1/66 by three passages in eggs at 25°C (4-6), revealed that this isolate was dissimilar to MDV-B at seven of these nine amino acid positions (Table 2). The wt5 isolate, which expresses non-ts and non-att phenotypes, was identical to the consensus wt sequences at seven loci, indicating that the progenitor of MDV-B had a genetic constitution reflective of other wt influenza B viruses. These sequence analyses indicated that one or more of the PA, NP, and M gene segments were likely involved in controlling the ts and att phenotypes of MDV-B.

TABLE 2.

Amino acids unique to the internal genes of MDV-B compared to wt influenza B virus sequences^a

Gene segment	No.	Position	ca B/Ann Arbor/1/66		Consensus sequences (wt viruses)		wt5 B/Ann Arbor/1/66		No. of aligned sequences
			Amino acid	Codon	Amino acid	Codon	Amino acid	Codon
PB1	0								23
PB2	0b	630	Arg	AGA	Ser	AGC	Ser	AGC	23
PA	2	431	Met	ATG	Val	GTG	Val	GTG	23
		497	His	CAT	Tyr	TAT	Tyr	TAT
NP	4	55	Ala	GCC	Thr	ACC	Ala	GCC	26
		114	Ala	GCG	Val	GTG	Val	GTG
		410	His	CAT	Pro	CCT	Pro	CCT
		509	Thr	ACT	Ala	GCT	Ala	GCT
M1	2	159	Gln	CAA	His	CAT	Gln	CAA	24
		183	Val	GTG	Met	ATG	Met	ATG
BM2	0								24
NS1	0								80
NS2	0								80

^aThe sequence of the internal gene segments of MDV-B was used in a BLAST search. Amino acid positions different between MDV-B and the aligned sequences are shown. The nucleotides in the codons that are underlined represent the substituted positions.

^bThe residue Arg630 of PB2 did not meet the criteria of unique to MDV-B; this amino acid was present in one wt influenza strain, B/Harbin/7/94 (AAF89735). All other wt strains, including wt5, encoded Ser at this position.

Altering eight amino acid positions of MDV-B reverts the ts phenotype.

A set of plasmids was constructed in which the eight residues of the PA, NP, and M1 genes were changed by site-directed mutagenesis to the consensus wt amino acids and the resulting recombinant viruses are listed in Table 3. A recombinant with all eight consensus wt changes, designated R8, was generated by cotransfection of the appropriate plasmids onto cocultured COS7-MDCK cells.

TABLE 3.

Genetic composition of recombinant viruses

Designation	Amino acida
	PA		NP				M		PB2
	431	497	55	114	410	509	159	183	630
MDV-B	M	H	A	A	H	T	Q	V	R
R1 (MDV-B)	M	H	A	A	H	T	Q	V	R
R2 (WT PA)	V	Y	A	A	H	T	Q	V	R
R3 (WT NP)	M	H	T	V	P	A	Q	V	R
R4 (WT M)	M	H	A	A	H	T	H	M	R
R5 (WT PA/M)	V	Y	A	A	H	T	H	M	R
R6 (WT NP/M)	M	H	T	V	P	A	H	M	R
R7 (WT PA/NP)	V	Y	T	V	P	A	Q	V	R
R8 (WT PA/NP/M)	V	Y	T	V	P	A	H	M	R
R9	V	Y	T	A	H	T	Q	V	R
R10	V	Y	A	V	H	T	Q	V	R
R11	V	Y	A	A	P	T	Q	V	R
R12	V	Y	A	A	H	A	Q	V	R
R13	V	Y	A	V	P	T	Q	V	R
R14	V	H	T	V	P	A	H	M	R
R15	M	Y	T	V	P	A	H	M	R
R16	V	H	A	V	P	T	Q	V	R
R17	V	H	A	V	P	T	H	V	R
R18	V	H	A	V	P	T	Q	M	R
R19	V	H	A	V	P	T	H	M	R
R20	V	Y	T	V	P	A	H	M	S

^aThe amino acid at the indicated position is given for each gene segment. Wild-type residues are indicated in boldface.

In order to characterize the effect of these genetic changes on temperature-dependent replication, viruses in the supernatants of the transfections were titrated on MDCK cells by plaque assay and PCK cells by TCID₅₀ at 33 and 37°C. As shown in Fig. 1, R1 (rMDV-B) expressed the ts phenotype in both MDCK and PCK cells, the observed difference in titer between the two temperatures was >3.0 log₁₀. In contrast, introduction of all eight wt amino acids into MDV-B reverted the ts phenotype; the difference in titer between 33 and 37°C for R8 was ≤1.0 log₁₀ in both PCK and MDCK cells (Fig. 1). The R8 recombinant expressed the non-ts phenotype (Fig. 1) and, in sharp contrast to R1 (rMDV-B), R8 produced large and clear plaques at 37°C (Fig. 2). These results showed that alteration of the eight amino acids within the PA, NP, and M gene segments was sufficient to revert the ts phenotype and generate a non-ts virus.

FIG. 1.

Temperature sensitivity of recombinant viruses. Recombinant viruses were generated by cotransfection of eight plasmids. R1 (rMDV-B) was constructed by transfection of eight plasmids corresponding to the consensus sequence of ca B/Ann Arbor/1/66. The gene segment listed below each set of bars denotes the recombinant virus designation (see Table 3) and wt gene segment(s) in the recombinant virus containing wt consensus nucleotide replacements. The colored bars beneath each label represent the derivation of the PA, NP, and M gene segments in each recombinant (red, wt residues; blue, MDV-B). Each recombinant virus was titrated in PCK cells (dark gray) by TCID₅₀ and in MDCK cells (light gray) by plaque assay at 33 and 37°C. The difference in the log₁₀ titer at these two temperatures (titer log₁₀ 33°C-titer log₁₀ 37°C) is displayed on the y axis. The dotted horizontal line represents a difference in which the titer at 33°C is 2.0 log₁₀ greater than the titer at 37°C. The error bars represent ± the standard deviation from the mean for the TCID₅₀ assays; the light gray bars (plaque assay) represent a typical experiment.

FIG. 2.

Plaque morphology of recombinant MDV-B viruses. Recombinant viruses with wt amino acids in NP, PA, and M were plated on MDCK cells and incubated at either 33 or 37°C for 3 days; the cell monolayer was fixed and immunostained with anti-MDV-B sera. Refer to Table 3 for the designation of the recombinant viruses.

Three of the eight amino acid differences are required to express the ts phenotype.

To identify the gene segments that controlled the ts phenotype, a series of recombinants were constructed. Recombinant viruses harboring only one of the wt gene segments were generated, and all of these recombinants exhibited growth restriction at 37°C in MDCK cells and in PCK cells (Fig. 1), indicating that wt amino acid changes in any one gene segment were not sufficient to revert the ts phenotype.

Recombinant viruses containing multiple wt gene segments were constructed to define which combination of segments could revert the ts phenotype. Recombinant viruses that carried both the wt NP and wt M (R6) gene segments also retained the ts phenotype (Fig. 1). These data demonstrated that the MDV-B PA gene segment was sufficient to express the ts phenotype even in the presence of two other wt gene segments. Similarly, the R5 recombinant harboring both the wt PA and wt M gene segments also expressed the ts phenotype, indicating the MDV-B NP gene segment was also sufficient for expression of the ts phenotype.

Only the combination of wt PA and wt NP gene segments together resulted in reversion of the ts phenotype. Both recombinants R7 and R8 were no longer temperature sensitive in either PCK or MDCK cells. The R8 recombinant virus containing the wt PA NP and M gene segments was less restricted at 37°C in PCK cells (Fig. 1) and had a slightly larger plaque size on MDCK cells (Fig. 2) than R7, indicating that the M gene segment may have had some impact on expression of the ts phenotype. The results of these studies demonstrated that the PA and NP gene segments were the major controlling elements of the ts phenotype. The MDV-B PA and NP segments could independently confer the ts phenotype, and a combination of wt amino acids in both these segments was required for reversion of this phenotype.

To determine the contribution of each residue of the four amino acids in the NP protein and two in the PA protein to the ts phenotype, a detailed mutational analysis of the PA and NP gene segments was performed (Fig. 3). A set of recombinant viruses was constructed in which each virus contained a PA gene with the two wt consensus amino acids and one of the four wt amino acids in the NP gene. Recombinant viruses R9 and R12 with single wt amino acids in NP, T55 or A509, respectively, continued to express the ts phenotype, indicating that these amino acid positions had little role in controlling expression of this phenotype. In contrast, both R10 and R11 containing single wt residues in NP, V114 and P410, resulted in a non-ts recombinant, indicating that these two amino acid positions were critical for control of the ts phenotype. A recombinant virus (R13) which combined both of these changes (V114 and P410) along with the two wt residues in PA also expressed the ts phenotype.

FIG. 3.

Contribution of amino acid residues in NP and PA for temperature sensitivity. Viruses were generated by cotransfection of six MDV-B plasmids, together with PA and NP plasmids with the indicated amino acid changes. Gray boxes represent wt amino acid residues present in each recombinant. The TCID₅₀ of the different recombinants was measured at both 33 and 37°C and the difference between the log₁₀ values of these titers was plotted on the y axis. The dotted horizontal line represents a difference in which the titer at 33°C is 2.0 log₁₀ greater than the titer at 37°C. The error bars represent ± the standard deviation from the mean for the TCID₅₀ assays.

A similar strategy was used to dissect the contribution of the two amino acids in the PA gene. A set of recombinants was constructed, each harboring an NP gene segment with four wt consensus amino acids and a PA gene with only one of the two consensus wt amino acids. Substitution of H497 with Y497 (R15) remained ts (Fig. 3), demonstrating that this locus had little impact on expression of the phenotype. In contrast, substitution of M431 with V431 (R14) resulted in reversion of the ts phenotype. These results show that amino acids A114 and H410 in MDV-B NP and M431 in MDV-B PA are the major determinants controlling temperature sensitivity of MDV-B.

In order to confirm that these three residues are critical in controlling the expression of the ts phenotype, recombinant R16 was constructed containing only PA (V431) and NP (V114 and P410) in MDV-B. This recombinant replicated efficiently at both 33 and 37°C; the difference in titer was 0.3 ± 0.1 log₁₀ PFU/ml on MDCK cells. Therefore, R16 did not express the ts phenotype. These data demonstrated that these three residues were the minimal requirements for reverting the ts phenotype of MDV-B.

The att phenotype of MDV-B requires loci in addition to those for expression of ts.

MDV-B and its 6:2 derivatives are not detected in lung tissues of ferrets after infection by the intranasal route, whereas nonattenuated influenza B viruses are detected in lungs and may occasionally cause other signs of influenza-like illness in these animals (24). To determine whether the eight wt amino acid changes had an impact on the att phenotype, ferrets were infected intranasally with R1 (rMDV-B) or R8 (wt PA/NP/M). Three days after infection, lung tissues were harvested, and virus was detected by using an egg infectivity assay. As expected, no virus was detected from the lung tissues of ferrets infected with R1. In contrast, the lungs of animals infected with R8 contained approximately 3 log₁₀ EID₅₀/g virus (Table 4). Thus, changing the eight unique amino acids in PA, NP, and M gene segments to wt consensus residues was sufficient to revert the att phenotype to a non-att phenotype.

TABLE 4.

Attenuation studies of recombinant viruses in ferrets

Study no.a	Recombinant virus	Phenotype	Nasal turbinatesb (log10 PFU/g)	Lung tissue (log10 EID50/g ± SD) (n/N)c
A	R1 (MDV-B)	ts	4.0	<1.5 (0/4)
	R3 (WT NP)	ts	4.1	<1.5 (0/4)
	R4 (WT M)	ts	4.2	<1.5 (0/4)
	R7 (WT NP/PA)	Non-ts	4.7	<1.5 (0/4)
	R8 (WT NP/PA/M)	Non-ts	4.7	2.8 ± 0.6 (4/4)
B	R16	Non-ts	5.0	<1.5 (0/4)
	R17	Non-ts	5.3	2.2 ± 0.9 (2/4)
	R18	Non-ts	5.7	2.8 ± 0.5 (3/4)
	R19	Non-ts	6.2	2.3 ± 0.6 (4/4)
	R20	Non-ts	6.1	3.6 ± 0.4 (4/4)
A	recYam	Non-ts	4.9	3.7 ± 1.1 (4/4)
	rec6Yam	ts	4.0	<1.5

^aGroups of four ferrets were inoculated with 6 log₁₀ PFU/ml (study A) or 7 log₁₀ PFU/ml (study B) Recombinant virus designations are listed in Table 3.

^bThree days after infection virus titers of the nasal turbinates and lung tissues were determined; the average titer is shown.

^c“<1.5” indicates that no virus was detected. n/N denotes number of animals with virus detected in the lung tissues (n) compared to the total number of inoculated animals (N).

In order to determine the minimal number of gene segments required to revert the att phenotype, ferrets were infected with recombinants R3 (wt NP) and R4 (wt M) or R7 (wt PA and NP). Virus was recovered from the nasal turbinates of all of the animals, indicating that each animal had been successfully infected; however, in contrast to R8, virus was not detected in the lung tissues (Table 4). These data demonstrated that none of the single gene segments was capable of reverting the att phenotype, neither was the combination of PA and NP together. All three gene segments—PA, NP, and M—required the presence of wt amino acid residues to revert the att phenotype

To further define the minimal number of changes required to revert the att phenotype, a second study was performed with recombinant viruses containing subsets of wt residues in the PA, NP, and M gene segments. Recombinant R16 contained one wt residue in PA (V431) and two wt residues in NP (V114 and P410); these three changes together were sufficient to revert the ts phenotype. As expected from the previous experiment, R16 was not detected in the lung tissues of ferrets 3 days after infection. Recombinant viruses R17 and R18 were constructed such that each contained one of the wt amino acids found in M in addition to the changes in R16. Both R17 and R18 were detected in the lung tissues of ferrets; however, only two or three of the animals had detectable virus in the lung, respectively. Recombinant R19, which contained both wt M amino acids (H159 and M183) in addition to PA (V431) and NP (V114 and P410), was detected in the lung tissues of all four inoculated animals (Table 4) and replicated to levels only slightly lower than R20 that contained all nine wt amino acids.

Transferring the genetic signature for att to a heterologous influenza strain.

Our results with the MDV-B backbone indicated that five amino acids in PA (V431), NP (V114 and P410), and M (H159 and M183) were sufficient to revert both the ts and att phenotypes. In order to determine whether changes at these loci could transfer the biological traits to a heterologous nonattenuated wt influenza B virus, the five MDV-B amino acids PA (M431), NP (A114 and H410), and M (Q159 and V183) plus an additional residue in PA (H497) were introduced into wt B/Yamanashi/166/98.

Recombinant wt B/Yamanashi/166/98 (recYam) (8) replicated well at both 37 and 33°C; the difference in titer was 0.17 log₁₀ between these two temperatures. In contrast, rec6Yam, the derivative recombinant virus containing the six MDV-B amino acids, was clearly ts: the difference in virus titer between 37 and 33°C was 4.6 log₁₀. Both viruses replicated efficiently in the nasal turbinates of ferrets; however, only wt recombinant recYam was recovered from the lung tissues (Table 4). Thus, introducing the ts/att loci from MDV-B into a divergent wt strain was sufficient to transfer the ts and att phenotypes.

DISCUSSION

These studies demonstrated that the PA and NP gene segments control the expression of the ts phenotype of MDV-B. Previously, the PA M431 amino acid of MDV-B had been shown to contribute to the control of the ts phenotype, but those studies concluded that other residues were also likely to contribute to expression and reversion of this trait (5, 6). The findings reported here show that three amino acid loci of MDV-B—PA M431, NP A114, and NP H410—were the minimal elements that controlled the ts phenotype, and reversion of all three loci in combination was required to revert the ts phenotype. Although the exact mechanistic roles of PA and NP have not been fully elucidated, there are multiple stages in viral replication that depend on either PA or NP proteins, including those mediated by the viral RNA polymerase complex such as vRNA→mRNA (transcription), vRNA→cRNA (first step in replication), and cRNA→vRNA (second step in replication) or the switch between transcription and replication (18, 26). Since the PA and NP segments were shown to independently impart the ts phenotype, viral replication may be blocked at several different points of the replication cycle at 37°C. Restriction of activity at multiple points may explain the several order of magnitude reduction in replication at this higher temperature.

Ferrets have body temperatures ranging from 38 to 40°C. Although the inability to replicate at the higher temperatures certainly may contribute to expression of the att phenotype, this defect may not be the sole mechanism for restriction of replication in the lung tissues of these animals. In order to revert the att phenotype, the three changes required to revert the ts phenotype were required in addition to at least one change in the M gene segment. Both changes in the M gene segment encoded changes in the M1 protein. The M1 protein has many different motifs, including a highly conserved zinc finger, and multiple activities during viral replication and packaging (2, 7, 12, 19, 25, 32). In addition, the M1 and BM2 products of the influenza B M gene segment are critical for efficient budding of virus particles at the cell membrane (10), the presence of Q159 or V183 in the M1 protein may result in a less efficient production of progeny viruses, resulting in a less efficient spread of virus in the infected host; however, the mechanism for the M gene's contribution to the att phenotype remains to be investigated.

Studies from a variety of negative-sense RNA viruses demonstrate that the genes important for replication and/or transcription can be associated with temperature sensitivity and attenuation. Several previous studies and a recent study using recombinant derivatives of the FluMist influenza A MDV (MDV-A) have mapped the amino acids responsible for expression of ts to five sites: three in PB1, one in PB2, and one in NP (11). For RNA viruses with nonsegmented genomes, such as human parainfluenza virus and respiratory syncytial virus, attenuating lesions were found in the phosphoprotein P and the large polymerase L (13-15, 20, 30). Both proteins are components of the viral RNA polymerase. The precise mapping of the specific residues in PA and NP for MDV-B may now facilitate the elucidation of the underlying molecular mechanisms of those temperature-dependent defects.

The genetic complexity of these MDV-B phenotypes potentially reveals the mechanism of genetic stability of the ts and att phenotypes. Studies in immunocompromised hamsters, as well as seronegative chimpanzees and humans, have shown that the ts and att phenotypes of MDV-B and its 6:2 derivatives are stable after many rounds of replication in the host (24, 28). In addition, the uniqueness of these amino acid changes in MDV-B and the observed stability may have resulted from the constant selection of viruses that replicated efficiently as the growth temperatures were lowered. Further genetic and biochemical analyses could elucidate the mechanism of action of these vaccine strains. The results reported here demonstrated that expression of the att phenotype is more complex and partially separable from ts.

Transferring the genetic loci responsible for ts and att of MDV-B to the non-ts, non-att recombinant wt B/Yamanashi/166/98 resulted in transferring the att and ts biological traits to the resulting recombinant. These data confirm that the minimal changes required to express the ts and att phenotypes reside within these loci. The genomes of ca B/Ann Arbor/1/66 and B/Yamanashi/166/98 are 96% identical and nucleotide alignments of these two viruses revealed >350 nucleotide differences (>50 amino acid differences) between these two strains. Nevertheless, introduction of the MDV-B ts and att loci was tolerated and conferred the appropriate biological properties.

The sequence analysis strategy used here to find unique residues in influenza type B sequences important for ts and att should be helpful to find yet-unknown att/ts residues in other attenuated viruses, such as ca A/Leningrad/134/57 or B/USSR/69 (16, 17, 27). It is noteworthy that the two wt residues (NP P410 and M1 H159) are conserved within both wt influenza type A and B viruses. The conservation of these residues across this evolutionary distance (∼30 to 40% amino acid identity between type A and B proteins) may underline the potential importance of these residues for protein structure and function. Our results demonstrate the utility of reverse genetics in combination with bioinformatics tools to find and characterize the effect of specific residues for the growth of influenza B virus. This genome-wide analysis is important because most of the determinants for biological properties of influenza A and B viruses are multigenic. Thus, this approach provides not only a powerful framework for understanding and potentially enhancing master donor viruses but can also be used for detecting and testing residues of all eight segments important for virus growth, pathogenesis, and transmission.