Avian Influenza A Virus Genes in Attenuation of Wild-Type

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Mar. 1992, p /92/ $02.00/0 Copyright C) 1992, American Society for Microbiology Vol. 30, No. 3 Use of Single-Gene Reassortant Viruses To Study the Role of Avian Influenza A Virus Genes in Attenuation of Wild-Type Human Influenza A Virus for Squirrel Monkeys and Adult Human Volunteers MARY LOU CLEMENTS,l 2* E. KANTA SUBBARAO,3 LOUIS F. FRIES,"2 RUTH A. KARRON,"4 WILLIAM T. LONDON,5 AND BRIAN R. MURPHY3 Center for Immunization Research and Division of Vaccine Sciences, Department of International Health, Johns Hopkins University School of Hygiene and Public Health, 1 and Departments of Medicine2 and Pediatrics, 4 Johns Hopkins University School of Medicine, Baltimore, Maryland 21205; Respiratory Viruses Section, Laboratory of Infectious Diseases, National Institute ofallergy and Infectious Diseases, Bethesda, Maryland ; and Retroviral Pathogenesis Section, Section of Molecular Virology and Immunology, Georgetown University, Rockville, Maryland Received 28 August 1991/Accepted 19 December 1991 The transfer of six internal RNA segments from the avian influenza A/Mallard/New York/6750/78 (H2N2) virus reproducibly attenuates human influenza A viruses for squirrel monkeys and adult humans. To identify the avian influenza A virus genes that specify the attenuation and host range restriction of avian-human (ah) influenza A reassortant viruses (referred to as ah reassortants), we isolated six single-gene reassortant viruses (SGRs), each having a single internal RNA segment of the influenza A/Mallard/New York/6750/78 virus and seven RNA segments from the human influenza A/Los Angeles/2/87 (H3N2) wild-type virus. To assess the level of attenuation, we compared each SGR with the A/Los Angeles/2/87 wild-type virus and a 6-2 gene ah reassortant (having six internal RNA segments from the avian influenza A virus parent and two genes encoding the hemagglutinin and neuraminidase glycoproteins from the wild-type human influenza A virus) for the ability to replicate in seronegative squirrel monkeys and adult human volunteers. In monkeys and humans, replication of the 6-2 gene ah reassortant was highly restricted. In humans, the NS, M, PB2, and PB1 SGRs each replicated significantly less efficiently (P < 0.05) than the wild-type human influenza A virus parent, suggesting that each of these genes contributes to the attenuation phenotype. In monkeys, only the NP, PB2, and possibly the M genes contributed to the attenuation phenotype. These discordant observations, particularly with regard to the NP SGR, indicate that not all genetic determinants of attenuation of influenza A viruses for humans can be identified during studies of SGRs conducted with monkeys. The PB2 and M SGRs that were attenuated in humans each exhibited a new phenotype that was not observed for either parental virus. Thus, it was not possible to determine whether the avian influenza virus PB2 or M gene itself or a specific constellation of avian and human influenza A virus genes specified restriction of virus replication in humans. The genes of avian influenza A viruses have nucleotide sequences that are significantly different from those of the corresponding genes of human influenza A viruses (1, 24). The avian strains replicate efficiently in avian hosts but are moderately restricted in their replication in the lower respiratory tracts of nonavian hosts, such as monkeys, and as a consequence they are attenuated in primates (14, 15). Thus, these viruses possess the properties of host range restriction and attenuation. The A/Mallard/New York/6750/78 (A/Mallard/NY/78) (H2N2) virus, which possesses these phenotypes, has been used as a donor of genes to produce live attenuated avian-human (ah) influenza A reassortant viruses (referred to as ah reassortants) (16). These candidate vaccines contain the two genes encoding the hemagglutinin (HA) and neuraminidase (NA) surface glycoproteins of a human influenza A wild-type virus, and the remaining genes (i.e., six internal RNA segments) are derived from the avian influenza A donor virus. Clinical studies have demonstrated that the transfer of the six internal RNA segments of influenza A/Mallard/NY/78 virus can reliably attenuate vir- * Corresponding author. ulent human influenza A viruses for susceptible monkeys and adult humans (2, 9, 11, 13, 16, 25, 28, 29). To study the genetic determinants of the avian influenza virus that specify the host range restriction of replication and the attenuation phenotypes of the ah reassortant viruses, we isolated single gene substitution reassortant viruses, referred to as SGRs. Each of these SGRs acquired only one RNA segment from the influenza A/Mallard/NY/78 donor virus and the remaining RNA segments from a human influenza A wild-type virus. We had previously isolated PB1, PA, NP, M, and NS SGRs from the influenza A/Mallard/NY/78 and human influenza A/Udorn/72 parent viruses and demonstrated that the NP and M SGRs were attenuated for squirrel monkeys (33). In the present study, we generated a second set of six SGRs from the A/Mallard/NY/78 (H2N2) and human influenza A/Los Angeles/2/87 (H3N2) parent viruses for two reasons. First, in the earlier study an A/Udorn/72 SGR containing the PB2 gene of the A/Mallard/NY/78 parent was not obtained and we sought to isolate this reassortant in the current experiment. Second, we wished to compare the levels of replication of the SGRs in both monkeys and humans to determine how accurately the response of the monkeys reflects the human experience with these SGRs 655

2 656 CLEMENTS ET AL. and, for the human studies, we needed a set of six SGRs that contained HA and NA glycoproteins from a recent epidemic strain of influenza A virus in order to perform studies in immunologically susceptible adult volunteers. Therefore, we were able to compare the levels of replication of each SGR, the 6-2 gene ah influenza A/Los Angeles/2/87 x A/Mallard/ NY/78 reassortant virus, and the influenza A/Los Angeles/ 2/87 wild-type virus parent in seronegative (having a hemagglutination inhibition (HAI) titer c 1:8) squirrel monkeys and adult human volunteers. a C: a 0 x an f- A X~ X X PBA- 0 -J : z : _3 PJA- -& :N.:i J. CLIN. MICROBIOL. Xt co 0n < z EL -J to* -P82 -PB _o -PA MATERLALS AND METHODS NP Viruses. The passage history of the avian influenza A/Mallard/NY/78 (H2N2) virus parent has been described previously (14). The human influenza A/Los Angeles/2/87 (H3N2) wild-type virus was initially isolated by H. F. Maassab (University of Michigan) in primary chick kidney (PCK) cell cultures from a throat wash obtained from an adult female with influenza illness. This wild-type isolate was then passaged four times in the allantoic cavities of avian leukosis virus-free embryonated hens' eggs (SPAFAS, Inc., Norwich, Conn.) at 35 C and was then plaque purified two times in PCK cells at 39 C. The clone (clone 3) was amplified by passage in SPAFAS eggs at 35 C and was used as the seed virus to produce lot E-272, the wild-type virus suspension used as the parent for production of reassortants as described below and for studies with monkeys and humans. The avian influenza A virus parent was mated with the human influenza A wild-type virus parent at a multiplicity of infection of approximately 1 during incubation at 37 C for 24 h. The A/Los Angeles/2/87 x A/Mallard/NY/ gene ah reassortant virus (clone 6-1-2, lot E-276) was then selected by passage at 39 C in the presence of antiserum against the HA and NA of the avian virus parent and biologically cloned by plaque-to-plaque passage as previously described (33). From this mating, a reassortant containing the avian PB2, PA, NP, and NS genes was derived, and this was backcrossed with the A/Los Angeles/2/87 wild-type virus to yield NS and PA SGRs. The PB1, NP, and M SGRs were produced by coinfecting PCK cell cultures with the A/Los Angeles/2/87 wild-type virus and the 6-2 gene ah A/Mallard/ NY/78 x A/Los Angeles/2/87 reassortant virus at 37 C at a multiplicity of infection of 10. Production of the PB2 SGR from the mating of the influenza A/Los Angeles/2/87 wildtype virus with the 6-2 gene ah reassortant virus required an additional step in which a clone of the progeny virus which possessed a mixed constellation of internal genes was backcrossed with the wild-type parent virus at a multiplicity of infection of 10. Each SGR was biologically cloned by plaqueto-plaque passage or by terminal dilution (M SGR) in PCK monolayer cultures. The final virus suspensions administered to volunteers were grown in 9-day-old specific-pathogen-free eggs (SPA- FAS, Inc.) and tested for the presence of adventitious agents by Louis Potash (Flow Laboratories, Inc., McLean, Va.). The virus suspensions administered to volunteers in this study and their infectivity titers in Madin-Darby canine kidney (MDCK) monolayers were as follows: wild-type influenza A/Los Angeles/72 virus (E-272), /ml 50% tissue culture infective doses (TCID50) per ml; PB1 SGR (E-286), TCID50 per ml; PA SGR (E-277), TCID50 per ml; NP SGR (E-281), TCID50 per ml; M SGR (E-285), TCID50 per ml; NS SGR (E-280), TCID50 per ml. In each case, the SGRs are designated by the gene derived from the avian influenza A donor virus, and E-272, NS- _ < a: C: 3j ( O X U) cn i 2 m m 0 -r ti a. i _ ^ _ ^ ^ _u D C 0 Ocr) (N a a: X U) X en < < z -J. &,x, -NS A. _ -HA SNA I -NP FIG. 1. Polyacrylamide gel electrophoresis of virion RNAs from human A/Los Angeles/2/87 (LA wt), avian A/Mallard/NY/78 x A/Los Angeles/2/87 (6-2) reassortant with six internal genes derived from A/Mallard/NY/78, and the HA and NA genes derived from A/Los Angeles/2/87 and SGRs with the indicated gene derived from the avian virus parent and the remaining genes derived from the human virus parent. The 16-cm gels contained 2.6% acrylamide and either 5.5 M (A) or 6.0 M (B) urea. Electrophoresis was carried out for 18 to 24 h at 0 C and 100 V. The gene assignments on the left side of the gels indicate the positions of the avian (Mal) virus genes, and those on the right indicate the positions of the human (LA wt) virus genes. E-280, etc., represent lot designations. The infectivity titer of the PB2 SGR (E-287) in embryonated eggs was % egg infective doses per ml, and that of the 6-2 gene ah reassortant (E-276) in PCK cells was TCID50/ml. Genotype of reassortant virus. The parental origins of the RNA segments of each reassortant virus were determined by comparison of their migration in polyacrylamide gel electrophoresis with that of the corresponding parental genes. Viruses were propagated and purified and RNA was extracted as previously described (10). Purified viral RNA was analyzed by electrophoresis for 18 to 24 h at 100 V and at 0 C with polyacrylamide gels containing 2.6% acrylamide and 6.0 or 5.5 M urea (Fig. 1). RNA segments were visualized by ammoniacal silver staining (24). Sequence analysis. The nucleotide sequence of the polymerase genes was determined by the dideoxynucleotide chain termination technique (23) by using synthetic oligonucleotide primers and genomic virion RNA as described previously (3). Oligonucleotides were synthesized on an Applied Biosystems model 380A DNA synthesizer. The

3 VOL. 30, 1992 GENES THAT ATTENUATE AVIAN-HUMAN INFLUENZA A VIRUSES 657 following primers were used to sequence the genes of A/ Mallard/NY/78 x A/Los Angeles/2/87 PB2 SGR: (i) PB2 5'- nucleotide [nt] 2206-GGGCAAGGAGACGTG and 5'-nt 1- ATTCAATATGGAGAGAATAAAGGA and (ii) PB1 5'-nt 1595-GCATTGGAGTAACAG. The following additional primers were used to sequence the genes of the A/Pintail/ Alberta/119/79 x AlWashington/897/80 reassortant clone 224: (i) PB2 5'-nt 530-GT'l'TCCCAAATGAA, (ii) PB1 5'-nt 946-AAATGGAATGAAAAT, and (iii) PA 5'-nt 721-GTGG ATGGATTCGAA. Studies with squirrel monkeys. The duration and level of replication of the parental viruses and their reassortant progeny in squirrel monkeys were evaluated as previously described (29, 33). Briefly, 32 monkeys housed in Horsfal isolation units were inoculated intratracheally with 107 TCID50 of virus in a 0.5-ml inoculum. Nasopharyngeal swab specimens were obtained daily for 10 days postinoculation, and tracheal lavage fluids were obtained on days 2, 4, 6, and 8 postinoculation. Aliquots of these specimens were frozen for subsequent titration on MDCK tissue cultures or in eggs. Two separate studies with 16 monkeys each were performed. In each study, two monkeys received human influenza A/Los Angeles/2/87 wild-type virus, two received the 6-2 gene ah reassortant virus, and three groups of four monkeys each received one of three SGRs. The mean duration and peak titer of virus shedding were determined for each group of four monkeys which received an SGR virus, and these values were compared with those obtained from each group of four monkeys which received either the wild-type human influenza A virus or the 6-2 gene ah reassortant virus. In one of these studies, the nasopharyngeal swab fluids from animals given the PB2 SGR were titrated in embryonated hens' eggs and MDCK cell culture. We found that the 50% egg infective dose in eggs was quite similar to the TCID50 estimated in MDCK cells (data not shown) for viruses capable of replicating in both substrates. Studies with seronegative adult human volunteers. Study protocols were approved by the institutional review boards of Towson State University and the Johns Hopkins Medical Institutions. Healthy adults between the ages of 18 and 40 years who were not taking medication, did not have a history of influenza vaccination, and had an HAI antibody titer in serum of 1:8 or less were recruited from community members in Baltimore, Md. A total of eight inpatient studies were performed. In these studies, the control groups which received wild-type virus were studied concurrently in a double-blind fashion with the groups which received the SGR viruses. Volunteers who received a wild-type or SGR virus were inoculated intranasally with 107 TCID50 of virus in a 0.5-ml inoculum and were housed in the Center for Immunization Research isolation unit at Francis Scott Key Medical Center, where they were observed daily for 10 consecutive days after inoculation. We also conducted two outpatient studies in which volunteers were inoculated intranasally with TCID50 of the 6-2 gene ah influenza A/Los Angeles/2/87 reassortant virus in a 0.5-ml inoculum. Outpatient volunteers recorded their vital signs four times per day and were observed once daily for 4 to 7 days. The criteria used to define illness in volunteers are stated in Table 3, footnote e. Laboratory assays. HAI antibodies in serum were measured as previously described (18), with influenza A/Los Angeles/2/87 (H3N2) virus used as antigen. A fourfold increase in titer was considered significant. Serum immunoglobulin G (IgG) and nasal wash IgA antibodies to influenza A/Los Angeles/2/87 virus were also measured by kinetic enzyme-linked immunosorbent assay (KELISA) by using a modification of the method of Snyder et al. (26) as previously described (20, 30). Sera were studied at a final dilution of 1:4,000, and preconcentrated nasal wash specimens were studied at a dilution of 1:8; these dilutions gave KELISA color development rates well within the dynamic range of the assay. Nasal wash specimens were also subjected to protein determinations by using a microplate modification of the BCA assay (Pierce Chemical Co., Rockford, Ill.) (31), and IgA KELISA rate results were normalized to an arbitrary protein content of 350,ug/ml (approximately the median value for all specimens) by using the following formula: normalized KELISA rate = specimen KELISA rate x (350/specimen protein concentration). We defined significant responses in the serum IgG KELISA by simultaneously studying the screening serum specimen which established volunteer eligibility and the immediate preinoculation serum specimen. These specimens covered an interval of 1 to 2 months outside the influenza virus epidemic season and therefore reflect short-term variability in influenza A/Los Angeles/2/87 virus KELISA reactivity in a given individual not exposed to antigen. For each subject, a ratio of the two results was derived; the ratios were transformed to logarithmic values. The mean plus two standard deviations were calculated, and the antilog of this value (1.23) was set as the upper limit of the ratio between a pre- and postinoculation specimen above which a significant response was said to have occurred. Lacking a screening nasal wash specimen, we generated an upper limit for the nonspecific variability of normalized IgA responses by a method similar to that outlined for sera by Snyder et al. (26). All possible normalized values, based on duplicate protein and triplicate KELISA rate determinations, were generated for each specimen. The distribution of possible ratios between these values was then characterized as above, and an upper limit (1.80) for ratios between repeated determinations on the same specimen was calculated. Increases of more than 1.8-fold between pre- and postinoculation normalized nasal wash IgA KELISA rates were thus considered significant. Nasal wash specimens for virologic study were collected before virus inoculation and daily for the duration of observation. The method used for virus inoculation and isolation from MDCK cells was published previously (11). For monkeys and humans administered with the PB2 SGR, specimens for virus quantitation were inoculated into 9-day-old embryonated eggs (Truslow Farms, Chestertown, Md.). After incubation at 32 C for 3 days, the allantoic fluid was tested for hemagglutination activity. Isolation of virus from nasal wash specimens and/or a significant increase in influenza virus-specific antibody titer was considered evidence of infection. Statistical methods. Data from all infected subjects were used in the calculations of mean duration and mean peak titer of virus shedding. Comparisons of group data were performed by using the Mann-Whitney U test, and Fisher's exact test was used for comparison of proportions; all tests were two tailed. RESULTS Isolation, genotype, and phenotype of reassortant viruses. Reassortant viruses, each of which acquired one or more RNA segments from the avian influenza A/Mallard/NY/78 donor virus, were isolated from PCK cells after coinfection

4 658 CLEMENTS ET AL. J. CLIN. MICROBIOL. TABLE 1. Nucleotide sequence homology between polymerase genes of SGR influenza viruses and their avian or human influenza A virus parents Comparison with avian Comparison with human Reassortant virus Gene virus parent virus parent Parental origin sequenced analyzeda ~~~~~~~~~~~~~~~~~of gene in No. of bases % No. of bases % reassortant analyzed Homology analyzed Homology A/Mallard/NY/78 x A/Los Angeles/2/87, PB2 SGR PB Avian A/Mallard/NY/78 x A/Los Angeles/2/87, PB2 SGR PB Human A/Pintail/Alberta/119/79 x A/Washington/897/80, clone 224 PB Avian A/Pintail/Alberta/119/79 x A/Washington/897/80, clone 224 PB Human A/Pintail/Alberta/119/79 x A/Washington/897/80, clone 224 PA Human with the human influenza A/Los Angeles/2/87 wild-type virus and the six-gene ah reassortant or with its reassortant progeny. In the present study, all matings and passages of reassortant viruses were performed in PCK tissue culture for two reasons. First, PCK derived from specific-pathogen-free chickens is an acceptable substrate for viruses that are to be administered to humans. Second, in our previous mating of the avian influenza A/Mallard/NY/78 x human influenza A/Udorn/72 viruses, the cells used for the matings and further passaging were of a mammalian cell line, MDCK. Since that time, we have isolated ah reassortants that are restricted in replication in MDCK tissue culture cells but replicate efficiently in PCK tissue culture cells (27). It was possible that our failure to isolate a PB2 SGR in the previous mating occurred because this reassortant would not grow in MDCK cells. This indeed turned out to be the case (see below), and by using PCK cells, we were able to isolate a full set of six SGRs (Fig. 1). We confirmed that the PB2 SGR had received only the PB2 gene from the avian virus parent by sequence analysis during which the nucleotide sequences of the PB1 and PB2 genes of the SGR were compared with the sequences of the same genes from the avian and human influenza A virus parents (Table 1). We also sought to confirm the origin of the polymerase genes in clone 224 of the A/Pintail/Alberta/119/79 x A/Washington/897/80 mating (27), which also has a host range phenotype of restricted replication identical to that of the PB2 SGR isolated in the present study. We previously reported that clone 224 had received the avian PB1 gene and the human PB2 and PA genes (27). Because of the similar phenotype of the A/Mallard/NY/78 x A/Los Angeles/2/87 PB2 SGR and clone 224, we sought to confirm the genotype of this virus. In contrast to our previous observation, we found that the PB2 gene of clone 224 was derived from the avian virus parent and the PB1 and PA genes were derived from the human virus parent (Table 1). The sequence of A/Mallard/NY/78 and A/Pintail/Alberta/119/79 PB2 genes was 6.2% different over the limited area sequenced, indicating that the two avian influenza virus PB2 genes are indeed different. Thus, two distinct avian influenza virus PB2 genes in the context of polymerase genes from two different human influenza H3N2 wild-type viruses specified the same host range restriction of replication, suggesting that this might be a more general phenomenon than previously thought. Tissue culture growth phenotypes of the parent viruses and the SGRs were compared in MDCK and PCK monolayer cultures. The SGRs containing the PB2 gene and the M gene each demonstrated new phenotypes in vitro that had not been observed with either the 6-2 gene ah reassortant or the wild-type human virus parent. The wild-type virus, the 6-2 gene ah reassortant virus, and each SGR, except the PB2 SGR, formed plaques with high efficiency on MDCK tissue culture. The SGR containing the PB2 gene from the avian parent virus failed to replicate in MDCK tissue (HA titer, <1:2) but replicated to a high titer in PCK cells or eggs (HA titer, 1:128), like the 6-2 gene ah reassortant and the wildtype human virus parent. Each of the viruses formed large plaques in PCK cells (range for plaque diameter for the human parent virus, the 6-2 gene reassortant, and the PB2 SGR was 2.4 to 3.5 mm) except the M SGR, which formed small plaques (mean diameter, 1.75 mm). Thus, the acquisition of the M gene specified a small plaque phenotype in PCK cells and the PB2 gene specified a host range restriction phenotype in MDCK cells. Level of replication in squirrel monkeys. Squirrel monkeys infected with wild-type virus exhibited a high level of virus replication in both the upper and lower respiratory tracts (Table 2). In contrast, the ah reassortant containing six internal avian RNA segments was attenuated, as shown by its greatly reduced replication in both the upper and lower respiratory tracts. The level of replication of each A/Los Angeles/2/87 x A/Mallard/NY/78 SGR virus in the nasopharynges and tracheas was then compared with that of the wild-type virus parent to assess the degree of growth restriction specified by the internal avian influenza virus gene (Table 2). We found that the PB2 SGR did not replicate to detectable levels in the upper or lower respiratory tracts of squirrel monkeys. In addition, the replication of the NP SGR was restricted 100-fold in the upper respiratory tracts and 10-fold in the lower respiratory tracts. In contrast, the M SGR exhibited a 10-fold reduction in replication in the upper respiratory tracts but none in the lower respiratory tracts. There was no evidence of reduced replication of the PB1, PA, or NS SGRs compared with the wild-type virus. All of the ah influenza A reassortant viruses except for the PB2 SGR induced moderate to high levels of HAI antibodies in serum. The PB2 SGR, which was the most highly attenuated SGR, induced low titers (1:4 to 1:8) of HAI antibodies in only two of four squirrel monkeys tested. Studies with adult human volunteers. The frequency and degree of illness produced by infection with wild-type virus were too low to allow a meaningful comparison of clinical virulence of the SGRs (Table 3), but the level of virus replication in the nasal wash specimens of the volunteers served as a measure of the virulence of the viruses. It is reasonable to compare levels of replication of two viruses because previous studies with humans have clearly shown that there is a correlation between the magnitude of virus replication and the development of symptomatic illness in experimentally infected volunteers (22). In the present study, we found that infected volunteers shed a small quantity of wild-type virus (mean, 10' 5 TCID50) over a

5 VOL. 30, 1992 GENES THAT ATITENUATE AVIAN-HUMAN INFLUENZA A VIRUSES 659 TABLE 2. Replication of influenza A/Los Angeles/2187 ah reassortant or wild-type virus in the upper and lower respiratory tracts of squirrel monkeysa Virus shedding in: No. tested Nasopharyngeal swab fluids Tracheal lavage fluids Reciprocal mean Virus (no. shedding A t viu) Mean peak titer Madutin Mean peak titer serumduratse ± SE (logl0 Mean duration ±SE Mea (logl0 duration (lg2) TCID50WMI) ±Sdy) TCID50WMI) ±SE(as A/Mallard/NY/78 x A/Los 4 (3) 1.2 ± ± b ± 1.3 Angeles/2/ SGR PB2C 4 (0) <05b ob <0.5b ob 1.8 ± Od SGR NP 4 (4) 1.8 ± ± 1.Ob 2.3 ± ± ± 1.0 SGR M 4 (4) 2.7 ± ± ± ± ± 0.8 SGR PA 4 (4) ± ± ± ± 0.6 SGR NS 4 (4) ± ± ± ± 1.4 SGR PB1 4 (4) 3.3 ± ± ± ± ± 0.5 A/Los Angeles/2/87 (wild-type) 4 (4) 3.5 ± ± ± ± ± 1.3 a Seronegative (HAI titer, < 1:4) squirrel monkeys were inoculated intratracheally with of ah or wild-type virus in a 0.5-ml inoculum. Viral genotypes are described in the text. Nasopharyngeal swab specimens were collected daily for 10 days postinoculation, and tracheal lavage fluids were obtained on days 2, 4, 6, and 8 postinoculation. b p < 0.03 compared with peak titer of virus shed or duration of virus shedding in the group which received the wild-type virus, by Mann-Whitney U two-tailed test. c Nasopharyngeal swab and tracheal lavage fluid specimens from squirrel monkeys inoculated with this virus were titrated in eggs and are presented as 50% egg infective doses per milliliter. d One of four monkeys inoculated mounted a fourfold increase in HAI antibody titer in serum. short period (mean, 2.6 days). This suggests that these volunteers were partially immune to this contemporary strain of influenza A (H3N2) virus even though they had a low level of or no HAI antibody in their serum before challenge. Nonetheless, we were able to detect statistically significant differences in the mean duration and peak titer of virus shedding in volunteers infected with a reassortant virus TABLE 3. from those observed in volunteers infected with wild-type virus. This comparison allowed us to assess the level of attenuation of each reassortant. We found that each of the SGRs, with the exception of the NP SGR, demonstrated a slight to marked reduction in peak titer and duration of virus shedding of reassortant virus in the upper respiratory tracts of volunteers compared with the Replication of A/Los Angeles/2/87 (H3N2) ah reassortant or wild-type virus in seronegative adult human volunteersa Virus shedding seing Reciprocal mean HAI % of volun- ~~~~~~titerin serum ± SE teers with ill- No. of (nasal washes) (log2) % With Wild-type or ness volun- Inetd Mean9pak titrmmea reassortant virus ter netd enpa ie enre-d teers ~~~%of Vol- (logl0 TCID_,,J duration Pre- Post- sponsed Fever Any reassortantvu tml)c (days)c inoculation inoculation ollny A/Mallard/NY/78 x A/ 13 46f of < ± ± " Los Angeles/2V87 (6-2 gene) PB2" 8 50 Of < ± M 12 50f of < ± ± NS f g 0.2 ± 0.2g 3.1 ± ± PB f 0.75 ± 0.12g g 3.4 ± ± PA ± ± ± ± NP ± ± ± ± A/Los Angeles/2/87 wild-type ± ± ± ± a Seronegative (HAI titer, <1:8) volunteers received 1070 TCID50 virus in a 0.5-ml inoculum. Viral genotypes are described in the text and are shown in Fig. 1. After inoculation, volunteers were examined and nasal wash specimens for virus culture were collected daily for 4 to 7 (6-2 gene ah virus) or 10 (other groups) days. b Virus recovery or a significant increase in antibody titer signified infection. cdata from each infected volunteer were used for calculations. Data are expressed as means + standard errors of the means. d An immune response was documented by a fourfold or higher increase in HAI titer in serum or a significant increase in serum IgG KELISA or nasal wash IgA KELISA antibody level, as defined in the text. e Volunteers were considered ill if they developed any of the following symptoms: fever (oral temperature >37.7C), systemic illness (myalgia), upper respiratory tract illness (rhinitis, pharyngitis, or both on two or more consecutive days), or lower respiratory tract illness (persistent cough on two or more consecutive days). f P < 0.05 compared with the proportion of volunteers infected or those who shed virus in the group which received wild-type virus, by Fisher's exact two-tailed test. g P < 0.05 compared with peak titer of virus shed or duration of virus shedding in the group which received wild-type virus, by Mann-Whitney U two-tailed test. h Nasal washes from subjects inoculated with this virus were titrated in eggs and are presented as 50% egg infective doses per milliliter.

6 660 CLEMENTS ET AL. J. CLIN. MICROBIOL. TABLE 4. Comparison of the host range restriction phenotypes observed in vitro and attenuation phenotypes in vivo of ah influenza A/ Los Angeles/2/87 (H3N2) reassortants Re Reassortant Sml lqe replication PCK cells' cells monkeys volunteers tsmall plaque Restricted Attenuation in": virus phenotype in in MDCK Squirrel Adult human Interpretation A/Mallard/NY/78 x A/Los Concordant results in primates and humans Angeles/2/87 (6-2 gene) PB PB2 gene contributes to attenuation and host range phenotypes in primates and humans NP 0 0 +C 0 Discordant results between primates and humans M + 0 +d + + M gene contributes to small plaque and attenuation phenotypes in humans NS Discordant results between primates and humans PB Discordant results between primates and humans PA PA gene does not specify attenuation a Normal plaque size in PCK cells ranged from a diameter of 2.4 to 3.5 mm, whereas the diameter of the small plaques of the M SGR was on average 1.75 mm. b Attenuation (indicated on a scale of 0 to + +) was denoted by restriction of virus replication of ah reassortants compared with that of wild-type virus. ' The NP gene single gene reassortant also specified restriction in virus replication in squirrel monkeys in a d previous study (14). The M gene single gene reassortant specified greater restriction of replication in a previous study (14) than in the present study. wild-type virus. The PB2 and M SGRs and the 6-2 gene ah reassortant failed to replicate to detectable levels in the upper respiratory tracts of volunteers, suggesting that the PB2 and M SGRs were fully attenuated for humans. Compared with the duration and peak titer of wild-type virus shed, the NS, PB1, PB2, and M SGRs each replicated significantly less (P < 0.02) and for a shorter period (P < 0.05). Since the PB1, PA, and NS SGR viruses replicated to a higher titer than the 6-2 gene ah reassortant virus, the acquisition of each of these genes apparently conferred only partial attenuation to the wild-type virus. DISCUSSION Two previous studies of ah reassortant viruses which derived one or more internal RNA gene segments from the avian influenza A/Mallard/NY/78 or A/Pintail/Alberta/119/79 virus and the remaining gene segments from a human H3N2 influenza wild-type virus identified the NP gene as the major avian influenza virus gene responsible for the attenuation phenotype exhibited by the 6-2 gene ah influenza A reassortant viruses in squirrel monkeys (27, 33). Studies with squirrel monkeys (33) also showed that the NS gene of the A/Mallard/NY/78 virus, which had been assigned by nucleotide sequence analysis to the A allele of the influenza virus NS RNA segment (34), did not exert an attenuating effect on the human influenza A wild-type virus for squirrel monkeys. On the basis of the findings with squirrel monkeys, it was predicted that the NP gene would be a major genetic determinant of restriction of replication of ah influenza A reassortant viruses in the respiratory tracts of humans and the NS gene would not be expected to affect viral replication (27, 33). Therefore, we were surprised to find in the present study that the levels of replication of the NP and NS SGRs derived from the A/Mallard/NY/78 donor virus and the A/Los Angeles/2/87 wild-type virus in the respiratory tracts of squirrel monkeys and humans were opposite (Table 4). In seronegative squirrel monkeys, the avian NP gene, not the avian NS gene, conferred the attenuation phenotype to the human influenza A/Los Angeles/2/87 wild-type virus. Conversely, in seronegative adults, the avian influenza virus NS gene, not the NP gene, appeared to contribute to the attenuation phenotype. These discordant results summarized in Table 4 indicate that it is not valid to extrapolate the results of studies of influenza virus SGRs with squirrel monkeys to humans. In addition, the avian influenza virus PB2 and M genes each conferred full attenuation, as indicated by our inability to detect virus shedding in the upper respiratory tracts of adult volunteers following intranasal inoculation with the PB2 or M SGRs. The PB1 SGR was partially attenuated for humans, as shown by a 50% reduction in virus replication compared with the wild-type virus. Neither the NP gene nor the PA gene contributed independently to the attenuation phenotype of human influenza A wild-type virus for humans. Clearly, future studies to identify avian influenza A virus genes that attenuate human influenza A wild-type viruses will have to be conducted with adult human volunteers, not squirrel monkeys. Results of earlier studies (30, 31) demonstrated that the avian influenza A/Mallard/NY/78 x human influenza A/Bethesda/1/85 (H3N2) reassortant virus was both attenuated and safe in seronegative adult volunteers, children, and infants, but the A/Mallard/NY/78 x human influenza A/Kawasaki/9/86 (H1N1) reassortant virus caused an unacceptably high rate of febrile reactions when given to seronegative infants and children. To explain the difference in reactogenicity between these HlNl and H3N2 ah reassortants, we had postulated, on the basis of the association of the avian M and NP genes with attenuation of human influenza A wildtype virus for squirrel monkeys (33), that the M and NP proteins of the A/Mallard/NY/78 virus probably interacted more efficiently with the HA and NA glycoproteins of the HlNl parent virus than those of the H3N2 parent (30). This more efficient interaction with the proteins of the HlNl wild-type virus resulted in enhanced replication and the mild, but unacceptable, reactogenicity of the HlNl ah reassortant that was observed previously (30). Evidence from the present study that the NP gene is not responsible for the attenuation of human influenza A wild-type virus for humans led to further clarification of the mechanism responsible for the enhanced reactogenicity. It is now suggested that the interaction between the A/Mallard/NY/78 M protein and the surface glycoproteins of the HlNl wild-type virus may have led to a more efficient budding process, resulting in a higher level of replication of the HlNl ah reassortant virus and a higher frequency of illness. In the present study, the M and PB2 SGRs, both of which were highly attenuated in humans, expressed phenotypes in

7 VOL. 30, 1992 GENES THAT ATTENUATE AVIAN-HUMAN INFLUENZA A VIRUSES 661 vitro that were not shown by either parental virus or the other reassortant viruses. The M SGR formed small plaques in PCK cells, and the PB2 SGR exhibited complete restriction of replication in MDCK cells. These findings indicate that a specific combination of avian and human influenza A virus genes is responsible for each of these phenotypes. It is therefore impossible to conclude from the results of the in vivo evaluation of these SGRs whether a constellation of avian and human influenza A virus genes or a defective PB2 or M gene accounts for the observed restricted replication of the reassortant viruses in humans and in tissue culture cells. There is additional evidence from other studies with a variety of viruses (5-9, 12, 21, 27, 32) that a constellation of genes can specify a phenotype in the progeny recombinant or reassortant virus not exhibited by either parent virus. For influenza A viruses, reassortment of polymerase genes from two divergent influenza A viruses resulted in a constellation of polymerase genes that altered viral replication and virulence in vivo (5, 19, 27). Florent found this to be the case with the influenza A virus reassortants that were tested as experimental live attenuated virus vaccines for humans (5). Reassortant viruses which acquired all six internal RNA segments from the A/Puerto Rico/8/34 virus and the HA and NA genes from a human influenza A H3N2 wild-type virus retained moderate virulence when given to volunteers and thus resembled their H3N2 wild-type parent viruses. However, each of six reassortant virus vaccines which acquired a mixed constellation consisting of two polymerase genes from the A/Puerto Rico/8/34 virus and the remaining polymerase and all other genes from the wild-type virus was highly restricted in replication and was attenuated in humans. This restriction of replication in humans probably reflected the incompatibility of the PR8 and H3N2 virus polymerase gene products and human host cell factors since the reassortants replicated efficiently in eggs but poorly in humans. This previous observation is clearly analogous to our findings in the present study with the PB2 SGR. The virological data from the studies in vivo with PB2 SGRs derived from two different avian viruses, the A/Mallard/NY/78 (H2N2) virus and the A/Pintail/Alberta/119/79 (H4N6) virus, provide strong evidence that the avian PB2 gene in the context of human H3N2 PB1, PA, and NP genes results in restriction of replication of the reassortant virus. Such PB2 SGRs have at least two properties desirable for a live attenuated influenza A virus vaccine: efficient replication in eggs and restriction of replication in mammalian cells in vitro and in vivo. Thus, a reassortant possessing the avian influenza PB2 gene and human influenza A virus H3N2 PB1 and PA genes might be considered for use as an attenuated donor virus for the production of live attenuated influenza A virus vaccines. In our study, the A/Mallard/NY/78 PB2 SGR appeared overly attenuated for humans since it induced immune responses in only half of the seronegative volunteers following intranasal inoculation, but it was as immunogenic as the 6-2 reassortant in these partially immune volunteers. It is possible that the PB2 SGR would have the desired balance between attenuation and immunogenicity for seronegative infants and children. As mentioned earlier, influenza A reassortant virus vaccines having one polymerase protein gene from the attenuated donor virus A/PR/8/34 and the remaining RNA gene segments from the human influenza A wild-type virus were satisfactorily attenuated and immunogenic for humans (5). This suggests that if the specific sequences in the avian influenza A/Mallard/NY/78 PB2 gene that result in the host range and attenuation phenotypes can be identified, it should be possible, by employing in vitro mutagenesis techniques (4, 17), to construct avian influenza A viruses with mutations in the appropriate regions of the PB2 gene that selectively control replication in the upper and lower respiratory tracts. Such a genetically altered avian influenza A virus PB2 gene could then be transferred to a human influenza wild-type virus to produce live attenuated influenza A reassortant virus vaccines with an appropriate balance between attenuation and immunogenicity. Clearly, a better understanding of the genetic determinants of the host range phenotype of avian influenza viruses would have important implications for vaccine development. ACKNOWLEDGMENTS This research was supported by Public Health Service contract NO1 Al from the National Institute of Allergy and Infectious Diseases. We thank Lorena Downs, Karen Christina, Bhavin Thumar, and Joan Stewart from the Center for Immunization Research and Frank Wood from the National Institutes of Health for assistance with this study. We also thank Alan Leach for preparation and Robert Chanock for critical review of the manuscript. REFERENCES 1. Buckler-White, A. J., C. W. Naeve, and B. R. Murphy Characterization of a gene coding for the M proteins which is involved in the host range restriction of an avian influenza A virus in monkeys. J. Virol. 57: Clements, M. L., S. D. Sears, K. Christina, B. R. Murphy, and M. H. Snyder Comparison of the virologic and immunologic responses of volunteers to live avian-human influenza A H3N2 reassortant virus vaccines derived from two different avian influenza virus donors. J. Clin. Microbiol. 27: Coelingh, K. J., C. C. Winter, B. R. Murphy, J. M. Rice, P. C. Kimball, R. A. Olmsted, and P. L. Collins Conserved epitopes on the hemagglutinin-neuraminidase proteins of human and bovine parainfluenza type 3 viruses: nucleotide sequence analysis of variants selected with monoclonal antibodies. J. Virol. 60: Enami, M., W. Luytjes, M. Krystal, and P. Palese Introduction of site-specific mutations into the genome of influenza virus. Proc. Natl. Acad. Sci. USA 87: Florent, G Gene constellation of live influenza A vaccines. Arch. Virol. 64: Halliburton, I. W., R. W. Honess, and R. A. Killington Virulence is not conserved in recombinants between herpes simplex virus types 1 and 2. J. Gen. Virol. 68: Javier, R. T., R. L. Thompson, and J. G. Stevens Genetic and biological analyses of a herpes simplex virus intertypic recombinant reduced specifically for neurovirulence. J. Virol. 61: Johnson, V. H., and B. L. Semler Defined recombinants of poliovirus and coxsackievirus: sequence-specific deletions and functional substitutions in the 5'-noncoding regions of viral RNAs. Virology 162: Lomniczi, B., S. Watanabe, T. Ben-Porat, and A. S. Kaplan Genetic basis of the neurovirulence of pseudorabies virus. J. Virol. 52: Massicot, J. G., B. R. Murphy, F. Thierry, L. Markoff, K.-Y. Huang, and R. M. Chanock Temperature-sensitive mutants of influenza virus: identification of the loci of the two ts lesions in the Udorn-ts-lA2 donor virus and the correlation of the presence of these two ts lesions with a predictable level of attenuation. Virology 101: Murphy, B. R., A. J. Buckler-White, W. T. London, J. Harper, E. L. Tierney, N. T. Miller, L. J. Reck, R. M. Chanock, and V. S. Hinshaw Avian-human reassortant influenza A viruses derived by mating avian and human influenza A viruses. J. Infect. Dis. 150: Murphy, B. R., E. G. Chalhub, S. R. Nusinoff, J. Kasel, and R. M. Chanock Temperature-sensitive mutants of influ-

8 662 CLEMENTS ET AL. enza virus. III. Further characterization of the ts-1[e] influenza A recombinant (H3N2) virus in man. J. Infect. Dis. 128: Murphy, B. R., M. L. Clements, E. L. Tierney, R. E. Black, J. Steinberg, and R. M. Chanock Dose response of influenza A/Washington/897/80 (H3N2) avian-human reassortant virus in adult volunteers. J. Infect. Dis. 152: Murphy, B. R., V. S. Hinshaw, D. L. Sly, W. T. London, N. T. Hosier, F. T. Wood, R. G. Webster, and R. M. Chanock Virulence of avian influenza A viruses for squirrel monkeys. Infect. Immun. 37: Murphy, B. R., D. L. Sly, N. T. Hosier, W. T. London, and R. M. ChanockL Evaluation of three strains of influenza A virus in humans and in owl, cebus, and squirrel monkeys. Infect. Immun. 28: Murphy, B. R., D. L. Sly, E. L. Tierney, N. T. Hosier, J. G. Massicot, W. T. London, R. M. Chanock, R. G. Webster, and V. S. Hinshaw Reassortant virus derived from avian and human influenza A viruses is attenuated and immunogenic in monkeys. Science 118: Muster, T., E. K. Subbarao, M. Enami, B. R. Murphy, and P. Palese An influenza A virus containing influenza B virus 5' and 3' noncoding regions on the neuraminidase gene is attenuated in mice. Proc. Natl. Acad. Sci. USA 88: Noble, G. R., H. S. Kaye, W. B. Yarborough, B. K. Fiedler, C. J. Reed, M. B. Felker, A. P. Kendal, and W. R. Dowdle Measurement of hemagglutination-inhibiting antibody to influenza virus in the 1976 influenza vaccine program: methods and test reproducibility. J. Infect. Dis. 136(Suppl.): Palese, P The genes of influenza virus. Cell 10: Powers, D. C., L. F. Fries, B. R. Murphy, B. Thumar, and M. L. Clements In elderly persons live attenuated influenza A virus vaccines do not offer an advantage over inactivated virus vaccine in inducing serum or secretory antibodies or local immunologic memory. J. Clin. Microbiol. 29: Reuer, Q., R. J. Kuhn, and E. Wimmer Characterization of poliovirus clones containing lethal and nonlethal mutations in the genome-linked protein VPg. J. Virol. 64: Richman, D. D., and B. R. Murphy The association of the temperature-sensitive phenotype with viral attenuation in animals and humans: implications for the development and use of live virus vaccines. Rev. Infect. Dis. 1: Sanger, F., S. Nicklen, and A. R. Coulson DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74: Scholtissek, C., and V. von Hoyningen Genetic relatedness of the gene which codes for the nonstructural (NS) protein of different influenza A strains. Virology 102: Sears, S. D., M. L. Clements, R. F. Betts, H. F. Maassab, B. R. J. CLIN. MICROBIOL. Murphy, and M. H. Snyder Comparison of live, attenuated HlNl and H3N2 cold-adapted and avian-human influenza A reassortant viruses and inactivated virus vaccine in adults. J. Infect. Dis. 158: Snyder, M. H., S. Banks, and B. R. Murphy Determination of antibody response to influenza virus surface glycoproteins by kinetic enzyme-linked immunosorbent assay. J. Clin. Microbiol. 26: Snyder, M. H., A. J. Buckler-White, W. T. London, E. L. Tierney, and B. R. Murphy The avian influenza virus nucleoprotein gene and a specific constellation of avian and human virus polymerase genes each specify attenuation of avian-human influenza A/Pintail/79 reassortant viruses for monkeys. J. Virol. 61: Snyder, M. H., M. L. Clements, R. F. Betts, R. Dolin, A. J. Buckler-White, E. L. Tierney, and B. R. Murphy Evaluation of live avian-human reassortant influenza A H3N2 and HlNl virus vaccines in seronegative adult volunteers. J. Clin. Microbiol. 23: Snyder, M. H., M. L. Clements, D. Herrington, W. T. London, E. L. Tierney, and B. R. Murphy Comparison by studies in squirrel monkeys, chimpanzees, and adult humans of avianhuman influenza A virus reassortants derived from different avian influenza virus donors. J. Clin. Microbiol. 24: Steinhoff, M. C., N. A. Halsey, L. F. Fries, M. H. Wilson, J. King, B. A. Burns, R. K. Samorodin, V. Perkis, B. R. Murphy, and M. L. Clements The A/Mallard/6750/78 avian-human, but not the A/Ann Arbor/6/60 cold-adapted, influenza AIKawasaki/86 (HlNl) reassortant virus vaccine retains partial virulence for infants and children. J. Infect. Dis. 163: Steinhoff, M. C., N. A. Halsey, M. H. Wilson, B. A. Burns, R. K. Samorodin, L. F. Fries, B. R. Murphy, and M. L. Clements Comparison of live attenuated cold-adapted and avianhuman influenza A/Bethesda/85 (H3N2) reassortant virus vaccines in infants and children. J. Infect. Dis. 162: Thompson, R. L., and J. G. Stevens Biological characterization of a herpes simplex virus intertypic recombinant which is completely and specifically nonneurovirulent. Virology 131: Tian, S.-F., A. J. Buckler-White, W. T. London, L. J. Reck, R. M. Chanock, and B. R. Murphy Nucleoprotein and membrane protein genes are associated with restriction of replication of influenza A/Mallard/NY/78 virus and its reassortants in squirrel monkey respiratory tract. J. Virol. 53: Treanor, J. J., M. Snyder, W. London, and B. R. Murphy The B allele of the NS gene of avian influenza viruses, but not the A allele, attenuates a human influenza A virus for squirrel monkeys. Virology 171:1-9.