Single gene reassortants identify a critical role for PB1, HA, and NA in the high virulence of the 1918 pandemic influenza virus

Claudia Pappas*†, Patricia V. Aguilar†, Christopher F. Basler†, Alicia Solo´ rzano†, Hui Zeng*, Lucy A. Perrone*, Peter Palese†‡§, Adolfo Garcı´a-Sastre†‡¶, Jacqueline M. Katz*, and Terrence M. Tumpey*§

*Influenza Division, Mailstop G-16, National Center for Immunization and Respiratory Diseases, Coordinating Center for Infectious Diseases, Centers for Disease Control and Prevention, 1600 Clifton Road Northeast, Atlanta, GA 30333; and Departments of †Microbiology and ‡Medicine and ¶Emerging Pathogens Institute, Mount Sinai School of Medicine, New York, NY 10029

Contributed by Peter Palese, December 17, 2007 (sent for review December 4, 2007) The 1918 influenza pandemic was exceptionally severe, resulting in acquired a neuraminidase (NA) gene of avian origin (7). The NA the death of up to 50 million people worldwide. Here, we show protein of influenza A virus functions largely in the release of which virus genes contributed to the replication and virulence of progeny particles (8), thus promoting viral spread, and a pan- the 1918 influenza virus. Recombinant viruses, in which genes of demic virus bearing a novel avian NA may have a selective the 1918 virus were replaced with genes from a contemporary advantage by evading immune detection and maintaining com- human H1N1 influenza virus, A/Texas/36/91 (Tx/91), were gener- patibility with the newly acquired avian HA (13, 14). Much less ated. The exchange of most 1918 influenza virus genes with is known about the origin of the H1N1 virus responsible for the seasonal influenza H1N1 virus genes did not alter the virulence of catastrophic influenza pandemic of 1918. Although virus se- the 1918 virus; however, substitution of the hemagglutinin (HA), quencing and phylogenetic analysis suggest it to be an avian-like neuraminidase (NA), or polymerase subunit PB1 genes significantly influenza virus (15), there are still unanswered questions about affected the ability of this virus to cause severe disease in mice. The the origin of the virus and the molecular properties that confer 1918 virus virulence observed in mice correlated with the ability of its exceptional virulence. 1918 recombinant viruses to replicate efficiently in human airway To better understand the molecular basis for the unprece- cells. In a second series of experiments, eight 1918 1:7 recombi- dented virulence of the 1918 pandemic virus, a reconstructed nants were generated, in which each Tx/91 virus gene was indi- influenza virus containing eight 1918 virus genes was generated vidually replaced by a corresponding gene from 1918 virus. Rep- in cultured cells (16). The reconstructed 1918 pandemic virus lication capacity of the individual 1:7 reassortant viruses was displays a high-growth phenotype in human bronchial epithelial assessed in mouse lungs and human airway cells. Increased virus cells and replicates efficiently in mice, ferrets, and macaques, titers were observed among 1:7 viruses containing individual 1918 causing death in all three species (16–19). Mice intranasally HA, NA, and PB1 genes. In addition, the 1918 PB1:Tx/91 (1:7) virus infected with the 1918 virus had a sudden onset of severe illness showed a distinctly larger plaque size phenotype than the small and succumbed to infection as early as 3 days after infection (16). plaque phenotype of the 1918 PA:Tx/91 and 1918 PB2:Tx/91 1:7 Accompanying the high virus titers were significant increases in reassortants. These results highlight the importance of the 1918 the numbers of macrophages and neutrophils detected in the HA, NA, and PB1 genes for optimal virus replication and virulence mouse lung after 1918 virus infection and their sustained pres- of this pandemic strain. ence in lung tissue distinguished this virus from a contemporary A/Texas/36/91 (Tx/91) virus infection (L.A.P., A.G.-S., J.M.K. human airway cells ͉ mice ͉ pathogenesis and T.M.T., unpublished data). In mice, the virulence of the 1918 virus was largely determined by the HA and, to a lesser extent, nfluenza A viruses regularly circulate in humans, causing by the polymerase gene complex (16, 17, 21). However, it is not Iannual epidemics and sporadic pandemics. Interpandemic known which of the three polymerase genes contribute to the (seasonal) influenza results in over 200,000 hospitalizations and exceptional virulence of the 1918 virus or whether other virus causes an average of 36,000 deaths each year in the United States genes also contribute to its virulence. alone (1, 2). During pandemic years, the death rate is signifi- During the peak of the 1918 pandemic, attending pathologists cantly higher. It is estimated that the ‘‘Spanish’’ influenza noted severe disease of the respiratory tree, and, for the majority pandemic of 1918 was responsible for more than 20 million of cases, confirmed death was due to pneumonia and respiratory deaths worldwide, a vastly higher number than the approximate failure (22, 23). Although most deaths were attributed to sec- combined total of 100,000 for the Asian influenza pandemic of ondary bacterial pneumonia (24), the disease process most likely 1957 and the Hong Kong influenza pandemic of 1968 (3–5). The began with severe acute viral infection resulting from efficient genetic composition of the influenza A viruses responsible for replication of the pandemic strain. A thorough understanding of the 1957 (H2N2) and 1968 (H3N2) human pandemics is largely the molecular mechanisms involved in virus replication of the known based on retrospective sequencing and phylogenic anal- 1918 virus may help reveal virulence factors used by other yses (6). Of importance, both pandemic viruses acquired a novel influenza viruses with pandemic potential. Our approach has hemagglutinin (HA) gene and a polymerase subunit PB1 gene of wild waterfowl origin (7). The major functions of the HA Author contributions: C.P., C.F.B., H.Z., P.P., J.M.K., and T.M.T. designed research; C.P., H.Z., protein are the receptor host cell binding and subsequent fusion L.A.P., and T.M.T. performed research; C.P., P.V.A., C.F.B., A.S., P.P., and A.G.-S. contributed of virus and host membranes in the endosome after the virus has new reagents/analytic tools; C.P., C.F.B., and T.M.T. analyzed data; and C.P., H.Z., J.M.K., been taken up by endocytosis (8). A novel HA protein has a and T.M.T. wrote the paper. replication advantage in a population immunologically naive to Freely available online through the PNAS open access option. the antigen (8). The PB1 subunit is a key component of the viral The authors declare no conflict of interest. RNA polymerase complex and catalyzes the sequential addition §To whom correspondence may be addressed. E-mail: [email protected] or tft9@ of nucleotides during RNA chain elongation (8–12). In addition cdc.gov. to the HA and PB1 proteins, the 1957 pandemic influenza virus © 2008 by The National Academy of Sciences of the USA

3064–3069 ͉ PNAS ͉ February 26, 2008 ͉ vol. 105 ͉ no. 8 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0711815105 Downloaded by guest on October 1, 2021 Table 1. Properties of 7:1 recombinant and parental influenza viruses used in this study

† ‡ Virus stock* Stock titer, pfu/ml Weight loss, % LD50

1918 4.5 ϫ 107 22.8 3.25 Tx PA:1918 7.8 ϫ 107 20.8 3.5 Tx PB1:1918 1.2 ϫ 108 10.5 5.5 Tx PB2:1918 1.8 ϫ 107 18.1 3.75 Tx HA:1918 4.6 ϫ 107 7.3 Ͼ6 Tx NP:1918 1.9 ϫ 107 22.5 3.5 Tx NA:1918 3.0 ϫ 107 15.3 5.5 Tx M:1918 2.1 ϫ 107 20.2 3.5 Tx NS:1918 8.8 ϫ 108 18.5 3.25 Tx/91 2.1 ϫ 108 3.7 Ͼ6

*Prepared on MDCK cells with trypsin and incubated for 48–72 h. †Maximum percent weight loss (mean of six mice per group) after infection with 105 pfu of virus. ‡ Expressed as the log10 pfu required to give 1 LD50.

been to reconstruct recombinant viruses, in which genes of the 1918 virus are replaced with genes from a contemporary human influenza (Tx/91) virus in attempts to understand which of the eight virus gene segments contribute to its high virulence. In a reciprocal experimental approach, eight Tx/91-like recombinant viruses were generated, in which each Tx/91 virus gene was individually replaced by a corresponding gene from the 1918 virus. The rescued viruses, referred to as 7:1 or 1:7 recombinant viruses, have revealed the importance of the 1918 HA, NA, and PB1 genes for optimal virus replication of this pandemic strain. The identification of the precise pandemic virus genes associated with replication may help elucidate virulence factors for other influenza viruses with pandemic potential and, thereby, help identify targets for drug intervention.

Results MICROBIOLOGY Construction and Characterization of Recombinant Viruses with 1918 Fig. 1. Pathogenicity of 7:1 recombinant H1N1 influenza viruses. Compari- son of lung virus titers (A), lethality (B), and weight loss (C) in mice (12 per virus Influenza Virus Genes. To determine which of the eight 1918 virus group) intranasally inoculated with 105 pfu of 1918 (ࡗ), Tx/91 PA:1918 (ᮀ), gene segments contribute to its high virulence, we generated a Tx/91 PB2:1918 (‚), Tx/91 PB1:1918 (E), Tx/91 HA:1918 (■), Tx/91 NP:1918 (छ), series of eight single-gene recombinant influenza viruses, each Tx/91 NA:1918 (ϩ), Tx/91 M:1918 (X), Tx/91 NS:1918 (F), or Tx/91 virus (*) virus. possessing seven gene segments of the 1918 virus and one Three mice from each virus-inoculated group were killed on the days indi- segment from a seasonal human H1N1 influenza virus, Tx/91. cated, and individual lungs were removed. Virus endpoint titers are expressed 1.5 The 1918 (7:1) recombinant viruses were compared with the as mean log10 EID50 per milliliter Ϯ SD. The limit of virus detection was 10 parental eight-gene 1918 virus for replication efficiency in EID50/ml. The remaining nine mice from each group were observed for mor- human respiratory cells and for virulence in mice. In a similar bidity and mortality. In A, the asterisk indicates virus lung titers significantly Ͻ fashion, we generated eight 1918 (1:7) recombinant viruses, in different (P 0.05) from the 1918 virus titers as determined by analysis of which each Tx/91 virus gene was individually replaced by a variance. corresponding gene from 1918 virus. All 1918 recombinant viruses had high infectivity titers (Ͼ107 pfu/ml) in MDCK cells, Table 2. Properties of 1:7 recombinant and parental influenza similar to the parental rescued 1918 and Tx/91 viruses (Tables 1 viruses used in this study and 2). We initially evaluated the pathogenicity of each 1918 7:1 † ‡ Virus stock* Stock titer, pfu/ml Weight loss, % LD50 recombinant virus by intranasally inoculating BALB/c mice with 5 1918 PA:Tx 5.8ϫ108 4.7 Ͼ6 10 pfu of virus for determination of morbidity (measured by 1918 PB1:Tx 3.4ϫ108 9.2 Ͼ6 weight loss), mortality, and virus replication. We also deter- 1918 PB2:Tx 1.1ϫ108 11.8 Ͼ6 mined the LD50 titers in mice infected with each 1918 7:1 1918 HA/NA:Tx 2.1ϫ107 24.6 4.25 recombinant virus and compared them with titers in groups of 1918 HA:Tx 2.1ϫ107 23.7 4.25 mice infected with the parental eight-gene 1918 virus or Tx/91 1918 NP:Tx 1.1ϫ108 5.7 Ͼ6 H1N1 virus, which previously were shown to be of high and low 1918 NA:Tx 9.2ϫ107 13.1 Ͼ6 virulence, respectively (16). Infection of mice with the eight-gene 1918 M:Tx 1.1ϫ108 3.3 Ͼ6 1918 virus resulted in lung virus titers on days 3 and 5 postin- 1918 NS:Tx 3.2ϫ107 5.3 Ͼ6 oculation (p.i.) that were at least 15,000-fold higher than those Tx/91 2.1ϫ108 6.1 Ͼ6 of mice infected with the Tx/91 virus (Fig. 1A). However, virus *Prepared on MDCK cells with trypsin and incubated for 48–72 h. replication in the lungs of mice inoculated with single-gene †Maximum percent weight loss (mean of six mice per group) after infection recombinant viruses containing NP, PB2, PA, M, or NS from with 106 pfu of virus. Tx/91 did not differ significantly from that in mice inoculated ‡ Expressed as the log10 pfu required to give 1 LD50. with the parental 1918 virus (Fig. 1A). Furthermore, each of the

Pappas et al. PNAS ͉ February 26, 2008 ͉ vol. 105 ͉ no. 8 ͉ 3065 Downloaded by guest on October 1, 2021 Fig. 3. Plaque morphology and release of 1:7 recombinant H1N1 viruses from apically infected NHBE cells. Confluent monolayers of MDCK cells were inoculated with 1:7 polymerase recombinants and Tx/91 virus, and plaque morphology was visualized at 48 h p.i. (A). NHBE cells were grown on Trans- Fig. 2. Release of 7:1 recombinant H1N1 influenza viruses from apically well inserts, and viral growth kinetics was determined (B and C). Cells were infected NHBE cells (derived from autopsy specimens from adults) and from infected with Tx/91 (ࡗ), 1918 PA:Tx/91 (ᮀ), 1918 PB1:Tx/91 (‚), or 1918 Detroit 562 epithelial cells. NHBE (A) and Detroit 562 (B) cells were grown on PB2:Tx/91 (E) virus at an moi of 0.01. Apical and basolateral (data not shown) Transwell inserts. Cells were infected with 1918 (ࡗ), Tx/91 PA:1918 (ᮀ), Tx/91 supernatants were collected, and virus content was determined in a standard PB2:1918 (‚), Tx/91 PB1:1918 (E), Tx/91 HA:1918 (■), Tx/91 NP:1918 (छ), Tx/91 plaque assay. Values represent the mean virus titer of fluids from three NA:1918 (ϩ), Tx/91 M:1918 (X), Tx/91 NS:1918 (F), or Tx/91 (*) virus at an moi replicate infected cultures. The asterisk indicates that Tx/91 virus titers are of 0.01 for1hat37°C. Unbound virus was removed by washing the cells three significantly different (P Ͻ 0.05) from those of all other virus infection groups times. Apical and basolateral (data not shown) supernatants were collected at as determined by analysis of variance. the indicated times, with or without (data not shown) addition of trypsin. The values shown represent the mean virus titer of supernatants from three replicate infected cultures. The asterisk indicates that the 1918 virus titers are progeny virus progressively increased during the first 24 h p.i., significantly different (P Ͻ 0.05) from those of all other virus infection groups and virus was detected almost exclusively in the apical superna- as determined by analysis of variance. tant (Fig. 2A). Only trace amounts of each virus tested (Ͻ101.2 pfu/ml) were released into the basolateral reservoir. All 1918 recombinant viruses replicated with similar kinetics and to these five 7:1 viruses caused severe signs of illness and weight loss similar titers in polarized NHBE and Detroit 562 cells with (Fig. that were comparable with those caused by the parental 1918 2) or without (data not shown) the addition of exogenous trypsin, virus (Fig. 1 B and C and Table 1). Mice infected with the Tx suggesting that these epithelial cells possess protease activity that PB2:1918, Tx PA:1918, or Tx M:1918 virus displayed a 2- to 3-day can support H1N1 virus HA cleavage. Regardless of the pres- delay in death compared with mice infected with the eight-gene ence or absence of trypsin, infectivity titers of Tx/91 HA:1918, 1918 virus; however, these 7:1 viruses were still considered highly Tx/91 NA:1918, and Tx/91 PB1:1918 viruses, like the parental virulent, and all mice succumbed to infection by day 9 p.i. (Fig. Tx/91 virus, were significantly lower than virus titers detected in 1C). In contrast to the lethal outcome in mice infected with the the 1918 virus-infected cultures at 12, 24, and 48 h p.i. (Fig. 2A). 1918 virus, mice infected with the 7:1 recombinant viruses Next, the replication kinetics of 7:1 viruses was determined in containing the Tx/91 HA or NA were significantly attenuated, Detroit 562 cell cultures. For comparison, cultures were infected 5 and none of the mice infected with 10 pfu died (Fig. 1 B and C with an H5N1 virus isolated in 2004, A/Viet Nam/1203/04 and Table 1). The reduced disease and weight loss of the Tx/91 (VN/1203), which was cultivated from a patient who died (27). HA:1918- and Tx/91 NA:1918-infected mice correlated with the Remarkably, the H5N1 virus replicated in human airway cells lower levels of virus replication in mouse lungs on days 3 and derived from the upper respiratory tract to titers similar to those 5 p.i. (Fig. 1A). Of interest, replacing the 1918 PB1 gene resulted of the 1918 virus (Fig. 2B). At 12, 16, and 24 h p.i., 1918 and in an attenuated virus (Tx/91 PB1:1918) with a lethality (LD50, VN/1203 virus production was at least 7-fold higher than that 105.5) that was 170 times less than that of the 1918 virus (Table observed in cultures infected with Tx/91 or the 1918 7:1 virus 1). Tx/91 PB1:1918-infected mice lost significant weight (10.5%) containing the PB1 from Tx/91 virus. As was observed in NHBE by day 8 p.i., but most of the mice infected with 105 pfu of virus cells, infectivity titers of Tx/91 HA:1918 and Tx/91 NA:1918 survived the infection (Fig. 1C). viruses were significantly lower than virus titers released in the 1918 virus-infected cultures at 12, 24, and 48 h p.i. (data not The 1918 HA, NA, and PB1 Genes Contribute to Replication Efficiency shown), further demonstrating that the 1918 HA, NA, and PB1 in Human Airway Cells. To determine whether the 1918 virus genes are essential for maximal replication of the pandemic virus in responsible for the high virulence in vivo also contribute to human cells derived from upper and lower respiratory tract optimal virus replication in human airway cells, we assessed the epithelium. growth and release of virus in primary human bronchial epithe- lial (NHBE) cells (25) and in Detroit 562 epithelial cells derived Generation of 1918 Recombinant 1:7 Viruses. To further prove that from a human pharyngeal carcinoma (26). Culture medium from the HA, NA, and PB1 genes contribute to the high virulence of virus-inoculated NHBE cells was collected from the apical and the 1918 pandemic virus, we generated reciprocal recombinant basolateral chambers of the cell monolayers at different times (1:7) viruses in which each of the contemporary Tx/91 virus after inoculation and examined for virus production in the genes were replaced with the corresponding gene from the 1918 presence and absence of trypsin. With all viruses tested, titers of virus. The reciprocal constellation of 1918 PB1 on the parental

3066 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0711815105 Pappas et al. Downloaded by guest on October 1, 2021 1918 virus gene was tested for its ability to enhance the repli- cation efficiency of a contemporary human H1N1 virus. These data demonstrated that the HA, NA, and PB1 virus genes are essential for maximal replication and virulence of the 1918 virus. Because human airway epithelium is the primary site for infection and replication of influenza viruses, we have used two airway epithelial (Detroit 562 and NHBE) cells to determine the replication efficiency of the 1918 recombinants. NHBE cells have been more recently used as an in vitro human airway epithelium model to evaluate interactions between influenza A virus and the human host (25, 30). The concept that efficient viral growth in the upper respiratory tract of humans can Fig. 4. Replication of 1:7 recombinant H1N1 influenza viruses in mouse facilitate virus excretion by coughing and sneezing, prompted us 6 lungs. Comparison of lung virus titers in mice infected with 10 pfu of the to evaluate the growth and release of the virus in polarized indicated 1:7 virus. For comparison, a group of mice was infected with the 2:6 oropharyngeal (Detroit 562) cells. We found that the replication virus containing the 1918 HA and NA genes (1918 HA/NA:Tx/91 virus). Three mice from each virus-infected group were killed on the days indicated, and efficiency of the 1918 recombinants in mouse lungs correlated to individual lungs were titered and expressed as mean log10 EID50 per milliliter Ϯ the replication efficiency in both airway epithelial cells. In SD. The asterisk indicates that virus lung titers are significantly different (P Ͻ addition, we also observed that an avian influenza H5N1 virus 0.05) from the Tx/91 virus titer as determined by analysis of variance. isolated from a human in 2004 displayed a high-growth pheno- type in Detroit 562 cells derived from the upper respiratory tract. It has been postulated that the lack of sustained human to human Tx/91 background genes (1918 PB1:Tx/91 virus) displayed larger transmission of avian H5N1 viruses is due to their ␣2,3 sialic acid plaque size morphology on MDCK cells (Fig. 3A) and greater receptor binding preference and thus to the presumed inability virus replication in NHBE cells compared with that displayed by of the virus to replicate efficiently at this site (31, 32). However, the 1918 PA:Tx/91, 1918 PB2:Tx/91, and Tx/91 viruses (Fig. 3B). we found that the H5N1 virus replicated to similarly high As early as 12 h p.i., the 1918 PB1:Tx/91 virus produced 8-fold infectivity titers as the 1918 virus in these cells. These data are greater virus release in apical supernatants compared with virus in agreement with a recent report that demonstrated efficient production by the other polymerase reassortants and the wild- H5N1 virus replication in ex vivo cultures of human nasopha- type Tx/91 virus. Virus replication in culture supernatants from ryngeal tissues (33), suggesting that avian H5N1 viruses may NHBE cells inoculated with 1:7 recombinant viruses (in which have a broader tropism for the human respiratory tract than the Tx/91 NP, PB2, M, and NS genes were substituted with the initially reported. corresponding 1918 virus genes) was not significantly different Infection with 1:7 reassortants in which the 1918 NP, M, NS, from virus replication in cells inoculated with the parental Tx/91 or polymerase subunits PB2, PA genes were individually substi- virus (data not shown). However, the Tx/91 virus with the tuted into the background of the Tx/91 virus did not result in recombinant virus expressing the 1918 NA or HA, in the increased virus replication compared with the parental Tx/91

presence and absence of 1918 NA, increased the replication virus. The NS1 protein was of particular interest because it has MICROBIOLOGY efficiency of the Tx/91 virus in NHBE cells (Fig. 3C) and in been shown to antagonize type I IFN production (34) leading to mouse lungs (Fig. 4). Furthermore, both 1918 HA:Tx/91 and the concept that a strong IFN-antagonist NS1 protein may 1918 HA/NA:Tx/91 viruses caused severe illness and death, contribute to enhanced influenza virus virulence, in general and whereas the remaining 1:7 viruses did not cause death in mice, in particular to the exceptional virulence of the 1918 virus. even at the highest obtainable titer used for virus inoculation However, the introduction of the 1918 virus NS1 gene into a (106 pfu). Taken together, these results demonstrate that the WSN virus background (1918 NS1:WSN) resulted in a virus that HA, NA, and PB1 genes of the 1918 influenza virus contribute was not more virulent in mice (35). Similarly, in the current to the virulence of this virus in mice. Moreover, the virulence study, 1918 NS1:Tx/91 virus did not confer a more virulent virus observed in mice correlates with the ability of the 1918 recom- in mice or increase the replication efficiency of the parental binant viruses to replicate efficiently in mouse lungs and human Tx/91 virus in human airway cells. These data suggested that the airway cells. NS1 protein is not a crucial virulence factor of the 1918 virus or that the model systems used in this study were not ideal to study Discussion human NS1 virulence. The 1918 virus genes that were able to The emergence of another pandemic virus is considered likely, enhance replication efficiency of the Tx/91 virus were the HA, if not inevitable (28). The molecular characterization of the NA, or PB1 genes. Interestingly, among all eight gene segments reconstructed 1918 pandemic influenza virus may shed light on tested in the current study, the HA was the only 1918 virus gene the threat posed by new influenza virus strains with pandemic able to confer a virulent phenotype when rescued in the genetic potential. The factors responsible for the high lethality associ- background of Tx/91 virus. The 1918 HA gene was shown to be ated with the 1918 virus are complex and poorly understood; essential for maximum virus replication and for eliciting a however, the ability of the virus to replicate efficiently in the host heightened host inflammatory response (16, 17, 21). The con- most likely contributed to its unusual virulence. Because the tribution of the 1918 NA gene may, in part, represent the need coding sequences of the 1918 viral RNA segments did not reveal for optimal balance between sialidase and 1918 HA receptor obvious genetic features that have been associated with viru- binding activities and/or previously observed HA cleavage prop- lence (15, 29), it is crucial to study 1918 recombinant viruses in erties (13, 14). The increased replication efficiency of the 1918 suitable animal models to better understand the genetic markers PB1:Tx/91 (1:7) virus, observed in NHBE cells, was also ob- responsible for virus replication and virulence of this pandemic served in MDCK cells as distinctly larger plaque size phenotype strain. Our results suggest that multiple 1918 virus genes con- in comparison with the small plaque phenotype of the 1918 tribute to optimal virus replication efficiency in human airway PA:Tx/91 and 1918 PB2:Tx/91 (1:7) reassortants. The PB1 cells and in lungs of mice. By comparing the highly replication- subunit is a key component of the viral RNA polymerase competent eight-gene 1918 virus with the single-gene 7:1 re- complex and contains multiple active sites critical for the poly- combinant viruses, we determined the role of each 1918 virus merization of RNA chains and also for association with PA and gene in replication and in virulence. Similarly, the ability of each PB2 to form a heterotrimer (9–11). The contribution of the PB1

Pappas et al. PNAS ͉ February 26, 2008 ͉ vol. 105 ͉ no. 8 ͉ 3067 Downloaded by guest on October 1, 2021 gene is particularly significant in the context of the 1957 and 1968 gram. This research was done by staff taking antiviral prophylaxis and using pandemic viruses, which each acquired PB1 together with HA, stringent biosafety precautions to protect the researchers, the environment, NA, or both genes from the avian gene pool in wild ducks by and the public. The identity of the 1918 influenza virus genes in the recom- genetic reassortment, retaining other virus genes from circulat- binant viruses was confirmed by RT-PCR and sequencing. ing human strains (6, 7). The 1918 PB1 protein differs from the Infection of Mice. Female BALB/c mice, 6–8 weeks old (Charles River Labora- conserved avian influenza consensus sequence by only seven tories), were anesthetized with an i.p. injection of 0.2 ml of Avertin (Aldrich) amino acid residues (15) and an avian-like PB1 gene may provide and 50 ␮l of infectious virus diluted in PBS was inoculated intranasally (i.n.) increased transcriptional activity of the RNA-dependent RNA (43). LD50 titers were determined by inoculating groups of three mice i.n. with polymerase (12) and greater virus replication. serial 10-fold dilutions of virus. LD50 titers were calculated by the method of The increased virulence associated with the 1918 PB1 might Reed and Muench (20) and are expressed as the log10 pfu required to give 1 be due to the PB1-F2 protein generated by an alternate reading LD50. For comparison of morbidity (measured by weight loss), mortality, and lung virus titers, additional mice were infected with inoculating doses of 105 frame (36). PB1-F2 is truncated in Tx/91 and all contemporary 6 human H1N1 viruses but is functional in the 1918 virus and may or 10 pfu of virus. On days 3 and 5 p.i., three mice from each group were killed and whole lungs were removed and homogenized in 1 ml of sterile PBS. Fifty contribute to virulence by functioning as a proapoptotic protein percent egg infectious dose (EID50/ml) titers were calculated by the method of (37, 38). Recent work where a point mutation was introduced at Reed and Muench (20). amino acid 66 of PB1-F2 revealed the importance of this protein in the virulence of a H5N1 virus and the 1918 pandemic virus Human Airway Cells and Viral Infection. Primary human bronchial epithelial (37). A better understanding of the contribution of polymerase (NHBE) cells (Cambrex Bio Science) (25) and Detroit 562 epithelial cells (Amer- proteins in virulence will aid in designing drugs that target the ican Type Culture Collection) (26), were grown in MEM as described in ref. 25. key intersubunit binding sites of the polymerase complex and Briefly, cells (5 ϫ 105) were seeded onto Corning 24-mm diameter semiper- ␮ diminish the high replication efficiency of pandemic virus strains meable membrane inserts with 0.4- m pore size and cultured for 1 week to achieve a stable transepithelial resistance of Ͼ1,000 ⍀⅐cm2. Monolayers were (39). For currently circulating H5N1 influenza viruses, the PB2 washed with MEM supplemented with 0.3% BSA (MEM/BSA). Virus was polymerase subunit appears to play a greater role in the high diluted in MEM/BSA and added to the apical surface of cells at a multiplicity growth phenotype and increased virulence associated with these of infection (moi) of 0.01 for1hat37°C. Monolayers were then washed and highly pathogenic viruses in mammals (40, 41). 2 ml of MEM/BSA was added to both apical and basolateral reservoirs. Cultures were set up with or without TPCK-treated trypsin (1 ␮g/ml; Sigma). Apical and Methods basolateral supernatants were collected at the indicated times, and virus Generation of 1918 Recombinant Viruses by Reverse Genetics. Genes encoding content was determined in a standard plaque assay (25). The values shown the 1918 pandemic influenza virus were reconstructed from deoxyoligonu- represent the mean virus titer of supernatants from three replicate infected cleotides corresponding to the reported 1918 virus coding sequences. The cultures. One avian H5N1 virus isolated from fatal human case in early 2004, noncoding regions of each segment are identical to that of the corresponding A/Vietnam/1203/2004 (VN/1203), was used in this study; virus stocks were segment of influenza A/WSN/33 (H1N1) virus. All 1918 recombinant viruses grown as described in ref. 27. were generated by using 1918 virus gene cDNAs described in refs. 16–18 and the described reverse genetics system (35, 42, 43). Transfection supernatants ACKNOWLEDGMENTS. This work was partially supported by National Insti- were passaged onto MDCK cells and virus stock prepared and titrated. The tutes of Health Grant P01 AI058113 (to A.G.-S.), the Northeast Biodefense 1918 viruses were handled under biosafety level 3 enhanced (BSL-3ϩ) con- Center Grant U54 AI057158, and the Center for Investigating Viral Immunity and Antagonism Grant U19 AI62623. C.F.B. was supported by Northeast tainment in accordance with guidelines of the National Institutes of Health Biodefense Center-Lipkin Grant U54 AI057158, Center for Investigating Viral and the Centers for Disease Control and Prevention (CDC) (available at http:// Immunity and Antagonism-Moran Grant U19 AI62623, and National Institutes www.cdc.gov/OD/ohs/biosfty/bmbl5/bmbl5toc.htm) and in accordance with of Health Grant P01 AI058113. P.V.A. was supported by a fellowship awarded requirements of the U.S. Department of Agriculture/CDC Select Agent Pro- by Northeast Biodefense Center-Lipkin Grant U54 AI057158.

1. Centers for Disease Control and Prevention (2007) Key Facts About Seasonal Influenza 16. Tumpey TM, et al. (2005) Characterization of the reconstructed 1918 Spanish influenza (Flu). Available at http://www.cdc.gov/flu/keyfacts.htm. Accessed November 11, 2007. pandemic virus. Science 310:77–80. 2. Thompson WW, et al. (2003) Mortality associated with influenza and respiratory 17. Kash JC, et al. (2006) Genomic analysis of increased host immune and cell death syncytial virus in the United States. J Am Med Assoc 289:179–186. responses induced by 1918 influenza virus. Nature 443:578–581. 3. Burnet F, Clark F (1942) Influenza Survey of the Last 50 Years in the Light of Modern 18. Kobasa D, et al. (2007) Aberrant innate immune response in lethal infection of Work on the Virus of Epidemic Influenza (MacMillan, Melbourne). macaques with the 1918 influenza virus. Nature 445:319–323. 4. Grove RD, Hetzel AM (1968) Vital Statistics Rates in the United States: 1940–1960 (U.S. 19. Tumpey TM, et al. (2007) A two-amino acid change in the hemagglutinin of the 1918 Government Printing Office, Washington, DC). influenza virus abolishes transmission. Science 315:655–659. 5. Rosenau MJ, Last JM (1980) Maxcy-Rosenau Preventative Medicine and Public Health 20. Reed LJ, Muench H (1938) A simple method for estimating fifty percent endpoints. Am J (Appleton-Centry-Crofts, New York). Hyg 27:493–497. 6. Webster RG, Bean WJ, Gorman OT, Chambers TM, Kawaoka Y (1992) Evolution and 21. Kobasa D, et al. (2004) Enhanced virulence of influenza A viruses with the haemag- ecology of influenza A viruses. Microbiol Rev 56:152–179. glutinin of the 1918 pandemic virus. Nature 431:703–707. 7. Kawaoka Y, Krauss S, Webster RG (1989) Avian-to-human transmission of the PB1 gene 22. LeCount ER (1919) The pathologic anatomy of influenza bronchopneumonia. JAm of influenza A viruses in the 1957 and 1968 pandemics. J Virol 63:4603–4608. Med Assoc 72:650–652. 8. Palese P, Shaw ML (2006) Orthomyxoviridae: The viruses and their replication. Fields 23. Wolbach SB (1919) Comments on the pathology and bacteriology of fatal influenza Virology, Knipe DM, Howley PM, eds (Lippincott Williams & Wilkins, Philadelphia), 5th cases, as observed at Camp Devens, Mass. Johns Hopkins Hospital Bulletin 30:104. Ed, pp 1648–1689. 24. Beveridge WI (1991) The chronicle of influenza epidemics. Hist Philos Life Sci 13:223–234. 9. Biswas SK, Nayak DP (1996) Influenza virus polymerase basic protein 1 interacts with 25. Zeng H, et al. (2007) Highly pathogenic avian influenza H5N1 viruses elicit an atten- influenza virus polymerase basic protein 2 at multiple sites. J Virol 70:6716–6722. uated type I interferon response in polarized human bronchial epithelial cells. J Virol 10. Gonzalez S, Zurcher T, Ortin J (1996) Identification of two separate domains in the 81:12439–12449. influenza virus PB1 protein involved in the interaction with the PB2 and PA subunits: 26. Temonen M, et al. (1993) Susceptibility of human cells to Puumala virus infection. J Gen a model for the viral RNA polymerase structure. Nucleic Acids Res 24:4456–4463. Virol 74:515–518. 11. Perez DR, Donis RO (2001) Functional analysis of PA binding by influenza a virus PB1: 27. Maines TR, et al. (2005) Avian influenza (H5N1) viruses isolated from humans in Asia in effects on polymerase activity and viral infectivity. J Virol 75:8127–8136. 2004 exhibit increased virulence in mammals. J Virol 79:11788–11800. 12. Naffakh N, Massin P, Escriou N, Crescenzo-Chaigne B, van der Werf S (2000) Genetic 28. Webby RJ, Webster RG (2003) Are we ready for pandemic influenza? Science 302:1519–1522. analysis of the compatibility between polymerase proteins from human and avian 29. Taubenberger JK (2006) The origin and virulence of the 1918 ‘‘Spanish’’ influenza virus. strains of influenza A viruses. J Gen Virol 81:1283–1291. Proc Am Philos Soc 150:86–112. 13. Gong J, Xu W, Zhang J (2007) Structure and functions of influenza virus neuraminidase. 30. Hatta M, et al. (2007) Growth of H5N1 influenza A viruses in the upper respiratory tracts Curr Med Chem 14:113–122. of mice. PLoS Pathog 3:1374–1379. 14. Mitnaul LJ, et al. (2000) Balanced hemagglutinin and neuraminidase activities are 31. Shinya K, et al. (2006) Avian flu: influenza virus receptors in the human airway. Nature critical for efficient replication of influenza A virus. J Virol 74:6015–6020. 440:435–436. 15. Taubenberger JK, et al. (2005) Characterization of the 1918 influenza virus polymerase 32. van Riel D, et al. (2006) H5N1 virus attachment to lower respiratory tract. Science genes. Nature 437:889–893. 312:399.

3068 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0711815105 Pappas et al. Downloaded by guest on October 1, 2021 33. Nicholls JM, et al. (2007) Tropism of avian influenza A (H5N1) in the upper and lower 38. McAuley JL, et al. (2007) Expression of the 1918 influenza A virus PB1–F2 enhances the respiratory tract. Nat Med 13:147–149. pathogenesis of viral and secondary bacterial pneumonia. Cell Host Microbe 2:240–249. 34. Garcia-Sastre A, et al. (1998) Influenza A virus lacking the NS1 gene replicates in 39. Ghanem A, et al. (2007) Peptide-mediated interference with influenza A virus poly- interferon-deficient systems. Virology 252:324–330. merase. J Virol 81:7801–7804. 35. Basler CF, et al. (2001) Sequence of the 1918 pandemic influenza virus nonstructural 40. Hatta M, Gao P, Halfmann P, Kawaoka Y (2001) Molecular basis for high virulence of gene (NS) segment and characterization of recombinant viruses bearing the 1918 NS Hong Kong H5N1 influenza A viruses. Science 293:1840–1842. genes. Proc Natl Acad Sci USA 98:2746–2751. 41. Salomon R, et al. (2006) The polymerase complex genes contribute to the high 36. Chen W, et al. (2001) A novel influenza A virus mitochondrial protein that induces cell virulence of the human H5N1 influenza virus isolate A/Vietnam/1203/04. J Exp Med death. Nat Med 7:1306–1312. 203:689–697. 37. Conenello GM, Zamarin D, Perrone LA, Tumpey T, Palese P (2007) A single mutation in 42. Fodor E, et al. (1999) Rescue of influenza A virus from recombinant DNA. J Virol 73:9679–9682. the PB1–F2 of H5N1 (HK/97) and 1918 influenza A viruses contributes to increased 43. Tumpey TM, et al. (2002) Existing antivirals are effective against influenza viruses with virulence. PLoS Pathog 3:1414–1421. genes from the 1918 pandemic virus. Proc Natl Acad Sci USA 99:13849–13854. MICROBIOLOGY

Pappas et al. PNAS ͉ February 26, 2008 ͉ vol. 105 ͉ no. 8 ͉ 3069 Downloaded by guest on October 1, 2021