View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Leading Edge Essay

Peering into the Crystal Ball: Influenza Pandemics and Vaccine Efficacy

Matthew S. Miller1,* and Peter Palese1,2,* 1Department of Microbiology 2Department of Medicine Division of Infectious Diseases, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA *Correspondence: [email protected] (M.S.M.), [email protected] (P.P.) http://dx.doi.org/10.1016/j.cell.2014.03.023

The looming threat of a new influenza virus pandemic has fueled ambitious efforts to devise more predictive parameters for assessing the risks associated with emergent virus strains. At the same time, a comprehensive understanding of critical factors that can accurately predict the outcome of vaccination is sorely needed in order to improve the effectiveness of influenza virus vaccines. Will new studies aimed at identifying adaptations required for virus transmissibility and systems-level analyses of influenza virus vaccine responses provide an improved framework for predictive models of viral adaptation and vaccine efficacy?

Introduction sure to evade the pre-existing immunity ‘‘Follow the Leader’’ The development of effective vaccines afforded by vaccines. This has necessi- The most challenging issue facing IAV has altered the course of modern civiliza- tated painstaking efforts to identify and vaccinologists has always been the ne- tion by alleviating the scourges of target conserved epitopes of these vi- cessity to predict the antigenic character- humankind’s most devastating patho- ruses (Julien et al., 2012). (2) There is istics of vaccine strains months in gens. Illnesses caused by variola (small- an astonishing paucity of robust, predic- advance of the actual influenza season pox), Corynebacterium diphtheriae (diph- tive immunological markers of vaccine in order to allow sufficient time for vaccine theria), Clostridium tetani (tetanus), efficacy. This, in turn, has precluded production and distribution. This can be yellow fever virus (yellow fever), Borde- a comprehensive, mechanistic under- equated to a game of virological ‘‘follow tella pertussis (whooping cough), polio vi- standing of what differentiates success- the leader,’’ wherein the medical and rus (polio), and measles virus (measles) ful vaccines from those that fail. The scientific communities are constantly have become as foreign to our youngest recognition of these challenges has chasing the unpredictable evolutionary generations as telegrams and type- ignited ambitious efforts to predict more trajectory of the virus. As a consequence writers. Indeed, the success of vaccines accurately the behavior of the aforemen- of this guesswork, antigenic mismatch to date is truly remarkable when con- tioned pathogens and the vaccines de- between strains included in the vaccine sidered in light of the rudimentary signed to protect against them. Here, and the strains that eventually circulate principles that guided their historical we review recent advances within the is a regular occurrence. This can severely design (Stern and Markel, 2005). Ironi- field of influenza virus research that are limit the effectiveness of a given vaccine. cally though, triumphs in the modern attempting to provide a more predictive Of equal concern though is the sub- era of ‘‘rationale vaccine design’’ have basis for assessing the consequences optimal vaccine efficacy reported even been few and far between. Pathogens of viral adaptations and the efficacy of during seasons in which near-perfect such as Mycobacterium tuberculosis vaccines, and we highlight the scientific matches are achieved. Adding to this (tuberculosis), Plasmodium spp. (causa- and regulatory boundaries that must be complexity are the vastly different quali- tive agents of malaria), human immuno- overcome to achieve these goals. Two ties of responses elicited by the available deficiency virus (HIV), and influenza new studies focused on these important influenza vaccine formulations, most A virus (IAV) continue to elude broad topics appear in this issue of Cell. notably those observed for inactivated, and highly efficacious vaccine-mediated Linster, van Boheemen, and colleagues split vaccines in comparison to live protection, exerting devastating human (Linster et al., 2014) define a minimal attenuated vaccines (Osterholm et al., and economic tolls. The factors that set of mutations (and their associated 2012). Recognition of the limitations that have limited the successful design of phenotypes) that confer H5N1 viruses plague our current seasonal influenza vaccines against these pathogens are with the propensity to transmit in ferrets, virus vaccine approaches has catalyzed complex. However, two prominent and Tsang and colleagues (Tsang et al., renewed efforts to understand and iden- barriers stand out: (1) highly mutable/ 2014) use a systems biology approach tify factors that may more accurately adaptable pathogens such as HIV and to identify baseline immunological pre- predict vaccine responses. Nevertheless, IAV evolve under immunological pres- dictors of vaccine responses. all of these issues pale in comparison to

294 Cell 157, April 10, 2014 ª2014 Elsevier Inc. Figure 1. Predictive Factors of Influenza Virus Transmission A number of adaptations (shown in blue) in H5N1 IAV were observed to confer the ability to transmit from ferret to ferret. This gain of function in transmission was accompanied by a loss of function in virulence. Predicting the pandemic risk associated with future outbreaks caused by novel IAV strains in humans will require a comprehensive understanding of which strains are capable of zoonotic transmission, the adaptations required for sustained human-to-human transmission, and the consequences of those adaptations on properties such as virulence and fitness. Siaa2,3Gal and Siaa2,6Gal = a 2,3-linked sialic acids and a 2,6-linked sialic acids, respectively. the ever-present threat of a new and un- flurry of human cases caused by H7N9 isolate. Further analysis demonstrated expected IAV pandemic. This has led to IAV in China. Both of these viruses are of that the N224K/Q226L mutations alone a redoubling of efforts to both detect avian origin and are mainly transmitted were primarily responsible for the switch and assess the risks associated with iso- to humans through exposure to high titers from Siaa2,3Gal to Siaa2,6Gal binding lated epidemics caused by ‘‘exotic’’ IAV of virus from infected poultry (To et al., (Imai et al., 2012). strains. 2013). The H5 HA mutations that conferred The recent threat of a potentially devas- altered receptor binding characteristics Predicting Pandemics? tating H5N1 pandemic prompted the NIH were then rescued in the pandemic Emergence of the ‘‘swine-origin’’ H1N1 to fund a series of proposals aimed at A/California/04/2009 (Cal/09) H1N1 virus IAV strain in 2009 that swept through identifying viral molecular determinants background to assess how a recombi- Mexico and went on to cause the first that might indicate adaptations that would nant of an avian H5N1 virus with the pandemic of the 21st century served as confer the ability to transmit efficiently be- circulating Cal/09 H1N1 virus might a direct example of the inherent limita- tween mammals. This work, carried out behave. This strategy closely resembles tions precluding the accurate prediction by the groups of Yoshiro Kawaoka at previous studies performed by the Perez of IAV dynamics (Girard et al., 2010). The University of Wisconsin, Madison/ lab in the context of H9N2 IAV (Kimble Never before had an IAV pandemic been University of Tokyo, Japan and Ron et al., 2011). Despite causing a switch caused by a virus carrying the same hem- Fouchier at Erasmus University Rotter- to human-like receptor binding, both agglutinin (HA) and neuraminidase (NA) dam was first published in the journals E119G/V152I/N224K/Q226L and N224K/ subtypes as one of the circulating sea- Nature and Science in May and June of Q226L mutations tended to attenuate sonal strains. This served as an impetus 2012, respectively (Herfst et al., 2012; virus replication in the respiratory tract to not only bolster surveillance efforts, Imai et al., 2012). The Kawaoka group of ferrets. However, a secondary N158D but to also more thoroughly understand used an unbiased genetic approach to or N158K mutation (which abolishes viral molecular determinants of virulence generate a library of viruses containing a glycosylation site at position 158) and transmission. mutations in the globular head domain that appeared in animals inoculated Despite the fact that at least 18 sub- of the HA from the A/Vietnam/1203/ with the N224K/Q226L variant improved types of IAV hemagglutinin (HA, H) and 2004 H5N1 isolate in a nonpathogenic, virus replication in the upper respiratory 11 subtypes of neuraminidase (NA, N) mouse-adapted IAV background (PR8). tract. The combination of N158D/N224K/ have so far been identified (Tong et al., Despite having been derived from a virus Q226L mutations conferred the ability to 2013), only IAVs carrying H1, H2, or H3 isolated from a human, this HA maintains transmit between ferrets via respiratory and N1 or N2 have proven capable of an avian-like preference for binding to droplets. A fourth mutation, which sustaining transmission among humans a2,3-linked sialic acids (Siaa2,3Gal), occurred naturally during the transmis- and of causing pandemics. However, iso- whereas IAVs that circulate in humans sion experiment (T318I), enhanced the lated outbreaks caused by other subtypes typically exhibit a preference for a2,6- stability of the HA molecule and further of IAV occur sporadically, igniting fears linked sialic acids (Siaa2,6Gal). The enhanced transmission (Imai et al., 2012). that a new pandemic may arise. Notable authors identified a virus containing HA The Fouchier group used a similar but modern examples of such outbreaks mutations E119G/V152I/N224K/Q226L more classical virological approach to have been those caused by H5N1 IAV in that exhibited a receptor-binding profile force adaptation of the A/Indonesia/5/ (primarly) Southeast Asia and a recent resembling that of a seasonal human IAV 2005 strain of H5N1 for the mammalian

Cell 157, April 10, 2014 ª2014 Elsevier Inc. 295 respiratory tract. Both wild-type A/ more complete picture of the phenotypic (Herfst et al., 2012). This clearly demon- Indonesia/5/2005 and a ‘‘pre-adapted’’ properties of changes that may constitute strates that the gain of function (GOF) recombinant virus containing a series increased risk of H5N1 transmissibility with regard to transmissibility was of introduced mutations (HA Q222L/ (Figure 1). accompanied by a loss of function with re- G224S is involved with switching receptor Taken together, these results sug- gard to virulence. Thus, assessment of binding preference from Siaa2,3Gal to gested that a re-assortant virus carrying how adaptations in ferrets affect viral Siaa2,6Gal; PB2 E627K is related to a the HA of A/Vietnam/1203/2004 H5N1 fitness, virulence, and transmission (both temperature adaptation that facilitates and the remaining segments of A/ in birds and other mammalian species) is replication in the human upper respiratory California/04/2009 H1N1 would require sorely needed to gain a truly holistic tract) were serially passaged ten times in as few as four amino acid substitutions perspective of the likelihood that these ferrets. Replication of the pre-adapted in the HA molecule to become transmis- viruses might cause a pandemic and virus improved with passaging, while sible in mammals, whereas wild-type what characteristics such a pandemic replication of the wild-type virus remained A/Indonesia/5/2005 H5N1 might require might exhibit. Indeed, studies in mice unchanged. The only mutation common only five mutations (three in HA, one in (Zaraket et al., 2013) have clearly demon- among the wild-type and pre-adapted PB1, and one in PB2) (Herfst et al., 2012; strated that the H5N1 adaptations that viruses after passaging was HA T156A. Imai et al., 2012). A meta-analysis of avail- increase fitness in mammals (HA-K582I) Strikingly, this mutation, like the N158D/ able H5N1 surveillance data revealed have severe consequences on viral K mutation observed by the Kawaoka that, although many of the substitutions fitness in waterfowl, and studies in guinea group, abolished a putative N-linked found to confer transmissibility among pigs (a favored model for the study of glycosylation site. The passage 10 virus ferrets are rare in nature, HA substitutions IAV transmission) have also revealed populations derived from animals inocu- N154D and T156A, which destroy an striking differences in the phenotypes lated with the pre-adapted virus were N-linked glycosylation site, and the associated with H5N1 infections when also capable of airborne transmission PB2 E627K temperature adaptation are compared to mice (Gao et al., 2009). between ferrets, whereas the wild-type much more common and regularly occur This brings us to the issue of regulatory virus was not. All animals from which together. Various mathematical models constraints. Understanding how factors transmitted virus was recovered main- were developed to predict the probability such as virulence, transmissibility, and tained the mutations introduced during that these substitutions would arise viral fitness interconnect will require pre-adaptation and also consistently together during a natural infection and GOF experiments, the use of which has harbored two novel amino acid sub- concluded that such a scenario was recently been the cause of extensive con- stitutions in HA, H103Y (located at the indeed probable. However, the proportion troversy under the new ‘‘dual-use trimer interface) and T156A (located of viruses carrying the transmissible research of concern (DURC)’’ guidelines proximal to the receptor binding site). genotype as a function of the total virus (Wolinetz, 2012). GOF experiments are However, the lowest number of sub- population within a single host was (and have always been) a fundamental stitutions relative to wild-type found in extremely low, which would likely present pillar of scientific inquiry and are essential an isolated transmissible virus was nine a formidable barrier to efficient trans- to the rigorous execution of the scientific (PB2-E627K, PB1-H99K, PB1-I368V, mission (Russell et al., 2012). method. Indeed, the sensationalization HA-H103Y, HA-T156A, HA-Q222L, HA- Though these studies have undoubt- and reactionary blow-back sparked by G224S, NP-R99K, and NP-S345N), which edly enhanced our understanding of the original H5N1 transmission studies led to uncertainty regarding the minimal IAV transmissibility, their interpretability stem largely from scientific ignorance set of substitutions necessary to confer is constrained by several scientific and with regard to how transmission, viru- transmissibility (Herfst et al., 2012). regulatory boundaries. Let us focus first lence, and fitness interrelate. Ironically, In this issue of Cell, Linster, van Bohee- on some crucial scientific considerations. the only way to address this uncertainty men, et al. expand upon this earlier The process of viral adaptation (particu- is to move forward with GOF studies that work by elucidating the minimal set larly via serial passaging in a new host) will serve to contextualize how adap- of substitutions required for airborne has historically been exploited to achieve tations that mediate mammalian trans- transmission of A/Indonesia/5/2005 attenuation of strains to be used for missibility affect other properties of the H5N1 among ferrets and by describing vaccine formulation. This concept relies virus (such as the dramatic reduction the phenotypes associated with each on the biological principle that the adap- in virulence observed by the Fouchier change. In addition to the PB2-E627K, tation of new traits (i.e., transmissibility) group, for example). These studies HA-Q222L/G224S, and HA-T156A substi- is frequently accompanied by a loss would serve to demystify the risks and tutions described above, the authors of other traits for which there is less consequences of viral adaptations that found that HA-H103Y (stabilized HA with selective pressure (i.e., host range and/ lead to mammalian transmissibility and respect to high temperature and low pH) or virulence). Indeed, the Fouchier should therefore not be restricted. and PB1-H99Y (increased polymerase group observed that, although ferrets Certainly, any future hopes for developing activity) constituted the minimal set of succumbed to intranasal inoculation with a predictive model for pandemic risk mutations required to transmit virus wild-type H5N1 at a dose of 1 3 106 assessment will rely on understanding among ferrets (Linster et al., 2014). Criti- TCID, the majority of the animals infected the sum of these properties. It is important cally, this type of analysis provides a with the transmissible virus survived to consider that, in the context GOF

296 Cell 157, April 10, 2014 ª2014 Elsevier Inc. experiments related to H5N1 signaling and antigen pro- transmission, studies are cessing/presentation, already being performed in whereas genes upregulated stringently regulated bio- later in the response were safety level 3 (BSL3) facilities more likely to be asso- by highly skilled individuals. ciated with cellular prolifera- This is a responsible, pre- tion and protein biosynthesis. cautionary approach that Further refinement showed ensures the safety of both that STAT1 upregulation the scientists performing was most pronounced 24 hr the studies, and the general after vaccination in the high public. responder group, whereas E2F2 was downregulated Vexing Vaccines postvaccination, most promi- The uncertain principles that nently on day 3 in the high determine the pandemic risk responder group. Remark- of IAV strains are mirrored by ably, the postvaccination similar uncertainties regard- expression profile of these ing the predictability of vac- two genes alone (STAT1 and cine responses designed to E2F2) was sufficient to differ- prevent IAV infections. A entiate high and low re- mechanistic understanding sponders (Bucasas et al., of the genetic and environ- Figure 2. Predictive Factors of Influenza Virus Vaccine Efficacy 2011)(Figure 2). These results mental factors that account Systems biology has facilitated the identification of pre- and postvaccination demonstrated the potential for the heterogeneous nature correlates of efficacy (examples of positive correlates shown in green; negative power of systems biology ap- correlates, red). Elucidation of the mechanistic properties of these markers will of individual responses to the allow for personalized approaches that maximize vaccine effectiveness and proaches but also highlighted same vaccine has long eluded will guide the development of next-generation vaccines. the need for even greater researchers, which in turn has levels of resolution. For which caused vaccine efficacy to cell types within the PBMC suffer. Though this problem is not unique to evaluate vaccine responses was pub- population do these markers primarily to IAV vaccines, it is additionally compli- lished in 2009 and explored the response apply? What mechanisms are respon- cated by the distinctive qualities of re- of humans to the highly effective yellow sible for the observed outcomes? Are sponses elicited by each vaccine formula- fever virus YF-17D vaccine (Querec et al., these responses specific to TIV, or do tion (for example: split, inactivated versus 2009). In this issue of Cell, Tsang and they apply more generally to other routes live-attenuated). Therefore, the identifica- colleagues (Tsang et al., 2014) describe, of vaccination? tion of predictive markers of vaccine effi- for the first time, a set of baseline pre- Elegant work by the Pulendran lab cacy is a pressing need. This type of infor- vaccination parameters that were found addressed many of these issues by mation would not only aide in enhancing to be predictive of postvaccination analyzing the immune response to both the efficacy of current vaccines through antibody responses (Figure 2). The chal- TIV and live-attenuated influenza virus more personalized approaches but would lenge moving forward will be to distill vaccine (LAIV) in healthy adults over also provide a rational basis for the re- the vast quantities of information obtained three consecutive seasons (Nakaya et al., sponses required during development of through these studies in order to assign 2011). Consistent with previous studies next-generation vaccines. biological significance to the observed (Sasaki et al., 2007), the authors found Targeted, conventional approaches trends. Indeed, the fusion of systems- that vaccination with TIV elicited the have had limited success in capturing level approaches with classical reduc- expansion of IgG-secreting plasmablasts truly predictive markers that determine tionist methods holds the promise more efficiently than administration the outcome of vaccination. However, of revolutionizing the rational design of of LAIV. The PBMC transcriptional sig- recent advances in systems biology vaccines. natures associated with each vaccine offer the opportunity to undertake much One of the earliest systems biology were also unique. The authors observed more powerful multifactorial analyses. studies of influenza virus vaccine re- marked transcriptional changes in genes These efforts are beginning to yield sponses focused on transcriptional related to innate immunity, a trend con- tremendous amounts of data regarding analyses of peripheral blood mono- sistently observed upon transcriptional the complex orchestration of factors that nuclear cells (PBMCs) derived from analyses of influenza virus vaccinees. govern the host response to vaccines, adult males who received the trivalent More recent work has demonstrated some of which may have predictive inactivated influenza virus vaccine (TIV) that neutrophils and monocytes are value with respect to vaccine efficacy. (Bucasas et al., 2011). Genes upregulated primarily responsible for contributing to The seminal study that substantiated the at early time points postvaccination this early IFN-related gene signature utility of the systems biology approach were largely involved in interferon (IFN) (Obermoser et al., 2013). Upregulation

Cell 157, April 10, 2014 ª2014 Elsevier Inc. 297 of genes involved in the type I IFN populations expanded most on day 1, required for optimal responses, while pathway were especially pronounced for whereas adaptive cell types, includ- baseline predictors can be exploited individuals who received LAIV, likely a ing plasmablasts, B cell subsets, and to assess optimal vaccine routes/formula- reflection of its replicative capacity. Inter- T cells, were representative of the day 7 tions on an individual basis. Coupled with estingly, whereas transcriptional changes response. ambitious ongoing efforts to elicit more associated with B cells correlated posi- Interestingly, high baseline antibody broadly protective responses against tively with antibody titers, T-cell-asso- titers were found to be inversely corre- influenza viruses (Krammer and Palese, ciated signatures exhibited a negative lated with the postvaccination response. 2014), the hope for an effective ‘‘univer- correlation with the antibody response. Baseline gene expression and pathway sal’’ influenza virus vaccine may soon be Using discriminant analysis via mixed activity were not robust predictors realized. integer profiling (DAMIP), the authors of postvaccination response; however, were able to identify and validate a mini- several pre-vaccination B and T cell ACKNOWLEDGMENTS mal gene signature capable of accurately subpopulations were found to predict We apologize to our colleagues whose work could predicting the antibody responses to TIV. endpoint antibody responses. Naive, not be cited due to space limitations. M.S.M. is Critically, to confirm the utility of their transitional, and memory B cell popula- supported, in part, by a Canadian Institutes of analyses, the authors selected one of tions were all positively correlated with Health Research Postdoctoral Fellowship. P.P. is the markers identified in the DAMIP antibody endpoint titers, whereas effector supported, in part, by NIH P01 AI097092-01 and analysis (calcium/calmodulin-dependent memory CD4+ T cell and perforin+ CD8+ HHSN266200700010C. protein kinase IV [CamkIV]) for functional T cell populations exhibited a negative REFERENCES validation. CamKIV expression at day 3 correlation (Figure 2). Retrospective post-TIV vaccination negatively corre- analyses using these predictive cell Bucasas, K.L., Franco, L.M., Shaw, C.A., Bray, lated with the magnitude of the anti- populations could then be used to assign M.S., Wells, J.M., Nin˜ o, D., Arden, N., Quarles, body response (Figure 2). Vaccination functional significance to the gene J.M., Couch, R.B., and Belmont, J.W. (2011). Early of CamKIVÀ/À mice with TIV recapitulated expression signatures observed from patterns of gene expression correlate with the this effect, inducing higher antibody titers whole PBMCs. It should be noted that humoral immune response to influenza vaccination 203 than in control mice (Nakaya et al., 2011). the baseline state of antibody titers and in humans. J. Infect. Dis. , 921–929. The biological signatures identified cellular composition of PBMCs among Gao, Y., Zhang, Y., Shinya, K., Deng, G., Jiang, Y., Li, Z., Guan, Y., Tian, G., Li, Y., Shi, J., et al. (2009). in these and other studies have sig- individuals is likely to reflect age, at least Identification of amino acids in HA and PB2 critical nificantly advanced our understanding to some extent. Thus, the strength with for the transmission of H5N1 avian influenza vi- of the specific responses elicited by which these predictors apply to individ- ruses in a mammalian host. PLoS Pathog. 5, various vaccines. They have also facili- uals in different age categories remains e1000709. tated the identification of postvaccina- to be determined. Girard, M.P., Tam, J.S., Assossou, O.M., and tion markers, such as STAT1, E2F2 Kieny, M.P. (2010). The 2009 A (H1N1) influenza vi- (Bucasas et al., 2011), CamkIV (Nakaya Looking into the Future rus pandemic: A review. Vaccine 28, 4895–4902. et al., 2011), and CD4+ T cell levels Experimental assessment of the adap- Herfst, S., Schrauwen, E.J.A., Linster, M., Chutini- (Nayak et al., 2013; Spensieri et al., tions required for IAV transmission mitkul, S., de Wit, E., Munster, V.J., Sorrell, E.M., Bestebroer, T.M., Burke, D.F., Smith, D.J., et al. 2013), which can accurately predict among mammals and the consequences (2012). Airborne transmission of influenza A/H5N1 the magnitude of the antibody response of these adaptations on other properties virus between ferrets. Science 336, 1534–1541. elicited by influenza virus vaccination of the virus (especially virulence and Imai, M., Watanabe, T., Hatta, M., Das, S.C., (Figure 2). However, the ability to predict fitness) promise to provide a more robust Ozawa, M., Shinya, K., Zhong, G., Hanson, A., Kat- the outcome of vaccination based on framework for assessing the pandemic sura, H., Watanabe, S., et al. (2012). Experimental baseline immunological markers has risk of novel IAV strains (Linster et al., adaptation of an influenza H5 HA confers respira- remained elusive. Tsang and colleagues 2014). However, this goal will require a tory droplet transmission to a reassortant H5 486 provide a first glimpse of the potential regulatory environment that dissolves HA/H1N1 virus in ferrets. Nature , 420–428. for systems biology to address this the unnecessary barriers that hinder Julien, J.-P., Lee, P.S., and Wilson, I.A. (2012). Structural insights into key sites of vulnerability problem. In agreement with the work much-needed progress while maintaining on HIV-1 Env and influenza HA. Immunol. Rev. described above (Bucasas et al., 2011; a reasonable level of oversight that en- 250, 180–198. Nakaya et al., 2011; Obermoser et al., courages public trust. In the meantime, Kasturi, S.P., Skountzou, I., Albrecht, R.A., Koutso- 2013), postvaccination PBMC gene ex- the use of systems biology techniques nanos, D., Hua, T., Nakaya, H.I., Ravindran, R., pression profiles were characterized by (Tsang et al., 2014) to assemble a more Stewart, S., Alam, M., Kwissa, M., et al. (2011). strong enrichment of IFN-related gene comprehensive picture of the factors Programming the magnitude and persistence of expression at day 1 postvaccination, that govern the immune response to antibody responses with innate immunity. Nature 470 which then shifted to adaptive pathways vaccination should serve as a guide for , 543–547. associated with antibody production the design of more efficacious vaccines. Kimble, J.B., Sorrell, E., Shao, H., Martin, P.L., and Perez, D.R. (2011). Compatibility of H9N2 avian by day 7. These results correlated well Identifying predictive postvaccination influenza surface genes and 2009 pandemic with changes in cell subset frequencies; markers of efficacy will inform the H1N1 internal genes for transmission in the ferret CD40+ and CD86+ monocytes and development of novel adjuvants (Kasturi model. Proc. Natl. Acad. Sci. USA 108, 12084– IFNa+ plasmacytoid dendritic cell (DC) et al., 2011) that stimulate pathways 12088.

298 Cell 157, April 10, 2014 ª2014 Elsevier Inc. Krammer, F., and Palese, P. (2014). Universal Osterholm, M.T., Kelley, N.S., Sommer, A., and and predict antibody responses. Proc. Natl. influenza virus vaccines: need for clinical trials. Belongia, E.A. (2012). Efficacy and effectiveness Acad. Sci. USA 110, 14330–14335. 15 Nat. Immunol. , 3–5. of influenza vaccines: a systematic review and Stern, A.M., and Markel, H. (2005). The history of 12 Linster, M., van Boheemen, S., de Graaf, M., meta-analysis. Lancet Infect. Dis. , 36–44. vaccines and immunization: familiar patterns, new Schrauwen, E.J.A., Lexmond, P., Ma¨ nz, B., Beste- Querec, T.D., Akondy, R.S., Lee, E.K., Cao, W., challenges. Health Aff. (Millwood) 24, 611–621. broer, T.M., Baumann, J., van Riel, D., Rimmelz- Nakaya, H.I., Teuwen, D., Pirani, A., Gernert, K., To, K.K.W., Chan, J.F.W., Chen, H., Li, L., and waan, G.F., et al. (2014). Identification, character- Deng, J., Marzolf, B., et al. (2009). Systems bio- Yuen, K.-Y. (2013). The emergence of influenza A ization, and natural selection of mutations driving logy approach predicts immunogenicity of the H7N9 in human beings 16 years after influenza A airborne transmission of A/H5N1 virus. Cell 157, yellow fever vaccine in humans. Nat. Immunol. H5N1: a tale of two cities. Lancet Infect. Dis. 13, this issue, 329–339. 10, 116–125. 809–821. Nakaya, H.I., Wrammert, J., Lee, E.K., Racioppi, L., Russell, C.A., Fonville, J.M., Brown, A.E.X., Burke, Tong, S., Zhu, X., Li, Y., Shi, M., Zhang, J., Bour- Marie-Kunze, S., Haining, W.N., Means, A.R., D.F., Smith, D.L., James, S.L., Herfst, S., geois, M., Yang, H., Chen, X., Recuenco, S., Go- Kasturi, S.P., Khan, N., Li, G.-M., et al. (2011). van Boheemen, S., Linster, M., Schrauwen, E.J., mez, J., et al. (2013). New world bats harbor Systems biology of vaccination for seasonal influ- et al. (2012). The potential for respiratory droplet- diverse influenza A viruses. PLoS Pathog. 9, enza in humans. Nat. Immunol. 12, 786–795. transmissible A/H5N1 influenza virus to evolve in e1003657. Nayak, J.L., Fitzgerald, T.F., Richards, K.A., Yang, a mammalian host. Science 336, 1541–1547. Tsang, J.T., Schwartzberg, P.L., Kotliarov, Y., H., Treanor, J.J., and Sant, A.J. (2013). CD4+ Sasaki, S., Jaimes, M.C., Holmes, T.H., Dekker, Biancotto, A., Xie, Z., Germain, R.N., Wang, E., T-cell expansion predicts neutralizing antibody C.L., Mahmood, K., Kemble, G.W., Arvin, A.M., Olnes, M.J., Narayanan, M., Golding, H., et al. responses to monovalent, inactivated 2009 and Greenberg, H.B. (2007). Comparison of (2014). Global analyses of human immune varia- pandemic influenza A(H1N1) virus subtype H1N1 the influenza virus-specific effector and memory tion reveal baseline predictors of postvaccination vaccine. J. Infect. Dis. 207, 297–305. B-cell responses to immunization of children and responses. Cell 157, this issue, 499–513. Obermoser, G., Presnell, S., Domico, K., Xu, H., adults with live attenuated or inactivated influenza Wang, Y., Anguiano, E., Thompson-Snipes, L., virus vaccines. J. Virol. 81, 215–228. Wolinetz, C.D. (2012). Implementing the new U.S. 336 Ranganathan, R., Zeitner, B., Bjork, A., et al. Spensieri, F., Borgogni, E., Zedda, L., Bardelli, M., dual-use policy. Science , 1525–1527. (2013). Systems scale interactive exploration Buricchi, F., Volpini, G., Fragapane, E., Tavarini, S., Zaraket, H., Bridges, O.A., and Russell, C.J. (2013). reveals quantitative and qualitative differences Finco, O., Rappuoli, R., et al. (2013). Human The pH of activation of the hemagglutinin protein in response to influenza and pneumococcal circulating influenza-CD4+ ICOS1+IL-21+ T cells regulates H5N1 influenza virus replication and vaccines. Immunity 38, 831–844. expand after vaccination, exert helper function, pathogenesis in mice. J. Virol. 87, 4826–4834.

Cell 157, April 10, 2014 ª2014 Elsevier Inc. 299