Induction of proinflammatory cytokines in human macrophages by influenza A (H5N1) viruses: a mechanism for the unusual severity of human disease?

Transcription

1 Mechanisms of disease Induction of proinflammatory cytokines in human macrophages by influenza A (H5N1) viruses: a mechanism for the unusual severity of human disease? C Y Cheung, L L M Poon, A S Lau, W Luk, Y L Lau, K F Shortridge, S Gordon, Y Guan, J S M Peiris Summary Background In 1997, the first documented instance of human respiratory disease and death associated with a purely avian H5N1 influenza virus resulted in an overall casefatality rate of 33%. The biological basis for the severity of human H5N1 disease has remained unclear. We tested the hypothesis that virus-induced cytokine dysregulation has a role. Methods We used cdna arrays and quantitative RT-PCR to compare the profile of cytokine gene epression induced by viruses A/HK/486/97 and A/HK/483/97 (both H5N1/97) with that of human H3N2 and H1N1 viruses in human primary monocyte-derived macrophages in vitro. Secretion of tumour necrosis factor (TNF ) from macrophages infected with the viruses was compared by ELISA. By use of naturally occurring viral reassortants and recombinant viruses generated by reverse genetic techniques, we investigated the viral genes associated with the TNF- response. Findings The H5N1/97 viruses induced much higher gene transcription of proinflammatory cytokines than did H3N2 or H1N1 viruses, particularly TNF and interferon beta. The concentration of TNF- protein in culture supernatants of macrophages infected with these viruses was similar to that induced by stimulation with Escherichia coli lipopolysaccharide. The non-structural (NS) gene-segment of H5N1/97 viruses contributed to the increase in TNF induced by the virus. Interpretation The H5N1/97 viruses are potent inducers of proinflammatory cytokines in macrophages, the most notable being TNF. This characteristic may contribute to the unusual severity of human H5N1 disease. Lancet 22; 36: See Commentary page 181 Departments of Microbiology (C Y Cheung MPhil, L L M Poon DPhil, W Luk BSc, Prof K F Shortridge PhD, Y Guan PhD, Prof J S M Peiris DPhil) and Paediatrics (A S Lau MD, Prof Y L Lau MD), University of Hong Kong, Queen Mary Hospital, Hong Kong SAR, China; and Sir William Dunn School of Pathology, University of Oford, Oford, UK (Prof S Gordon PhD) Correspondence to: Prof J S M Peiris, Department of Microbiology, University of Hong Kong, University Pathology Building, Queen Mary Hospital, Pokfulam, Hong Kong SAR, China ( malik@hkucc.hku.hk) Introduction The H5N1 avian flu in Hong Kong in 1997 was the first documented instance of human respiratory disease and death associated with a purely avian influenza virus 1,2 and was thought to be an incipient pandemic situation. The human disease was unusually severe, with an overall case-fatality rate of 33%. Ecept for one patient who was substantially immunocompromised, underlying disorders did not eplain the severity of the course of the disease. 3 The basis for the unusual clinical presentation and severity of illness in patients with H5N1 disease remains unknown. Although the major determinants of pathogenicity of influenza viruses in avian species are well defined, those in mammals are unclear. Eperimental H5N1/97 infection in mice, ferrets, and macaques has given conflicting results. In eperimentally infected BALB/c mice, some human H5N1 viruses (eg, 483/97) led to disseminated disease but others (eg, 486/97) did not; these differences were attributed to mutations in the polymerase basic protein 2 (PB2) and HAEMAGGLUTININ (HA) genes. 4 However, differences in virus dissemination between these viruses were not demonstrable in eperimentally infected ferrets. 5 No evidence of productive virus replication outside the respiratory tract was seen in limited autopsy studies of patients dying from H5N1 disease 6 or after eperimental infection of macaques in which pathological features similar to those of human H5N1 disease were reproduced. 7 Patients with H5N1 disease had a primary viral pneumonia complicated by syndromes of acute respiratory distress and multiple organ dysfunction. Lymphopenia and haemophagocytosis were notable findings. 3,6 Haemophagocytosis and the syndromes of acute respiratory distress and multiple organ dysfunction are associated with cytokine dysregulation. 8,9 We suggest that the clinical features of severe human H5N1 disease are compatible with virus-induced cytokine dysregulation. To test our hypothesis, we used human primary monocyte-derived macrophages as an in-vitro model. The cytokine gene and protein epressions induced by H5N1/97 influenza viruses were compared with those of contemporary human H1N1 or H3N2 viruses and those of avian origin. Methods Viruses and cells Viruses, their abbreviations, origins, and passage history are shown in table 1. An influenza A (H5N1/97) isolate from a patient with fatal H5N1 disease (483/97) and one from a patient with mild disease (486/97) were compared with contemporary human H1N1 (54/98) and H3N2 (1174/99) viruses. Avian viruses genetically related or unrelated to H5N1/97 were included to help identify the genes causing the biological effects associated with H5N1/97 viruses. These viruses are believed to be natural reassortants that THE LANCET Vol 36 December 7,

2 GLOSSARY HAEMAGGLUTININ (HA) AND NEURAMINIDASE (NA) The two surface glycoproteins of the influenza A virus. Influenza A viruses are separated into subtypes on the basis of these glycoproteins. There are 15 HA (H1 15) and nine NA (N1 9) subtypes recognised at present. A virus is designated on the basis of HA and NA, for eample H3N2. HOUSEKEEPING GENES Genes that are constitutively epressed in virtually all cells, since they are fundamental to all cellular functions. The epression of such genes can serve as a reference against which the epression of other genes can be normalised. REVERSE GENETICS The generation of recombinant viruses containing a genome derived from cloned cdnas. derived their HA gene from a goose virus Gs/Gd/96-like H5N1 and their internal genes (PB2, PB1, polymerase acidic protein [PA], nucleoprotein [NP], matri protein [M], non-structural protein [NS]) from G1/97-like H9N2 or W312/97-like H6N1 virus lineages. 1 Representatives of each of these lineages (437.6/99, H5N1; G1/97, H9N2; and W312/97, H6N1, respectively) were included in the analysis. 11 The Y28/97 (H9N2) virus, which shares no close evolutionary relation with H5N1, was used as a control (table 1). 1 The viruses isolated from human beings 483/97 (H5N1), 486/97 (H5N1), 54/98 (H1N1), and 1174/99 (H3N2) were cloned by limiting dilution, and seed virus stocks were prepared solely in MDCK (Madin-Darby canine kidney) cells. The other avian viruses were initially isolated and cloned by limiting dilution in 9 1-day-old fertilised hens eggs, but the final eperimental stock virus was prepared in MDCK cells. Two viruses, 437.6/99 (H5N1) and W312/97 (H6N1) did not replicate to high titre in MDCK cells, and virus stocks prepared from infected egg allantoic fluid were used for the eperiments. Virus infectivity was assessed by titration in MDCK cells. The virus titres of 437.6/99 (H5N1) and W312/97 (H6N1) were measured by titration in fertilised hens eggs. To avoid the effects of pre-eisting cytokines and other factors (eg, double-stranded RNA) that might be present in the virus culture supernatant, the preparations were purified by adsorption to and elution from turkey red blood cells, 12 unless otherwise indicated. Influenza virus infection of macrophages Peripheral-blood leucocytes were separated from buffy coats of healthy blood donors (from the Hong Kong Red Cross Blood Transfusion Service) by centrifugation on a Ficoll-Paque density gradient (Pharmacia Biotech, Uppsala, Sweden) and purified by adherence. 13 The research protocol was approved by the ethics committee of the University of Hong Kong. Macrophages were seeded onto tissue culture plates in RPMI 164 medium supplemented with 5% heatinactivated autologous plasma. The purity of the monocyte preparations was confirmed by immunostaining for CD14 (BD Biosciences, San Diego, CA, USA). The cells were allowed to differentiate for 14 days in vitro. Differentiated macrophages (from monocytes seeded at cells per well in 24-well tissue-culture plates) were infected at a multiplicity of infection of two unless otherwise indicated. After 3 min of virus adsorption, the virus inoculum was removed, and the cells were washed with warm culture medium and incubated in macrophage SFM medium (GIBCO BRL, Gaithersburg, MD, USA) supplemented with 6 mg/l penicillin, 6 mg/l streptomycin, and 2 mg/l N-p-tosyl-Lphenylalaninechloromethyl ketone-treated trypsin (Sigma, St Louis, MO, USA). Escherichia coli serotype O55:B5 lipopolysaccharide (Sigma) was used at 1 g/l as a positive control. Samples of culture supernatant were collected for virus titration and cytokine analysis. RNA was etracted from cells for analysis of cytokine gene epression. 8 h after infection, replicate cell monolayers were fied and analysed by immunofluorescent staining specific for influenza virus nucleoprotein (DAKO Imagen, Dako Diagnostics Ltd, Ely, UK). Macrophage response profiling with a cytokine cdna array 3 h after infection with virus at multiplicity of infection of two, RNA was etracted from macrophages seeded on si-well plates, and phosphorus-32-labelled cdna was hybridised onto Atlas human cytokine/receptor cdna epression arrays (Clontech, Palo Alto, CA, USA) according to the manufacturer s instructions. Autoradiograms were scanned (Microteck Scan Maker 4) and analysed with AtlasImage 1. software (Clontech). Gene epression was normalised to a panel of nine HOUSEKEEPING GENES and epressed as a ratio over that in mock-infected cells. Quantification of mrna by real-time quantitative RT-PCR DNase-treated total RNA was isolated by means of RNeasy Mini kit (Qiagen, Hilden, Germany). The cdna was synthesised from mrna with poly(dt) primers and Superscript II reverse transcriptase (Life Technologies, Rockville, MD, USA) and quantified by real-time PCR analysis with a LightCycler (Roche, Mannheim, Germany). The methods used for quantifying tumour necrosis factor (TNF ) and -actin mrna have been described previously. 14 Other cytokine mrnas were assayed by commercial kits (Search LC GmbH, Germany). Quantification of cytokines by ELISA The concentrations of TNF in macrophage supernatant were measured by a specific ELISA assay (R&D Systems, Minneapolis, MN, USA). Samples of culture supernatant Virus and subtype Abbreviation Source Passage history from Gene segments with close genetic relatedness to initial isolation 483/97 and viruses* A/Hong Kong/483/97 (H5N1) 483/97 (H5N1/97) Human MDCK 7 Not applicable A/Hong Kong/486/97 (H5N1) Human MDCK 7 Not applicable A/Hong Kong/54/98 (H1N1) 54/98 (H1N1) Human MDCK 7 None A/Hong Kong/1174/99 (H3N2) 1174/99 (H3N2) Human MDCK 7 None A/Quail/Hong Kong/G1/97 (H9N2) G1/97 (H9N2) Quail faeces Egg 2; MDCK 3 PB1, PB2, PA, NP, M, NS A/Duck/Hong Kong/Y28/97 (H9N2) Y28/97 (H9N2) Duck faeces Egg 2; MDCK 3 None A/Goose/Hong Kong/437.6/99 (H5N1) 437.6/99 (H5N1) Goose faeces Egg 3 HA A/Teal/Hong Kong/W312/97 (H6N1) W312/97 (H6N1) Teal faeces Egg 3 PB1, PB2, PA, NP, NA, M, NS NA=NEURAMINIDASE. *See references 1 and 11. Antigenic characterisation: A/Hong Kong/54/98 is A/Johannesburg/82/96-like; A/Hong Kong/1174/99 is A/Sydney/5/97-like. Table 1: Viruses, source, passage history, and genetic homology with H5N1/97 viruses 1832 THE LANCET Vol 36 December 7, 22

3 were irradiated with ultraviolet light (CL-1 Ultra Violet Cross linker) for 15 min before the ELISAs were done, to inactivate any infectious virus. Previous eperiments had confirmed that the dose of ultraviolet light used did not affect cytokine concentrations as measured by ELISA (data not shown). Recombinant viruses generated by reverse genetics Recombinant viruses were generated by a recently established REVERSE GENETICS system. 15 RT-PCR products of viral RNA segments were cloned into plasmid phw2. 15 The newly introduced viral RNA in the recombinant viruses was sequenced for confirmation. All procedures involving live H5N1 viruses and recombinant viruses were carried out in a facility of biosafety level 3. Role of the funding source The sponsors of the study had no role in study design, collection, analysis, or interpretation of data, or in the writing of the report. Results The human and avian viruses replicated to similar titres (about 1 4 log 1 tissue-culture infectious dose [TCID] 5 per ml) after infection of macrophages at a multiplicity of infection of two, apart from the human H1N1 virus, which replicated to slightly higher titre (1 5 log 1 TCID 5 per ml). 8 h after infection, similar proportions (9 1%) of macrophages had evidence of viral antigen (nucleoprotein) epression as assessed by immunofluorescence. Several cytokine genes were upregulated when assessed 3 h after infection with influenza virus (table 2). However, upregulation of mrna for TNF, interferon beta, RANTES (regulated on activation, normal T cell Gene product Ratio of epression over mock-infected cells 486/97 483/ /99 54/ /99 (H5N1/97) (H5N1/97) (H5N1) (H1N1) (H3N2) Interleukin Interleukin 1 2 7* Interleukin Interleukin * Interleukin Interleukin * 2 * Interleukin Interleukin 1 2 7* Interleukin 12 2 * subunit Interleukin * 2 * * subunit Interleukin Interleukin Macrophage 6 6* 6 9* * inflammatory protein 1 Macrophage 6 4* 7 * inflammatory protein 1 Monocyte 2 4* 2 5* chemotactic protein 1 RANTES 4 2* 2 5* Interferon alpha 2 3* Interferon beta 34 * 4 * 14 * 6 6* 1 * TNF 25 * 3 * 5 7* 4 * 5 6* Signals were normalised to a panel of housekeeping genes and epressed as a ratio over mock-infected cells. *Upregulation by two or more times over that of mock infection. Downregulation by two or more times below that of mock infection. Table 2: cdna array profile of cytokine gene epression of primary human macrophages 3 h after infection with H5N1/97 or other human and avian influenza viruses epressed and secreted), macrophage inflammatory protein (MIP) 1 and 1, and monocyte chemotactic protein 1 (MCP-1) was much greater after infection with the H5N1/97 viruses 486/97 and 483/97 than after infection with H3N2 or H1N1 viruses. In addition, one or the other H5N1/97 virus also differentially upregulated interferon alpha and interleukins 12, 1, 1, and 4. The cytokine mrna profile elicited by 437.6/99 (H5N1), which shares a common evolutionary origin for only one (ie, HA) of eight genes with H5N1/97 viruses, 11 was similar to that of human H1N1, rather than of H5N1/97 viruses. The cytokine genes found to be differentially upregulated by the H5N1/97 viruses in the cdna array were studied by quantitative RT-PCR. The cytokine transcription profiles at 3 h and 6 h after infection essentially confirmed the findings of the cdna array (figure 1). In comparison with 54/98 (H1N1) virus, both H5N1/97 viruses differentially upregulated transcription of TNF, interferon beta, interleukin 1, MCP-1, MIP-1, RANTES, interferon alpha, interleukin 12 (figure 1), and MIP-1 (data not shown). Broadly, there were two patterns of kinetics of cytokine gene transcription: cytokines progressively upregulated from 1 h after infection onwards (eg, TNF, interferon beta, interleukin 1, interferon alpha) and those with a late upregulation at 6 h after infection (eg, MCP-1, RANTES, interleukin 12). The avian G1/97 (H9N2) virus, which shares a common evolutionary origin with H5N1/97 for the si internal genes, 1 showed a cytokine profile similar to that of the H5N1/97 viruses. By contrast, cells infected with Y28/97 (H9N2), which does not share close genetic relation with H5N1/97, showed a cytokine profile similar to human 54/98 (H1N1) virus (figure 1). Since interleukin 6 is part of the cytokine cascade triggered through TNF, 16 mrna for interleukin 6 was also quantified. 6 h after infection, the concentrations of interleukin 6 mrna in macrophages infected with, 483/97 (H5N1/97), and G1/97 (H9N2) were over si times greater than those found with cells infected with 54/98 (H1N1) or Y28/97 (H9N2) (data not shown). TNF- mrna concentrations after infection with 486/97 and 483/97 (H5N1/97) viruses and G1/97 (H9N2) virus were similar to those resulting from stimulation of macrophages with E coli lipopolysaccharide (figure 1). The overall observations remained valid whether the TNF- mrna data were analysed with or without normalisation for -actin mrna concentrations. The differential effect between H5N1/97 and 54/98 (H1N1) viruses on TNF- mrna upregulation remained even when cells were treated with cycloheimide (1 mg/l) from 1 h before infection until the end of the eperiment (data not shown). An increase in the multiplicity of infection of 54/98 (H1N1) virus from two to 2 did not result in TNF- concentrations similar to those induced by H5N1/97 (data not shown). TNF has well-documented immunopathological potential, so concentrations of the protein were measured in culture supernatants of infected macrophages by ELISA. The concentrations peaked h after infection (figure 2). Supernatants from macrophage cultures infected with, G1/97 (H9N2), and W312/97 (H6N1) viruses had much higher concentrations of TNF than those infected with 54/98 (H1N1), 437.6/99 (H5N1), or Y28/97 (H9N2). To confirm the reproducibility of these results, macrophages derived from three donors were studied in parallel eperiments (figure 3). The macrophages infected with or G1/97 (H9N2) showed TNF- secretion similar to those stimulated with THE LANCET Vol 36 December 7,

4 6 4 TNF 483/97 (H5N1/97) G1/97 (H9N2) Y28/97 (H9N2) 54/98 (H1N1) Mock infection Lipopolysaccharide Interferon beta Copies of mrna per 1 4 copies of -actin mrna > Interleukin 1 MCP Interferon alpha MIP RANTES 6 Interleukin 12 p Time since infection (h) Figure 1: Cytokine gene epression profile of influenza-virus-infected human macrophages by quantitative RT-PCR Cytokine mrna concentrations were assayed 1 h, 3 h, and 6 h after infection with the influenza viruses. The concentrations of cytokine mrna were normalised to those of -actin mrna in the corresponding sample. Means of duplicate assays are shown. lipopolysaccharide. The amount of TNF secreted by cells infected with was significantly higher (Student s t test, p= 2) than that secreted by cells infected with 54/98 (H1N1). Thus, the ELISA results for TNF- secretion corroborate the findings on TNF- mrna concentrations observed in the quantitative RT-PCR eperiments as well as the cdna array analysis. These differences in TNF- gene epression were not related to the numbers of cells infected (as judged by immunofluorescence for viral antigen), differences in virus replication competence, or differences in passage history (table 1). Cytokines or other factors carried over in the virus inoculum were unlikely to have confounded the results, since the results were essentially the same when purified or unpurified virus was used as virus inoculum (data not shown). Furthermore, 1834 THE LANCET Vol 36 December 7, 22

5 TNF released (ng/l) MECHANISMS OF DISEASE W312/97 (H6N1) G1/97 (H9N2) Y28/97 (H9N2) inactivation of the virus by ultraviolet irradiation before infection of the macrophages abolished TNF- induction (data not shown). Thus, replicating virus was required for a TNF- response to be elicited. In view of previous evidence that 483/97 virus was more virulent than 486/97 in BALB/c mice, 4 we directly compared TNF- gene transcription and protein epression of these two viruses. Although 483/97 virus showed slightly higher concentrations of TNF- mrna 1 h after infection, the concentrations were similar at 3 h, as was TNF- protein epression. Similar results were obtained for gene transcription of interferon beta (data not shown). The TNF- transcription and secretion assays showed that G1/97 (H9N2) and W312/97 (H6N1) viruses (which share the internal gene constellation of H5N1/97 viruses 1 ) also share the high-tnf- phenotype. By contrast, the 437.6/99 (H5N1) and Y28/97 (H9N2) viruses (which do not share internal genes with a close evolutionary relation to H5N1/97 11 ) are low-tnf- 54/98 (H1N1) Mock infection 437 6/99 (H5N1) Time after infection (h) Figure 2: TNF- production by primary macrophages Means of duplicate assays at various times after infection are shown. TNF released (ng/l) <8 6 h after infection 12 h after infection LPS G1/97 (H9N2) Y28/97 (H9N2) 54/98 (H1N1) 1174/99 (H3N2) Mock Figure 3: TNF- production by macrophages from three donors Concentrations in culture supernatants from virus-infected or lipopolysaccharide-stimulated (LPS) macrophages collected at 6 h or 12 h after infection; means (SE) of three eperiments from different macrophage donors were calculated. TNF released (ng/l) 8 6 h after infection 7 12 h after infection <8 WSN/33 (H1N1) rwsn H5-NS rwsn G1-NS G1/97(H9N2) rwsn Y28-NS Y28/97 (H9N2) Figure 4: TNF in culture supernatants of macrophages infected with recombinant viruses Means (SD) of duplicate cultures are shown. inducers. These results pointed to the internal genes of H5N1/97 as crucial to the high-tnf- phenotype. The NS1 gene of influenza viruses is important in regulating epression of type 1 interferon. 17 Hence, the role of the NS gene segment was further investigated by constructing recombinant viruses that carried this gene segment from, G1/97 (H9N2), or Y28/97 (H9N2) virus in a background of A/WSN/33. A plasmid-derived A/WSN/33 wild-type was used as a control. Non-purified recombinant and wild-type viruses were tested on macrophages with and without ultraviolet inactivation to control for any biologically active mediators that could have been carried over in the infecting inoculum. Macrophage cultures infected with the reassembled WSN virus produced low concentrations of TNF. The recombinant viruses containing the NS gene segment from 486/97 (H5N1; rwsn H5-NS) or G1/97 (H9N2; rwsn G1-NS) produced higher concentrations of TNF (figure 4), although the concentrations were lower than those of the wild-type virus. The recombinant containing the NS gene segment of Y28/97 (H9N2; rwsn Y28- NS) produced little TNF. Ultraviolet irradiation of the virus inoculum before infection abolished TNF- secretion in all instances (data not shown). The titres of infectious virus in the macrophage culture supernatants infected with WSN/33 (wild type) and the recombinants were similar (data not shown), confirming the replication competence of the recombinant viruses. Discussion Previous studies have shown that human influenza viruses infect macrophages inducing the release of TNF, interleukins 6 and 1, interferons alpha and beta, and chemokines such as RANTES, MIP-1 and MIP-1, MCP-1, MCP-3, and interleukin 1. 18,19 The cytokine mrna profile and the kinetics of TNF- production from macrophages in response to infection with human H1N1 virus in our eperiments (figures 1 3) were similar to those documented previously. However, we found that H5N1/97 viruses were more potent upregulators of cytokine transcription and led to TNF- secretion similar to that after stimulation with lipopolysaccharide (figures 1 and 3). These novel findings may eplain the unusual clinical presentation and disease severity associated with human H5N1/97 infection. 3 THE LANCET Vol 36 December 7,

6 Cytokine responses in vivo result from a comple network of autocrine and paracrine interactions involving many cell types. In this investigation, we chose to study a central cell of the innate immune response the macrophage. We used the monocyte-derived macrophage to represent the lung macrophage, a potent source of cytokines. Both cell types are recently derived from peripheral-blood monocytes and are similar physiologically. 2 Similar eperiments in, for eample, primary bronchial epithelial cells, are warranted. Which of these cytokine responses are primary (ie, directly induced by the virus) and which are the secondary cytokine cascade resulting from paracrine responses is unclear. We used a high multiplicity of infection, resulting in the infection of almost all cells in the first round of viral replication. Thus, the mrna concentrations 1 3 h after infection (eg, the differential responses in TNF and interferon beta mrna after H5N1/97 and G1/97) are more likely to reflect primary events in response to the virus. The cytokines upregulated later (eg, MCP-1, RANTES, interleukin 12) could well reflect the secondary cascade of cytokine responses. Primary human macrophages produce MCP-1, MIP-1, MIP-1, and RANTES in response to stimulation with TNF. 21 Taken together with the result that the H5N1-mediated differential upregulation of TNF- transcription is observed in cells infected in the presence of cycloheimide (an inhibitor of protein synthesis), these findings suggest that the TNF- response we observed is largely a primary virus-mediated effect. Some cytokines are regulated at both the transcriptional and the post-transcriptional level. Post-transcriptional effects on cytokine regulation are not detectable by gene-epression profiling alone. For one cytokine (TNF ) that is pathogenetically relevant, we have shown that the H5N1/97-associated differential upregulation is also reflected at the level of protein secretion. The systemic inflammatory response, multiorgan dysfunction, and acute respiratory distress syndrome, and reactive haemophagocytosis were distinctive features in patients with severe H5N1 disease. 3,6 TNF is involved in the pathogenesis of haemophagocytosis. 8 TNF, and the cascade of chemotactic cytokines it induces, result in the accumulation of neutrophils in the lung, induce the respiratory burst response and neutrophil degranulation, and are implicated in the pathogenesis of the acute respiratory distress syndrome. 9 Thus, our finding that H5N1/97 infection leads to eaggerated TNF and other cytokine responses from primary human macrophages in vitro is relevant to the clinical manifestations of severe H5N1 disease. Adult monocytes produce higher concentrations of TNF in response to lipopolysaccharide stimulation than monocytes derived from cord blood. 22 This difference could eplain the age-related severity of disease observed in individuals with H5N1 disease. 3 In a mouse model of infection with influenza virus and respiratory syncytial virus, inhibition of TNF reduced the pulmonary recruitment of inflammatory cells and the severity of illness. 23 In uncomplicated H1N1 influenza in human beings, TNF, interleukin 6, and interferon alfa are detectable in the nasal lavage fluid, and the concentrations of some of these are related to disease severity. 24 The cytokine profile in the lungs of patients with H5N1 disease is unknown, but concentrations of TNF, interleukin 6, interferon gamma, and soluble interleukin-2 receptor were higher than normal in serum samples taken 7 11 days after the onset of illness. 6 However, cytokine responses are largely compartmentalised, and serum cytokine concentrations are a poor reflection of the response in the local tissue and microenvironment. 25 Comparative studies with avian influenza viruses with gene segments that were genetically closely related to H5N1/97 allowed us to conclude that the viral internal genes, rather than surface glycoprotein genes, are important for the high-tnf- phenotype. Eperiments with recombinant viruses carrying the NS gene segment of, G1/97 (H9N2), and Y28/97 (H9N2) viruses in a background of A/WSN/33 led us to conclude that the NS gene of H5N1/97 and related viruses (G1/97) contributes, at least partly, to the increased epression of TNF in macrophages (figure 4). The role of the NS gene or gene products on TNF- secretion has not previously been documented. The NS gene segment of influenza A virus codes for NS1 and viral nuclear eport protein (NEP). The NS1 gene or gene product is believed to be crucial in allowing the virus to evade the host interferon response and is therefore a determinant of virus virulence. 17,26 This evasion may occur through avoiding activation of the cell-signalling pathways (interferon regulatory factor-3 and nuclear factor B) and protein kinases associated with interferon induction. 26 Alternatively, there could be action at the posttranscriptional level to inhibit the maturation of cellular pre-mrna. This inhibitory effect on the maturation of antiviral mrna can be counteracted by introduction of a point mutation that abolishes the binding of the NS1 to a subunit of the cellular 3 end processing machinery. 27 NS1 of H5N1/97 might act through one or more of these mechanisms or via a novel pathway in leading to the high epression of TNF. The precise molecular mechanism by which the NS gene of H5N1/97 mediates the differential increase of TNF requires further study. Furthermore, our results do not eclude contributions from other viral internal genes or gene products to the manifestation of the high-tnf- phenotype. Others have reported that H5N1/97 viruses are resistant to the antiviral effects of cytokines. 28 Our finding that H5N1/97 viruses differentially upregulate cytokine production is not simply a consequence of this resistance to the antiviral effects of cytokines. We used purified virus (no preformed cytokines in the inoculum) and a high multiplicity of infection (most cells were infected in the first cycle of infection) and measured outcome (cytokine mrna and TNF- secretion) early in the first cycle of virus infection. Clearly, increased resistance to the antiviral effects of cytokines is likely to be an advantageous adaptation, especially to a virus that is inherently a high inducer of cytokines. Thus, the two processes could be synergistic in leading to cytokine dysregulation and disease pathogenesis. 3 The relevance of these mechanisms to the virulence of the H1N1 pandemic virus of 1918 also deserves to be eplored. Avian G1/97-like (H9N2) viruses, which induce high TNF- production in macrophages, have caused mild influenza-like disease in children. 29 H5N1/97 disease in children was also mild, adults being more likely to have a severe outcome. The G1/97-like (H9N2) viruses prevalent in quail and other poultry in southern China 3 and beyond must therefore be regarded with some concern. Although a pandemic may have been averted in 1997, another influenza pandemic is a certainty. Influenza infection in primary macrophages provides a biologically relevant human model in which to investigate the viral gene mutations that lead to the high and low TNF- induction phenotype through influenza reverse genetics and to study the cell-signalling pathways that are activated 1836 THE LANCET Vol 36 December 7, 22

7 by these viral gene products. A better understanding of influenza virulence and pathogenesis might lead to novel therapeutic options and will help us to be better prepared for the net influenza pandemic. Contributors J S M Peiris conceived the study and, with C Y Cheung and S Gordon, planned the overall eperimental design. C Y Cheung carried out the eperiments, L L M Poon did the influenza reverse genetic work, A S Y Lau and Y L Lau contributed to planning of eperiments and interpretation of cytokine data, L L M Poon and W Luk contributed to optimisation of the cytokine assays, K F Shortridge provided viruses and reagents, and Y Guan was involved in phylogenetic analysis. The report was written jointly by J S M Peiris, C Y Cheung, and L L M Poon. All the investigators were involved in interpretation of the data and revision of the report and agreed to its final form. Conflict of interest statement None declared. Acknowledgments We thank R G Webster for providing the plasmids for the reverse genetics eperiments, O K Wong and H Y Ng for ecellent technical assistance, and W H S Wong for help in preparing the figures. This work was supported by research grants to J S M Peiris from the Research Grants Council of Hong Kong (HKU 792/99M) and the UK/Hong Kong Joint Research Scheme (/5). The cytokine quantification was partly supported by a research grant to L L M Poon from the Seed Funding for Basic Research, University of Hong Kong. References 1 Claas EC, Osterhaus ADME, van Beek R, et al. Human influenza A H5N1 virus related to a highly pathogenic avian influenza virus. Lancet 1998; 351: Subbarao K, Klimov A, Katz J, et al. Characterization of an avian influenza A (H5N1) virus isolated from a child with a fatal respiratory illness. Science 1998; 279: Yuen KY, Chan PKS, Peiris M, et al. Clinical features and rapid viral diagnosis of human disease associated with avian influenza A H5N1 virus. Lancet 1998; 351: Hatta M, Gao P, Halfmann P, Kawaoka Y. Molecular basis for high virulence of Hong Kong H5N1 influenza in mammals. Science 21; 293: Zitzow LA, Rowe T, Morken T, Shieh EJ, Zaki S, Katz J. Pathogenesis of avian influenza A (H5N1) viruses in ferrets. J Virol 22; 76: To KF, Chan PKS, Chan KF, et al. Pathology of fatal human infection associated with avian influenza A H5N1 virus. J Med Virol 21; 63: Rimmelzwaan GF, Kuiken T, van Amerongen G, et al. Pathogenesis of influenza A (H5N1) virus infection in a primate model. J Virol 21; 75: Fisman DN. Hemophagocytic syndromes and infection. Emerg Infect Dis 2; 6: Headley AS, Tolley E, Meduri GU. Infections and the inflammatory response in acute respiratory distress syndrome. Chest 1997; 111: Webby RJ, Webster RG. Emergence of influenza A viruses. Philos Trans R Soc Lond B Biol Sci 21; 356: Cauthen AN, Swayne DE, Shultz-Cherry S, Perdue ML, Suarez DL. Continued circulation in China of highly pathogenic avian influenza viruses encoding the hemagglutinin gene associated with the 1997 H5N1 outbreak in poultry and humans. J Virol 2; 74: Hirst GK. Adsorption of influenza hemagglutinins and virus by red blood cells. J Ep Med 1942; 76: Montaner LJ, Collin M, Herbein G. Human monocytes: isolation, cultivation, and its applications. In: Herzenberg LA, ed. Weir s handbook of eperimental immunology, 5th edn, vol IV. Cambridge, MA: Blackwell Science, 1996: Blaschke V, Reich K, Blaschke S, Neumann C. Rapid quantitation of proinflammatory and chemoattractant cytokine epression in small tissue samples and monocyte-derived dendritic cells: validation of a new real-time RT-PCR technology. J Immunol Methods 2; 246: Hoffmann E, Neumann G, Kawaoka Y, Hobom G, Webster RG. A DNA transfection system for generation of influenza A virus from eight plasmids. Proc Natl Acad Sci USA 2; 97: Aggarwal BB, Samanta A, Feldman M. TNF-alpha. In: Oppenhein JJ, Feldmann M, Durum SK, eds. Cytokine reference: a compendium of cytokines and other mediators of host defense, vol 1. San Diego: Academic Press, 21: Garcia-Sastre A, Egorov A, Matassov D, et al. Influenza A virus lacking the NS1 gene replicates in interferon-deficient systems. Virology 1998; 252: Hofmann P, Sprenger H, Kaufmann A, et al. Susceptibility of mononuclear phagocytes to influenza A virus infection and possible role in the antiviral response. J Leukocyte Biol 1997; 61: Matikainen S, Pirhonen J, Miettinen M, et al. Influenza A and sendai viruses induce differential chemokine gene epression and transcription factor activation in human macrophages. Virology 2; 276: Gordon S, Hughes DA. Macrophages and their origins: heterogeneity in relation to tissue microenvironment. In: Lipscomb M, Russell S, eds. Lung macrophages and dendritic cells in health and disease. NewYork: Marcel Dekker, 1997: Ciesielski CJ, Andreakos E, Fowell BMJ, Feldmann M. TNF- induced macrophage chemokine secretion is more dependent on NF-κB epression than lipopolysaccharides-induced macrophage chemokine secretion. Eur J Immunol 22; 32: Liu E, Tu W, Law HK, Lau YL. Changes of CD14 and CD1a epression in response to IL-4 and granulocyte-macrophage colony stimulation factor are different in cord blood and adult blood monocytes. Pediatr Res 21; 5: Hussell T, Pennycook A, Openshaw PJM. Inhibition of tumor necrosis factor reduces the severity of virus-specific lung immunopathology. Eur J Immunol 21; 31: Hayden FG, Fritz R, Lobo MC, Alvord W, Strober W, Straus SE. Local and systemic cytokine responses during eperimental human influenza A virus infection. J Clin Invest 1998; 11: Dehou MS, Boutten A. Compartmentalized cytokine production within the human lung in unilateral pneumonia. Am J Respir Crit Care Med 1994; 15: Garcia-Sastre A. Inhibition of interferon-mediated antiviral responses by influenza A viruses and other negative stranded RNA viruses. Virology 21; 279: Kim MJ, Latham AG, Krug RM. Human influenza viruses activate an interferon-independent transcription of cellular antiviral genes: outcome with influenza A virus is unique. Proc Natl Acad Sci USA 22; 99: Seo SH, Hoffmann E, Webster RG. Lethal H5N1 influenza viruses escape host anti-viral cytokine responses. Nat Med 22; 8: Peiris M, Yuen KY, Leung CW, et al. Human infection with influenza H9N2. Lancet 1999; 354: Guan Y, Shortridge KF, Krauss S, et al. H9N2 viruses possessing H5N1-like internal genomes continue to circulate in poultry in southeastern China. J Virol 2; 74: THE LANCET Vol 36 December 7,