© 2006 Nature Publishing Group

Antiviral treatment is more effective than smallpoxvaccination upon lethal monkeypox virus infectionKoert J. Stittelaar1, Johan Neyts3, Lieve Naesens3, Geert van Amerongen1,4, Rob F. van Lavieren5, Antonin Holy6,Erik De Clercq3, Hubert G. M. Niesters1, Edwin Fries1, Chantal Maas1, Paul G. H. Mulder2,Ben A. M. van der Zeijst4 & Albert D. M. E. Osterhaus1

There is concern that variola virus, the aetiological agent ofsmallpox, may be used as a biological weapon. For this reasonseveral countries are now stockpiling (vaccinia virus-based) small-pox vaccine. Although the preventive use of smallpox vaccinationhas been well documented, little is known about its efficacy whenused after exposure to the virus. Here we compare the effectivenessof (1) post-exposure smallpox vaccination and (2) antiviral treat-ment with either cidofovir (also called HPMPC or Vistide) or witha related acyclic nucleoside phosphonate analogue (HPMPO–DAPy) after lethal intratracheal infection of cynomolgus monkeys(Macaca fascicularis) with monkeypox virus (MPXV). MPXVcauses a disease similar to human smallpox1 and this animalmodel can be used to measure differences in the protectiveefficacies of classical and new-generation candidate smallpoxvaccines2. We show that initiation of antiviral treatment 24 hafter lethal intratracheal MPXV infection, using either of theantiviral agents and applying various systemic treatment regi-mens, resulted in significantly reduced mortality and reducednumbers of cutaneous monkeypox lesions. In contrast, whenmonkeys were vaccinated 24 h after MPXV infection, using astandard human dose of a currently recommended smallpoxvaccine (Elstree-RIVM), no significant reduction in mortalitywas observed. When antiviral therapy was terminated 13 daysafter infection, all surviving animals had virus-specific serumantibodies and antiviral T lymphocytes. These data show thatadequate preparedness for a biological threat involving smallpoxshould include the possibility of treating exposed individualswith antiviral compounds such as cidofovir or other selectiveanti-poxvirus drugs.

The eradication of variola virus by a worldwide vaccinationcampaign led by the World Health Organization (WHO) in the1970s (ref. 3), together with the subsequent discontinuation of masssmallpox vaccination with vaccinia virus, have resulted in low ornon-existent smallpox immunity in today’s human population4. Thepossible introduction of variola virus or MPXV as a biologicalweapon, or the zoonotic spread of MPXV3,5, is of growing concern.Given the relatively high rate of severe side effects associated withclassical smallpox vaccination, reintroduction of general preventivesmallpox vaccination has not been recommended6. Current plans donot envisage mass vaccination with classical smallpox vaccines untilafter an outbreak has been detected.

Although the effectiveness of classical smallpox vaccination whenused preventively has been well documented, and post-exposurevaccination is part of the WHO eradication strategy, little is knownabout its efficacy when used after exposure to variola virus. There-fore, treatment with antiviral drugs such as cidofovir or related

compounds that have long-lasting antiviral activity should be con-sidered as an alternative or additional strategy for the prevention orpost-exposure treatment of smallpox.

Cidofovir, a nucleotide analogue that is licensed for the treatmentof cytomegalovirus retinitis in patients with HIV infection, has beenshown to be active against a number of poxviruses, includingvaccinia-, variola-, monkeypox- and cowpox virus, and has shownefficacy in immunodeficient patients with severe molluscum con-tagiosum virus and orf virus infections7. Concern that the virulenceof variola virus and other poxviruses could be increased by geneticmodification would be another reason for developing antiviral

LETTERS

Figure 1 | Kaplan–Meier plot showing the effect of different post-exposuretreatments on the survival of monkeys. Survival expressed as percentagesafter intratracheal MPXV infection. After virus exposure, macaques weresham-treated (group I), vaccinated with smallpox vaccine Elstree-RIVM(group II), treated with 5 doses of cidofovir (group III), with 6 doses ofcidofovir (group IV), with 5 doses of HPMPO-DAPy (group V) or with 6doses of HPMPO-DAPy (group VI). Arrowheads indicate times ofvaccination and antiviral treatment. *The control group is supplementedwith 11 animals from previous experiments. Statistical analysis of survivalrates between groups is shown. NA, not applicable.

1Department of Virology and 2Department of Epidemiology and Biostatistics, Erasmus MC, 3000 DR Rotterdam, The Netherlands. 3Rega Institute for Medical Research, K.U.Leuven, B-3000 Leuven, Belgium. 4Netherlands Vaccine Institute, 3720 AL Bilthoven, The Netherlands. 5ViroClinics B.V., 3000 DR Rotterdam, The Netherlands. 6Institute ofOrganic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, 16610 Prague 6, Czech Republic.

Vol 439|9 February 2006|doi:10.1038/nature04295

745

© 2006 Nature Publishing Group

strategies against poxvirus infections in humans3,8,9. In several mousemodels for poxvirus infections, it has been shown that cidofovirtreatment may delay or reduce mortality10–16 (see refs 17–19 for areview). Although use of cidofovir has been proposed for treatmentof human smallpox20, its effectiveness against orthopoxvirus infec-tions in primates, including humans, has not been addressed to date.Although other anti-poxvirus compounds with different modes ofanti-poxvirus activity have been described, their clinical use is notexpected in the near future21–24.

Here we compare the efficacy of (1) smallpox vaccination with theclassical vaccinia virus-based smallpox vaccine currently stockpiledin The Netherlands (Elstree-RIVM) and (2) antiviral treatmentwith either cidofovir or a related experimental analogue (HPMPO-DAPy)25,26, against a lethal respiratory MPXV infection of cynomol-gus monkeys (Macaca fascicularis). This is the most prevalent naturalroute of infection and the model is considered to be an appropriatesurrogate for human smallpox, which meets regulatory requirementsoutlined by the so-called ‘animal rule’27. It has been successfully usedfor comparing the efficacy and safety of classical and new-generationcandidate smallpox vaccines, including Elstree-RIVM2: Indeed, pre-ventive Elstree-RIVM vaccination was highly effective against thischallenge2. Historical and anecdotal evidence suggests that classicalsmallpox vaccination may be successful up to 4–5 days afterexposure28. This ‘window of opportunity’ is probably related to therelatively long incubation time (7–14 days) for smallpox undernatural conditions with relatively low infection doses28. On thebasis of body temperature measurement, the incubation time forthe MPXV-macaque model is about three days, and is probably dueto the high lethal respiratory dose of MPXV given directly into therespiratory tract. This may mimic exposure of the first humanindividuals during a bioterrorist attack2.

We infected six groups of macaques with a lethal respiratoryinfection dose of 107 plaque-forming units (p.f.u.) MPXV peranimal, and initiated different antiviral intervention strategies 24 hlater. Group I (n ¼ 6) served as an untreated control. Group II(n ¼ 6) was vaccinated intracutaneously with a human dose ofElstree-RIVM using a bifurcated needle. Group III (n ¼ 6) andgroup IV (n ¼ 6) were given 5 mg kg21 cidofovir intraperitoneally(i.p.) on days 1, 3, 7, 10 and 13 after infection or on days 1, 3, 5, 7, 10and 13 after infection, respectively. Group V (n ¼ 4) and group VI(n ¼ 6) were given 5 mg kg21 HPMPO-DAPy i.p. on days 1, 3, 7, 10and 13 after infection or on days 1, 3, 5, 7, 10 and 13 after infection,respectively. As the pharmacokinetics of both antiviral compounds inmonkeys proved to be largely similar to those of cidofovir in humans(see Supplementary Information), we chose antiviral doses of cido-fovir and HPMPO-DAPy equivalent to those recommended forcidofovir in humans. All monkeys that received antiviral treatmentalso received supportive treatment with 25 mg kg21 probenecid,administered orally. Cidofovir should always be given under thecover of probenicid, a competitive inhibitor of organic anion trans-port, to prevent cidofovir nephrotoxicity29. Probenecid should beused with caution in children, pregnant women and persons withsulfa drug allergy. In addition, monkeys received 30 ml of Ringer’ssolution i.p. when antiviral treatment was given, to improvehydration.

All six animals in the control group died within 15 days ofinfection. For statistical comparisons of mortality rates betweenthe groups, 11 control animals from previous experiments wereincluded in the analyses, resulting in a mean day of death of 12.6 days(range 9–19) (Fig. 1). From a total of 34 animals, 17 died as a result ofMPXV infection (Fig. 1). Severe morbidity, extensive cutaneousmonkeypox lesions, dyspnea and low blood oxygen (O2) saturationwere observed in all control animals (group I; Fig. 2 and Supplemen-tary Information). Vaccination with RIVM-Elstree (group II) pro-tected one animal from MPXV-related death (Figs 1, 2) and severemorbidity. In contrast, cidofovir treatment protected five animals ingroup III (five-dose cidofovir regimen) and four animals in group IV

(six-dose regimen) from severe morbidity and death (Fig. 1). Despitecidofovir treatment, mild morbidity was observed in five animals ingroup III and two animals in group IV (Fig. 2 and SupplementaryInformation). Similarly, treatment with HPMPO-DAPy protectedtwo animals in group V (five-dose regimen, group size n ¼ 4) andfive animals in group VI (six-dose regimen, group size n ¼ 6) fromsevere morbidity and death (Figs 1, 2 and Supplementary Infor-mation). Monitoring of plasma creatinine levels indicated thatnephrotoxicity did not contribute to the mortality rate of themonkeys treated with the antiviral compounds (SupplementaryInformation).

Plasma MPXV loads were measured in all of the animals through-out the course of the experiment, using real-time polymerase chainreaction (PCR) (Fig. 3a). Viral loads largely paralleled the severity ofdisease and clinical signs. Five days after infection with MPXV, nosignificant differences in plasma viral load were found betweencontrol (group I) and vaccinated animals (group II, P ¼ 0.982;Fig. 3b). Animals treated with cidofovir (groups III and IV) hadsignificantly lower plasma viral loads than control and the vaccinatedanimals (P ¼ 0.021 and 0.002, respectively). Plasma viral loads inanimals treated with HPMPO-DAPy (groups V and VI) weresignificantly lower than those of all the other groups (P , 0.005–0.041). Seven days after infection, plasma viral loads in animalstreated six times with either cidofovir (group IV) or with HPMPO-DAPy (group VI) were significantly lower than those in the control orvaccinated animals (P , 0.005 and P ¼ 0.007, respectively). Theadditional dose of cidofovir and HPMPO-DAPy given to groupsIV and VI compared with groups III and V resulted in a significant

Figure 2 | Blood O2 saturation profiles. Monkeys treated as described inFig. 1. Individual monkeys (1–34) represented by symbols as indicated.Horizontal bar shows percentage of animals with cutaneous monkeypoxlesions: none (white), 16–33% (light grey), 33–66% (dark grey), 66–100%(black). The cumulative number of cutaneous monkeypox lesions in allanimals per group is shown in the bars. Crosses indicate time of death ofindividual animals. (See Supplementary Information for more details.)

LETTERS NATURE|Vol 439|9 February 2006

746

© 2006 Nature Publishing Group

additional reduction in plasma viral loads (cidofovir, P ¼ 0.002;HPMPO-DAPy, P ¼ 0.007). This was further corroborated by com-paring the area under the curves for the respective groups until day 13after infection, when all animals from groups III–VI were still alive(Fig. 3c; cidofovir, P ¼ 0.048; HPMPO-DAPy, P ¼ 0.026).

Antiviral treatment was stopped 13 days after infection, at whichpoint all surviving animals had MPXV-specific plasma IgG anti-bodies, as measured by enzyme-linked immunosorbent assay(ELISA), as well as MPXV-specific T cells in their peripheral bloodmononuclear cells, as measured by interferon-g ELISPOT assay(Supplementary Information).

Taken together, our data show that post-exposure antiviral treat-ment with cidofovir or the related analogue HPMPO-DAPy, initiated24 h after lethal respiratory infection of monkeys with MPXV, ismore effective at reducing the mortality and number of cutaneousmonkeypox lesions than standard vaccination with a classical variolavaccine administered 24 h after infection. Although mortality is arelevant read-out parameter, it should be noted that this is influencedlargely by factors related to individual variation in pathogenesis andpathology. Given the relatively low numbers of animals that can betested in non-human primate experiments, we also used other read-out criteria, including clinical, virological and immunologicalparameters. Of these, the number of cutaneous monkeypox lesionsobserved and the viral load data in the respective groups are probablythe most relevant.

As the experimental compound HPMPO-DAPy was at least aseffective as cidofovir in reducing numbers of cutaneous monkeypoxlesions and viral loads, the potential usefulness of this compound

warrants further study. In this study, the monkeys received cidofoviror HPMPO-DAPy doses equivalent to those of cidofovir treatmentsgiven to humans, as well as the necessary supportive treatment withprobenecid and extra hydration. We are therefore confident that theexperimental conditions closely mimicked those for treatment ofman after poxvirus infection29.

In conclusion, this proof-of-concept study indicates that, in anoutbreak situation, the use of cidofovir or a related nucleotideanalogue with an improved bioavailability and toxicity profile,should be seriously considered an alternative or additional strategyfor vaccination of individuals who may have been exposed tosmallpox or other poxvirus infections such as monkeypox. Ourstudy emphasizes the potential of cidofovir in the short-term treat-ment of acute poxvirus infections. For use against such infections inman, the treatment schedules used in this experimental model willrequire further optimization. Clinical experience with cidofovir ismainly based on long-term treatment of herpesvirus infections.

METHODSAnimals, MPXV infection, vaccination, antiviral treatment and sampling.The studies were carried out in 34 captive-bred subadult (2–4 yr) healthycynomolgus macaques (Macaca fascicularis) in accordance with the institutionalguidelines for care and use of laboratory animals, which comply with legalrequirements in The Netherlands. MPXV strain MSF#6, which was obtainedfrom a fatally infected human in Congo, was provided by H. Meyer30. The viruswas passaged in African green monkey kidney cells: twice in MA104 cells andtwice in Vero cells. The challenge virus stock had a titre of 2 £ 108 p.f.u. per ml inVero cell monolayers. Monkeys were inoculated intratracheally with 107 p.f.u. ofthe virus in a volume of 5 ml PBS. Animals died naturally (animals 2, 3, 8, 9, 10,

Figure 3 | Viral load profiles after infection with MPXV. a, Plasma MPXVloads of monkeys treated as described in Figs 1, 2. b, Plasma MPXV loads ofmonkeys (treated as described in Fig. 1) 5 or 7 days after infection. Data showaverage ^ s.e.m. Statistical analysis of the pairwise comparisons of survival

rates between groups is also shown. NA, not applicable. c, Area under thecurve (AUC) measurements of plasma MPXV loads between days 0 and 13after infection of monkeys treated as described in Fig. 1. Data showaverage ^ s.e.m.

NATURE|Vol 439|9 February 2006 LETTERS

747

© 2006 Nature Publishing Group

11, 18, 22, 23, 26, 28 and 31) or were killed for ethical reasons when they weremoribund (animals 1, 4, 5, 6 and 12).

The classical smallpox vaccine (Lot. SP003, production date July 2002),referred to as Elstree-RIVM, was produced by the The National Institute forPublic Health and the Environment (Bilthoven, The Netherlands). Elstree-RIVM was reconstituted as recommended by the manufacturer to give a finalconcentration of 108 TCID50 ml21 (TCID50, 50% tissue-culture infectious dose).A bifurcated needle was used to deliver the human equivalent dose of 2.5 £ 105

TCID50 intracutaneously. Cidofovir (provided by Gilead Sciences) andHPMPO-DAPy [(R)-2,4-diamino-6-[3-hydroxy-2-(phosphonomethoxy)pro-poxy]pyrimidine were weighed and dissolved in Dulbecco’s modified Eagle’smedium (Cambrex Bio Science) to yield a dose of 5 mg per kg body weight thatwas administered i.p. (after adjusting the pH with NaOH) in a volume of 2 mlper kg body weight on days 1, 3, 7, 10 and 13 after infection (five-dose regimen)or on days 1, 3, 5, 7, 10 and 13 after infection (six-dose regimen).

Additionally, 30 ml Ringer’s solution was administered i.p. on each of thesedays to the animals treated with antivirals. These animals were also given oralprobenecid (25 mg per kg body weight; Duchefa Biochemie B.V.) on days 0, 1, 3,5 and 7 after infection. Probenecid was not administered on days 10 and 13because of the risk of aspiration pneumonia in animals that showed severedyspnea. Throat swabs and EDTA-anticoagulated blood samples were collectedon days 21, 0, 1, 3, 5, 7, 10, 13, 18, 21 and 28 after infection.Blood O2 saturation. Blood (haemoglobin) O2 saturation was measured using apulse oximeter (Novametrix Medical Systems Inc) at the times of samplecollection, and expressed as the percentage of haemoglobin carrying O2.Virus quantification by real-time PCR. MPXV DNA was quantified in plasmaand throat swabs using a LightCycler (Roche) as previously described2.Statistical analysis. Data were analysed using SPSS v.11. The outcome variablesof viral load (at 5 and 7 days after infection) and the area under the curve of viralload (from 0 to 13 days after infection) were analysed after logarithmic (base 10)transformation. We assumed that after log transformation, viral loads had anormal distribution in the separate groups as well as in the combined groups.This assumption was verified by inspection of the distribution (mainly con-cerning symmetry) and using the Kolmogorov–Smirnov test, yielding P-valuesof about 0.70. We therefore used a one-way analysis of variance (ANOVA) toanalyse the data.

There were six groups of monkeys in total. For viral load 5 days after infection,four groups were distinguished: groups III and IV were combined, and groups Vand VI were combined because until day 5 after infection these animals receivedexactly the same treatments. Four groups were also considered for area under thecurve measurements: groups I and II were excluded from this analysisbecause most of the animals had died by day 13 after infection. In the one-wayANOVA we first tested homogeneity of the within-group variances across thegroups. Then, depending on the appropriate assumption of either homogeneousor heterogeneous variances, differences in means between the groups weretested. If and only if the overall group effect (that is, comparing allgroups simultaneously) turned out to be significant (P , 0.05), a number ofrelevant pairwise comparisons were tested. Differences in survival between thegroups were tested using the log-rank test. Again, pairwise comparisons wereonly tested if the overall P value from comparing all groups was significant(P , 0.05).

Received 10 September; accepted 29 September 2005.Published online 11 December 2005.

1. Zaucha, G. M., Jahrling, P. B., Geisbert, T. W., Swearengen, J. R. & Hensley, L.The pathology of experimental aerosolized monkeypox virus infection incynomolgus monkeys (Macaca fascicularis). Lab. Invest. 81, 1581–-1600 (2001).

2. Stittelaar, K. J. et al. Modified vaccinia virus Ankara protects macaquesagainst respiratory challenge with monkeypox virus. J. Virol. 79, 7845–-7851(2005).

3. Smith, G. L. & McFadden, G. Smallpox: anything to declare? Nature Rev.Immunol. 2, 521–-527 (2002).

4. Breman, J. G. & Arita, I. The confirmation and maintenance of smallpoxeradication. N. Engl. J. Med. 303, 1263–-1273 (1980).

5. Lewis-Jones, S. Zoonotic poxvirus infections in humans. Curr. Opin. Infect. Dis.17, 81–-89 (2004).

6. Grabenstein, J. D. & Winkenwerder, W. Jr US military smallpox vaccinationprogram experience. J. Am. Med. Assoc. 289, 3278–-3282 (2003).

7. De Clercq, E. & Neyts, J. Therapeutic potential of nucleoside/nucleotideanalogues against poxvirus infections. Rev. Med. Virol. 14, 289–-300 (2004).

8. Finkel, E. Australia. Engineered mouse virus spurs bioweapon fears. Science 291,585 (2001).

9. Stanford, M. M. & McFadden, G. The ‘supervirus’? Lessons fromIL-4-expressing poxviruses. Trends Immunol. 26, 339–-345 (2005).

10. Bray, M. & Roy, C. J. Antiviral prophylaxis of smallpox. J. Antimicrob. Chemother.54, 1–-5 (2004).

11. Buller, R. M. et al. Efficacy of oral active ether lipid analogs of cidofovir in alethal mousepox model. Virology 318, 474–-481 (2004).

12. Quenelle, D. C., Collins, D. J. & Kern, E. R. Efficacy of multiple- or single-dosecidofovir against vaccinia and cowpox virus infections in mice. Antimicrob.Agents Chemother. 47, 3275–-3280 (2003).

13. Roy, C. J., Baker, R., Washburn, K. & Bray, M. Aerosolized cidofovir is retainedin the respiratory tract and protects mice against intranasal cowpox viruschallenge. Antimicrob. Agents Chemother. 47, 2933–-2937 (2003).

14. Smee, D. F., Bailey, K. W. & Sidwell, R. W. Comparative effects of cidofovir andcyclic HPMPC on lethal cowpox and vaccinia virus respiratory infections inmice. Chemotherapy 49, 126–-131 (2003).

15. Neyts, J., Leyssen, P., Verbeken, E. & De Clercq, E. Efficacy of cidofovir in amurine model of disseminated progressive vaccinia. Antimicrob. AgentsChemother. 48, 2267–-2273 (2004).

16. Smee, D. F., Bailey, K. W., Wong, M. H., Wandersee, M. K. & Sidwell, R. W.Topical cidofovir is more effective than is parenteral therapy for treatment ofprogressive vaccinia in immunocompromised mice. J. Infect. Dis. 190, 1132–-1139(2004).

17. De Clercq, E. Cidofovir in the therapy and short-term prophylaxis of poxvirusinfections. Trends Pharmacol. Sci. 23, 456–-458 (2002).

18. Bronze, M. S. & Greenfield, R. A. Therapeutic options for diseases due topotential viral agents of bioterrorism. Curr. Opin. Investig. Drugs 4, 172–-178(2003).

19. Prichard, M. N. & Kern, E. R. Orthopoxvirus targets for the development ofantiviral therapies. Curr. Drug Targets Infect. Disord. 5, 17–-28 (2005).

20. Wharton, M. et al. Recommendations for using smallpox vaccine in a pre-eventvaccination program. Supplemental recommendations of the AdvisoryCommittee on Immunization Practices (ACIP) and the Healthcare InfectionControl Practices Advisory Committee (HICPAC). MMWR Recomm. Rep. 52,1–-16 (2003).

21. Yang, G. et al. An orally bioavailable anti-poxvirus compound (ST-246) thatinhibits extracellular virus formation and protects mice from lethalorthopoxvirus challenge. J. Virol. 79, 13139–-13149 (2005).

22. Reeves, P. M. et al. Disabling poxvirus pathogenesis by inhibition of Abl-familytyrosine kinases. Nature Med. 11, 731–-739 (2005).

23. Yang, H. et al. Antiviral chemotherapy facilitates control of poxvirus infectionsthrough inhibition of cellular signal transduction. J. Clin. Invest. 115, 379–-387(2005).

24. McFadden, G. Gleevec casts a pox on poxviruses. Nature Med. 11, 711–-712(2005).

25. Balzarini, J. et al. 6-[2-phosphonomethoxy)alkoxy]-2,4-diaminopyrimidines: anew class of acyclic pyrimidine nucleoside phosphonates with antiviral activity.Nucleosides Nucleotides Nucleic Acids 23, 1321–-1327 (2004).

26. Neyts, J., Leyssen, P., De Clercq, E. & Holy, A. Effects of (S)-HPMPDAP and(S)-HPMPO-DAPym on vaccinia virus infections in mice. Antiviral Res. 62, 109(2004).

27. Rosenthal, S. R. & Clifford, J. C. Development of vaccines for bio-warfareagents. Dev. Biol. (Basel) 110, 99–-105 (2002).

28. Massoudi, M. S., Barker, L. & Schwartz, B. Effectiveness of postexposurevaccination for the prevention of smallpox: results of a delphi analysis. J. Infect.Dis. 188, 973–-976 (2003).

29. Cundy, K. C. Clinical pharmacokinetics of the antiviral nucleotide analoguescidofovir and adefovir. Clin. Pharmacokinet. 36, 127–-143 (1999).

30. Meyer, H. et al. Outbreaks of disease suspected of being due to humanmonkeypox virus infection in the Democratic Republic of Congo in 2001. J. Clin.Microbiol. 40, 2919–-2921 (2002).

Supplementary Information is linked to the online version of the paper atwww.nature.com/nature.

Acknowledgements We are grateful to F. Pistoor, J. Rimmelzwaan-Offutt,S. Slabbekoorn-Romijn, R. D. d’Ullois and J. Guldemeester for assistance and toR. Gruters for comments. This work in Leuven and Prague was supported by anNIAID/NIH grant to J.N. and A.H.. This study is a part of an IOCB (Prague)research programme and project.

Author Information Reprints and permissions information is available atnpg.nature.com/reprintsandpermissions. The authors declare competingfinancial interests: details accompany the paper on www.nature.com/nature.Correspondence and requests for materials should be addressed to A.D.M.E.O.(a.osterhaus@erasmusmc.nl).

LETTERS NATURE|Vol 439|9 February 2006

748