Proc. Natl. Acad. Sci. USAVol. 94, pp. 11808–11812, October 1997Biochemistry

Structural evidence for a second sialic acid binding site in avianinfluenza virus neuraminidases

JOSEPH N. VARGHESE*, PETER M. COLMAN, ALBERTUS VAN DONKELAAR, TONY J. BLICK, ANJALI SAHASRABUDHE,AND JENNIFER L. MCKIMM-BRESCHKIN

Biomolecular Research Institute, 343 Royal Parade, Parkville, 3052 Australia

Edited by Don C. Wiley, Harvard University, Cambridge, MA, and approved August 18, 1997 (received for review July 16, 1997)

ABSTRACT The x-ray structure of a complex of sialic acid(Neu5Ac) with neuraminidase N9 subtype from AyternyAustraliayG70Cy75 inf luenza virus at 4°C has revealed thelocation of a second Neu5Ac binding site on the surface of theenzyme. At 18°C, only the enzyme active site contains boundNeu5Ac. Neu5Ac binds in the second site in the chair confor-mation in a similar way to which it binds to hemagglutinin.The residues that interact with Neu5Ac at this second site aremostly conserved in avian strains, but not in human and swinestrains, indicating that it has some as-yet-unknown biologicalfunction in birds.

Influenza, an orthomyxo virus, has a negative-stranded RNAsegmented genome that codes for two surface glycoproteins(1). One of the genes codes for neuraminidase (NA) (2), aglycoprotein found on the surface of the influenza virusparticle. The NA is an enzyme that cleaves terminal sialic acid(Neu5Ac) from glycoconjugates found on the surface ofmolecules on target cells in the upper respiratory tract of somesusceptible mammals, including humans (3, 4). These mole-cules with terminal Neu5Ac are also the target receptors forthe viral hemagglutinin (HA) (5), the major surface glycopro-tein on the viral particle surface. NA destroys these HAreceptors, allowing progeny virus particles, budding frominfected cell surfaces, to be released (6). It also is thought thatNA facilitates passage of virus through the protective mucincovering target cells by desialylation of the Neu5Ac-rich mucin(7) and prevents aggregation by HA of freshly synthesized viralglycoproteins via sialylated carbohydrates.

The x-ray structure of influenza virus NA has been deter-mined (8, 9) for type A subtype N2 (10), N9 (11, 12), and typeB (13), together with their complexes with Neu5Ac (14) andother NA inhibitors (15–17). These structural studies have ledto a renewed interest in NA inhibitors as a means of controllinginfluenza virus infections. Potent NA inhibitors now have beendeveloped (15, 18, 19), and one of them, 4-guanidino-Neu5Ac2en (GG167, Zanamivir) (15, 20), now is undergoingphase III clinical trials.

HA activity has been reported for the N9 subtype of NA ofinfluenza type A virus (21) at 4°C, which was not related toaberrant NA activity, but was associated with a second Neu5Acbinding site on the surface of NA away from the active site(22). The residues on the surface of NA responsible for thehemabsorbing (HB) activity have been identified by monoclo-nal variants that lost capacity to bind red blood cells (22), andthe activity has been successfully transferred to the N2 subtypeof NA (23) by site-directed mutagenesis. Furthermore, it wasshown that N9 NA activity did not remove the putativeNeu5Ac-related moiety that bound red blood cells to this HBsite (24) as the agglutination was restored on cooling to 4°C.

An HB site also has been discovered on the NA of AyFPVyRostocky34 H7N1 (25), which appears to have the samelocation on the NA surface as in N9 NA. However previousattempts to observe this site by x-ray diffraction were unsuc-cessful. Here we report conditions for visualizing, by x-raydiffraction, Neu5Ac bound to the HB site of N9 NA and showthat the sequence signature of side chains, which interact withthis Neu5Ac, is largely conserved in all avian influenza viruses.

MATERIALS AND METHODS

Data Collection. The NA enzyme was purified from influ-enza virus AyternyAustraliayG70Cy75 virus as described (26)and crystallized by established procedures (21). The crystalswere transferred to 20% glycerol while maintaining the con-centration of the phosphate buffer before freezing in a coldstream of nitrogen gas at 2166°C. X-ray diffraction data werecollected on a R-axis IV Image Plate x-ray detector mountedon a MAC Science SRA M18XH1 rotating anode x-raygenerator, operating at 47 kV and 60 mA with focusingmirrors. Two x-ray data sets were collected, namely wild-typeNA complexed with Neu5Ac at 4°C (N9–4C) and 18°C(N9–18C). All crystals were soaked with 20 mM Neu5Ac for4 hr and flash-frozen to 2166°C before data collection. Listed onTable 1 are the data collection statistics for the two data sets.

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked ‘‘advertisement’’ inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

© 1997 by The National Academy of Sciences 0027-8424y97y941-5$2.00y0PNAS is available online at http:yywww.pnas.org.

This paper was submitted directly (Track II) to the Proceedings office.Abbreviations: NA, neuraminidase; Neu5Ac, sialic acid; HA, hemag-glutinin; HB, hemabsorbing.Data deposition: The atomic coordinates reported in this paper havebeen deposited in the Protein Data Bank, Brookhaven NationalLaboratory, Upton, NY 11973 (reference no. 1MWE for the AyternyAustraliayG70Cy75 NAyNeu5Ac complexed at 4°C).*To whom reprint requests should be addressed. e-mail: jose.

varghese@bioresi.com.au.

Table 1. X-ray data collection statistics for crystals ofAyternyAustraliayG70Cy75 N9 NA

Crystal N9-18C N9-4C*

Neu5Ac soaking temp.°C† 18 4Total observations 164,615 291,377Unique intensities 38,810 45,647IP frames 80 60Rmerge 8.8% 7.6%^Rsym&‡ 8.9% 8.5%Resolution 1.8 Å 1.7 Å

Completeness total 83.6% 96.6%Last shell 68.2% 91.8%Range 1.92-1.8 Å 1.8-1.7 Å

^Iys& 9.8 9.6

*N9-4C refined to R 5 0.178 (Rfree 5 0.205) with 369 waters added,for data between 6-1.7 Å with rmsd from ideal geometry of 0.016 Åand 1.90° for bonds and angles, and rms B value of 13.4. Eighty-twopercent of residues were in the most favorable region on a Ra-machandran Plot, with no outliers.

†20 mM for 4 hr, then flash-frozen to 2166°C.‡Averaged over the number of frames.

11808

Structure Refinement. The refined structure at 2166°C ofwild type (27) was used as the reference structure. The locationof the Neu5Ac moieties were obtained by examination of thedifference Fourier of the complexed and the uncomplexed NAx-ray data using the phases of the uncomplexed refined atomicmodel with all active site water molecules removed. Both datasets revealed Neu5Ac in the active site of NA in a twisted-boatconformation, and active site water molecules as determinedpreviously (14). The N9–4C complex showed the secondNeu5Ac binding site clearly in the difference Fouriers. Thesemoieties appeared as large positive features greater than 5 sin the difference Fouriers, whereas for the N9–18C complexthe difference Fouriers indicated only noise peaks (less than 3

s) at the second site. The orientation of the Neu5Ac moietyin the chair conformation was unambiguous, and atomicmodels were built into these difference Fouriers.

The structure then was refined to 1.7 Å resolution forN9–4C using the refined wild-type structure and models forthe Neu5Ac moieties at the two sites as a starting atomicmodel, using XPLOR (28). Water molecules in the active siteand elsewhere were added by examining the difference Fourierpeaks of observed and calculated structure factors of theatomic model that were greater than 5 s. In the above XPLORrefinement all charges in charged amino acids were set to zero,as well as charged groups on the inhibitors. Crystallographic-based stereo chemical restraints (29) were used, and the

FIG. 1. A molecular surface rendered image (42) of a tetrameric head of AyternyAustraliayG70Cy75 N9 NA viewed from above the molecule.The active site residues interacting with the Neu5Ac moiety (twist-boat conformation) in the catalytic sites are show in green. The residues thatinteract with the Neu5Ac moiety (chair conformation) in the HB sites are colored yellow (conserved in all avian strains) and blue (conserved inN9 stains). (Inset) A magnified region (31.6) in the vicinity of these two sites of one subunit, illustrating the deep pocket of the catalytic site andthe flat surface of the HB site of the enzyme.

Biochemistry: Varghese et al. Proc. Natl. Acad. Sci. USA 94 (1997) 11809

structure retained good stereo geometry (see Table 1 forrefinement statistics).

RESULTS

X-ray diffraction analysis of crystals of AyternyAustraliayG70Cy75 N9 NA, soaked with 20 mM Neu5Ac soaked for 4 hrat 4°C, and subsequently flash-frozen to 2166°C, reveals twoNeu5Ac moieties bound to the NA. One is in the conservedactive site in a twisted boat conformation as reported previ-ously (14), the other is observed nearby about 14 Å away (theketosidic oxygens of the two Neu5Ac moeties are about 21.3Å apart), in the direction of Arg-371 and on the other side ofthe 367–372 loop (Figs. 1 and 2). This second site is in a chairconformation similar to what is found in the Neu5Ac bindingsites of HA (30). A similar analysis with crystals soaked with20 mM Neu5Ac soaked for 4 hr at 18°C and then flash-frozen

to 2166°C indicates that this second Neu5Ac binding site isunoccupied.

The location of this second site in the N9–4C complex (Fig.2) is precisely where it was predicted by site-directed mutagen-esis (22) in wild-type AyternyAustraliayG70Cy75 N9 NA andin AyFPVyRostocky34 N1 NA (25). One of the oxygens of thecarboxylate group of Neu5Ac has a hydrogen bond interactionwith the hydroxyl oxygen of Ser-367 and the other carboxylateoxygen hydrogen bonds to the amide on the side chain ofAsn-400. The main chain carbonyl oxygen of Asn-400 interactswith both the 4-hydroxyl oxygen and the nitrogen of the5-acetamido group of Neu5Ac. This nitrogen also interactswith the hydroxyl oxygen of Ser-372. The methyl carbon of the5-acetamido group has a hydrophobic interaction with Trp-403, sitting 3.5 Å above the six-membered ring of the trypto-phan similar to an interaction in the HA Neu5Ac binding site(31) with the plane of the acetamido group lying almost

FIG. 2. (A) A stereo drawing (43) of the refined x-ray atomic model of the Neu5Ac (orange) bound in the HA site of AyternyAustraliayG70Cy75NA with Neu5Ac soaked at 4°C, showing all the amino acids (green), and water molecules (red) making contact with the moiety. Atomic interactionsare shown in broken lines. Oxygen, nitrogen, and carbon atoms are colored red, blue, and black, respectively. The protein backbone is representedby a yellow tube. (B) The two-Fo-Fc electron density map (blue caged-mesh contour at 1.6 s level) of the Neu5Ac in the same orientation, usingthe refined phases of the complex, where Fo and Fc are the observed and calculated structure factors, respectively.

11810 Biochemistry: Varghese et al. Proc. Natl. Acad. Sci. USA 94 (1997)

parallel to the plane of the tryptophan rings. The distalhydroxyl oxygen (O9) of the 6-triol group of Neu5Ac interactswith the « amino of Lys-432, the next hydroxyl (O8) interactswith the hydroxyl oxygen of Ser-370, and the proximal hydroxyloxygen is exposed to the solvent.

Thus three loops are on NA that primarily are responsiblefor the formation of this HB site. The first loop containingresidues 367-SIASRS-372 is involved in three interactions withNeu5Ac through the three serine residues 367, 370, and 372.A second loop containing residues 400-NTSW-403 is involvedwith three interactions with Neu5Ac through Asn-400 sidechain and main chain carbonyl oxygen, and a hydrophobicinteraction with Trp-403. A third loop containing Lys-432interacts with the distal hydroxyl of the 6-triol of Neu5Ac viathe «-amino nitrogen of the side chain.

A water molecule (Fig. 2) was found hydrogen-bonded toboth carboxylate oxygens of Neu5Ac, and another to both theproximal hydroxyl oxygen of the 6-triol group of Neu5Ac andto the ring oxygen. A third water molecule interacts with thecarbonyl oxygen of the 5-acetamido group of Neu5Ac, andanother with the ketosidic oxygen of Neu5Ac.

DISCUSSION

We have shown that six residues on three separate loops of N9NA interact directly with the Neu5Ac in the second Neu5Acbinding site. These are the three serine residues (367, 370, and372) in the loop containing residues 367-SIASRS-372, Asn-400 and Trp-403 in the loop containing 400-NTSW-403, andLys-432 in a third loop. Not all of these six residues areconserved in the NAs, which have been shown to possess theHB activity (22, 23, 25). Thus the Lys-432 interaction is notabsolutely necessary for binding, as a site-directed mutant ofAyTokyoy3y67 N2 NA gained the HB site activity with aglutamine at this position (23). It was shown (22) that antigenicvariants with amino acid substitutions at 432 had a minimal

effect on HA activity on N9. Furthermore AyFPVyRostocky34 N1NA has HB activity (25) with an asparagine atposition 432, and an isoleucine at position 400. This wouldindicate that the interactions with the distal hydroxyl (O9) ofthe 6-triol group (which interacts with Lys-432 in N9 NA)could interact with the protein via a water molecule as in theactive site of bacterial sialidases (32) or some as-yet-unidentified residue on the AyFPVyRostocky34 N1 NA andAyTokyoy3y67 N2 NA. Similarly the carboxylate oxygen ofNeu5Ac that interacts with Asn-400 in N9 NA would have tointeract with AyFPVyRostocky34 NA in a different manner.

In summary, the sequence variation at residues 400 and 432in NAs with known HB activity indicates that these residues arenot essential for HB activity, and the triple serine SxxSxS loopof 367 to 372 and Trp-403 is a minimal signature for HBbinding in NA.

This is consistent with the observation that mutations (22)at residues 367, 370, and 372 markedly reduced or abolishedHB activity and that an A369D mutation would interfere withthe HB site in N9 NA. Also the insensitivity of the mutationI368R to HB activity arises because the side chain at 368 pointsaway from the HB site (see Fig. 2). It therefore would requirea L370S and a R403W mutation as a minimum requirement fortransfer of HB activity (23) to AyTokyoy3y67 NA.

Sequence comparisons of the different subtypes of NA in theregion of these three loops indicate that only the N9 subtypehas all six contact residues as defined here. It should be notedthat Gly-373, which is involved in a turn between two b-strandson the fifth b-sheet positions the invariant catalytic residueArg-371, and Ser-404, which positions Trp-403 in the HB site,both are conserved in all type A NAs. However, if a compar-ison is made using the triple serine and Trp-403 signature(Table 2), all known strains from N1 to N9, which carry thesignature, are avian (or equine), with the exception of the twohuman N2 strains RI51y57 and Leningrady134y57. Further-more, all avian sequences carry this signature, and in general

Table 2. A comparison of sequences of influenza virus isolates (* avian strains) that contain the HBsignature (large bold)

366–373 399–404 430–433 Influenza strain Subtype

KSTSSRSG AITDWS RPNH AyFPVyRostocky34 N1*KSTSSRSG AITDWS RPKE AyParrotyUlstery73 N1*I SKESRSG DNNNWS RPQE AyRIy51y57 N2I SKDSRSG DNNNWS RPQE AyLeningrady134y57 N2I SKDSRSG DNNNWS RPQE AyduckyHongKongy24y76 N2*I SKDSRSG DNNNWS RPQE AyChickenyPennsylvaniay1370y83 N2*I SVSSRSG NNKNWS RKQE AyDuckyAlbertay60y76 N5*I SVSSRSG NNKNWS RKQE AyShearwatery72 N5*I SPRSRSG DNSNWS RPEE AyFPVyWeybridge N7*I SRTSRSG DNLNWS RKQE AyEquineyMiamiy1y63 N8I SRTSRSG DNLNWS RKQE AyDuckyUkrainey1y63 N8*I SRTSRSG DNLNWS RKQE AyQuailyItalyy1117y65 N8*I SRTSRSG DNLNWS RKQE AyEquineySaoPauloy6y69 N8I SRTSRSG DNLNWS RKQE AyEquineyAlgiersy72 N8I SRTSRSG DNLNWS RKQE AyDuckyChabarovsky1610.72 N8*I SRTSRSG DNLNWS RKQE AyDuckyMemphisy928y74 N8*I SRTSRSG DNLNWS RKQE AyTurkeyyMinnesotay501y78 N8*I SRTSRSG DNLNWS RKQE AyDuckyHokkaidoy8y80 N8*I SRTSRSG DNLNWS RKQE AyGuineaFowlyNYy4-3587y84 N8*I SRTSRSG DNLNWS RKQE AyDuckyBurjatiay652y88 N8*I SRTSRSG DNLNWS RKQE AyHerring GullyDEy677y88 N8*I SRTSRSG DNLNWS RKQE AyEquineyJiliny1y89 N8I SRTSRSG DNLNWS RKQE AyMullardyEdmontony220y90 N8*I SRTSRSG DNLNWS RKQE AyEquineyAlaskay1y91 N8I SIA SRSG LNTDWS RPKE AyternyAustraliayG70Cy75 N9*I SIA SRSG LNTDWS RPKE AywhaleyMainey1y84 N9I STASRSG LNTDWS RPKE AyRuddyTurnstoneyNJy60y85 N9*

Only the three loops that interact with the Neu5Ac HB site of G70C N9 are listed. In bold are residuesin contact with Neu5Ac in N9. The sequence data are from the GenBank database.

Biochemistry: Varghese et al. Proc. Natl. Acad. Sci. USA 94 (1997) 11811

human and swine NAs do not. We propose that the signaturehas a functional relevance to avian influenza.

It has been shown (24) that the HB site on AyternyAustraliayG70Cy75 NA binds a moiety on red cells that is notcleavable by AyternyAustraliayG70Cy75 NA, a consequenceof the moiety either being Neu5Ac in a noncleavable linkageor something other than Neu5Ac. The biological significanceof this HB site has not been determined, but the above resultswould indicate that it may function as a lectin to some aviancell receptor, which is unaffected by influenza NA activity.Influenza is normally an asymptomatic infection in birds (33),but replicates preferentially in the cells lining the intestinaltract of waterfowl (34, 35), where all of the different subtypesof influenza A have been isolated. This has led to the prop-osition that waterfowl are the primary vectors for the globalspread of the disease (36) through fecal droppings. Thisavirulent adaptation of the virus to avian species enables it tosurvive and persist in a vast global reservoir.

If this HB signature relates to some as-yet-unidentified rolein avian influenza, then it could be inferred that the firsthuman 1957 pandemic N2 strains carrying the HB signature(RI51y57 and Leningrady134y57) most probably were de-rived from a genetic reassortment event with an avian strain(37). This HB signature since has disappeared under antigenicdrift in post-1957 human and swine stains, indicating that itserves no biological function in pathogenesis of influenza inhumans and pigs.

One of the puzzling aspects of this study is the relationshipof this site to equine strains. Equine strains, which appearacross all subtypes, almost always carry this HB signature, andit is possible they share with avian species similar carbohy-drates that are the natural ligands for this HB site. Further-more, the signature is present in the RI51y57 viral strain,which had high affinity to normal horse serum (38), but isabsent in the RI5-y57 strain, which had a low affinity to normalhorse serum. The RI51y57 strain was selected as a result ofserial passaging in chicken embryo. Although circumstantialevidence indicated that the difference between the RI51y57and RI5-y57 strains probably was due to different affinity ofthe HA for the inhibitor in horse serum (38–40), it wouldappear that the differences in the NA at the HB site could bean important factor in the different properties of the twostrains. When horse serum was treated with Vibrio choleraextract, RI51y57 virus retained binding to horse serum,whereas RI5-y57 virus did not do so, further indicating thateither the HA was different from most HAs that don’t bind toV. cholera-treated serum, or that this HB site on the NA wasresponsible for this difference.

Sequences of the paramyxovirus HN protein, which carriesboth NA and HA function, were examined based on a se-quence alignment that is consistent with the viral NA fold (41).This HB signature was not able to be identified.

We thank Pat Pilling for technical assistance and Colin Ward andBrian Smith for helpful discussions and reading this manuscript.

1. Lamb, R. A. (1989) in The Influenza Viruses ed. Krug, R. M.(Plenum, New York), pp. 1–87.

2. Colman, P. M. (1994) Protein Sci. 3, 1687–1696.3. Klenk, E., Faillard, H. & Lempfrid, H. (1955) Z. Physiol. Chem.

301, 235–246.4. Gottschalk, A. (1957) Biochim. Biophys. Acta 23, 645–646.5. Ward, C. W. (1981) Curr. Top. Microbiol. Immunol. 94y95, 1–74.6. Palese, P., Tobita, K., Ueda, M. & Compans, R. W. (1974)

Virology 61, 397–410.7. Klenk, H. D. & Rott, R. (1988) Adv. Virus. Res. 34, 247–281.8. Varghese, J. N., Laver, W. G. & Colman, P. M. (1983) Nature

(London) 303, 35–40.

9. Colman, P. M., Varghese, J. N. & Laver, W. G. (1983) Nature(London) 303, 41–44.

10. Varghese, J. N. & Colman, P. M. (1991) J. Mol. Biol. 221,473–486.

11. Baker, A. T., Varghese, J. N., Laver, W. G., Air, G. M. & Colman,P. M. (1987) Proteins 1, 111–117.

12. Tulip, W. G., Varghese, J. N., Baker, A. T., van Donkelaar, A.,Laver, W. G., Webster, R. G. & Colman, P. M. (1992) J. Mol. Biol.221, 487–497.

13. Burmeister, W. P., Ruigrok, R. W. H. & Cusack, S. (1992) EMBOJ. 11, 49–56.

14. Varghese, J. N., McKimm-Breschkin, J. L., Caldwell, J. B., Kortt,A. A. & Colman, P. M. (1992) Proteins 14, 327–332.

15. von Itzstein, M., Wu, W.-Y, Kok, G. B., Pegg, M. S., Dyson, J. C.,Jin, B., VanPhan, T., Symthe, M. L., White, H. F., Oliver, S. W.,Colman, P. M., Varghese, J. N., Ryan, D. M., Woods, J. M.,Bethell, R. C., Hotham, V. J., Cameron. J. M. & Penn, C. R.(1993) Nature (London) 363, 418–423.

16. Varghese, J. N., Epa, V. C. & Colman, P. M. (1995) Protein Sci.4, 1081–1087.

17. Smith, P. W., Sollis, S. L., Howes, P. D., Cherry, P. C., Cobley,K. N., Taylor, H., Whittington, A. R., Scicinski, J., Bethell, R. C.,Taylor, N., Skarzynski, T., Cleasby, A., Singh, O., Wonacott, A.,Varghese, J. & Colman, P. (1996) Bioorg. Med. Chem. Lett. 6,2931–2936.

18. Sollis, S. L., Smith, P. W., Howes, P. D., Cherry, P. C. & Bethell,R. C. (1996) Bioorg. Med. Chem. Lett. 6, 1805–1808.

19. Kim, C. U., Lew, W., Williams, M. A., Liu, H., Zhang, L.,Swaminathan, S., Bischofberger, N., Chen, M. S., Mendel, D. B.,Tai, C. Y., Laver, W. G. & Stevens, R. C. (1997) J. Am. Chem.Soc. 119, 681–690.

20. von Itzstein, M., Wu, W.-Y. & Jin, B. (1994) Carbohydr. Res. 259,301–305.

21. Laver, W. G., Colman, P. M., Webster, R. G., Hinshaw, V. S. &Air, G. M. (1984) Virology 137, 314–323.

22. Webster, R. G., Air, G. M., Metzger, D. W., Colman, P. M.,Varghese, J. N., Baker, A. T. & Laver, W. G. (1987) J. Virol. 61,2910–2916.

23. Nuss, J. M. & Air, G. M. (1991) Virology 183, 496–504.24. Air, G. M. & Laver, W. G. (1995) Virology 211, 278–284.25. Hausmann, J., Kretzschmar, E., Garten, W. & Klenk, H.-D.

(1995) J. Gen. Virol. 76, 1719–1728.26. McKimm-Breschkin, J. L., Caldwell, J. B., Guthrie, R. E. & Kortt,

A. A. (1991) J. Virol. Methods 32, 121–124.27. Blick, T. J., Tiong, T., Sahasrabudhe, A., Varghese, J. N., Colman,

P. M., Hart, G. J., Bethell, R. C. & McKimm-Breschkin, J. L.(1995) Virology 214, 475–484.

28. Brunger, A. T., (1992) X-PLOR, Version 3.3: A System for X-rayCrystallography and NMR (Yale Univ. Press, New Haven, CT).

29. Engh, R. A. & Huber, R. (1991) Acta Cryst. A 47, 392–400.30. Sauter, N. K., Glick, G. D., Crowther, R. L., Park, S.-J., Eisen,

M. B., Skehel, J. J., Knowles, J. R. & Wiley, D. C. (1992) Proc.Natl. Acad. Sci. USA 89, 324–328.

31. Weiss, W., Brown, J. H., Cusack, S., Paulson, J. C., Skehel, J. J.& Wiley, D. C. (1988) Nature (London) 333, 426–431.

32. Crennell, S. J., Garman, E. F., Philippon, C., Laver, W. G., Vimr,E. R. & Taylor, G. L. (1996) J. Mol. Biol. 259, 264–80.

33. Alexander, D. J. (1986) in Proceedings of the 2nd InternationalSymposium on Avian Influenza (U.S. Animal Health Assoc.,Richmond, VA), pp. 4–13.

34. Slemons, R. D., Johnson, D. C., Osborn, J. S. & Hayes, F. (1974)Avian Dis. 18, 119–125.

35. Hinshaw, V. S., Webster, R. G. & Turner, B. (1980) Can. J.Microbiol. 26, 622–629.

36. Webster, R. G., Bean, W. J., Gorman, O. T., Chambers, T. M. &Kawaoka, Y. (1992) Microbiol. Rev. 56, 152–179.

37. Skehel, J. J. (1974) Symp. Soc. Gen. Microbiol. 24, 321–342.38. Choppin, P. W. & Tamm, I. (1960) J. Exp. Med. 112, 895–920.39. Sugiura, A., Shimojo, I. & Enomoto, C. (1961) Japan J. Exp. Med.

31, 159–167.40. McCahon, D. & Schild, G. C. (1971) J. Gen. Virol. 12, 207–219.41. Colman, P. M., Hoyne, P. A. & Lawrence, M. C. (1993) J. Virol.

67, 2972–2980.42. Nicholls, A., Sharp, K. & Honig B. (1991) Proteins 11, 281–296.43. Kraulis, P. J. (1991) J. Appl. Cryst. 24, 946–950.

11812 Biochemistry: Varghese et al. Proc. Natl. Acad. Sci. USA 94 (1997)