VIROLOGY

183,496-504

(1991)

Transfer of the Hemagglutinin Activity of Influenza Virus Neuraminidase Subtype N9 into an N2 Neuraminidase Background JACQUELINE M. NUSS AND GILLIAN M. AIR’ Department

of Microbiology,

University Received

of Alabama January

28,

at Birmingham, 199 1; accepted

UAB Station, March

Birmingham,

Alabama

35294

9, 199 1

It has previously been shown that influenza virus neuraminidase (NA) of the N9 subtype is unusual in that it possesses hemagglutinin activity as well as NA activity. Loss of red cell binding in certain escape mutants suggested that the hemagglutinating site is separate from the NA active site and involves at least two of the polypeptide loops found on the surface of the molecule (Webster et al., 1987. J. Viral. 61,291 O-291 6). We have used site-directed mutagenesis to transfer the amino acids in these loops at positions 368-370 and 399-403 of N9 NA (A/tern/Australia/G7Oc/75), separately and together, into subtype N2 NA (A/Tokyo/3/67). The three mutant proteins were expressed from an SV40 promoter and were also under the control of the bacteriophage T7 promoter for use in the vaccinia virus T7 polymerase transient expression system (Fuerst et al., 1986. Proc. Nat/. Acad. Sci. USA. 83, 6122-8126). The mutant which contained both loops of N9 NA had acquired the hemagglutinin activity of N9. The agglutinated red cells are released by the enzyme activity of N9 NA, indicating that the agglutination involves binding to sialic acid in the same configuration as does the parental N9 NA, and an inhibitor of NA did not affect hemagglutination, indicating that this site is separate from the NA site as in parental N9. o 1991 Academic PWSS. h.

The three-dimensional structure of N9 NA at 3.0 A resolution (Baker et a/., 1987) showed that its polypeptide chain is folded similarly to that of subtype N2 NA (Varghese et al., 1983). Analysis of antibody-selected escape mutants of N9 NA revealed that single changes in amino acids located on two of the NA surface loops which make up the rim of the enzyme active site crater cause loss of the HA activity without affecting NA activity. Side chains of these amino acids point away from the enzyme active site and appear to form a shallow pocket which was suggested to be the HA site (Webster et a/., 1987). We wanted to investigate the possibility of transferring the HA activity of N9 NA into N2 NA and started with transfer of these two surface loops. Despite 50% sequence differences between the two subtypes, many residues flanking the escape mutant sites on these loops are identical, which permitted us to utilize oligonucleotide-directed site-specific mutagenesis to alter those amino acid residues in N2 which varied from those in N9. This paper reports the construction of three N2 mutants by this technique and describes the properties of one of these mutants, TokC, in which the HA activity of N9 has been acquired.

INTRODUCTION

The two surface glycoproteins of influenza Aviruses, neuraminidase (NA) and hemagglutinin (HA), have distinct roles in the virion life cycle. The HA mediates virus entry by attachment to sialic acid receptors on the host cell surface and subsequent fusion between viral and host cell membranes. The NA, at least in cultured infected cells, permits release of progeny virus from the infected host and prevents self-aggregation during spread of infection (Palese et a/., 1974). HA and NA molecules are further classified into subtypes on the basis of lack of antigenic cross-reactivity. There are currently nine recognized subtypes of influenza A neuraminidase (WHO, 1980). It has been previously reported (Laver et al., 1984) that neuraminidase subtype N9 isolated from influenza virus A/tern/Australia/G70c/75 also possesses hemagglutinin activity. This HA activity has been found in other N9 NAs (Air eta/., 1987) but not in other subtypes of NA. The hemagglutinin activity of N9 NA was shown to be threefold higher than that of H3 HA at 4°C (Laver et a/., 1984). This HA activity of N9 NA appeared to be localized at a separate site on the protein molecule from its enzyme active site since an inhibitor of NA, 2-deoxy-2,3-dehydro-N-acetyl-neuraminic acid, had no effect on the ability of the NA to agglutinate red blood cells. ’ To whom 0042.6822191

requests

for reprints

$3.00

Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form resewed.

should

MATERIALS Vectors,

AND

METHODS

ceils, and virus stocks

The cloning of full-length copies of the N2 and N9 NA genes isolated from reassortant influenza viruses

be addressed. 496

TRANSFER

OF

HA ACTIVITY

A/NWS/33(Hl)-AfTokyo/3/67(N2) and A/NWS/33(Hl)-A/tern/Australia/G70c/75(N9) has been described (Lentz et a/., 1984; Air er al., 1985). The Escherichia co/i strains TGl and RR1 were used to grow up the NA plasmids. The CV-1 African green monkey kidney cell line was used for expression of the wild-type and mutant NA genes, and these cells were maintained in Dulbecco’s modified Eagle medium (GIBCO) containing 10% supplemented calf serum (HyClone), 50 U/ml penicillin, and 50 pg/ml streptomycin. All DNA manipulations were done as described (Lentz and Air, 1986). Mutagenic oligonucleotides were synthesized using an Applied Biosystems DNA synthesizer. Mutagenesis

Oligonucleotide-directed site-specific mutagenesis was performed (Zoller and Smith, 1983) with modifications as previously described (Lentz and Air, 1986; Norris et a/., 1983). The sequences of N2 and N9 NAs have been previously reported (Lentz eta/., 1984; Air er al., 1985) and were used to design the mutagenizing primers. The 26mer mutagenic oligo N2-1112M, 5’ ACAATCAGCATCGCTTCACGCTCAGG 3’, was used to generate mutant TokA. Mutagenic oligo N2-1202M, 5’ CAAGTCATAGT-TCTCAATACTGAT~GGTCAGGT-T-A 3’, is a 35mer and was used to generate mutant TokB, and both oligos were used in generating mutant TokC. The positions of the mismatches used to obtain the desired amino acid changes are indicated by the italicized nucleotides. Primer N2-1 112M creates three amino acid changes in N2 at positions 368-370 (KDL to IAS). Primer N2-1202M creates five amino acid changes at residues 399-403 (DSDNR to LNTDW). After mutagenesis, the reaction mix was diluted 1:lO, transformed into E. co/i strain TGl, and plated. Candidate clones were selected using the plaque hybridization technique (Zoller and Smith, 1983) and a single room temperature wash. The normal procedure involves sequential washes at increasing temperatures, but due to the high number of mismatches in our mutagenic oligonucleotides, room temperature washes were sufficient to discriminate mutants from wild type. The mutations were confirmed by sequencing using the dideoxy chain-termination method (Sanger et a/., 1977). The mutant NA genes were expressed from the T7 promoter of the vector Bluescript (Stratagene) in the vaccinia virus T7 polymerase transient assay system (Fuerst et a/., 1986). Briefly, justconfluent CV-1 cells on 35-mm plates were infected at 37°C with the recombinant vaccinia virus expressing T7 polymerase (wTF7-3) at an m.o.i. of -5 PFU/cell diluted in CaMgPBS -t 0.1% BSA. After the virus was allowed to adsorb for 1 hr, the plates were washed

FROM

NA

N9 TO

497

N2

twice with serum-free DMEM and then immediately transfected with the Bluescript-NA plasmid (Felgner et al., 1987; Chang and Brenner, 1988) using the following modifications. Fifteen micrograms of the desired clone DNA and 12 ~1 lipofectin (BRL) were each diluted in separate polystyrene tubes into 330 ~1 DMEM (serum-free). The medium-lipofectin mix was added dropwise to the medium-DNA mix, incubated for 15 min at room temperature, and then added to cells infected with wTF7-3. After l-2 hr at 37”C, 330 ~1 DMEM (serum-free) was added to bring the final volume to 1 ml. Cells were assayed at 16 hr post-transfection for enzyme activity and red cell binding. In addition, recombinant SV40 stocks were made of wild-type N2 and N9 NA and mutant TokC NA as previously described (Lentz and Air, 1986). These were transfected into CV-1 cells along with the helper virus d11055. Titers of recombinant SV40 virus stock were achieved such that a 1:5 dilution would allow moderate to high cytopathic effects 48-72 hr post-transfection. These SV40 stocks were used in the neuraminidase inhibition (NI) assays and red cell binding assays. Analysis of proteins

For immunoprecipitation, vaccinia virus-infected CV1 cells were transfected with Bluescript containing wild-type N2, N9, and mutant N2 NAs. At 12 hr posttransfection, the cells were starved 1 hr for methionine and cysteine, labeled 1 hr with EXPRE35S35S-Protein Labeling mix (NEN), and chased 2 hr with cold DMEM containing additional methionine (1.5 mg/lOO ml) and cysteine (2.4 mg/lOO ml). The cells were scraped, lysed, and immunoprecipitated (Lentz and Air, 1986) for 20 min at 37°C with a rabbit antiserum to A/Tok/Y 67 NA (1 ~1of a 1: 10 dilution). Immuno-complexed proteins were recovered by binding to Protein A-Sepharose (45 min at RT), and l/i 0th of the SDS-eluted samples were run on a 12% polyacrylamide gel. NA and HA assays

NA assays were performed by standard procedures (Aymard-Henry et a/., 1973) using 50 ~1 of either 40 mg/ml fetuin or 2 mM N-acetylneuramin lactose (NANL) as substrate. NI assays followed the same procedure except that 2 I.CIof a 1 :lO dilution of monoclonal antibody or 10 mM inhibitor was added 10 min prior to addition of fetuin substrate. Red cell binding assays were done at 0°C as previously described (Laver eta/., 1984). Monoclonal

antibodies

and inhibitors

Monoclonal antibody (mAb) 113/2 (Webster et a/., 1984) which recognizes an epitope on N2 NA, and

498

NUSS A

0.9

0.8 i

AND

AIR

+ w control oNN2 -*-TokA * TokB -* TokC

68k 43k

,

0/* 0

0.5

1

1.5

2

2.5

3

3.5

-4,

4

18k 14k

Time (hours)

FIG. 1. (A) Neuraminidase activity in wild-type and mutant NAs expressed in w-T7 transient assay system. N2 wild-type = N2 NA from AITokyo/3/67. TokA,B,C = N2 NA mutants containing N9 residues in HA site (see text). The absorbance at 549 nm is a measure of the level of enzyme activity from 10% of an infected 35-mm plate 16 hr post-transfection. TheXaxis reflects the length of time samples were incubated with the substrate fetuin. (B) lmmunoprecipitation of N2 NA using polyclonal antisera to AAok/3/67 (see Materials and Methods for description). For comparison to NA assays, l/l 0th of labeled proteins from a 35-mm plate was loaded on the gel.

mAbs NC47, NC45, NC42, NC34, NC31, NC20, NC1 1, which all recognize epitopes on N9 NA, have been described previously (Webster et al., 1987) and were generously provided by Dr. Webster. The inhibitor 2-deoxy-2,3-dehydro-N-acetyl-neuraminic acid (DANA) was obtained from Sigma Chemical Co. and was used at a concentration of 10 mlVI.

three mutants are enzymatically active when using either fetuin or NANL as the substrate. To determine whether the reduced NA activity of the mutants was due to decreased expression, an immunoprecipitation was performed using sample amounts equivalent to those in the enzyme assay. Figure 1B reveals that levels of protein in the wild-type and mutant N2 NAs do not differ significantly.

RESULTS Expression

of mutant

N2 neuraminidases

Oligonucleotide-directed site-specific mutagenesis was performed on a copy of the N2 NA gene of influenza virus AAokyo/3/67 cloned in Ml 3mpl8 to obtain three mutants. Mutant TokA contained N9 residues 368-370 in the N2 background, mutant TokB contained N9 residues 399-403, and mutant TokC contained N9 residues at both of these locations. Once the mutants were confirmed by sequence analysis, these genes were subcloned into a Bluescript plasmid vector for transient expression in a vaccinia virus T7 polymerase system (Fuerst et al., 1986). Neuraminidase activity assays were performed to determine if the mutant N2 NAs were being properly expressed, folded into tetramers, and transported to the cell surface. Results of these experiments are shown in Fig. 1A. Although NA activity of the mutants is reduced in comparison to that of wild-type N2, all

Ability of mutant

NAs to agglutinate

red blood cells

Next we tested the mutant N2 NAs for their ability to agglutinate human erythrocytes. These assays were carried out at 0°C as previously described (Laver et a/., 1984) to prevent the NA activity from clipping sialic acids off the surface of the red cells and releasing agglutination. Figure 2 shows that wild-type N2 and mutants TokA and TokB, each containing a single loop of N9 sequence, were unable to agglutinate the red blood cells (RBCs), whereas wild-type N9 exhibited a high level of HA activity. However, mutant TokC, with both N2 loops replaced by N9 residues, had clearly acquired the hemagglutinin activity of N9. This result was reproduced in every experimental trial and confirmed the location of the HA site in N9 NA. To show that the specificity of HA binding is the same in both the wildtype N9 and TokC mutant, RBCs were added at 0°C to infected cells, the binding was photographed, and the

TRANSFER

OF

HA ACTIVITY

FROM

NA

N9

TO

N2

TokA

N2

TokB

N9

TokC

FIG;. 2. Red cell binding N2, Iv A from AAokyo/3/67;

to CV-1 cells cotransfected with wTF7-3 N9, NA from AItern/Australia/G70c/75;

an d Bluescript-NA clone 1-okA,B,C, N2 mutants

were incubated at 37°C for 30 min. Results of I-expressed NA clones are illustrated in the top ‘ows of Fig. 3. After warming to 37”C, nearly all of two the ?BCs that were bound at 0°C had been eluted. Thu i the mutant is binding to sialic acid in a configuration i/vhich can be cleaved by NA as in N9. 3

sv4

DNAs. VV, Cells infected with recombinan with residue changes to N9 (see text).

It w only;

The HA activity of TokC is not due to crosscontamination with wild-type N9 To demonstrate that the acquired HA activity of mutant TokC was not due to contamination with re combinant wild-type N9 SV40 stock, NI assays wer’ e done

N9

O0

37”

Inh

Ab

TokC

TRANSFER OF HA ACTIVITY FROM NA N9 TO N2

% A C : V I G

501

60%

60%

40%

N2

N2+ 113l2

N2+ DANA

N9

N9+ 113/2

N9+ DANA

TokC

TokC+ 11312

TokC+ DANA

FIG. 4. Inhibition of mutant TokC NA activity. NA activities of wild-type N2, N9, and TokC are shown in the absence or presence of either N2 monoclonal antibody 11312 (Webster eta/., 1984) or the inhibitor DANA. Ab or DANA was incubated with infected cells prior to addition of fetuin substrate. The Y axis depicts the percentage of NA enzyme activity.

using an N2 mAb directed against N2 NA and several anti-N9 NA mAbs previously described (Webster era/., 1987). As depicted in Fig. 4, N2 mAb 113/2, which recognizes an epitope on the opposite side of the NA molecule from the region changed by mutagenesis, was able to inhibit the enzyme activity of TokC, yet had no effect on wild-type N9 activity. Figure 5 shows that monoclonal antibodies to N9 had no effect on the mutant TokC’s enzyme activity, whereas they were able to severely limit or completely abolish N9 enzyme activity. In addition, N9 mAb NC20 was incubated with TokCinfected CV-1 cells prior to addition of erythrocytes. The bottom panels of Fig. 3 show that while NC20 was able to inhibit N9 HA activity, it had no effect on the HA activity of TokC. The HA activity of N9 NA is separate enzyme active site

addition, when the N2 mAb 113/2 (Webster et a/., 1984), which inhibits NA activity (Fig. 4) was added to CV-1 cells cotransfected with wTF7-3 and TokC NA prior to addition of red cells, no inhibition of HA activity was observed (fig. 6). These experiments suggest that the HA activity of both N9 and TokC is not due to increased substrate binding by the enzyme, but must exist at a separate site that is still available for binding to red cell receptors when NA is blocked. We also tested for inhibition of HA activity of the expressed wild-type N9 and TokC mutant by the smaller NA substrate NANL but no inhibition was seen (data not shown). NANL has been shown not to affect RBC binding by the hemagglutinin molecule itself (Weis et al,, 1988).

from its

We examined the effect of an NA inhibitor, DANA, on the HA activity of both N9 and TokC NAs. DANA was added to infected cells at 0°C for 10 min prior to addition of erythrocytes. Results show that the HA activity of TokC was not inhibited by DANA (Fig. 3), whereas the NA activity of both wild-type and mutant NAs was inhibited (Fig. 4). This result is the same as that previously reported and seen here for N9 HA activity. In

DISCUSSION It was previously reported (Laver et al., 1984) that while the inhibitor DANA was able to completely inhibit N9 NA activity, it had no effect on the HA activity of this molecule. This observation led to the hypothesis that the HA activity of N9 must be located at a site distinct from its enzyme activity. Further support for this hypothesis was shown by Webster et al. (1987), who reported that some mAbs to N9 were able to inhibit its

FIG. 3. Inhibition of red cell binding. Both wild-type N9 and TokC NAs are able to bind red blood cells at 0°C. Top four panels: RBCs bound at 0°C on the surface of CV-1 cells infected with SV40 containing N9 NA and TokC NA can be eluted at 37°C. Inh, infected CV-1 cells were first incubated with DANA prior to addition of RBCs. Ab, N9 mAB NC20 was added prior to addition of erythrocytes. N9 mAb NC31 was also tested and gave identical results (data not shown),

502

NUSS

%

AND

AIR

80%

; T

60%

\: I T Y

No ab

NC47

NC45

NC42

NC34

FIG. 5. Inhibition of mutant TokC NA activity by antibodies against NC20, and NC1 1 (Webster et al., 1987) were added to infected cells

NC31

N9 NA. N9 monoclonal prior to addition of fetuin

NA activity, but had no effect on its HA activity. In addition, some escape mutants selected by N9 mAbs had lost their HA activity, while still retaining NA activity. Amino acid changes in these escape mutants were located on two of the surface loops surrounding the active site crater. Single changes occurring at amino

NC20

NC11

antibodies substrate.

NC47,

NC45

NC42,

NC34,

NC31,

acids 367, 369,370,372, and 400 resulted in partial or complete loss of HA activity, whereas a change at residue 368 had no effect on hemagglutination by N9 NA. An examination of these residues in N2 NA, which lacks any HA activity, revealed differences at positions 368,369,370, and 400, whereas amino acids 367 and

+ Ab FIG. 6. CV-1 cells cotransfected with presence (+ Ab) of N2 mAb 11312.

vvTF7-3

and mutant

TokC

NA were

tested

for inhibition

of TokC

HA activity

in the absence

(NO Ab) or

TRANSFER

OF

HA

ACTIVITY

368 369 370

399-

67

\-402 401

FIG. 7. Three-dimensional model of a monomer of N2 NA. Positions changed in N2 mutants are indicated. The NA active site is indicated by the shaded area. Mutant TokC contains all of these changes and has acquired HA activity.

372 were identical between the two subtypes, as were other flanking residues. Therefore site-directed mutagenesis was utilized to transfer all the amino acids of the loops implicated in HA binding of N9 into the N2 molecule. Figure 7 illustrates the location of the changes on a model of N2 NA; this did indeed create a new HA site on the N2 framework. All of the mutants had NA activity at reduced levels compared to those observed with wild-type N2 NA. The mutants are either defective in binding or cleaving substrate or do not fold as efficiently as wild-type N2 NA. This is understandable since although the changes made are not in the active site pocket, they are located on surface loops known from crystal structure determination to extend into the active site. Thus any distortion of the loops caused by the substitutions could also affect the active site and presumably have some destabilizing effect. Similarly, the decreased inhibition of the mutant TokC’s NA activity with N2 mAb 113/2 is likely to be due to decreased binding affinity caused by some distortion of the epitope. The successful transfer of the N9 HA site into the TokC mutant suggests that this site is relatively small and may only involve the few residues which form a shallow depression on the surface of the NA. A comparison of crystal structures of the N2 and N9 NA molecules reveals that they are folded similarly, but the side chains of residues which are identical between the two in the HA site are in slightly different orientations (Webster et a/,, 1987). It is unknown whether transfer of the differing residues on these loops in N2 results in a conformational change which permits TokC to bind sialic

FROM

NA

N9

TO

N2

503

acid residues as in N9 or if it is these amino acid side chains themselves which are important for this interaction to occur, as suggested by escape mutant data. Higher refinement of crystallographic models at this site in the two structures will assist in clarification of this problem (Bossart, Babu, Luo, Laver, and Air, unpublished results). Some escape mutants selected by N9 monoclonal antibodies were mapped to the polypeptide loops which surround the enzyme active site and include the two loops which were altered in N2 to create TokC. We wanted to determine if certain N9 mAbs could recognize TokC since their epitopes are known to include at least those loops we changed to N9 sequence. Figure 5 shows that there is no inhibition of TokC NA activity by any of the N9 mAbs tested. These results indicate that while the HA activity of N9 has been acquired, epitopes for N9 mAbs have not been transferred and emphasize the point that antibody epitopes encompass large areas on the protein surface and are not just small interactions with local residues. The three-dimensional structure of N9 NA complexed with mAb NC41 has been determined (Tulip et al., 1989). Escape mutant mapping had identified three loops on top of the NA monomer for the NC41 epitope, whereas the structural data showed that antibody NC41 contacts five loops. The function of the HA activity of N9 is unknown. We have shown the location of the N9 HA site by its successful transfer into neuraminidase subtype N2. The N2 structure is similar enough to N9 that transfer of activity was easily accomplished without destroying NA activity and therefore may indicate some biological significance. Perhaps the site in N2 binds something other than sialic acid.

ACKNOWLEDGMENTS We thank Dr. Robert G. Webster, St Jude Children’s Research Hospital (Memphis, TN) for his generous supply of N9 and N2 monoclonal antibodies, and Kimberly Rivera for excellent technical assistance. We also thank Dr. Gail We& for the modified transfection procedure. Mutagenic oligonucleotides were synthesized by Dr. Jeffrey Engler, University of Alabama at Birmingham. This work was supported in part by Grants Al-l 9084, CA-l 3 148, and Training Grant AI-07 150 from the National Institutes of Health.

REFERENCES AIR, G. M., LENT?!, M. R., RITCHIE, L. R., WEI, X., and HAMILTON, A.-M. (1987). Expression and mutation of influenza antigen genes using SV40 vectors. In “The Biology of Negative Strand Viruses” (M. Kolakofsky, Ed.), Elsevier, New York. AIR, G. M., RITCHIE, L. R., LOVER, W. G., and COLMAN, P. M. (1985).

504

NUSS

Gene and protein sequence of an influenza neuraminidase with hemagglutinin activity. Virology 145, 1 17-122. AYMARD-HENRY, M., COLEMAN, M. T., DOWDLE, W. R., LAVER, W. G., SCHILD, G. C., and WEBSTER, R. G. (1973). Influenzavirus neuraminidase and neuraminidase inhibition test procedures. Bull. WHO 48, 199-202. BAKER, A. T., VARGHESE, I. N., LAVER, W. G., AIR, G. M., and COLMAN, P. M. (1987). Three-dimensional structure of neuraminidase of subtype N9 from an avian influenza virus. Profeins 2, 11 l-l 17. CHANG, A. C. Y., and BRENNER, D. G. (1988). Cationic liposome-mediated transfection: A new method for the introduction of DNA into mammalian cells. Focus 10, 66-69. FELGNER, P. L., GADEK, T. R., HOLM, M., ROMAN, R., CHAN, H. W., WENZ, M., NORTHROP, J. P., RINGOLD, G. M., and DANIELSEN, M. (1987). Lipofectin: A highly efficient, lipid-mediated DNA-transfection procedure. Proc. Natl. Acad. Sci. USA 84, 7413-7417. FUERST, T. R., NILE.% E. G., STUDIER, F. W., and Moss, B. (1986). Eukaryotic transient-expression system based on recombinant vaccinia virus that synthesizes bacteriophage T7 RNA polymerase. Proc. Natl. Acad. Sci. USA 83, 8 122-8 126. LAVER, W. G., COLMAN, P. M., WEBSTER, R. G., HINSHAW, V. S., and AIR, G. M. (1984). Influenza virus neuraminidase with hemagglutinin activity. Virology 137, 314-323. LENTZ, M. R., and AIR, G. M. (1986). Loss of enzyme activity in a site-directed mutant of influenza neuraminidase compared to expressed wild-type protein. Virology 148, 74-83. LENTZ, M. R., AIR, G. M., LAVER, W. G., and WEBSTER, R. G. (1984). Sequence of the neuraminidase gene of influenza virus AITokyo/3/ 67 and previously uncharacterized monoclonal variants. Virology 135, 257-265.

AND

AIR NORRIS, K., NORRIS, F., CHRISTIANSEN, L., and FIIL, N. (1983). Efficient site-directed mutagenesis by simultaneous use of two primers. Nucleic Acids Res. 11, 5 103-5 112. PALESE, P., TOBITA, K., UEDA, M., and COMPANS, R. W. (1974). Characterization of temperature-sensitive influenza virus mutants defective in neuraminidase. Virology 61, 397-410. SANGER, F., NICKLEN, S., and COULSON, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Nat/ Acad. Sci. USA 74, 5463-5467. TULIP, W. R., VARGHESE, J. N., WEBSTER, R. G., AIR, G. M., LAVER, W. G., and COLMAN, P. M. (1989). Crystal structures of neuraminidase-antibody complexes. Co/d Spring Harbor Symp. Quanf. Biol. 54,257-263. VARGHESE, J. N., LAVER, W. G., and COLMAN, P. M. (1983). Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9A resolution. Nature 303, 35-40. WEBSTER, R. G., AIR, G. M., METZGER, D. W., COLMAN, P. M., VARGHESE, 1. N., BAKER, A. T., and LAVER, W. G. (1987). Antigenic structure and variation in an influenza N9 neuraminidase. 1. Viral. 61, 2910-2916. WEBSTER, R. G., BROWN, L. E., and LAVER, W. G. (1984). Antigenic and biological characterization of influenza virus neuraminidase (N2) with monoclonal antibodies. Virology 135, 30-42. WEIS, W., BROWN, J. l-l., CUSACK, S., PAULSON, J. C., SKEHEL, J. J., and WILEY, D. C. (1988). Structure of the influenza virus hemagglutinin complexed with its receptor, sialic acid. Nature 333, 426-431. WHO (1980). Memorandum. A revision of the system of nomenclature for influenza viruses. Bull. WHO 58, 585-591, ZOLLER, M., and SMITH, M. (1983). Oligonucleotide-directed mutagenesis of DNA fragments cloned into Ml 3 vectors. In “Methods in Enzymology,” Vol. 100, pp. 468-500. Academic Press.