- Email: [email protected]
Location of antigenic sites on the three-dimensional structure of the influenza N2 virus neuraminidase
VIROLOGY
145,237-248
(1985)
Location of Antigenic Sites on the Three-Dimensional Structure of the influenza N2 Virus Neuraminidase G. M. AIR,*” M. C. ELS,* L. E. BROWN,? W. G. LAVER,*
AND
R. G. WEBSTERO
*Department of Microbiology, University of A&ma at Birmingham, Birmin&m, Alabama 35294 TMicrobiology Department, University of Melbmwne, Parkville, Victoria 5052,Australia; $John Curtin School of Medical Research, Canberra City, 2601,Australia; and Wt. Jude Childrens Research Hospital, Memphis, Tennessee, 3%101 Received March 20,19%5;accepted June ~$1985 Sequence analysis of the neuraminidase (NA) genes of influenza virus X-i’(F1) and of 12 variants selected with monoclonal antibodies has been used to define in physical terms the antigenic structure of this NA, which was operationally established by R. G. Webster, L. E. Brown, and W. G. Laver (1934, Virologg 135,30-42). X-‘7(Fl) is a reassortant virus containing the NA of the early Asian (H2N2) isolate A/RI/5+/57, and the results of antigenic and sequence analysis of X-7(Fl) and of variants selected with monoclonal antibodies have been combined with a similar analysis of the A/Tokyo/3167 NA (H2N2, M. R. Lentz, G. M. Air, W. G. Laver, and R. G. Webster (1984), Virology 135,257-265) to obtain a model of antibody binding to N2 NAs. The selection process was biased, however, since only those monoclonal antibodies which inhibited NA activity could be used to select variants. Most of the changes in the variants selected with monoclonal antibodies occur in those parts of the polypeptide chain which encircle the enzyme active site pocket in the threedimensional structure (P. M. Colman, J. N. Varghese, and W. G. Laver (1983), Nature (London) 303,41-44). The results suggest that in general the antibody binds to a site on the NA which includes those amino acid side chains which are altered in monoclonal variants. There are, however, several aspects of the antigen-antibody interaction which are not easily explained, and which will probably only be fully elucidated by X-ray cryso 1985 Academic press. ITIC. tallographic analysis of NA-antibody complexes. INTRODUCTION
The neuraminidase (NA) is one of the two projecting surface glycoproteins of influenza A and B viruses, the other being the hemagglutinin (HA). Although the precise role of the NA in viral infection is not known, it appears to be important in facilitating movement of the virus through mucin to host cells, and assists in release of progeny viruses by cleaving sialic acid from cell receptors and from the HA. Antibody to the NA does not neutralize virus infectivity except at high concentrations, but it does slow the release of virus from infected cells, giving reduced plaque sizes (Kilbourne et al, 1968). The NA of type A influenza viruses uni Author addressed.
to whom requests for reprints
should be
dergoes antigenic shift and drift, as does the HA. In 1957 the HlNl viruses which had been circulating in the human population since 1933 when the virus was first isolated were replaced by viruses of the Asian subtype, H2N2. Between 1957 and 1968 the H2 HA and N2 HA underwent antigenic drift, before the HA was replaced in 1968 in Hong Kong (H3N2) influenza. The N2 NA continued to drift along with the Hong Kong HA, and H3N2 variants are still circulating. The three-dimensional structure of the N2 NA has been solved by X-ray crystallography to 2.9A (Varghese et a& 1983).The structure determination was made using crystals of two different N2 NAs, representing early (A/RI/5+/57) and late (A/ Tokyo/3/67) Asian strains. The crystals were grown from NA released by pronase from reassortant viruses which contained
237
0042-6822185$3.00 Copyright All rights
0 1985 by Academic Press, Inc. of reproduction in any form reserved.
238
AIR
ET AL.
adsorption to and elution from chicken erythrocytes followed by density gradient centrifugation through lo-40% sucrose (Laver, 1969). Monoclonal antibodies and selection of variants. Hybridoma cell lines producing antibodies to the neuraminidase of X-7(Fl) were prepared by the method of Kiihler and Milstein (1976). Their preparation, properties, and the antigenic properties of variants selected with these antibodies in eggs or in MDCK cells have been described previously (Webster et aL, 1984). Serological assays. The reactivities of monoclonal antibodies with variant viruses were measured in an enzyme-linked immunosorbent assay (ELISA) as previously described (Kida et aL, 1982; Webster et ak, 1984). Chemical modi&cation of the neuraminidase. Pronase-released NA heads of Tokyo/67 and X-7(Fl) were reacted with 1-fluoro-2,4-dinitrobenzene (FDNB) or succinic anhydride at pH 8.5 as previously described for modification of the HA (Laver et CAL, 1981). In order to determine the overall degree of succinylation, the succinyl protein was dissolved in 6 M guanidine hydrochloride, reacted with FDNB, and hydrolyzed with 6 N HCl. The free lysine recovered was a measure of the degree of succinylation. Reactivity of individual lysine residues was measured by analysis of tryptic peptides separated by paper electrophoresis and chromatography (Laver, 1978). Sequencing of neuraminidase genes of X7(Fl) and its variants. Viral RNA was extracted from purified virions by proteinase K digestion in the presence of SDS folMETHODS lowed by phenol-chloroform extraction and Viruses. The parental virus used, X- ethanol precipitation (Both and Air, 1979). 7(Fl), was constructed by Kilbourne et al. Sequencing was by the dideoxy method (1967) in the following way. A reassortant (Sanger et aL, 1977) using reverse tranvirus designated X-7, containing the HA of scriptase (Life Sciences Inc.) to synthesize A/NWS/33 (HlNl) and the NA of A/RI/ the cDNA (Air, 1979). Primers for this 5+/57 (H2N2), was recombined again with transcription were the synthetic oligonuRI/5+/57 to yield X-7(Fl). X-7(Fl) was cleotides made for sequencing the NA gene cloned twice at limiting dilution before se- of A/Tokyo/3/67 (Lentz et aL, 1984). lection of variants with monoclonal anti- The radiolabel was deoxyadenosine 5’-[LYbodies (Webster et aL, 1984). Viruses were ?S]thiotriphosphate, and buffer gradient grown in the allantoic cavity of 11-day-old sequencing gels were used (Biggin et d, embryonated chicken eggs and purified by 1983).
the HA of NWS/33 (Laver, 1978). The N2 structure as determined by Varghese et al. (1983) is a composite of the two antigenitally different NAs, and the structural differences between them will become apparent when the structures are refined to higher resolution. A major reason why influenza epidemics cannot be controlled is that antigenic drift and shift allow viruses to escape from antibody neutralization. To understand the mechanism of this escape,we characterized monoclonal antibodies of Tokyo/67 NA (Webster et aL, 1982) and determined the sequence changes in variants selected by these antibodies (Laver et aL, 1982; Lentz et ah, 1984). The results were surprisingly restricted in that antibodies made against Tokyo/67 NA selected changes at only a single amino acid, residue 344. Although two other antibodies which bound to Tokyo/67 NA selected changes at two other sites, the resulting antigenic map was sparse and we could not evaluate the significance of our results to the overall immunogenicity and antigenicity of the N2 NA. Using a much more extensive panel of monoclonal antibodies, a detailed operational antigenic map of the NA of X-7(Fl) was recently compiled (Webster et aL, 1984). We have now been able to correlate this map with the physical structure of the NA by sequence analysis of variants of the X-7(Fl) NA selected with monoclonal antibodies, and to a limited extent by examining the effect of chemical modification of specific amino acid side chains on antigenic activity.
INFLUENZA
N2 NEURAMINIDASE
The 3’-terminal nucleotides of the vRNA were determined by direct RNA sequencing (Peattie, 1979).
239
RNA were retested antigenically by ELISA, and two were found to be different from the original variant stocks. These are identified by inverted commas in Table 1 RESULTS and are not considered further. The antigenic analyses and sequence Sequence of the X-7(Fl) NA Gene changes of the variants are shown in Table The oligonucleotide primers are spaced 2. There is good correlation between the to give overlapping sequences through the antigenic analysis and sequence differences entire NA gene. The dideoxy procedure oc- and similarities. casionally gives ambiguous results at particular nucleotides. In most instances these Chemical Modi$caticm “cross-bands” can be eliminated by changAlthough it seems likely that those ing hybridization conditions (1 hr at 60” or amino acids which are altered in variants the formamide mix as described by Berton selected with monoclonal antibodies are et aL, 1984, instead of the usual 1 min at 100” then quick chilling), or by altering the part of the site at which antibody binds, ratio of template to primer. However, in there is no evidence yet that this is the case. sequencing the NA gene of X-7(Fl) we were We were interested to know if lysines, which are on the surface of the NA tetranot able to resolve persistent ambiguities around nucleotide 1050. We therefore mer, are reactive with reagents which cloned part of a double-stranded cDNA modify amino groups, and if such modificopy (from a BamHI site at 1033 to a cation would affect binding of antibodies. The results of reacting the NA of Tokyo/ Hind111 site at 1428) of the X-7(Fl) NA 67 with FDNB and with succinic anhydride gene and those of several variants into the single-stranded phage vector M13mp19 are shown in Table 3. The total reaction (Norrander et aL, 1983). Dideoxy sequenc- with FDNB was 45% substitution, and ing using the common Ml3 primer gave peptide analysis gave reactivities for most clear results. Nucleotides l-35 were con- lysine residues. With succinic anhydride firmed by direct RNA sequencing (Peattie, there was considerably greater reaction (70%), and the resulting partial peptide 1979). The sequence of the X-7(Fl) NA gene digestion products were not so easily inand the protein coded by it are shown in terpreted. However, lysine 368, which changes to glutamic acid in one variant of Fig. 1. Tokyo167 NA (Laver et aZ., 1982) and in several X-7(Fl) variants (Table 1) reacts Sequence Changes in Variants Sekcted with only 20% with FDNB but 100% with sucMono&ma1 Antibodies cinic anhydride. The NA gene of each of the variants selected with monoclonal variants was seDISCUSSION quenced from nucleotide 255 to the end, thus the entire sequence coding for the NA Sequence of the NA Gene of X-?‘(Fl) head was determined. Only one or two The early strains of Asian (H2N2) influnucleotide changes were found in each enza viruses, isolated in 1957, are antigenvariant. ically closely related, but were found to The nucleotide and deduced amino acid have varying reactivities to antibodies, sequence changes in 14 variants selected serum inhibitors, and mucoprotein recepwith monoclonal antibodies are shown in tors. It was postulated that this variation Table 1. When the results of the sequencing might be due to the presence of two kinds were compared with the antigenic prop- of virus particles in varying proportions, erties reported by Webster et al (1984), and this was demonstrated to be the case some discrepancies were found. The viral with several New York isolates (Choppin preparations grown for sequencing the and Tamm, 1959). Antibody- and inhibitor-
AIR
240
ET AL.
IO ~NP~QKIIlIGSUSLlInTUCFLNQInILnTlUl TcnnnnonTnnTfimnTTGotTcTGTcTc 40 50
10 50 LHFKQHECOSPASHQUIIP~E nTTfmTTTTFmcntmT~TGconcTcccccocotmfrcc~TnftToccnToToftnccnnT~TnnT 130 110 I50 160 170
I
PronoBe starrage 00
20
60
TcTc~cnTToc~~TnT~TTccTcnT~~TTo~nTccTGoc~cT~TGTGnc 70 00 90 loo
pi 100
60 IlERtlllElUYLtlWTllE tlotwmwllTRIKInononlFloloTnllToFln~notl I90 200 210 220
loo
90
KEICPEUUEYRliUSKPQCQIlGFnPFSKOtlSlllLSnGGLIl WIAAORWITTTOCCCCOARGl~TO~l~nO~TlOGTC~CCOC~lOlC~ll~WITlT~~TlllTCl~~lTC~TCCO~lllClOCl~T~T 310 250 260 270 280 290 300
300
390
320
330
450
100 loo I70 HnTLLtlNELGUPFHLGTKQUCUnUSSSSCHOOKnULHUCU TCAeCtiARCCETRTTAAl~l~TTO~TOTTC~lTTCnlll~~~lOTOTOTAtCRlCGTCCRCCTCRAtTTGTeRCWIT66RIWWlllOlGT 550 500 490 500 510 5m 530 540
200
210 o~c~u~nT~~s~sonnrtTni TOOGRWCflCAGTflOTRTotlCloGl~ToCtll~ 7n 710 750
760
2oG lEECSCYPnYPOURClCROtlUKGStlFlPUlOlltHELl~SlOS TRTRG#iWKiTOTTM1TOTT~ClC~lnl~T~~l~lOl~lCl~~C~Cl66 050 ooo 070 ooo
360
so
470
IW
190
570
500
590
6OG
23Q
710
720
200 27ll LFIKEGK~~H~~PL~G~RQH CCOFITflCTAUV)TACTRTTCnTT AARGA66tiGRIWYITTGTCCnl~ll~C~llOl~l~TC~n 790 ooo 010 020 030 770 iao
MO
650
2M
210
350
A#ITnTCcTCAGWICCc#l~T~lTloTnlc~ ooo 090 730
660
670
230
70
310
220
210
TGOORllRTnSFlYOOALUOSlGSUSQlllLnlQESECUClH CAeT6666ATllR6AAATGT~T~lTCArTTT~l~O~llOT~~l~l~ 610 620 630 640
120
l5G#
im
400
110
110
Ii0
im I30 UUTREPYUSCOPGKCYQF~LGQO~~LOHKHS~~G~~~~OFI~P 3M
30
xl0
290
090
900
310
~OTT~T~l~Tnl~TTnl~nTl~TlC 910 9m
# 3m #33Ll GYUCSOLUOOTPRNOOSSSHSnCRDPnnERGWPGUtGUAF CGGCTATGTOTOCTCAGGGCTlOTl~~nCCC~~~~~~CTCl~C~T~OCRATTGW66611TCClAAe#IT IO30 PM 900 990 IWO 1010 1020
310
360 # 370# DHGOOUUllGATlllKESRSGYETFKUlOOUSlPHSKSOUHn TOACRAT66AUITGlllOTnT~T~~~lC~~T~OCl~GllnTOR(WICTTTCAAlffTCnTT66 1090 1100 1110 iim 1IM IIt0
300
II50
940
950
900
OWGWOMRTCCtlOOITOT-loooCCll lO5ll 1060 1070
IOOG
#
IMO
II66
930
3M
390 ~TIUNTTCCF~~~~TC~CW~TCFN~TR? 1170 1100 1190
# m 110 120 QUIUDt4lWSGYSGlFSUEOKSCl~RCF~UELlRo+~PQET ncRGOTCR7RGTT~AACARTllATTWiTETBOTTACTCTOGlnlllTCTClOlT~C#VYIGCTOCRlMW)TAGGTU1TTr~lOl~ll~l~~~ l2m 1230 I240 I210 I250 1260 i2m 1266
12al
1290
I300
1310
13m
I*10
I420
If30
Iqlo
460 1fO 450 nuuuTSWSlUUFCGTSGTYOlosupGGn~~nFnpl# T~TnTWiT~AeeTCC~TnlTGTlGlOllTlGlG~llC~T~TTnl~AReROGeTCnT66CCTGRlW~nT~~TTTcnT~TnTnT~TTTc~~T 13M
1340
TTTAG~TCCTTGlllCl~l 1450
1160
I350
1360
1370
l3OG
1390
IiMl
INFLUENZA
N2 NEURAMINIDASE
241
seen in the receptor-binding properties of the HA of the two substrains (Choppin and NUCLEO~DE AND AMINO ACID SEQUENCE CHANGES IN Tamm, 1964; Carroll et aL, 1981). The NAs MONOCLONAL VARIANTS OF X-?(F~)NEURAMINIDASE appear identical by antigenic analysis (Webster et ah, 1982). A reassortant virus M0nO&n.41 Amino acid Ab and was constructed which contained the HA Change in codon change variant of A/NWS/33 and the NA of A/RI/5+/57, and was designated X-7. When X-7 was 81/4 Vl 1226 TGG to CGG 403 Trp to Arg back crossed with RI/5+/57, a reassortant designated X-7(Fl) was isolated (Kilbourne 514/l Vl 1123 TCA to TTA 370 Ser to Leu et aL, 1967). X-7(Fl) has twice as many m220/2 Vl 1129 TCA to TTA 370 Ser to Leu neuraminidase molecules as X-7 (Webster 664/l Vl 1050 AGA to AAA 344 Arg to Lys et aL, 1968), and has therefore been used extensively for studies on the NA pro561/3 Vl 344 Arg to Lys 1050 AGA to AAA tein, including crystal structure analysis 1049 AGA to GGA 344 Arg to Gly 415/l Vl (Varghese et al, 1983). 112/2 Vl 1020 AAT to AGT 334 Asn to Ser Elleman et al. (1982) reported the se1121 AAA to GAA 368 Lys to Glu quence of a cloned cDNA copy of the NA gene of RI/5-/57. We have now determined “438/l Vl” 1020 AAT to AGT 334 Asn to Ser 1121 AAA to GAA 368 Lys to Glu the sequence of the NA gene of X-7(Fl) by dideoxy sequencing directly from the m474/1 Vl 1020 AAT to AGT 334 Asn to Ser vRNA template. In view of the passage and 368 Lys to Glu 1121 AAA to GAA reassorting history of this virus, we were m490/3 Vl 334 Asn to Ser 1020 AAT to AGT surprised to find only four nucleotide dif368 Lys to Glu 1121 AAA to GAA ferences between the NA genes of X-7(Fl) m193/2 Vl 1004 GAC to AAC 329 Asp to Asn and RI/5-/57. These are listed in the legend to Fig. 1. Two of them result in amino acid 467 CAT to AAT 150 His to Asn 145/l Vl differences, at positions 7 (Thr in RI/5-; 150 His to Gin 489/2 Vl 469 CAT to CAG Ile in X-7(Fl)) and 315 (Ser in RI/5-, Gly “509/1 Vl” No change in X-7(Fl)). Amino acid 7 is apparently the first amino acid of the trans-membrane Note m designates variants selected in MDCK cells rather anchor sequence (Blok and Air, 1982), and than in eggs. Antibody and variant number in inverted the sequence difference here is unlikely to commas have different antigenie properties when grown for have any biological significance. Since no sequence analysis compared with original properties reported differences in properties of the NAs of Xby Webster et aL (1994). 7(Fl) and RI/5- have been reported, the change at 315 is probably also biologically sensitive substrains were designated “+” silent. Our surprise at the similarity of the and insensitive substrains “-.” Two such two sequences is because of the many instances of differences in sequence when difsubstrains were A/RI/5+/57 and A/RI/5/ 57. Once separated, they were found to be ferent laboratory stocks of the “same” vistable on passage, retaining the + or - rus have been examined. The HA genes of character (Choppin and Tamm, 1960), and two stocks of B/Hong Kong/8173 showed were antigenically distinguishable (Chop- 16 differences (Krystal et a& 1983, Hovanec pin and Tamm, 1964). Differences could be and Air, 1984). The HAS of A/PR/8/34 TABLE 1
FIG. 1. Nucleotide and predicted amino acid sequence of the NA gene of influenza virus X-7(Fl). The primers used for sequencing by the dideoxy method are underlined. The site of pronase cleavage to generate NA heads (Laver et &, 1982) is indicated with an arrow, and residues which change in variants selected by monoclonal antibodies are shown (#). Nucleotide differences to the sequence of A/RI/5-/57 (Elleman et al, 1982) are at positions 39 (C in RI/5-), 962 (A), 1430 and 1431 (CG).
81/4
370 s to L
+
+
0
(+)
370 s to L
+
+
0
(+)
0
+
0
t
+
0
0
+
Vl m220/2
344 R to K
+
t
0
0
0
t
0
0
0
0
+
+
Vl 664/l
344 R to K
+
0 +
0
0
0 t
0
0
+ +
t
Vl 561/3
344 R to G
+ +
t
0
0
t" +
t
0 +
0
+
Vl 415/l
Variants/antibody
K to E
344, 368 N to S,
2B
(+) t t
334, 368 N to S, K to E
+
+ 0 0 0 + 0 0 0 0 0 + t
+ 0 0 0 t 0 0 0 0
N to S, K to E
344,368
329 D to N
+
A-
+
0 +
0 0
0 +
+
t +
+
+
Vl m193/2
0
0
0
0
0
t
0
0
0
Vl m490/3
Vl m474/1
Vl 112/2
used for selection
150 H to N
+ + t + t t + t + t 0 0
Vl 145/l
H to Q
150
2D
0
0
t
+ + t + t
Vl 480/2
- -.
- - -- - - ---_ --
-- --.
--
__~_
_.
_ -_
Note. +, binding of monoclonal antibodies to variant viruses in ELISA; OD* readings of approximately 0.450 as compared with control binding to BSAcoated wells. 0, antibody binding decreased to less than 50% of wild-type binding. (t), antibody binding decreased by 25-50s of wild-type binding. “In neuraminidase inhibition tests, monoclonal antibody 415/l inhibited the parental virus but failed to inhibit the enzyme activity of the variant (Webster et al, 1984); nevertheless, this antibody still bound to the variant as detected by ELISA.
403 W to R
2A
Site:
Change:
+
+ +
48012
145/i
+
49013 193/2
0
+
+
+
474/l
0
+
+
(+)
0
+
Vl 514/l
+ +
+ +
0 +
415/l 112/2
561/3
664/l
514/l 220/2
Vl 81/4
2
ANTIBODY BINDING BY VARIANTS OF X-‘I(Fl)NA
TABLE
_ -_.
9 z z F
INFLUENZA
N2 NEURAMINIDASE
TABLE 3 REACTIVITY OF LYSINES IN TOKYO/~~ NA
Percentage substitution with: Lysine residue number
FDNB
75 80 89 102 128 143 187 261 296 350 368 378 389 415 431 Overall (%):
100 196 50 0 196 166 ins01 30 -100 0 20 50 0 0 70 45
Succinic anhydride 100 166 ? 60? 166 100 ins01 100 160 0 100
100 30? 30? 106 70
Not-e.Treatment of the NA with succinic anhydride abolishes inhibition of enzyme activity by monoclonal antibodies in groups 2A and 2B, for details see Webster et al (1984).
from different stocks (designated “Cambridge” and “Mount Sinai”) were known to be antigenicaliy distinguishable, and 12 nucleotide sequence differences were found (Winter et aL, 1981; Caton et aL, 1982). Similarly, the sequence of the HA of A/ duck/Ukraine/63 was not the same from different laboratories (Fang et al, 1981; Ward and Dopheide, 1982). We cannot explain why RI/5+/57 and RI&/57 have been so stable on laboratory passage. When X-7(Fl) was cloned twice at limiting dilution for selection of the variants (Table l), a mutant which was missing nearly half of the stalk sequence was cloned out, and all the variants selected with monoclonal antibodies have retained this deletion, which has no effect on antigenic properties (Els et aft, 1985). We also sequenced the entire NA gene of the original reassortant, X-7, including the 5’-noncoding sequence which was determined by direct chemical RNA sequencing of the 3’ end of vRNA. The only clear difference was at nucleotide 17, where
243
A in the cDNA strand of X-7(Fl) is G in X-7. The initiating ATG codon starts at nucleotide 20. Antigenic Structure of the NZ Neuraminidams Figure 2 shows the amino acid substitutions found in variants of X-7(Fl) selected with monoclonal antibodies, compared with those we previously reported for A/Tokyo/3/67 NA variants (Laver et ak, 1982, Lentz et al, 1984). Although none of the monoclonal antibodies against the NA of one strain bound the other (Webster et al, 1982, 1984), the same substitutions at positions 344 and 368 are found in variants of both Tokyo/67 and X-7(Fl) selected with strain-specific monoclonal antibodies. In the X-7(Fl) variants the change of Lys to Glu at residue 368 was accompanied by a change at 334 in three independently isolated variants. The Tokyo variant “Pot” was determined to have a change at 368 by analysis of tryptic peptides (Laver et al., 1982). In case a change at 334 was missed in the peptide analysis, we have now sequenced the NA gene of “Pot” and confirmed that there is no change at 334. Table 3 shows that Lys 368 does not react with FDNB, which is difficult to explain if it is a surface side chain which it must be if it is accessible to antibody. Lysine 368 does react 100% with succinic anhydride, but this has little effect on binding anti-NA antibodies in rabbit antiserum (Laver et al, 1983), although monoclonal antibodies of groups 2B and 2A did not bind to succinylated NA. Multiple differences between the Tokyo/ 67 and X-7(Fl) NA sequences (Fig. 2) around several of the amino acid positions which change in monoclonal variants (150, 344, 368,403) presumably account for the lack of binding of Tokyo antibodies to X7(Fl) and vice versa. The antiJap antibody 113/2, which selected a variant changed at residue 253 in Tokyo/67 NA (Lentz et al, 1984), also binds to X-7(Fl) (Webster et al, 1982), so the substitution of Lys for Glu at 258 (Fig. 2) does not effect the binding of this antibody. On the other hand, antibody Texas 18/l, which selected a variant
244
AIR ET AL.
INFLUENZA
N2 NEURAMINIDASE
changed at residue 221, does not hind to X‘7(Fl), although the sequences are identical for some distance around 221. Presumably the tertiary structure contains changes which alter antibody binding. Carbohydrate attachment sites are conserved between X-7(Fl) and Tokyo/67 NAs. However, extra potential glycosylation sites (Asn-X-Ser/Thr) are created in some variants. In X-7(Fl) this occurred at 329 (Asp to Asn), and in Tokyo/67 at 344 (Arg to Thr and to Ser). These latter variants were both selected with the same antibody (Lent.2 et al, 1984), and it is not known whether addition of the extra carbohydrate (if it occurs) plays a role in the abolition of antibody binding. The antigenic map shows four overlapping antigenic sites (Webster,et al, 1984). Antibodies binding to sites 1 and 4 did not inhibit enzyme activity, and so could not be used to select variants. Although site 3 antibodies inhibited NA activity, no variants could be isolated. Therefore we have analyzed only variants selected by antibodies recognizing site 2, which can be subdivided into overlapping regions 2A, 2B, and 2D. No variants were selected with monoclonal antibodies of group 2C. Our results show that site 2D is well-separated from the others, but the position in the three-dimensional structure of the sequence change in the one variant selected with a site 2A antibody is so close to those selected by site 2B antibodies that it is not clear why site 2A is antigenically so distinct from site 2B (Table 2). Figure 3 shows the location in the threedimensional structure (Varghese et al, 1983) of the changes in the NA of variants of X-7(Fl) and Tokyo/67 selected with monoclonal antibodies. All the antibodies used to select variants were those which inhibited enzyme activity (Webster et aL, 1982, 1984), so it is not surprising that nearly all of the changes are located on the upper surface of the molecule, surrounding
245
the active site pocket (Colman et u& 1983). In general, there is a good correlation between the proximity of the change to the active site pocket and whether the antibody selecting that variant inhibits enzyme activity toward a small substrate, N-acetylneuraminyl lactose, which fits almost completely into the active site pocket, as well as toward the large substrate, fetuin (Webster et aL, 1984). The sequence changes correlate well with the antibody-binding ELISA assays (Webster et aL, 1984 and Table 2). Variants with identical sequence changes have the same reactivities (within experimental error, Table 2), and variants with a different change at the same site, or a change close by, have slightly different reactivities. The competition assays reported by Webster et al. (1984) are also in general agreement with the sequence results. Antibodies which select variants at the same or nearby sites compete with each other, but not with antibodies recognizing changes at widely different places in the three-dimensional structure. The one exception to perfect correlation between sequence change and biological properties in the NA inhibition assays, in the competition assays, and to a lesser extent in the ELISA assays, is the antibody 112/2. Although a variant selected with this antibody had the same changes at the same residues (334 and 368) as variants selected with antibodies m474/1 and m490/ 3,112/2 is antigenically rather distinct, and may be binding at the extremity of a site which includes residues 334 and/or 368. 112/2 inhibits NA activity on N-acetylneuraminyl lactose, which would suggest that it binds at 368 rather than 334 (Fig. 3). However, it is not completely competed by antibodies 561/3 and 22012 which recognize nearby changes at 344 and 370. A set of results which is difficult to explain is the HA inhibition shown on binding certain of the X-7(Fl) monoclonal an-
FIG. 2. Amino acid sequences of NA from X-7(Fl) and A/Tokyo/3167 (Lentz et aL, 1934), and changes in monoclonal variants selected from X-7(Fl) (above) and Tokyo/67 (below). Residues which differ between the two strains are boxed, and + indicates the two differences in RI/5- (residue 7, Thr and residue 315, Ser; Elleman et o& 1982) compared to X-7(Fl) (RI/5+).
246
AIR
ET
FIG. 3. Schematic and model of the N2 NA three-dimensional structure (Varghese et aL, 1983; Colman et al, 1983) showing the sites of change in monoclonal variants of Tokyo/67 and X-7(Fl).
tibodies. Webster et al (1934) hypothesized that those antibodies which cause HA inhibition might bind at the perimeter of the NA tetramer. Since the NA tetramer has circular fourfold symmetry (Varghese et c& 1983), all four subunits have the same residues at the perimeter. Those antibodies which select variants which are changed at peripheral residues, e.g., 370, sometimes inhibit hemagglutination and sometimes do not. Similarly, of the two antibodies which select variants changed at residue 150, far from the perimeter, one inhibits HA strongly while the other inhibits weakly. The results of sequence analysis of variants of N2 NA molecules selected with monoclonal antibodies correlate well
enough with the results of antigenic analysis reported by Webster et al. (1982,1934) to suggest that the operational map translates into a physical map of antibodybinding sites. However, antibody 112/2 clearly binds differently to the NA than do antibodies m474/1 and m490/3 (Table 2 and Webster et aL, 19&I), although all three select the same variant with changes at both 334 and 368. Antibodies SlO/l (Tokyo) and 514/l (X-7(Fl)) select variants at 368 and 370, respectively. Both can recognize a change in an adjacent loop of the structure of Arg 344 to Gly, but cannot distinguish the more conservative change of Arg 344 to Lys. The effects of different amino acid substitutions on binding of monoclonal anti-
INFLUENZA
N2 NEURAMINIDASE
bodies to the N2 NAs and variants may be more fully explained when structures of NA-antibody complexes become available (Colman et aL, 1981). ACKNOWLEDGMENTS
247
of influenza virus which differ in reactivity with receptors and antibodies. CIBA Foundation Symposium “Cellular Biology of Myxovirus Inhibitions” (G. E. W. Wolstenholme, ed.), pp. 218235. Churchill, London. COLMAN, P. M., GOUGH, K. H., LILL.EY, G. G., BLA-
GROVE,R. J., WEBSTER,R. G., and LAVER, W. G. This work was supported in part by Grants AI 19084, (1981). Crystalline monoclonal Fab fragment with AI 08831, and CA 13148from the National Institutes specificity towards an influenza virus neuraminiof Health. We thank Rodney Tucker and Lynn Ritchie dase. J. Mol. Bid 152.609-614. for excellent technical assistance. This collaborative COLMAN,P. M., VARGHESE,J. N., and LAVER, W. G. project was greatly assisted by international direct(1983). Structure of the catalytic and antigenic sites dialing facilities provided by the Australian Overseas in influenza virus neuraminidase. Nature (London) Telecommunications Commission. M.E. is an Emma 303,41-44. Smith Overseas Scholar. ELLEMAN,T. C., AZAD, A. A., and WARD, C. W. (1982). Neuraminidase gene from the early Asian strain of REFERENCES human influenza virus A/RI/5-/57 (H2N2). Nucleic AIR, G. M. (1979). Nucleotide sequence coding for the Acids Rea. 10,7005-7015. “signal peptide” and N-terminus of the hemagglu- ELS, M. C., AIR, G. M., MURTI, K. G., WEBSTER,R. G., tinin from an Asian (H2N2) strain of influenzavirus. and LAVER, W. G. (1985). An 18 amino acid deletion viTOlk$l~ 97,468-472. in an influenza neuraminidase. %rology 142, 241BERTON,M. T., NAEVE, C. W., and WEBSTER,R. G. 247. (1984). Antigen+ structure of the influenza B hem- FANG, R., MIN Jou, W., HUYLEBROEK,D., DEVOS,R., agglutinin: Nucleotide sequence analysis of antiand FIERS,W. (1981).Complete structure of A/duck/ genie variants selected with monoclonal antibodies. Ukraine163 influenza hemagglutinin gene: Animal J. Viral. 52,919-927. virus as progenitor of human H3 Hong Kong 1968 BIGGIN, M. D., GIBSON, T. J., and HONG, G. F. (1983). influenza hemagglutinin. Cell 25.315-323. Buffer gradient gels and %Slabel as an aid to rapid HOVANEC,D. L., and AIR, G. M. (1984). Antigenic DNA sequencing. Proc NatL Acad Sci. USA 80, structure of the hemagglutinin of influenza B/Hong 3963-3965. Kong18173 as determined from gene sequence BLOK, J., and AIR, G. M. (1982).Variation in membrane analysis of variants selected with monoclonal aninsertion and ‘stalk’ sequences in eight subtypes of tibodies. Virology 139,384-392. influenza type A virus neuraminidase. Biochemistry KIDA, H., BROWN,L. E., and WEBSTER,R. G. (1982). 21,4001-4007. Biological activity of monoclonal antibodies to opBOTH, G. W., and AIR, G. M. (1979). Nucleotide seerationally defined antigenic regions on the hemquence coding for the N-terminal region of the maagglutinin molecule of A/Seal/Massachusetts/l/80 trix protein of influenza virus. Eur. J. Rio&em 96, (H7N7) influenza virus. %oZogy 122,38-47. 363-373. KILBOURNE,E. D., LAVER, W. G., SCHULMAN,J. L., and CARROLL,S. M., HIGA, H. H., CAHAN,L. D., end PAULWEBSTER,R. G. (1968). Antiviral activity of antiSON,J. C. (1981). Different sialyloligosaccharide reserum specific for an influenza virus neuraminidase. ceptor determinants of antigenically related influJ. Viral 2,281-288. enza virus hemagglutinins. In “Genetic Variation KILBOURNE,E. D., LIEF, F. S., SCHULMAN,J. L., JAHIEL, among Influenza Viruses” (D. P. Nayak and C. F. R. I., and LAVER, W. G. (1967). Antigenic hybrids of influenza viruses and their implications. Perspect Fox, eds.), pp. 415-421. Academic Press, New York. CATON,A. J., BROWNLEE,G. G., YEWDELL,J. W., and ViroL 5,87-106. GERHARD,W. (1982). The antigenic structure of the KOHLER, G., and MILSTEIN, C. (1976). Derivation of influenza virus A/PR/8/34 hemagglutinin (Hl) specific antibody-producing tissue culture and tusubtype. Cell 31,417-427. mor lines by cell fusion. EUT. J. ImmunoL 6, 511CHOPPIN,P. W., and TAYM, I. (1959). Two kinds of 519. particles with contrasting properties in influenza KRYSTAL, M., YOTJNG,J. F., PALESE, P., WILSON, I. A., SKEHEL,J. J., and WILEY, D. C. (1983). Sequential A virus strains from the 1957 pandemic. Virology mutations in hemagglutinins of influenza B isolates: 8.539-542. Definition of antigenic domains. Proc. Nat1 Acad CHOPPIN, P. W., and TAMM, I. (1960). Studies of two sci. USA 80,4527-4531; correction: Proc Nat1 Aad kinds of virus particles which comprise influenza Sci USA 81,126l (1984). A2 virus strains. 1. Characterization of stable homogeneous substrains in reactions with specific an- LAVER, W. G. (1969). Purification of influenza virus. In “Fundamental Techniques in Virology” (K. Habel tibody, mucoprotein inhibitors, and erythrocytes. and N. P. Salzman, eds.), pp. 82-86. Academic Press, J. Eq. Med 112.895-920. New York. CHOPPIN,P. W., and TAMM, I. (1964). Genetic variants
248
AIR ET AL.
LAVER,W. G. (1978). Crystallization and peptide maps of neuraminidase “heads” from H2N2 and H3N2 influenza virus strains. Virology 86, 78-87. LAVER, W. G., AIR, G. M., and WEBSTER,R. G. (1981). Antigenicity of influenza virus hemagglutinin following chemical modification. Virology 111,538-548. LAVER, W. G., AIR, G. M., WEBSTER,R. G., and MARKOFF,L. J. (1982). Amino acid sequence changes in antigenic variants of type A influenza virus N2 neuraminidase. Virology 122,450-460. LAVER, W. G., WEBSTER,R. G., and AIR, G. M. (1983). Effect of chemical modification on the antigenic properties of A/Tokyo/67 neuraminidase. Effect of a single sequence change on the thermal stability of A/Tokyo/67 neuraminidase. 1% “The Origin of Pandemic Influenza Viruses” (W. G. Laver, ed.), pp. 113-120. Elsevier, Amsterdam/New York. LENTZ,M. R., AIR, G. M., LAVER, W. G., and WEBSTER, R. G. (1984). Sequence of the neuraminidase genes of A/Tokyo/67 influenza virus and of variants selected with monoclonal antibodies. Vim&y 135, 257-265. NORRANDER,J., KEMPE, T., and MESSING,J. (1983). Construction of improved Ml3 vectors using oligonucleotide-directed mutagenesis. Gene 26, 101106. PEATTIE,D. A. (1979). Direct chemical method for sequencing RNA. Proc Nat1 Acad Sti USA 76,17601764.
SANGER,F., NICKLEN, S. and COULSON,A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. NatL Acad Sci USA 74.5463-5467.
VARGHESE,J. N., LAVER, W. G., and COLMAN,P. M. (1983). Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9A resolution. Nature (Londm) 303,35-40. WARD, C. W., and DOPHEIDE,T. A. (1982). The Hong Kong hemagglutinin. Structural relationships between the human (H3) hemagglutinin and the hemagglutinin from the putative progenitor strain A/ duck/Ukraine/l/63 (HAv7). In “Genetic Variation among Influenza Viruses” (D. P. Nayak and C. F. Fox, eds.), pp. 323-340. Academic Press, New York. 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. WEBSTER,R. G., HINSHAW, V. S., and LAVER, W. G. (1982). Selection and analysis of antigenic variants of the neuraminidase of N2 influenza viruses with monoclonai antibodies. Virology 117,93-104. WEBSTER,R. G., LAVER, W. G., and KILBOURNE,E. D. (1968).Reactions of antibodies with surface antigens of influenza virus. J. Gen ViroL 3,315-326. WINTER, G., FIELDS, S., and BROWNLEE,G. G. (1981). Nucieotide sequence of the hemagglutinin gene of a human influenza virus Hl subtype. Nature (LUTZdon) 292,72-75.
145,237-248
(1985)
Location of Antigenic Sites on the Three-Dimensional Structure of the influenza N2 Virus Neuraminidase G. M. AIR,*” M. C. ELS,* L. E. BROWN,? W. G. LAVER,*
AND
R. G. WEBSTERO
*Department of Microbiology, University of A&ma at Birmingham, Birmin&m, Alabama 35294 TMicrobiology Department, University of Melbmwne, Parkville, Victoria 5052,Australia; $John Curtin School of Medical Research, Canberra City, 2601,Australia; and Wt. Jude Childrens Research Hospital, Memphis, Tennessee, 3%101 Received March 20,19%5;accepted June ~$1985 Sequence analysis of the neuraminidase (NA) genes of influenza virus X-i’(F1) and of 12 variants selected with monoclonal antibodies has been used to define in physical terms the antigenic structure of this NA, which was operationally established by R. G. Webster, L. E. Brown, and W. G. Laver (1934, Virologg 135,30-42). X-‘7(Fl) is a reassortant virus containing the NA of the early Asian (H2N2) isolate A/RI/5+/57, and the results of antigenic and sequence analysis of X-7(Fl) and of variants selected with monoclonal antibodies have been combined with a similar analysis of the A/Tokyo/3167 NA (H2N2, M. R. Lentz, G. M. Air, W. G. Laver, and R. G. Webster (1984), Virology 135,257-265) to obtain a model of antibody binding to N2 NAs. The selection process was biased, however, since only those monoclonal antibodies which inhibited NA activity could be used to select variants. Most of the changes in the variants selected with monoclonal antibodies occur in those parts of the polypeptide chain which encircle the enzyme active site pocket in the threedimensional structure (P. M. Colman, J. N. Varghese, and W. G. Laver (1983), Nature (London) 303,41-44). The results suggest that in general the antibody binds to a site on the NA which includes those amino acid side chains which are altered in monoclonal variants. There are, however, several aspects of the antigen-antibody interaction which are not easily explained, and which will probably only be fully elucidated by X-ray cryso 1985 Academic press. ITIC. tallographic analysis of NA-antibody complexes. INTRODUCTION
The neuraminidase (NA) is one of the two projecting surface glycoproteins of influenza A and B viruses, the other being the hemagglutinin (HA). Although the precise role of the NA in viral infection is not known, it appears to be important in facilitating movement of the virus through mucin to host cells, and assists in release of progeny viruses by cleaving sialic acid from cell receptors and from the HA. Antibody to the NA does not neutralize virus infectivity except at high concentrations, but it does slow the release of virus from infected cells, giving reduced plaque sizes (Kilbourne et al, 1968). The NA of type A influenza viruses uni Author addressed.
to whom requests for reprints
should be
dergoes antigenic shift and drift, as does the HA. In 1957 the HlNl viruses which had been circulating in the human population since 1933 when the virus was first isolated were replaced by viruses of the Asian subtype, H2N2. Between 1957 and 1968 the H2 HA and N2 HA underwent antigenic drift, before the HA was replaced in 1968 in Hong Kong (H3N2) influenza. The N2 NA continued to drift along with the Hong Kong HA, and H3N2 variants are still circulating. The three-dimensional structure of the N2 NA has been solved by X-ray crystallography to 2.9A (Varghese et a& 1983).The structure determination was made using crystals of two different N2 NAs, representing early (A/RI/5+/57) and late (A/ Tokyo/3/67) Asian strains. The crystals were grown from NA released by pronase from reassortant viruses which contained
237
0042-6822185$3.00 Copyright All rights
0 1985 by Academic Press, Inc. of reproduction in any form reserved.
238
AIR
ET AL.
adsorption to and elution from chicken erythrocytes followed by density gradient centrifugation through lo-40% sucrose (Laver, 1969). Monoclonal antibodies and selection of variants. Hybridoma cell lines producing antibodies to the neuraminidase of X-7(Fl) were prepared by the method of Kiihler and Milstein (1976). Their preparation, properties, and the antigenic properties of variants selected with these antibodies in eggs or in MDCK cells have been described previously (Webster et aL, 1984). Serological assays. The reactivities of monoclonal antibodies with variant viruses were measured in an enzyme-linked immunosorbent assay (ELISA) as previously described (Kida et aL, 1982; Webster et ak, 1984). Chemical modi&cation of the neuraminidase. Pronase-released NA heads of Tokyo/67 and X-7(Fl) were reacted with 1-fluoro-2,4-dinitrobenzene (FDNB) or succinic anhydride at pH 8.5 as previously described for modification of the HA (Laver et CAL, 1981). In order to determine the overall degree of succinylation, the succinyl protein was dissolved in 6 M guanidine hydrochloride, reacted with FDNB, and hydrolyzed with 6 N HCl. The free lysine recovered was a measure of the degree of succinylation. Reactivity of individual lysine residues was measured by analysis of tryptic peptides separated by paper electrophoresis and chromatography (Laver, 1978). Sequencing of neuraminidase genes of X7(Fl) and its variants. Viral RNA was extracted from purified virions by proteinase K digestion in the presence of SDS folMETHODS lowed by phenol-chloroform extraction and Viruses. The parental virus used, X- ethanol precipitation (Both and Air, 1979). 7(Fl), was constructed by Kilbourne et al. Sequencing was by the dideoxy method (1967) in the following way. A reassortant (Sanger et aL, 1977) using reverse tranvirus designated X-7, containing the HA of scriptase (Life Sciences Inc.) to synthesize A/NWS/33 (HlNl) and the NA of A/RI/ the cDNA (Air, 1979). Primers for this 5+/57 (H2N2), was recombined again with transcription were the synthetic oligonuRI/5+/57 to yield X-7(Fl). X-7(Fl) was cleotides made for sequencing the NA gene cloned twice at limiting dilution before se- of A/Tokyo/3/67 (Lentz et aL, 1984). lection of variants with monoclonal anti- The radiolabel was deoxyadenosine 5’-[LYbodies (Webster et aL, 1984). Viruses were ?S]thiotriphosphate, and buffer gradient grown in the allantoic cavity of 11-day-old sequencing gels were used (Biggin et d, embryonated chicken eggs and purified by 1983).
the HA of NWS/33 (Laver, 1978). The N2 structure as determined by Varghese et al. (1983) is a composite of the two antigenitally different NAs, and the structural differences between them will become apparent when the structures are refined to higher resolution. A major reason why influenza epidemics cannot be controlled is that antigenic drift and shift allow viruses to escape from antibody neutralization. To understand the mechanism of this escape,we characterized monoclonal antibodies of Tokyo/67 NA (Webster et aL, 1982) and determined the sequence changes in variants selected by these antibodies (Laver et aL, 1982; Lentz et ah, 1984). The results were surprisingly restricted in that antibodies made against Tokyo/67 NA selected changes at only a single amino acid, residue 344. Although two other antibodies which bound to Tokyo/67 NA selected changes at two other sites, the resulting antigenic map was sparse and we could not evaluate the significance of our results to the overall immunogenicity and antigenicity of the N2 NA. Using a much more extensive panel of monoclonal antibodies, a detailed operational antigenic map of the NA of X-7(Fl) was recently compiled (Webster et aL, 1984). We have now been able to correlate this map with the physical structure of the NA by sequence analysis of variants of the X-7(Fl) NA selected with monoclonal antibodies, and to a limited extent by examining the effect of chemical modification of specific amino acid side chains on antigenic activity.
INFLUENZA
N2 NEURAMINIDASE
The 3’-terminal nucleotides of the vRNA were determined by direct RNA sequencing (Peattie, 1979).
239
RNA were retested antigenically by ELISA, and two were found to be different from the original variant stocks. These are identified by inverted commas in Table 1 RESULTS and are not considered further. The antigenic analyses and sequence Sequence of the X-7(Fl) NA Gene changes of the variants are shown in Table The oligonucleotide primers are spaced 2. There is good correlation between the to give overlapping sequences through the antigenic analysis and sequence differences entire NA gene. The dideoxy procedure oc- and similarities. casionally gives ambiguous results at particular nucleotides. In most instances these Chemical Modi$caticm “cross-bands” can be eliminated by changAlthough it seems likely that those ing hybridization conditions (1 hr at 60” or amino acids which are altered in variants the formamide mix as described by Berton selected with monoclonal antibodies are et aL, 1984, instead of the usual 1 min at 100” then quick chilling), or by altering the part of the site at which antibody binds, ratio of template to primer. However, in there is no evidence yet that this is the case. sequencing the NA gene of X-7(Fl) we were We were interested to know if lysines, which are on the surface of the NA tetranot able to resolve persistent ambiguities around nucleotide 1050. We therefore mer, are reactive with reagents which cloned part of a double-stranded cDNA modify amino groups, and if such modificopy (from a BamHI site at 1033 to a cation would affect binding of antibodies. The results of reacting the NA of Tokyo/ Hind111 site at 1428) of the X-7(Fl) NA 67 with FDNB and with succinic anhydride gene and those of several variants into the single-stranded phage vector M13mp19 are shown in Table 3. The total reaction (Norrander et aL, 1983). Dideoxy sequenc- with FDNB was 45% substitution, and ing using the common Ml3 primer gave peptide analysis gave reactivities for most clear results. Nucleotides l-35 were con- lysine residues. With succinic anhydride firmed by direct RNA sequencing (Peattie, there was considerably greater reaction (70%), and the resulting partial peptide 1979). The sequence of the X-7(Fl) NA gene digestion products were not so easily inand the protein coded by it are shown in terpreted. However, lysine 368, which changes to glutamic acid in one variant of Fig. 1. Tokyo167 NA (Laver et aZ., 1982) and in several X-7(Fl) variants (Table 1) reacts Sequence Changes in Variants Sekcted with only 20% with FDNB but 100% with sucMono&ma1 Antibodies cinic anhydride. The NA gene of each of the variants selected with monoclonal variants was seDISCUSSION quenced from nucleotide 255 to the end, thus the entire sequence coding for the NA Sequence of the NA Gene of X-?‘(Fl) head was determined. Only one or two The early strains of Asian (H2N2) influnucleotide changes were found in each enza viruses, isolated in 1957, are antigenvariant. ically closely related, but were found to The nucleotide and deduced amino acid have varying reactivities to antibodies, sequence changes in 14 variants selected serum inhibitors, and mucoprotein recepwith monoclonal antibodies are shown in tors. It was postulated that this variation Table 1. When the results of the sequencing might be due to the presence of two kinds were compared with the antigenic prop- of virus particles in varying proportions, erties reported by Webster et al (1984), and this was demonstrated to be the case some discrepancies were found. The viral with several New York isolates (Choppin preparations grown for sequencing the and Tamm, 1959). Antibody- and inhibitor-
AIR
240
ET AL.
IO ~NP~QKIIlIGSUSLlInTUCFLNQInILnTlUl TcnnnnonTnnTfimnTTGotTcTGTcTc 40 50
10 50 LHFKQHECOSPASHQUIIP~E nTTfmTTTTFmcntmT~TGconcTcccccocotmfrcc~TnftToccnToToftnccnnT~TnnT 130 110 I50 160 170
I
PronoBe starrage 00
20
60
TcTc~cnTToc~~TnT~TTccTcnT~~TTo~nTccTGoc~cT~TGTGnc 70 00 90 loo
pi 100
60 IlERtlllElUYLtlWTllE tlotwmwllTRIKInononlFloloTnllToFln~notl I90 200 210 220
loo
90
KEICPEUUEYRliUSKPQCQIlGFnPFSKOtlSlllLSnGGLIl WIAAORWITTTOCCCCOARGl~TO~l~nO~TlOGTC~CCOC~lOlC~ll~WITlT~~TlllTCl~~lTC~TCCO~lllClOCl~T~T 310 250 260 270 280 290 300
300
390
320
330
450
100 loo I70 HnTLLtlNELGUPFHLGTKQUCUnUSSSSCHOOKnULHUCU TCAeCtiARCCETRTTAAl~l~TTO~TOTTC~lTTCnlll~~~lOTOTOTAtCRlCGTCCRCCTCRAtTTGTeRCWIT66RIWWlllOlGT 550 500 490 500 510 5m 530 540
200
210 o~c~u~nT~~s~sonnrtTni TOOGRWCflCAGTflOTRTotlCloGl~ToCtll~ 7n 710 750
760
2oG lEECSCYPnYPOURClCROtlUKGStlFlPUlOlltHELl~SlOS TRTRG#iWKiTOTTM1TOTT~ClC~lnl~T~~l~lOl~lCl~~C~Cl66 050 ooo 070 ooo
360
so
470
IW
190
570
500
590
6OG
23Q
710
720
200 27ll LFIKEGK~~H~~PL~G~RQH CCOFITflCTAUV)TACTRTTCnTT AARGA66tiGRIWYITTGTCCnl~ll~C~llOl~l~TC~n 790 ooo 010 020 030 770 iao
MO
650
2M
210
350
A#ITnTCcTCAGWICCc#l~T~lTloTnlc~ ooo 090 730
660
670
230
70
310
220
210
TGOORllRTnSFlYOOALUOSlGSUSQlllLnlQESECUClH CAeT6666ATllR6AAATGT~T~lTCArTTT~l~O~llOT~~l~l~ 610 620 630 640
120
l5G#
im
400
110
110
Ii0
im I30 UUTREPYUSCOPGKCYQF~LGQO~~LOHKHS~~G~~~~OFI~P 3M
30
xl0
290
090
900
310
~OTT~T~l~Tnl~TTnl~nTl~TlC 910 9m
# 3m #33Ll GYUCSOLUOOTPRNOOSSSHSnCRDPnnERGWPGUtGUAF CGGCTATGTOTOCTCAGGGCTlOTl~~nCCC~~~~~~CTCl~C~T~OCRATTGW66611TCClAAe#IT IO30 PM 900 990 IWO 1010 1020
310
360 # 370# DHGOOUUllGATlllKESRSGYETFKUlOOUSlPHSKSOUHn TOACRAT66AUITGlllOTnT~T~~~lC~~T~OCl~GllnTOR(WICTTTCAAlffTCnTT66 1090 1100 1110 iim 1IM IIt0
300
II50
940
950
900
OWGWOMRTCCtlOOITOT-loooCCll lO5ll 1060 1070
IOOG
#
IMO
II66
930
3M
390 ~TIUNTTCCF~~~~TC~CW~TCFN~TR? 1170 1100 1190
# m 110 120 QUIUDt4lWSGYSGlFSUEOKSCl~RCF~UELlRo+~PQET ncRGOTCR7RGTT~AACARTllATTWiTETBOTTACTCTOGlnlllTCTClOlT~C#VYIGCTOCRlMW)TAGGTU1TTr~lOl~ll~l~~~ l2m 1230 I240 I210 I250 1260 i2m 1266
12al
1290
I300
1310
13m
I*10
I420
If30
Iqlo
460 1fO 450 nuuuTSWSlUUFCGTSGTYOlosupGGn~~nFnpl# T~TnTWiT~AeeTCC~TnlTGTlGlOllTlGlG~llC~T~TTnl~AReROGeTCnT66CCTGRlW~nT~~TTTcnT~TnTnT~TTTc~~T 13M
1340
TTTAG~TCCTTGlllCl~l 1450
1160
I350
1360
1370
l3OG
1390
IiMl
INFLUENZA
N2 NEURAMINIDASE
241
seen in the receptor-binding properties of the HA of the two substrains (Choppin and NUCLEO~DE AND AMINO ACID SEQUENCE CHANGES IN Tamm, 1964; Carroll et aL, 1981). The NAs MONOCLONAL VARIANTS OF X-?(F~)NEURAMINIDASE appear identical by antigenic analysis (Webster et ah, 1982). A reassortant virus M0nO&n.41 Amino acid Ab and was constructed which contained the HA Change in codon change variant of A/NWS/33 and the NA of A/RI/5+/57, and was designated X-7. When X-7 was 81/4 Vl 1226 TGG to CGG 403 Trp to Arg back crossed with RI/5+/57, a reassortant designated X-7(Fl) was isolated (Kilbourne 514/l Vl 1123 TCA to TTA 370 Ser to Leu et aL, 1967). X-7(Fl) has twice as many m220/2 Vl 1129 TCA to TTA 370 Ser to Leu neuraminidase molecules as X-7 (Webster 664/l Vl 1050 AGA to AAA 344 Arg to Lys et aL, 1968), and has therefore been used extensively for studies on the NA pro561/3 Vl 344 Arg to Lys 1050 AGA to AAA tein, including crystal structure analysis 1049 AGA to GGA 344 Arg to Gly 415/l Vl (Varghese et al, 1983). 112/2 Vl 1020 AAT to AGT 334 Asn to Ser Elleman et al. (1982) reported the se1121 AAA to GAA 368 Lys to Glu quence of a cloned cDNA copy of the NA gene of RI/5-/57. We have now determined “438/l Vl” 1020 AAT to AGT 334 Asn to Ser 1121 AAA to GAA 368 Lys to Glu the sequence of the NA gene of X-7(Fl) by dideoxy sequencing directly from the m474/1 Vl 1020 AAT to AGT 334 Asn to Ser vRNA template. In view of the passage and 368 Lys to Glu 1121 AAA to GAA reassorting history of this virus, we were m490/3 Vl 334 Asn to Ser 1020 AAT to AGT surprised to find only four nucleotide dif368 Lys to Glu 1121 AAA to GAA ferences between the NA genes of X-7(Fl) m193/2 Vl 1004 GAC to AAC 329 Asp to Asn and RI/5-/57. These are listed in the legend to Fig. 1. Two of them result in amino acid 467 CAT to AAT 150 His to Asn 145/l Vl differences, at positions 7 (Thr in RI/5-; 150 His to Gin 489/2 Vl 469 CAT to CAG Ile in X-7(Fl)) and 315 (Ser in RI/5-, Gly “509/1 Vl” No change in X-7(Fl)). Amino acid 7 is apparently the first amino acid of the trans-membrane Note m designates variants selected in MDCK cells rather anchor sequence (Blok and Air, 1982), and than in eggs. Antibody and variant number in inverted the sequence difference here is unlikely to commas have different antigenie properties when grown for have any biological significance. Since no sequence analysis compared with original properties reported differences in properties of the NAs of Xby Webster et aL (1994). 7(Fl) and RI/5- have been reported, the change at 315 is probably also biologically sensitive substrains were designated “+” silent. Our surprise at the similarity of the and insensitive substrains “-.” Two such two sequences is because of the many instances of differences in sequence when difsubstrains were A/RI/5+/57 and A/RI/5/ 57. Once separated, they were found to be ferent laboratory stocks of the “same” vistable on passage, retaining the + or - rus have been examined. The HA genes of character (Choppin and Tamm, 1960), and two stocks of B/Hong Kong/8173 showed were antigenically distinguishable (Chop- 16 differences (Krystal et a& 1983, Hovanec pin and Tamm, 1964). Differences could be and Air, 1984). The HAS of A/PR/8/34 TABLE 1
FIG. 1. Nucleotide and predicted amino acid sequence of the NA gene of influenza virus X-7(Fl). The primers used for sequencing by the dideoxy method are underlined. The site of pronase cleavage to generate NA heads (Laver et &, 1982) is indicated with an arrow, and residues which change in variants selected by monoclonal antibodies are shown (#). Nucleotide differences to the sequence of A/RI/5-/57 (Elleman et al, 1982) are at positions 39 (C in RI/5-), 962 (A), 1430 and 1431 (CG).
81/4
370 s to L
+
+
0
(+)
370 s to L
+
+
0
(+)
0
+
0
t
+
0
0
+
Vl m220/2
344 R to K
+
t
0
0
0
t
0
0
0
0
+
+
Vl 664/l
344 R to K
+
0 +
0
0
0 t
0
0
+ +
t
Vl 561/3
344 R to G
+ +
t
0
0
t" +
t
0 +
0
+
Vl 415/l
Variants/antibody
K to E
344, 368 N to S,
2B
(+) t t
334, 368 N to S, K to E
+
+ 0 0 0 + 0 0 0 0 0 + t
+ 0 0 0 t 0 0 0 0
N to S, K to E
344,368
329 D to N
+
A-
+
0 +
0 0
0 +
+
t +
+
+
Vl m193/2
0
0
0
0
0
t
0
0
0
Vl m490/3
Vl m474/1
Vl 112/2
used for selection
150 H to N
+ + t + t t + t + t 0 0
Vl 145/l
H to Q
150
2D
0
0
t
+ + t + t
Vl 480/2
- -.
- - -- - - ---_ --
-- --.
--
__~_
_.
_ -_
Note. +, binding of monoclonal antibodies to variant viruses in ELISA; OD* readings of approximately 0.450 as compared with control binding to BSAcoated wells. 0, antibody binding decreased to less than 50% of wild-type binding. (t), antibody binding decreased by 25-50s of wild-type binding. “In neuraminidase inhibition tests, monoclonal antibody 415/l inhibited the parental virus but failed to inhibit the enzyme activity of the variant (Webster et al, 1984); nevertheless, this antibody still bound to the variant as detected by ELISA.
403 W to R
2A
Site:
Change:
+
+ +
48012
145/i
+
49013 193/2
0
+
+
+
474/l
0
+
+
(+)
0
+
Vl 514/l
+ +
+ +
0 +
415/l 112/2
561/3
664/l
514/l 220/2
Vl 81/4
2
ANTIBODY BINDING BY VARIANTS OF X-‘I(Fl)NA
TABLE
_ -_.
9 z z F
INFLUENZA
N2 NEURAMINIDASE
TABLE 3 REACTIVITY OF LYSINES IN TOKYO/~~ NA
Percentage substitution with: Lysine residue number
FDNB
75 80 89 102 128 143 187 261 296 350 368 378 389 415 431 Overall (%):
100 196 50 0 196 166 ins01 30 -100 0 20 50 0 0 70 45
Succinic anhydride 100 166 ? 60? 166 100 ins01 100 160 0 100
100 30? 30? 106 70
Not-e.Treatment of the NA with succinic anhydride abolishes inhibition of enzyme activity by monoclonal antibodies in groups 2A and 2B, for details see Webster et al (1984).
from different stocks (designated “Cambridge” and “Mount Sinai”) were known to be antigenicaliy distinguishable, and 12 nucleotide sequence differences were found (Winter et aL, 1981; Caton et aL, 1982). Similarly, the sequence of the HA of A/ duck/Ukraine/63 was not the same from different laboratories (Fang et al, 1981; Ward and Dopheide, 1982). We cannot explain why RI/5+/57 and RI&/57 have been so stable on laboratory passage. When X-7(Fl) was cloned twice at limiting dilution for selection of the variants (Table l), a mutant which was missing nearly half of the stalk sequence was cloned out, and all the variants selected with monoclonal antibodies have retained this deletion, which has no effect on antigenic properties (Els et aft, 1985). We also sequenced the entire NA gene of the original reassortant, X-7, including the 5’-noncoding sequence which was determined by direct chemical RNA sequencing of the 3’ end of vRNA. The only clear difference was at nucleotide 17, where
243
A in the cDNA strand of X-7(Fl) is G in X-7. The initiating ATG codon starts at nucleotide 20. Antigenic Structure of the NZ Neuraminidams Figure 2 shows the amino acid substitutions found in variants of X-7(Fl) selected with monoclonal antibodies, compared with those we previously reported for A/Tokyo/3/67 NA variants (Laver et ak, 1982, Lentz et al, 1984). Although none of the monoclonal antibodies against the NA of one strain bound the other (Webster et al, 1982, 1984), the same substitutions at positions 344 and 368 are found in variants of both Tokyo/67 and X-7(Fl) selected with strain-specific monoclonal antibodies. In the X-7(Fl) variants the change of Lys to Glu at residue 368 was accompanied by a change at 334 in three independently isolated variants. The Tokyo variant “Pot” was determined to have a change at 368 by analysis of tryptic peptides (Laver et al., 1982). In case a change at 334 was missed in the peptide analysis, we have now sequenced the NA gene of “Pot” and confirmed that there is no change at 334. Table 3 shows that Lys 368 does not react with FDNB, which is difficult to explain if it is a surface side chain which it must be if it is accessible to antibody. Lysine 368 does react 100% with succinic anhydride, but this has little effect on binding anti-NA antibodies in rabbit antiserum (Laver et al, 1983), although monoclonal antibodies of groups 2B and 2A did not bind to succinylated NA. Multiple differences between the Tokyo/ 67 and X-7(Fl) NA sequences (Fig. 2) around several of the amino acid positions which change in monoclonal variants (150, 344, 368,403) presumably account for the lack of binding of Tokyo antibodies to X7(Fl) and vice versa. The antiJap antibody 113/2, which selected a variant changed at residue 253 in Tokyo/67 NA (Lentz et al, 1984), also binds to X-7(Fl) (Webster et al, 1982), so the substitution of Lys for Glu at 258 (Fig. 2) does not effect the binding of this antibody. On the other hand, antibody Texas 18/l, which selected a variant
244
AIR ET AL.
INFLUENZA
N2 NEURAMINIDASE
changed at residue 221, does not hind to X‘7(Fl), although the sequences are identical for some distance around 221. Presumably the tertiary structure contains changes which alter antibody binding. Carbohydrate attachment sites are conserved between X-7(Fl) and Tokyo/67 NAs. However, extra potential glycosylation sites (Asn-X-Ser/Thr) are created in some variants. In X-7(Fl) this occurred at 329 (Asp to Asn), and in Tokyo/67 at 344 (Arg to Thr and to Ser). These latter variants were both selected with the same antibody (Lent.2 et al, 1984), and it is not known whether addition of the extra carbohydrate (if it occurs) plays a role in the abolition of antibody binding. The antigenic map shows four overlapping antigenic sites (Webster,et al, 1984). Antibodies binding to sites 1 and 4 did not inhibit enzyme activity, and so could not be used to select variants. Although site 3 antibodies inhibited NA activity, no variants could be isolated. Therefore we have analyzed only variants selected by antibodies recognizing site 2, which can be subdivided into overlapping regions 2A, 2B, and 2D. No variants were selected with monoclonal antibodies of group 2C. Our results show that site 2D is well-separated from the others, but the position in the three-dimensional structure of the sequence change in the one variant selected with a site 2A antibody is so close to those selected by site 2B antibodies that it is not clear why site 2A is antigenically so distinct from site 2B (Table 2). Figure 3 shows the location in the threedimensional structure (Varghese et al, 1983) of the changes in the NA of variants of X-7(Fl) and Tokyo/67 selected with monoclonal antibodies. All the antibodies used to select variants were those which inhibited enzyme activity (Webster et aL, 1982, 1984), so it is not surprising that nearly all of the changes are located on the upper surface of the molecule, surrounding
245
the active site pocket (Colman et u& 1983). In general, there is a good correlation between the proximity of the change to the active site pocket and whether the antibody selecting that variant inhibits enzyme activity toward a small substrate, N-acetylneuraminyl lactose, which fits almost completely into the active site pocket, as well as toward the large substrate, fetuin (Webster et aL, 1984). The sequence changes correlate well with the antibody-binding ELISA assays (Webster et aL, 1984 and Table 2). Variants with identical sequence changes have the same reactivities (within experimental error, Table 2), and variants with a different change at the same site, or a change close by, have slightly different reactivities. The competition assays reported by Webster et al. (1984) are also in general agreement with the sequence results. Antibodies which select variants at the same or nearby sites compete with each other, but not with antibodies recognizing changes at widely different places in the three-dimensional structure. The one exception to perfect correlation between sequence change and biological properties in the NA inhibition assays, in the competition assays, and to a lesser extent in the ELISA assays, is the antibody 112/2. Although a variant selected with this antibody had the same changes at the same residues (334 and 368) as variants selected with antibodies m474/1 and m490/ 3,112/2 is antigenically rather distinct, and may be binding at the extremity of a site which includes residues 334 and/or 368. 112/2 inhibits NA activity on N-acetylneuraminyl lactose, which would suggest that it binds at 368 rather than 334 (Fig. 3). However, it is not completely competed by antibodies 561/3 and 22012 which recognize nearby changes at 344 and 370. A set of results which is difficult to explain is the HA inhibition shown on binding certain of the X-7(Fl) monoclonal an-
FIG. 2. Amino acid sequences of NA from X-7(Fl) and A/Tokyo/3167 (Lentz et aL, 1934), and changes in monoclonal variants selected from X-7(Fl) (above) and Tokyo/67 (below). Residues which differ between the two strains are boxed, and + indicates the two differences in RI/5- (residue 7, Thr and residue 315, Ser; Elleman et o& 1982) compared to X-7(Fl) (RI/5+).
246
AIR
ET
FIG. 3. Schematic and model of the N2 NA three-dimensional structure (Varghese et aL, 1983; Colman et al, 1983) showing the sites of change in monoclonal variants of Tokyo/67 and X-7(Fl).
tibodies. Webster et al (1934) hypothesized that those antibodies which cause HA inhibition might bind at the perimeter of the NA tetramer. Since the NA tetramer has circular fourfold symmetry (Varghese et c& 1983), all four subunits have the same residues at the perimeter. Those antibodies which select variants which are changed at peripheral residues, e.g., 370, sometimes inhibit hemagglutination and sometimes do not. Similarly, of the two antibodies which select variants changed at residue 150, far from the perimeter, one inhibits HA strongly while the other inhibits weakly. The results of sequence analysis of variants of N2 NA molecules selected with monoclonal antibodies correlate well
enough with the results of antigenic analysis reported by Webster et al. (1982,1934) to suggest that the operational map translates into a physical map of antibodybinding sites. However, antibody 112/2 clearly binds differently to the NA than do antibodies m474/1 and m490/3 (Table 2 and Webster et aL, 19&I), although all three select the same variant with changes at both 334 and 368. Antibodies SlO/l (Tokyo) and 514/l (X-7(Fl)) select variants at 368 and 370, respectively. Both can recognize a change in an adjacent loop of the structure of Arg 344 to Gly, but cannot distinguish the more conservative change of Arg 344 to Lys. The effects of different amino acid substitutions on binding of monoclonal anti-
INFLUENZA
N2 NEURAMINIDASE
bodies to the N2 NAs and variants may be more fully explained when structures of NA-antibody complexes become available (Colman et aL, 1981). ACKNOWLEDGMENTS
247
of influenza virus which differ in reactivity with receptors and antibodies. CIBA Foundation Symposium “Cellular Biology of Myxovirus Inhibitions” (G. E. W. Wolstenholme, ed.), pp. 218235. Churchill, London. COLMAN, P. M., GOUGH, K. H., LILL.EY, G. G., BLA-
GROVE,R. J., WEBSTER,R. G., and LAVER, W. G. This work was supported in part by Grants AI 19084, (1981). Crystalline monoclonal Fab fragment with AI 08831, and CA 13148from the National Institutes specificity towards an influenza virus neuraminiof Health. We thank Rodney Tucker and Lynn Ritchie dase. J. Mol. Bid 152.609-614. for excellent technical assistance. This collaborative COLMAN,P. M., VARGHESE,J. N., and LAVER, W. G. project was greatly assisted by international direct(1983). Structure of the catalytic and antigenic sites dialing facilities provided by the Australian Overseas in influenza virus neuraminidase. Nature (London) Telecommunications Commission. M.E. is an Emma 303,41-44. Smith Overseas Scholar. ELLEMAN,T. C., AZAD, A. A., and WARD, C. W. (1982). Neuraminidase gene from the early Asian strain of REFERENCES human influenza virus A/RI/5-/57 (H2N2). Nucleic AIR, G. M. (1979). Nucleotide sequence coding for the Acids Rea. 10,7005-7015. “signal peptide” and N-terminus of the hemagglu- ELS, M. C., AIR, G. M., MURTI, K. G., WEBSTER,R. G., tinin from an Asian (H2N2) strain of influenzavirus. and LAVER, W. G. (1985). An 18 amino acid deletion viTOlk$l~ 97,468-472. in an influenza neuraminidase. %rology 142, 241BERTON,M. T., NAEVE, C. W., and WEBSTER,R. G. 247. (1984). Antigen+ structure of the influenza B hem- FANG, R., MIN Jou, W., HUYLEBROEK,D., DEVOS,R., agglutinin: Nucleotide sequence analysis of antiand FIERS,W. (1981).Complete structure of A/duck/ genie variants selected with monoclonal antibodies. Ukraine163 influenza hemagglutinin gene: Animal J. Viral. 52,919-927. virus as progenitor of human H3 Hong Kong 1968 BIGGIN, M. D., GIBSON, T. J., and HONG, G. F. (1983). influenza hemagglutinin. Cell 25.315-323. Buffer gradient gels and %Slabel as an aid to rapid HOVANEC,D. L., and AIR, G. M. (1984). Antigenic DNA sequencing. Proc NatL Acad Sci. USA 80, structure of the hemagglutinin of influenza B/Hong 3963-3965. Kong18173 as determined from gene sequence BLOK, J., and AIR, G. M. (1982).Variation in membrane analysis of variants selected with monoclonal aninsertion and ‘stalk’ sequences in eight subtypes of tibodies. Virology 139,384-392. influenza type A virus neuraminidase. Biochemistry KIDA, H., BROWN,L. E., and WEBSTER,R. G. (1982). 21,4001-4007. Biological activity of monoclonal antibodies to opBOTH, G. W., and AIR, G. M. (1979). Nucleotide seerationally defined antigenic regions on the hemquence coding for the N-terminal region of the maagglutinin molecule of A/Seal/Massachusetts/l/80 trix protein of influenza virus. Eur. J. Rio&em 96, (H7N7) influenza virus. %oZogy 122,38-47. 363-373. KILBOURNE,E. D., LAVER, W. G., SCHULMAN,J. L., and CARROLL,S. M., HIGA, H. H., CAHAN,L. D., end PAULWEBSTER,R. G. (1968). Antiviral activity of antiSON,J. C. (1981). Different sialyloligosaccharide reserum specific for an influenza virus neuraminidase. ceptor determinants of antigenically related influJ. Viral 2,281-288. enza virus hemagglutinins. In “Genetic Variation KILBOURNE,E. D., LIEF, F. S., SCHULMAN,J. L., JAHIEL, among Influenza Viruses” (D. P. Nayak and C. F. R. I., and LAVER, W. G. (1967). Antigenic hybrids of influenza viruses and their implications. Perspect Fox, eds.), pp. 415-421. Academic Press, New York. CATON,A. J., BROWNLEE,G. G., YEWDELL,J. W., and ViroL 5,87-106. GERHARD,W. (1982). The antigenic structure of the KOHLER, G., and MILSTEIN, C. (1976). Derivation of influenza virus A/PR/8/34 hemagglutinin (Hl) specific antibody-producing tissue culture and tusubtype. Cell 31,417-427. mor lines by cell fusion. EUT. J. ImmunoL 6, 511CHOPPIN,P. W., and TAYM, I. (1959). Two kinds of 519. particles with contrasting properties in influenza KRYSTAL, M., YOTJNG,J. F., PALESE, P., WILSON, I. A., SKEHEL,J. J., and WILEY, D. C. (1983). Sequential A virus strains from the 1957 pandemic. Virology mutations in hemagglutinins of influenza B isolates: 8.539-542. Definition of antigenic domains. Proc. Nat1 Acad CHOPPIN, P. W., and TAMM, I. (1960). Studies of two sci. USA 80,4527-4531; correction: Proc Nat1 Aad kinds of virus particles which comprise influenza Sci USA 81,126l (1984). A2 virus strains. 1. Characterization of stable homogeneous substrains in reactions with specific an- LAVER, W. G. (1969). Purification of influenza virus. In “Fundamental Techniques in Virology” (K. Habel tibody, mucoprotein inhibitors, and erythrocytes. and N. P. Salzman, eds.), pp. 82-86. Academic Press, J. Eq. Med 112.895-920. New York. CHOPPIN,P. W., and TAMM, I. (1964). Genetic variants
248
AIR ET AL.
LAVER,W. G. (1978). Crystallization and peptide maps of neuraminidase “heads” from H2N2 and H3N2 influenza virus strains. Virology 86, 78-87. LAVER, W. G., AIR, G. M., and WEBSTER,R. G. (1981). Antigenicity of influenza virus hemagglutinin following chemical modification. Virology 111,538-548. LAVER, W. G., AIR, G. M., WEBSTER,R. G., and MARKOFF,L. J. (1982). Amino acid sequence changes in antigenic variants of type A influenza virus N2 neuraminidase. Virology 122,450-460. LAVER, W. G., WEBSTER,R. G., and AIR, G. M. (1983). Effect of chemical modification on the antigenic properties of A/Tokyo/67 neuraminidase. Effect of a single sequence change on the thermal stability of A/Tokyo/67 neuraminidase. 1% “The Origin of Pandemic Influenza Viruses” (W. G. Laver, ed.), pp. 113-120. Elsevier, Amsterdam/New York. LENTZ,M. R., AIR, G. M., LAVER, W. G., and WEBSTER, R. G. (1984). Sequence of the neuraminidase genes of A/Tokyo/67 influenza virus and of variants selected with monoclonal antibodies. Vim&y 135, 257-265. NORRANDER,J., KEMPE, T., and MESSING,J. (1983). Construction of improved Ml3 vectors using oligonucleotide-directed mutagenesis. Gene 26, 101106. PEATTIE,D. A. (1979). Direct chemical method for sequencing RNA. Proc Nat1 Acad Sti USA 76,17601764.
SANGER,F., NICKLEN, S. and COULSON,A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. NatL Acad Sci USA 74.5463-5467.
VARGHESE,J. N., LAVER, W. G., and COLMAN,P. M. (1983). Structure of the influenza virus glycoprotein antigen neuraminidase at 2.9A resolution. Nature (Londm) 303,35-40. WARD, C. W., and DOPHEIDE,T. A. (1982). The Hong Kong hemagglutinin. Structural relationships between the human (H3) hemagglutinin and the hemagglutinin from the putative progenitor strain A/ duck/Ukraine/l/63 (HAv7). In “Genetic Variation among Influenza Viruses” (D. P. Nayak and C. F. Fox, eds.), pp. 323-340. Academic Press, New York. 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. WEBSTER,R. G., HINSHAW, V. S., and LAVER, W. G. (1982). Selection and analysis of antigenic variants of the neuraminidase of N2 influenza viruses with monoclonai antibodies. Virology 117,93-104. WEBSTER,R. G., LAVER, W. G., and KILBOURNE,E. D. (1968).Reactions of antibodies with surface antigens of influenza virus. J. Gen ViroL 3,315-326. WINTER, G., FIELDS, S., and BROWNLEE,G. G. (1981). Nucieotide sequence of the hemagglutinin gene of a human influenza virus Hl subtype. Nature (LUTZdon) 292,72-75.