VIROLOGY

189,79-87

Location MASAMI

(1992)

of Neutralizing MATSUZAKI,

Department

Epitopes on the Hemagglutinin-esterase

Protein of Influenza C Virus

KANETSU SUGAWARA, KAZUHITO ADACHI, SEIJI HONGO, HIDEKAZU NISHIMURA, FUMIO KITAME, AND KIYOTO NAKAMURA’ of Bacteriology,

Yamagata University

Received September

School of Medicine,

18. 199 1; accepted

lida-Nish,,

March

Yamagata 990-23, Japan

13, 1992

Neutralization-resistant variants of influenza C/Ann Arbor/l/50 virus were selected with monoclonal antibodies against four different antigenic sites on the hemagglutinin-esterase (HE) glycoprotein, and their HE genes were sequenced to identify amino acid residues important for the integrity of each site. Twelve different amino acid substitutions in a total of 18 antigenic variants were all located on the HE1 subunit. Although variants for antigenic site A-2 had a change at position 367, all substitutions in the variants for sites A-l, A-3, and A-4 occurred in the central region of the HE1 spanning amino acid positions 178 to 283. Furthermore, it was found that many of the substitutions in the variants selected with antibodies to sites A-l and A-3 were clustered within or near one of the three variable regions revealed previously by comparing amino acid sequences of the HEs among various influenza C isolates (Buonagurio, D. A., Nakada, B., Fitch, W. M., and Palese, P., Virology 146, 221-232, 1985). The antigenic variants were also examined for their ability to agglutinate chicken and human erythrocytes in order to obtain information concerning the receptor-binding site on the HE molecule. The results suggested that the amino acid changes at residues 178, 186, 187, 190, 206, 212, and 226 decreased the hemagglutinating activity whereas those at residues 245, 266, and 283 produced an opposite effect. 0 1992 Academic Press, Inc

not occur in influenza C viruses. Recently, however, we obtained evidence suggesting that influenza C variant, which had a close relatedness to the virus isolated in 1980 in the United States (G/Mississippi/l/80) and was dissimilar to any of the strains isolated in Japan before 1982, has spread rapidly in the Kinki district of Japan in 1982-l 983 (Adachi et al., 1989). This, together with the finding that the G/Mississippi/l/80-related viruses were distinguishable from previous isolates in serological tests with heterogeneous sera (Kawamura et a/., 1986; Adachi et al., 1989), raises the possibility that influenza C virus may be exposed to immune selection within the human population and that variants with a survival advantage in the population may arise through antigenic changes. Thus more information must be accumulated to fully understand the significance of immune selection in influenza C epidemiology. The HE glycoprotein of influenza C virus displays three biological activities, i.e., receptor-binding activity for 9-0-acetyl-N-acetylneuraminic acid, fusion with the host cell membrane, and receptor-destroying activity, which is a neuraminate-O-aceylesterase (Sugawara et al., 1985, 1986; Vlasak et al., 1987; Herrler el a/., 1988a; Formanowski and Meier-Ewert, 1988). This protein can be proteolytically cleaved into two subunits, HE1 and HE2, which are held together by disulfide bonds (Herrler et al., 1979; Sugawara et al., 1981). The N terminus of the smaller subunit (HE2) contains a stretch of hydrophobic amino acid residues, and this

INTRODUCTION Influenza C virus usually causes a mild upper respiratory illness characterized by fever and long-lasting nasal discharge (Katagiri et al., 1983, 1987) although it can also cause lower respiratory infections such as bronchitis and pneumonia (Moriuchi et a/., 1991). Antigenie analysis with monoclonal antibodies (MAbs) to the viral glycoprotein hemagglutinin-esterase (HE) revealed that there was antigenic variation among influenza C strains isolated at different times in different areas (Sugawara et al., 1986). Nevertheless, influenza C virus seems to be antigenically more stable over time than influenza A and B viruses since analysis with polyclonal antiviral sera showed a high degree of crossreaction among all the isolates examined (Czekalowski and Prasad, 1973; Chakraverty, 1974, 1978; MeierEwert et al., 1981; Kawamura et al., 1986). Previous studies on the RNA genomes of various influenza C isolates demonstrated that the extent of genetic difference did not correlate with the time of virus isolation and that variants from multiple evolutionary pathways cocirculated (Buonagurio et al., 1985, 1986; Kawamura et al., 1986), which led one to a speculation that in contrast to human influenza A viruses, selection of particular antigenic variants by immune pressure may

’ To whom reprint requests should be addressed 79

0042.6822/92

$5.00

CopyrIght 0 1992 by Academic Press, Inc. All rights of reproduction I” any form reserved

80

MATSUZAKI ET AL. TABLE 1 SUMMARY OF OPERATIONALANALYSISOF HE PROTEINWITHANTIGENICVARIANTSSELECTEDWITH NEUTRALIZINGMAbs”

Reactivity of MAbs with the following antigenic variant$ Antigenic site A-l

A-2 A-3 A-4

-

J14c

MAb

Vld,V4

J14 J9 Q5 u9 K16 Ul u2 D37

-t

J9 ___

Q5

K16

u9

Vl tov3

Vl

v2

vi

v3

-

+ -

+

-

t

Vl.V2

Ul

u2

vt

v3

f

f +

Vl

D37 V2,V3

vt,v3

+ f

-

’ Detailed data will be presented elsewhere (Sugawara et al., submitted for publication). b Determined by enzyme-linked immunosorbent assays according to the previously described procedures (Sugawara et a., 1988). No entry, titer with variant was identical to that with parental virus: +. titer with variant was lo- to 1OO-foldless than that with parental virus; -, titer with variant was 1OO-foldless than that with parental virus. ’ MAb used for selection of variants. d Antigenic variant number.

sequence has been suggested to be involved in membrane fusion (Herrler eta/., 1981; Pfeifer and Compans, 1984; Nakada et al., 1984). Affinity labeling with 3H-labeled diisopropylfluorophosphate has revealed that amino acid 71 of the larger subunit (HEl) is the active site serine of the acetylesterase of influenza C virus (Herrler et al., 1988b; Vlasak et a/., 1989). Little or no information has been presented, however, on the location of the receptor-binding site. We previously obtained seven MAbs specific for the HE of influenza C/Ann Arbor/l/50 virus, four with neutralization activity and three without the activity (Sugawara eta/., 1986, 1988). Recently, to expand the panel of MAbs, we attempted to prepare additional anti-HE MAbs and could obtain a total of 37 antibodies (including seven antibodies described above), eight of which were neutralizing. The operational analysis with antigenie variants selected with these neutralizing MAbs identified four nonoverlapping antigenic sites designated A-l, A-2, A-3, and A-4 on the HE protein (see Table 1). Competitive binding experiments suggested, however, that sites A-l and A-2 might topographically overlap each other (Sugawara et al., 1988). In the present study, to identify amino acids important for the integrity of each of the four antigenic sites, the HE gene sequences were determined for 18 neutralization-resistant variants selected in the presence of antibodies to each site. We also investigated the hemagglutinating properties of these MAb-selected variants to obtain information about the relationship between the antigenic sites and the receptor recognition site.

MATERIALS AND METHODS Production of MAbs and selection of antigenic variants Eight neutralizing MAbs (J14, J9, Q5, U9, K16, Ul, U2, and D37) directed to four different antigenic sites on the HE glycoprotein of influenza C/Ann Arbor/l/50 virus were produced according to the method of Kdhler and Milstein (1976) as described previously (Hong0 et a/., 1986; Sugawara et a/., 1991). Antigenic variants resistant to each of these MAbs, except for those resistant to 19, were prepared according to the procedures described elsewhere (Sugawara et a/., 1988). Briefly, serial lo-fold dilutions of the parental C/Ann Arbor/l/50 virus were mixed with the MAb-containing ascitic fluid diluted to l/l 0. After incubation for 30 min at room temperature, the variants that escaped neutralization were isolated by plaquing on LLC-MK, cells and then propagated in the amniotic cavities of g-day-old embryonated hen’s eggs. JS-resistant variants were isolated by inoculating the mixture of parent virus and this antibody into the amniotic cavities of embryonated eggs; no variants could be isolated when the mixture was inoculated onto LLC-MK, cells presumably because of their decreased capacity to replicate in the cells (data not shown). Virus purification and RNA extraction Virus was pelleted from the amniotic fluid and purified by two cycles of centrifugation in a discontinuous gradient consisting of 30 and 60% (w/w) sucrose as

ANTIGENIC

SITES IN INFLUENZA

described elsewhere (Kawamura et a/., 1986). Viral RNA was extracted from purified virions with SDSphenol according to the method described by Palese and Schulman (1976).

81

C VIRUS HE TABLE 2 NUCLEOTIDEAND AMINO ACID CHANGES IN INFLUENZA C/ANN AREORM~O ANTIGENICVARIANTS

Antigenic site

MAb

Variant

A-l

J14

114-Vl J14-v4 J9-Vl J9-v2 J9-v3 Q5-Vl Q5-V2 u9-Vl u9-v3 K16-Vl K16-V2 Ul-Vl Ul-v3 u2-Vl u2-v2

Nucleotide change (position)

Amino acid change (position)

Nucleotide sequence analysis The HE genes of antigenic variants were sequenced by the dideoxynucleotide chain-terminating procedure (Sanger et a/., 1977) with virion(v) RNA as template and synthetic oligonucleotide primers as described in detail elsewhere (Adachi et al., 1989). By using 17 primers listed in the previous report (Adachi et al., 1989) and three additional primers (1228GAATACCTCC1237, 1681AGCATAGGAT1690, and 1823AGTTCAACGA1832) for priming reverse transcription of vRNA, the nucleotide sequences of HE genes except for the first 63 nucleotides at the 5’terminal ends and 3’ noncoding regions (mRNA sence) could be determined.

19

Q5 u9 A-2

K16

A-3

Ul u2

A-4

Hemagglutination

titration

This was performed in microtiter plates in a final volume of 200 ~1, using a 0.5Ob suspension of chicken erythrocytes or a lo/o suspension of human erythrocytes. Measurement of receptor-destroying

activity

A chicken erythrocyte suspension (20/o, v/v) in 500 ~1 of phosphate-buffered saline (PBS), pH 7.2, was mixed with 20 /II of PBS containing 12 HAU of the virus to be tested, and the mixture was incubated at 37” for 2 hr. At the end of the incubation period, the erythrocytes were washed three times with PBS, and their residual virus-binding capacity was estimated by hemagglutination assays. RESULTS Nucleotide changes and deduced amino acid substitutions in variants selected with anti-HE MAbs The antigenic variants selected with anti-HE MAbs were not bound by the antibody used for selection except that variants Ul-Vl and U2-Vl fully retained the capacity to bind the selecting antibody despite their resistance to neutralization (see Table 1). To identify the amino acids important for the integrity of each antigenie determinant, the HE gene sequences of a total of 18 antigenic variants were determined and compared with the previously determined sequence (Matsuzaki et al., 1990) of the parental C/Ann Arbor/l/50 HE gene. The nucleotide changes and deduced amino acid sub-

D37

U2-V3 D37-Vl D37-V3

G(868) --, G(868) + A(581) + A(578) + A(578) + A(589) + G(754) + C(638) + G(754) -. G(1120) -+ G(llZO)+A T(67 1) + A(655) + T(554) + G(643) -+ A(655) + A(655) -. C(698) + C(698) + G(817) +

A A T G G T A T A A C G C A G G A A T

Asp(283) Asp(283) Asn(l87) Lys( 186) Lys(l86) Asn( 190) Glu(245) Ser(206) Glu(245) Asp(367) Asp(367) Phe(217) Lys(212) Leu( 178) Glu(208) Lys(2 12) Lys(212) Thr(226) Thr(226) Val(266)

+ + + --t --* -, --* --* * -+ -+ + -* + + -) + + + +

Asn Asn lie Arg Arg Tyr Lys Leu Lys Asn Asn Ser Glu Ser Lys Glu Glu Asn Asn Phe

stitutions in these variants are summarized in Table 2. Except for two cases (variants U2-V2 and D37-V3), the HE gene sequence of each variant contained a single point mutation that results in an amino acid substitution. Antigenic variants selected with a given MAb had amino acid changes at positions identical with or very close to each other when 114, J9, K16, or U 1 was used for selection. By contrast, Q5, U9, and U2 each selected variants with changes at positions that are 34 to 55 residues apart in the primary sequence. The antigenie variants having an identical amino acid change were sometimes obtained with two independent MAbs to the same antigenic site: the Glu to Lys change at residue 245 occurred in Q5-V2 and U9-V3 and the Lys to Glu change at residue 212 in Ul-V3, U2-V2, and U2-V3. In Figures 1 and 2, the positions of amino acid changes in the antigenic variants are shown in the hydropathy profile and in the primary sequence, respectively. The change of Val 266 to Phe found in variant D37-V3 is not shown in these figures since, as discussed below, this substitution is presumably not responsible for the antigenic alteration in this double mutant. Twelve different amino acid substitutions identified in a total of 18 variants were all located at the HE1 subunit and none at the HE2. It is also to be noted that the region consisting of about 100 amino acids, which spans positions 178 to 283 of the HE1 polypeptide,

82

MATSUZAKI ET AL.

HE1

-4.0

+---~--~-~.-~~

178

100

HE2

21

200

300

400

500

600

Amino acid number FIG. 1. Hydropathy profile of influenza C/Ann Arbor/l/50 virus HE protein showing the location of amino acid changes in neutralization-resis-

tant variants. The search length was seven amino acids. The N-terminal 13 amino acid residues were not sequenced for the HE of C/Ann Arbor/l/50 virus, and their hvdrooathv profile was determined based on the previously published sequence of the C/California/78 virus HE . (Nakada et al., 1984).

displayed a high frequency of substitutions:all variants for sites A-l, A-3, and A-4 had amino acid changes in this region. The amino acid relevant to site A-Z (Asp 367) was located to one of the major hydrophilic domains in the HE1 subunit, and the site A-4 substitution (Thr 226 + Asn) was mapped to a slightly hydrophilic region. The substitutions of Lys for Glu 245 and Asn for Asp 283, which occurred in some of the variants selected with antibodies to site A-l, were in the strongly hydrophilic domains of the HEl. However, the other amino acids relevant to this antigenic site (Lys 186, Asn 187, Asn 190, and Ser 206) were clustered at a less hydrophilic domain. Two of four amino acid changes for site A-3 (Glu 208 --* Lys and Lys2 12 + Glu) were within a moderately hydrophilic domain, whereas the remaining two (Leu 178 + Ser and Phe 2 17 + Ser) were mapped to the relatively hydrophobic areas that flank a weakly or moderately hydrophilic domain where six amino acid changes for either site A-l or site A-3 are clustered. By comparing amino acid sequences among the HE proteins of various influenza C strains, Buonagurio et al, (1985) revealed three clusters of amino acid substitutions in the HE1 portions of the molecules (positions 77-98, 180-214, and 331-346) and speculated that they might be associated with antigenic characteristics of the influenza C virus HE. As seen in Fig. 2, none of the amino acid substitutions in the laboratory-selected variants occurred in the first and third variable regions centering around residues 90 and 340, respectively. By contrast, most of the changes causing the loss of the recognition by the antibodies to sites A-l and A-3 did occur in or near the second variable region from position 180 to 2 14.

Biological properties of antigenic variants selected with anti-HE MAbs As one of the approaches to study the relationship between the antigenic determinants and the receptor recognition site, antigenic variants selected with antiHE MAbs were examined for their ability to agglutinate chicken or human erythrocytes at two different temperatures, 4” and 23”. It has been shown previously that although human erythrocytes contain influenza C virus-specific receptors on their surface, the number was much less than that on rat, mouse, and chicken erythrocytes (Nishimura et a/., 1988). As seen in Table 3, the parent virus agglutinated chicken erythrocytes at both temperatures. The virus also agglutinated human erythrocytes at 4” but did not at 23”. Among 18 antigenie variants tested, only four (K16-Vl , K16-V2, UlVl, and U2-V2) showed the same hemagglutination profile as that of the parent virus. Variant J9-Vl was much less able to hemagglutinate than was the parent virus: the variant agglutinated chicken erythrocytes only at 4” and did not at all human erythrocytes at either temperature. A significant decrease in the ability to agglutinate erythrocytes, though not so drastic as with J9-Vl, was also observed with J9-V2, J9-V3, Q5Vl , U9-Vl , Ul -V3, U2-Vl , UZ-V3, and D37-Vl. Bycontrast, variants J14-Vl , J14-V4, Q5-V2, U9-V3, and D37V3 exhibited the increased hemagglutinating activity compared with the parent virus. D37-V3 showed clear agglutination of human erythrocytes even at 23”, a temperature at which the cells could not be agglutinated by the parent virus. J14-Vl , J14-V4, Q5-V2, and U9-V3 displayed transient aggluination of human erythrocytes at 23”:agglutination was apparent after 30 min though no longer evident after 60 min.

ANTIGENIC

1

SITES IN INFLUENZA

83

C VIRUS HE

HE1 20

40

60

80

100

*************AEKIKICL~K~~FSLHMGFGG~~L~A~~~~FELVKP~GASVL~~IG~KS~~S~P~~V~~T~KFRSLSGG

101

120

140

160

180

200

SLMLSMFGPPGKVDYLYQGCGKHKVFYEGVPH~I~JCYR~IKLlJFQKMIYELASQSHCMSLVM~DKTIPLQ

0 t

201

220

240

260

280

300

301

320

340

360

380

400

480

500

580

600

VRSSPRFLLMPERSYCFDMKEKGLVTAVQSIWGKGRESDHCISQSG~SPF A

401

430

440

501

520

540

I’

HE2

460 ______---_-_--------. TEE~LL~PKFGRCP~KEESIPKI*DGLLIPTSATDTTVTKPKS~IFGIDDLIIGLLFVAIVEAGIGGYLLG.~~SGGGVTKES~KGF~KIG~~~Q~ -_____-----__-_--_--_1

560

LRSSTPlIAIEKL~lDRISHDEQAIRDLTLEIErJARSEALLGELGII~LVG~~IGLQESLWE~SEIT~~~GD~VEVSPG*~ID~~~ICDQSCQNF~F

620 640 601 KFNETAPVPTIPPLDTKIDLQSDPFYWGSSLGLAITAAISLAALVISGIAICRTK --------------------

655

FIG. 2. Locatron of the ammo acrd substitutions In MAb-resistant antrgenic vanants. The positrons of ammo acid substrtutrons in vanants for antigenic sites A-l, A-2, A-3, and A-4 are indicated by filled-in circles, filled-in trrangles, open crrcles. and open triangles, respectively. The heavy lines show the three variable regions revealed by the studres of Buonaguno and coworkers (1985). N-terminal signal sequence and C-terminal transmembrane domain are underlined by waved and dotted lines, respectively. The active site of acetylesterase IS boxed by solid lines and the membrane fusion-inducing HE2 N-terminus by dotted lines. The potential N-glycosylation sates are shadowed. The positions of amino acid substitutions that resulted in a decrease or an increase in the avidity of virus for erythrocyte receptors are marked by arrows or open arrows, respectively. Ammo acids that were not identified are indicated by asterisks.

Alterations in the hemagglutinating properties, observed with the majority of antigenic variants, might result from changes in the receptor-destroying characteristics rather than changes in the receptor-binding characteristics. To examine this possibility, the ability to destroy the receptors on chicken erythrocytes was compared between each antigenic variant and the parent virus according to the procedures described under Materials and Methods. The results obtained with three representative variants (K16-Vl , U9-V3, and D37-Vl) are given in Table 4. Chicken erythrocytes pretreated with the parent virus were no longer agglutinated by the parent virus itself, K16-Vl, or D37-Vl, but were still agglutinated to high titers by U9-V3. Essentially identical results were obtained with erythrocytes treated with K16-Vl having the hemagglutinating

properties similar to those of the parent virus. Pretreatment of erythrocytes with U9-V3, a variant shown to possess an increased hemagglutinating activity, rendered the cells completely resistant to agglutination by any of the viruses tested, suggesting that this variant was able to destroy chicken erythrocyte receptors more efficiently than was the parent virus. The D37Vl -treated erythrocytes, on the other hand, could still be agglutinated by the parent virus, K16-Vl , and U9-V3 although the titers were low when assayed by either of the former two viruses, suggesting that variant D37-Vl was inferior to the parent virus in both the abilities to agglutinate erythrocytes and to destroy their receptors. Thus differences in the hemaggluinating properties from the parent virus, detected in a number of antigenie variants, are not attributable to differences in

84

MATSUZAKI

ET AL.

TABLE 3 HEMAGGLUTINATINGPROPERTIESOF MAb-SELECTED ANTIGENICVARIANTS Hemagglutination

titers (HAU/ml)

Chicken erythrocytes 4O Antigenic site

MAb

A-l

J14 J9 Q5 u9

A-2 A-3

K16 Ul u2

A-4

D37

Human erythrocytes

23”

4”

23”

Variant

60 min

30 min

60 min

60 min

30 min

60 min

Parent J14-Vl, J14-V4 J9-Vl J9-v2,19-v3 Q5-Vl Q5-V2 u9-Vl u9-v3 K16-Vl , K16-V2 Ul-Vl Ul-v3 U2-Vl, U2-V3 u2-v2 D37-Vl D37-V3

4480 320 960 3840 2240 640 1280 640 1920 1920 1920 1920 1920 4480 3840

5120 960 e 3840 2560 1920 1920 1120 2280 2240 3840 3840 3840 3840 3840

5120 960 < < 2560 1920 < 960 2280 2240 < < 2560 < 2560

1920 640 < < 240 1920 480 1280 1280 2240 < < 2240 < 2560

< 960 < < < 1920 < 1920 < < < < < < 2560

< < < < < < < < < < < < < < 2560

a Less than 40.

their receptor-destroying properties. The data rather support the idea that the mutations in these variants resulted in changes in the receptor-binding properties, causing alterations in their ability to agglutinate erythrocytes. The positions of the deduced amino acid substitutions in the variants that exhibited differences from the parent virus in the hemagglutinating properties are indicated in Fig. 2. It should be noted that the substituTABLE 4

DISCUSSION

RECEPTOR-DESTROYING ACTIVITY OF MAb-SELECTED ANTIGENICVARIANTS~ Hemagglutination titers (HAWml) determined with erythrocytes pretreated by:

Virus used for hemagglutination assays

None

Parent

K16-Vl

Parent K16-Vl u9-v3 D37-Vl

4800 1920 1280 6400