Host cell-mediated variation in H3N2 influenza viruses

VIROLOGY

156, 386-395

(1987)

Host Cell-Mediated Variation in H3N2 Influenza Viruses JACQUELINE M. KATZ,’ CLAY-i-ON W. NAEVE, Department

of Virology

and Molecular

Biology,

Received

St. Jude Memphis,

August

Children’s Tennessee

6, 1986;

accepted

AND

Research 38 10 1 October

ROBERT G. WEBSTER Hospital,

28,

332 N. Lauderdale,

P. 0. Box

3 18,

1986

The influence of the host cell on the selection of antigenic variants of influenza A H3N2 viruses and the relevance of host cell selection to the induction of immunity by these viruses have been investigated. Influenza viruses were isolated from human clinical samples during a single epidemic, were passaged in mammalian Madin-Darby Canine Kidney (MDCK) cells or in embryonated hens eggs, and were tested for antigenic variability in the hemagglutinin (HA) molecule with a panel of monoclonal antibodies. In many cases, the HA of virus cultivated in eggs was antigenically distinct from the HA of virus from the same individual grown in mammalian cells. Viruses recovered from different individuals were antigenically similar to each other when grown in mammalian cell lines yet were antigenically heterogeneous when cultivated in eggs. The HA genes of viruses isolated from different individuals during the epidemic were shown, by sequence analysis, to differ from each other by five or six amino acid residues. Sequence analyses of the HA genes of MDCK cell-grown and egg-grown virus obtained from the same individual demonstrated that the molecular changes between antigenically distinct HAS of MDCK cell- and egg-grown A/Mem/l2/85 virus involved a single amino acid substitution at residue 156 in HA1 , which lies at the tip of the HA molecule and immediately adjacent to the receptor-binding site. However, the amino acid sequences of HAS from MDCK-grown and egg-grown viruses (A/Mem/2/85) isolated from a second individual were identical although these viruses exhibited antigenic differences when examined with anti-HA monoclonal antibodies. Therefore, single amino acid changes in the HA molecule may not be the sole cause of antigenic changes in the HA observed between pairs of MDCK cell-grown and egg-grown viruses and genes other than that encoding the HA may contribute to the host cell-mediated antigenic variation of these viruses. Nevertheless, antigenlc differences between viruses grown in eggs and MDCK cells did not influence their ability to protect, since ferrets infected with either live egg-grown or MDCK-grown virus were protected equally well from challenge with virus grown in either host cell type. o 1987 Academic press, IIIC.

cosylation site at the distal tip of the HA molecule of influenza B viruses adapted to growth in eggs (Robertson et al., 1985). These findings raise questions regarding the current use of embryonated eggs as the usual host for cultivation of influenza viruses from clinical and field isolates prior to antigenic characterization of the virus and use in epidemiological studies, but most importantly as the source of virus used in influenza vaccine preparations. One of the key unanswered questions is whether viruses grown in mammalian cells would provide a better antigenic match with the viruses replicating in humans and thus provide the basis for a more effective influenza vaccine. Preliminary results with Hl Nl viruses (Schild et al., 1984) indicate that the phenomenon of host cell selection is not restricted to type B viruses but extends also to type A influenza viruses. The aims of the present study are to determine if host cell-mediated variation also occurs in H3N2 influenza viruses, and whether the mammalian host cell influences selection of antigenitally distinguishable viruses. Additionally, crossprotection studies were performed in ferrets to determine whether viruses grown in eggs would protect

INTRODUCTION The recurrence of influenza virus infections in humans can be attributed largely to antigenic changes in the viral surface glycoproteins, in particular, the hemagglutinin (HA) molecule. An understanding of this antigenic variability is crucial for the control of the disease. In addition to antigenic drift and shift, evidence for another mechanism of variation has previously been reported for type B influenza viruses (Schild eta/., 1983). This mechanism, which is independent of the immunological pressures which provoke antigenic drift, is a result of the selection of virus subpopulations by the host cell in which virus is cultivated. Adaptation of influenza B virus to growth in embryonated eggs resulted in selection of variants which were antigenically and biologically distinguishable from viruses isolated from the same source in mammalian MDCK cells. The molecular basis of this variation was found to reside in a single amino acid substitution in the HA molecule. This substitution resulted in the loss or alteration of a gly-

’ To whom 0042-6622187

requests

for reprints

$3.00

Copyright 0 1997 by Academic Press, Inc. All rights of reproduction in any form reserved.

should

be addressed. 386

HOST CELL-MEDIATED

against challenge with influenza viruses grown in mammalian cells and vice versa. An outbreak in Memphis, Tennessee, of influenza A H3N2 virus in humans during the winter season of 1985 provided the opportunity of obtaining a large number of clinical specimens from which viruses were isolated by passage in either eggs or mammalian MDCK cells. These viruses provided the basis for the present study on the influence of the host cell on the selection of antigenic variants of H3N2 viruses and the relevance of the host cell to the induction of immunity by the variants. We report here that growth in eggs of H3N2 viruses isolated from various individuals resulted in viruses which were antigenically different from each other and frequently also from virus from the same source grown in mammalian cell culture. In contrast, MDCK-grown viruses isolated from different individuals were antigenitally similar to each other. The antigenic differences between MDCK- and egg-grown viruses, however, were shown to be irrelevant in terms of the induction of cross-protective immunity since ferrets were protected equally well by live virus grown in either host cell type. MATERIALS

AND

METHODS

Viruses The H3N2 influenza viruses used in this study were isolated during a single influenza outbreak which occurred in Memphis, Tennessee, over a 2-week period in January and February 1985. The viruses were isolated from throat swabs or gargles directly in MDCK cells (Madin and Darby, 1958) or in embryonated chicken eggs (eggs). After initial isolation of virus in eggs from amniotic fluid, virus was passaged further in the allantoic cavity. Viruses were concentrated in an Amicon system and were purified by differential sedimentation through 25709/0 sucrose gradient. Virion RNA was isolated by treatment of purified virus with proteinase K and sodium dodecyl sulfate, followed by extraction with phenol: chloroform (1: 1) as described previously (Bean et al., 1980). Monoclonal

antibodies

and polyclonal

sera

Monoclonal antibodies to the HA of egg-grown A/ Bangkok/l/79 and HA of MDCK-grown A/Memphis/2/ 85 (Mem/2) viruses were prepared according to Kohler and Milstein (1976) as described previously (Kida eta/., 1982). Mouse ascitic fluids were used as the source of monoclonal antibodies in enzyme-linked immuno-

VARIATION OF INFLUENZA

387

sorbent assay (ELISA), hemagglutination-inhibition (HI), and neutralization tests. Preimmune, postinfection, and postchallenge sera obtained from ferrets were treated with trypsin and periodate to remove inhibitors prior to use in HI tests. Serological

tests

HI tests were performed in microtiter plates by standard methods using 0.5% chicken erythrocytes. The ELlSAs were performed as described previously (Kida et a/., 1982) except that 50 hemagglutinating units (HAU) intact purified virus was used to coat wells for 2 hr at room temperature. Neutralization assays were done by mixing 200 tissue culture infectious doses (TCID)r,Ovirus and varying dilutions of monoclonal antibodies for 1 hr at 20”. The resulting mixtures were titrated for residual virus infectivity on MDCK monolayers prepared in 96-well tissue culture plates. After 3 days incubation at 37” in 5% CO,, neutralization titers were assessed by the presence of cytopathic effect on cell monolayers and hemagglutination activity in culture supernatants. Neutralization titers are expressed as the reciprocal of the antibody dilution which completely inhibited virus infectivity in 50% of quadruplicate cultures. Infection

of ferrets with influenza virus

Seronegative 4- to 6-month-old female ferrets were anesthetized with methoxyflurane and infected intranasally with 1 O6TCIDSO or egg infectious doses (EID)50 of virus. Nasal washes were collected after anesthesia with ketamine (Bristol Laboratories) which increased mucous secretion. The nasal washes were titrated for virus infectivity in both MDCK cells and in eggs. Ferrets were challenged with 1 O6 TCIDsO or EIDsO virus at the time intervals given in the text. RNA sequencing Twelve oligonucleotide primers complementary to the virion HA gene were synthesized on an Applied Biosystems Inc. 380A DNA synthesizer. Crude DNA was deprotected in NH,OH at 55” for 12 hr and the NH,OH was removed under vacuum in a Speed-Vat concentrator. The DNA was resuspended in 70% deionized formamide and electrophoresed on 20% acrylamide-7 M urea preparative gels. The DNA was visualized by uv absorbance, cut out of the gel, eluted by diffusion in 2 M NH,Ac, diluted fourfold, and loaded onto a l-ml DEAE-cellulose column equilibrated with 0.5 M NH,Ac. The column was washed with water, and the DNA was eluted with 30% (v/v) triethylammonium bicarbonate (pH 7.5) desalted under vacuum, and resuspended in water.

388

KATZ.

NAEVE,

AND

Nucleic acid sequences were determined by reverse transcription of virion RNA in the presence of synthetic primers and dideoxynucleotides using a modification of the “quasi-end labeling” method described by BinaStein et al. (1979). Reaction products were resolved on 8% polyacrylamide-7 M urea thin gels in TBE buffer (90 ml\/lTris-borate, pH 8.0; 1 mM EDTA) as described (Maxam and Gilbert, 1977). Sequence data were stored, edited, and compiled through the use of the “DB system” computer programs of Staden (1982).

isolates to a representative number (10 out of a total of 32 tested) of anti-HA monoclonal antibodies. Viruses that were isolated from different individuals and grown in MDCK cells showed a high degree of antigenic similarity. However the viruses isolated from the same clinical specimen and grown in eggs were more diverse in their reactivity with the panel of monoclonal antibodies. The reactivity patterns observed for egg-grown viruses fell into three distinct groups; viruses in group I were characterized by their reactivity with monoclonal antibodies 1716, 54/6, 73/l, and 75/3. The reactivity with these antibodies was unique to egg-grown viruses in group I (e.g., Mem/2); all other egg-grown viruses tested reacted weakly or not at all with these antibodies as did any of the virus isolates grown in MDCK cells, including MDCK-grown Mem/2 and Mem/30. In total, one-third of virus isolates showed this reactivity pattern when grown in eggs. The second group of egg-grown viruses (e.g., Mem/l2) were characterized by a marked loss of reactivity for the majority (29/32) of anti-Bangkok HA antibodies tested. The MDCK-adapated forms of these virus isolates, on the other hand, showed no such loss of reactivity. Therefore group II egg-grown viruses differed strikingly not only from other egg-grown viruses isolated in a single outbreak, but also from virus isolated from the same clinical specimen and grown in MDCK cells. Viruses isolated from four different individuals exhibited this reactivity pattern. The eight remaining viruses tested fell into a third category of reac-

RESULTS Antigenic analysis of MDCK cell-grown egg-grown influenza A (H3N2) viruses

and

A total of 18 viruses isolated from clinical specimens and grown in either MDCK cells or embryonated eggs were tested for their reactivity with polyclonal ferret reference sera and with a panel of monoclonal antibodies directed against the HA of the H3N2 egg-grown virus A/BangkokIl/79. All of these monoclonal antibodies failed to react with MDCK- and egg-grown Hl Nl viruses and, therefore, were not directed against hostspecified carbohydrate on the HA molecule. Studies with ferret antisera showed that the isolates were most closely related to the A/Mississippi/l/85 (H3N2) reference strain (results not shown). Table 1 shows the hemagglutination inhibition reaction of six different virus

TABLE HEMAGGLUTINATION-INHIEITION

REACTIONS OF ANTI-HA

MONOCLONAL

WEBSTER

1

ANTIBODIES

WITH INFLUENZA A (H3N2)

Monoclonal A/Memphis/85 isolate No.

2 30 12 13

6 19

VIRUSES GROWN IN DIFFERENT

antibodyb

Host celT

4/l

85/l

4914

106/2

4515

M E M E

512’ 512 256 256

256 128 256 256

512 1024 256 512

512 128 128 256

64 64 64 64

128 1024 32 256

M E M E

128

256

51; <

256 < 256 <

<

256 < 128 <

64 64 64 32

16 8 32 8

2 < 4 <

2 < 1 <

< < < <

2 < 2 <

M E M E

512 256 128 32

512 64 128 64

256 64 256 128

256 128 128 64

128 128 64 32

64 16 32 32

4 2 4 8

1 < 2 4

< < < <

2 < 2 4

51;

HOST CELLS

a Viruses were isolated and passaged in eggs (E) or MDCK cells (M). * Monoclonal antibodies were raised against the HA of egg-grown A/Bangkok/l/79 c Hemagglutination-inhibition titers represent the reciprocal of the highest dilution virus. d < Represents a titer of less than 100.

19/l

1716

5416

4 512

(H3N2). of monoclonal

1 512

4 128

73/l id 256

1 128

antibody

7513

Group

2 512

< 64

(Xl O-‘) inhibiting

I 2

128 II

III

4 HA units

of

HOST

VARIATION

CELL-MEDIATED

OF

389

INFLUENZA

tivity, showing no difference in reactivity for the monoclonal antibodies between virus grown in eggs and virus from the same source grown in MDCK cells (e.g., Mem/G).

virus with the cell surface receptors neutralization assays.

Antigenic analysis by ELISA binding and neutralization tests

To determine whether the antigenic differences observed between viruses from the same source grown in different host cells could be attributed solely to host cell-specified modification of the HA, such as glycosylation, antigenic analysis was done on Mem/2 and Mem/l2 viruses that had been isolated and passaged in one cell type and then subsequently grown for a number of passages in the alternate host cell. As Table 3 shows, both viruses originally cultivated in eggs and subsequently passaged in mammalian cells retained the antigenic characteristics of viruses grown in eggs. In contrast, repeated passage of MDCK-grown Mem/ 12 virus in eggs, resulted in the selection of a virus population which possessed the antigenic phenotype of egg-grown Mem/l2. After six serial passages in eggs, Mem/2 virus still retained the cell-grown phenotype. Hence, the mutations that result in the egggrown phenotypes of both viruses are stable even when egg-grown viruses are subsequently passaged in mammalian cell culture. However, the subsequent growth in eggs of virus originally isolated on MDCK cells may or may not lead to the selection of a virus population with the egg-grown phenotype. It was also determined that MDCK cells were typical of other mammalian cell lines in the antigenic profile of the virus they produced. Antigenically identical viruses were obtained when viruses from the same source were grown in MDCK, guinea pig kidney, and human rectal tissue cell lines (data. not shown).

Antigenicity of viruses grown in alternate host systems

The above results were based on HI tests with monoclonal antibodies prepared to egg-grown virus. Serological reactivity to a number of monoclonal antibodies raised against HA from MDCK-grown Mem/2 (2/2,2/4) as well as the HA from egg-grown A/Bangkok/ l/79 (73/l, 85/l) were determined in different assay systems (Table 2). For MemM virus, the reactivities previously observed with monoclonal antibodies directed against egg-grown HA were similar when tested in ELISA and the virus neutralization assay. Thus, monoclonal antibody 73/l which differentiated between MDCK- and egg-grown Mem/2 viruses by HI also gave 300-fold lower titers against MDCK-grown compared to egg-grown Mem/2 in ELISA and virus neutralization assays. In contrast, monoclonal antibodies 85/l, 2/2, and 2/ 4, which differentiated between MDCK-grown and egggrown Mem/l2 viruses in HI and virus neutralization tests (giving, in general, 1OO- to 1OOO-fold lower titers with egg-grown virus), showed no substantial difference in binding to these viruses in ELISA. These antibodies bound equally well to both viruses, regardless of the host cell in which virus was grown. These results indicate that the epitopes which these antibodies recognize are present on both egg-grown and MDCKgrown Mem/l2 viruses, but that the antibodies are unable to inhibit effectively the interaction of egg-grown TABLE SEROLOGICAL

REACTIVITY OF MDCK-GROWN

AND EGG-GROWN

2

A/MEMPHIS/~~

ISOLATES USING ANTI-HA

MONOCLONAL

Antibody HI titer (Xl O-‘)*

in HI and virus

Virus

neutralization

titer

(log,,)

ANTIBODIES@

binding titer by ELlSAd

(log,,,)

Host cell

73/l

85/l

212

214

73/t

85/l

212

214

73/l

85/l

212

214

Meml2

M E

< 256

256 128

256 256

256 256

<2.0 4.5

5.3 5.0

5.3 5.2

4.8 4.0

<2.5 4.8

5.3 5.2

4.3 4.6

4.2 4.1

Mem/l2

M E

< <

256 <

256 <

256 32

<2.0 12.0

5.3 2.5

5.3 2.7

5.8 2.8

<2.5 <2.5

5.5 5.7

4.4 4.0

4.3 4.2

Virus

a Monoclonal antibodies were raised in mice against the H3 HA of egg-grown A/Bangkok/l/79 (73/l, 85/l) and MDCK-grown Mem/2 (2/2, 2/4). * HI titers represent the reciprocal of the highest dilution of antibody (Xl Oe2) inhibiting 4 HA units of virus. ’ Neutralization titers are expressed as the reciprocal of the highest dilution of antibody (log,,,) which neutralized 200 IDSo virus in 50% of MDCK cultures. dThe antibody binding results are presented as the reciprocal of the dilution of antibody (loglo) which gave 50% of maximum level of binding to virus.

390

KATZ,

NAEVE,

AND

TABLE HEMAGGLUTINATION-INHIBITION

WEBSTER 3

REACTIONS OF A/MEMPHIS/85

VIRUSES PASSAGED IN ALTERNATE

Monoclonal Virus Mem/2

Mem/l2

Cell type used for isolation and passage8

4515

4914

M(X4) M(x4)-E(x1) M(x4)-E(x6) E(X3) E(x3)-M(x1) E&3)-M(x2)

>2056* 128 512 >2056 512 256

22056 >2056 >2056 >2056 >2056 >2056

M(X4) M(x4)-E(xl)

>2056 512

>2056 1028

M(x4)-E(x2) E(X3) E(x3)-M(x1) E(x3)-M(x2)

512 512 >2056 >2056

a Viruses were isolated and passaged three more times as indicated, in the alternate host b Hemagglutination-inhibition titers represent virus. c < Represents a titer of less than 100.

antibodies 5416

73/l <= < 1

>2056 >2056 >2056

>2056 >2056 1028

512b 256 8
1

<

<

<

Egg

1 < <

1 < <

2 < <

< < <

Egg Egg Egg

cell;

cl.E7

cl.E9

32 32

64 64 4 < 4 128

128 64 64 32 128 128

1 1

16

E = eggs)

antibody

and then

passaged

(Xl O-‘) inhibiting

one

or

4 HA units

of

required to isolate clones, virus which would grow in MDCK cells was present in the original specimen at a higher (lo- to lOO-fold) frequency than virus which would readily grow in eggs. All clones isolated in MDCK cells were antigenically homogeneous and identical to the total viral population. In contrast, virus from the same clinical specimen grown at limit dilution in eggs displayed two distinct antigenic profiles characterized by reactivity (E9, El 0) or lack of reactivity (E5, E7) for monoclonal antibodies 54/6, 73/l, and 75/3. The dom-

4

VIRUSES ISOLATED FROM A PRIMARY SPECIMEN

IN MDCK

CELLS AND EGGS”

MDCK-grown

cLE5

<

MDCK MDCK

<

Egg-grown

4515 4914 5416 73/l 7513 85/l

>2056 128

Egg Egg Egg

<

TABLE

Uncloned virus

MDCK MDCK MDCK

1

The observed antigenic heterogeneity of virus isolated during a single outbreak was further investigated to determine whether a clinical isolate obtained from an individual consisted of a homogeneous virus population or whether it contained two or more subpopulations of virus (Table 4). Based on the limit dilution

Monoclonal antibodv

>2056 >2056 256 >2056 >2056 >2056

1

or four times in the first cell type given (M = MDCK cell type. the reciprocal of the highest dilution of monoclonal

REACTIONS OF CLONED

Virus “phenotype”

85/l

1 4 4

Isolation of antigenically distinct virus subpopulations from the primary specimens in eggs

HEMAGGLUTINATION-INHIBITION

HOST CELL TYPE

cl.ElO 64 128 1028 128 256 256

Uncloned virus 128 12056 1 < 2 22056

cl.Ml

cLM3

cLM7

cl.ME

256 >2056

512 >2056

128 12056

256 >2056

1 < 1 1028

1 < 2 >2056

1

1 < 1

< 1 >2056

22056

B Clones were obtained by two limit dilution passages of a primary clinical specimen (A/Mem/24/85) in either MDCK cells or eggs. A total of 10 MDCK-selected and 10 egg-selected clones were obtained. Four of each type are presented here. The uncloned viruses were obtained after two passages at low dilution of the same clinical specimen. b Hemagglutination inhibition titers represent the reciprocal of the highest dilution of monoclonal antibody (X 1 O-‘) inhibiting 4 HA units of virus. c < Represents a titer of less than 100.

HOST

CELL-MEDIATED

VARIATION

Protection egg-grown

The entire HA genes of MDCK-grown and egg-grown Mem/2 and Mem/l2 viruses were sequenced by the dideoxy chain-termination technique. Despite repeated sequencing of the HA genes of MemM viruses, no amino acid sequence differences were detected between the HAS of MDCK- and egg-derived Mem/2 viruses (Table 5). This result contrasts with the antigenic differences detected between these two viruses using a panel of anti-HA monoclonal antibodies (Table 1). A single amino acid change in the HA1 polypeptide was detected between the HA of MDCK and egg-grown Mem/l2. This substitution occurred at residue 156 in HAl, where a glutamate in MDCK-grown virus sequence was substituted for a lysine in the egg-grown virus. Residue 156 is located at the tip of the HA molecule, within antigenic site B (Wiley er al., 1981) and in close proximity to the receptor-binding site as illustrated in Fig. 1. Comparison of the sequence of the HA from MDCKgrown Mem/2 and Mem/l2 shows five amino acid differences between two viruses isolated during the same influenza outbreak (Table 5 and Fig. 1). These occurred at residues 102, 219,246, and 280 in HA1 and 215 in HA2. These same five amino acid differences occurred between strains 2 and 12 grown in eggs along with the

TABLE ACID SEQUENCE OF HA OF MEM/~

studies in ferrets using live MDCK- and viruses

To determine whether the antigenic difference observed between influenza viruses grown in different host cells are significant in terms of protection, pairs of antigenically distinct MDCK cell-grown and egggrown viruses obtained from the same clinical isolate were tested for their ability to induce cross-protective immunity in ferrets, The central question addressed was whether immunity induced by egg-grown virus was as effective as that given by MDCK cell-grown virus in protecting animals from subsequent reinfection with mammalian cell-grown virus. An initial experiment was performed using MDCK- and egg-grown Mem/2 viruses. The infectivity titers of these viruses were MDCKgrown Mem/2, 107.5 TCID,dml in MDCK cells and 1 04.’ ElDJml in eggs; egg-grown Mem/2, lo’.’ TClD&ml in MDCK cells and lo*.’ EID,dml in eggs. Therefore MDCK-grown virus replicated to over 1OOO-fold lower titers in eggs compared to MDCK cells, whereas egggrown virus replicated equally well in either host cell type. Ferrets were infected intranasally with 1 O6TCIDBo of live MDCK cell-grown or 10” ElD50 egg-grown virus. Peak virus titers were recovered from ferrets on Days 1 and 3 p.i., with titers declining somewhat by Day 5. Generally, no virus was recovered in nasal washes obtained from ferrets 7 days p.i. (data not shown). Table 6 shows the results of an infection and challenge experiment performed with MDCK- and egg-grown Mem/ 2 viruses. Both MDCK-grown and egg-grown viruses replicated to high titers in the nasal mucosa of ferrets. Virus isolated from ferrets infected with MDCK-grown

Sequence analysis of the HA genes of MDCKgrown and egg-grown MemE and Mem/lZ viruses

OF AMINO

391

INFLUENZA

additional amino acid substitution at residue 156 in HA1 in egg-grown Mem/l2 virus.

inant phenotype of the total virus population, generated by two passages of virus at low dilution in eggs, was that of clones E5 and E7. However, upon subsequent passage at low dilution in eggs, the profile of the viral population changed to resemble that of clones E9 and El 0 (data not shown) suggesting that virus of this antigenic phenotype had become the dominant subpopulation of the total viral population.

COMPARISON

OF

5 AND MEM/~ 2 VIRUSES GROWN IN MDCK Amino

acid

sequence

CELLS OR EGGS’

at residue

HA1 Virus Mem/2b Memll2

Host isolate

cell used to and passage

HA2

102

156

219

246

280

215

Egg

Val Val

Glu Glu

Ser Ser

Ser Ser

Glu Glu

CYS CYS

MDCK Egg

W Glv

Glu LYS

Pro Pro

Ile Ile

W W

Trp Trp

MDCK

“The complete sequences of the HA genes of these viruses were determined by reverse transcription of virion synthetic primers and dideoxynucleotides using a modification of the “quasi-end labeling” method (Bina-Stein era/., ’ No amino acid sequence differences were observed between the HA of MDCK cell-grown and egg-grown Mem/2

RNA in the 1979). viruses.

presence

of

392

KATZ.

a156--

NAEVE.

AND

WEBSTER

viruses recovered from ferrets were antigenically identical to the MDCK- or egg-grown virus used to infect the animals. This was true regardless of whether specimens from ferrets were cultured in eggs or MDCK cells. Four weeks after infection, ferrets were challenged with 1O6 TCIDSO of live MDCK cell-grown Mem/2 virus. Ferrets which had been previously infected with either MDCK- or egg-grown virus were completely protected from challenge with MDCK-grown Mem/2 virus; no virus was recovered in nasal washes taken from ferrets on Days 1, 3, and 5 postchallenge, whereas uninfected control ferrets shed substantial amounts of virus over this period. No further increase in serum HI antibody was noted for ferrets previously infected with virus. Studies were also done with Mem/l2 virus in each host system with similar results. A summary of the protection studies done in ferrets with live H3N2 viruses is given in Table 7. In all cases, ferrets infected with either live egg-grown or MDCK-grown virus were protected equally well from challenge with virus grown in either host cell type. These results indicate that the antigenic differences between viruses grown in eggs and MDCK cells are not reflected in their ability to protect when used as live vaccines in ferrets.

/246

DISCUSSION

FIG. 1. Location of altered amino acid residues on the a-carbon tracing of the three-dimensional structure of the H3 HA molecule (Wilson ef a/., 1981). The amino acid substitution in the HA of egggrown vs MDCK-grown Mem/l2 virus is located at residue 156 in HAl. The amino acid residues which differ between Meml2 and

Memll2

viruses are shown at residue numbers 102, 219, 246, and

280 in HA1

The change

at residue

215 of HA2

is not shown

as this

residue is not present in the bromelain-cleaved HA for which the structure was determined. This figure was generated using an IBM version of FRODO (kindly provided by Drs. Keith Ward and Jim Wick, University of Wisconsin, Parkside) on an Evans and Sutherland PS340.

Mem/2 gave approximately 1OOO-fold lower titers when assayed in eggs compared to the homologous cell type and was therefore similar in growth characteristics to the infecting virus. Virus isolated from ferrets infected with egg-grown Mem/2, in this experiment, had a higher titer when assayed on MDCK cells compared to eggs. infection of ferrets with either MDCK- or egg-grown Mem/2 virus resulted in substantial serum HI antibody measured 14 days p.i. The antibody raised in response to MDCK- or egg-grown viruses cross-reacted with virus grown in the heterologous host cell type. Based on their reactivity for monoclonal antibodies in HI assay,

The antigenic analysis of a number of H3N2 influenza viruses isolated during an influenza outbreak has shown that host cell-mediated selection of antigenic variants frequently occurs in H3N2 viruses currently circulating in the human population. We have shown here that a clinical specimen isolated from an individual may contain at least two antigenically distinct subpopulations of virus. Isolation of virus in eggs from a clinical specimen occurred at a much lower frequency than isolation of virus from the same source in MDCK cells. Therefore, cultivation of such a mixed virus population in eggs may select for the growth of a minor variant subpopulation which has a growth advantage in this host cell type. The selection of such variants in eggs may explain why the viruses obtained from different individuals during the influenza outbreak were antigenically heterogeneous when grown in eggs, while the viruses from the same individuals were antigenically similar when cultivated in mammalian cells. The identification of distinct antigenically of virus within a single clinical

variant subpopulations isolate has also recently

been reported by Patterson and Oxford (1986). Antigenic heterogeneity among the HAS of viruses recovered within a single influenza epidemic has previously

been

observed

for type

B viruses

(Oxford

et a/.,

1983) and type A H 1 N 1 (Six et al,, 1983) viruses. In the present study, the antigenic variation between co-

HOST

CELL-MEDIATED

VARIATION TABLE

RESPONSE OF FERRETS INFECTED WITH LIVE MDCK-

OR EGG-GROWN

Response

Virus used for primary infection

Virus Ferret

No.

isolation

in

OF

393

INFLUENZA

6

MEM/2

VIRUS TO CHALLENGE

WITH LIVE MDCK-GROWN

MEM/~

VIRUS

Response to challenge with MDCK-grown virus”

to infection Serum HI antibodya to virus grown in

Virus MDCK

Egg

MDCK

Egg

MDCK

Egg

isolation

Serum HI antibody to virus grown in

in

MDCK

Egg

MDCK-grown MemI2

171 271

6.6c 6.1

3.5 3.0

2560 2560

2560 10240


< <

640 1280

640 1280

Egg-grown Meml2

449 454

5.8 5.3

4.0 4.4

1280 1280

5120 2560

< <

< <

2560 2560

2560 2560

None

457 466

-

-

-

-

6.7 6.4

3.3 2.8

10240 10240

2560 1280

a All ferrets had a serum HI antibody titer of 510 prior to infection with virus. b Ferrets were challenged 4 weeks after primary infection with 1 O6 TCIDSO MDCK-grown Mem/2 virus. ‘Values represent the amount of virus recovered in nasal washes on Day 3 postinfection or challenge and are expressed as the ID&ml (log,J obtained when nasal washes were titrated in either MDCK cells or eggs. Titers of virus recovered on Day 3 are representative of virus isolation from ferrets on Days l-5 postinfection or challenge. d < Indicates that no virus was isolated from 10-l or higher dilution of nasal washes titrated in MDCK cells or eggs.

circulating viruses appeared to be a consequence of cultivation of viruses in eggs. However, sequence analysis of viruses obtained from two individuals and grown either in MDCK cells or in eggs established a substantial number of amino acid differences in the HA molecule of viruses isolated during the same influenza outbreak. A total of five amino acid changes in the HA were observed between MDCK-grown Mem/2 and Mem/l2 viruses. Egg-grown Mem/2 and Mem/l2 viruses differed at these same five amino acid residues and in addition at residue 156 in HAl. This degree of difference between two viruses cocirculating during a single epidemic is as great as the sequence changes observed in the HA of H3N2 viruses isolated 1 or more years apart and from different parts of the world (Both et al., 1983; Skehel et al., 1983). Although the sequence analysis of the HA gene demonstrates the molecular heterogeneity between the HAS of cocirculating viruses, the amino acid sequence information does not indicate the actual substitutions responsible for the observed antigenic differences between these viruses. In fact, the HA of MDCK-grown Mem/2 and Mem/l2 viruses are antigenically identical when tested for their reactivities with monoclonal antibodies by HI assay, yet differ in amino acid sequence at five residues. This suggests that none of the five amino acid changes between Mem/2 and Mem/l2 affect the antigenic epitopes recognized by these monoclonal antibodies. MDCK- and egg-grown Mem/2 viruses have identical amino acid sequences, yet can be distinguished anti-

genitally, indicating that some post-translational modifications of viral HA may occur which results in the isolation of antigenically distinct viruses from cell culture and from eggs. Such modification does not appear to be host-specified, since the different reactivities of MDCK- and egg-grown viruses for monoclonal antibodies were independent of the last host cell type used for propogation. Alternatively, the antigenic differences between MDCK- and egg-grown Mem/2 may be the result of a mutation in a viral gene other than that which codes for the HA but which, nevertheless, exerts an effect on the antigenic structure of the HA molecule. Antigenic and biologic differences between influenza strains grown in eggs versus mammalian cell culture have been correlated with sequence changes in the TABLE SUMMARY OF PROTECTION LIVE MDCKHost

7

OF FERRETS OBTAINED OR EGG-GROWN H3N2

cell used to grow virus for

BY VACCINATION VIRUSES

No. ferrets

WITH

protected

Virus

Infection

Challenge

Memf2

MDCK EGG

MDCK MDCK

414 313

Mem/l2

MDCK EGG

MDCK MDCK

313 414

MemIl2

MDCK EGG

EGG EGG

212 212

No. ferrets

tested

394

KATZ,

NAEVE,

HA1 subunit which alter the glycosylation of the HA molecule. Adaptation of type B influenza viruses to growth in eggs selects for viruses which have a deleted or altered oligosaccharide side chain at the tip of the HA molecule in close proximity to the receptor-binding pocket (Robertson et al., 1985). In contrast, Deom et a/. (1986) have shown that loss of an oligosaccharide from the HA of AANSN influenza virus increases the ability of the mutant virus to bind to and grow in mammalian cell culture. In this case, it was not the absolute presence or absence of carbohydrate at a particular site but rather the composition of the oligosaccharide side chain that altered the affinity of the virus for host cell receptors. Viruses with HA containing more highly branched, complex oligosaccharides had a lowered affinity for erythrocyte and host cell receptors. Our study has shown that amino acid substitutions other than those which directly effect the glycosylation of the viral HA may also result in antigenic and biologic differences between virus grown in eggs versus that grown in mammalian cells. This substitution, at residue 156 of HA1 , occurs at a site immediately adjacent to the receptor-binding pocket of the HA and involves a change from an acidic residue (Glu) in MDCK-grown virus to a basic amino acid (Lys) in the egg-grown form of the virus. This could conceivably alter the conformation in or around the receptor-binding site, and lead to a change in the affinity of egg-grown virus for cell surface receptors and the observed antigenic differences between MDCK-grown and egg-grown Mem/l2 viruses. Other workers report similar findings for other currently circulating human H3N2 viruses (Dr. R. S. Daniels, personal communication). Differences in antigenicity between HAS of MDCK cell- and egg-grown viruses are accompanied by one or two amino acid changes in known antigenic areas of the HA which also delineate the periphery of the receptor-binding site in the tip of the HA molecule. However, on the basis of the novel results obtained with antigenically distinct MDCKgrown and egg-grown Mem/2 viruses for which, despite repeated sequencing, no amino acid sequence change in the HA could be found, the single amino acid change between the HAS of MDCK- and egg-grown Mem/l2 viruses may not be solely responsible for the antigenic differences between these viruses. The involvement of a gene(s) other than that encoding the HA cannot be ruled out. It is interesting that although MDCK- and egg-grown Mem/l2 viruses differed strikingly in their reactivity for monoclonal antibody in HI and virus neutralization tests, no such antigenic differences were detected between these viruses using the same monoclonal antibodies in a direct binding assay (ELISA). The inability of monoclonal antibody to inhibit hemagglutination or infection

AND

WEBSTER

by egg-grown Mem/l2 virus may be a result of an enhanced ability of the egg-grown HA to bind to host cell receptors, such that some monoclonal antibodies are less able to prevent egg-grown virus binding to chick erythrocytes or MDCK cells in the respective assays. In support of this theory, egg-grown Mem/l2 virus is capable of agglutinating neuraminidase-treated erythrocytes containing insufficient amounts of sialic acid to be agglutinated by MDCK-grown Mem/l2 virus (data not shown). This result is similar to the observations of Burnet et a/. (1946) that human influenza virus isolates can be arranged according to avidity for human erythrocytes into a receptor gradient. Any mutation in the HA which effects the receptor-binding site and thus increases the capacity of a variant to bind to certain cells would provide a selective growth advantage in that host cell type. In the case of Mem/l2 virus, the growth of virus in eggs appears to have selected for such a variant. Other workers have shown that closely related HAS which differ antigenically may also differ in their specificity for siayloligosaccharides present on cell surface receptors (Daniels eta/., 1984) and that a single amino acid substitution can alter the specificity of the receptorbinding site on the HA molecule (Rogers et a/., 1983). Differences in receptor specificity among influenza viruses have been correlated with sensitivity of viruses to inhibitors present in nonimmune horse serum (Rogers et al., 1983). The MDCK- and egg-grown viruses used in the present study showed no appreciable difference in their sensitivity to horse serum inhibitors or inhibitors found in sera from other mammalian or avian species (data not shown) and, on this basis, display no evidence of modification of receptor specificity in one or other host cell system. The present study has investigated the importance of host cell selection of antigenic variants of influenza virus in the induction of immunity through vaccination. Studies carried out using live influenza viruses indicated that antigenic differences between viruses grown in eggs and MDCK cells did not influence their ability to protect ferrets. Although levels of local antibody in the nasal mucosa of immune ferrets were not determined in the present study, it is likely that the induction of complete protection by virus grown in either host cell type reported here was a result of such antibody-mediated immunity (Fenton et a/., 1981). Although we have shown here that MDCK-grown and egg-grown viruses are equally effective as live vaccines, their relative effectiveness as inactivated vaccines, such as those currently used to control influenza infection, remains unknown. Based on serological studies on type B influenza viruses (Schild et al., 1983), type A Hl Nl viruses (Oxford et al., in press), and our own

HOST

CELL-MEDIATED

ob$.ervations with H3N2 viruses (data not shown), MDCK-grown, rather than egg-grown, viruses are more cross-reactive with viruses that grow in the respiratory tract of humans and induce antisera that is less discriminatory in its reactivity with related viruses. For these reasons, it is important to determine whether inactivated vaccine prepared from mammalian cell grown virus offers better protection than egg-grown vaccine against virus grown in mammalian cells. These studies are currently underway in this laboratory. ACKNOWLEDGMENTS This work was supported by U.S. Public Health Research Grant Al-20591 from the National Institute of Allergy and Infectious Diseases, Cancer Center Support (CORE) Grant CA 21765-l 0, Biomedical Research Support Grant RR 05584-21, and American Lebanese Syrian Associated Charities. The authors thank Drs. G. Schild, R. Daniels, J. Robertson, and J. Wood (National Institute for Biological Standards and Control, London, UK) for helpful discussions, and acknowledge the work of Karl Schadow (Dept. of Microbiology, University of Tennessee, Memphis, TN). Susan Channell, Lisa Newbeny, John Kayoma, Janet Winkler, and Deanna Williams provided excellent technical assistance, and we thank Lisa Wilson for typing the manuscript.

REFERENCES BEAN, W. J., !&RAM, G., and WEBSTER, R. G. (1980). Electrophoretic analysis of iodine-labeled influenza virus RNA segments. Anal. Biochem. 102, 228-232. BINA-STEIN, M., THOREN, M., SALZMAN, N., and THOMPSON, J. A. (1979). Rapid sequence determination of late simian virus 40 16s mRNA leader by using inhibitors of reverse transcriptase. Proc. Nat/. Acad. Sci. USA 76,731-735. BOTH, G. W., SLEIGH, M. J., Cox, N. J., and KENDAL, A. P. (1983). Antigenic drift in influenza virus H3 hemagglutinin from 1968 to 1980: Multiple evolutionary pathways and sequential amino acid changes at key antigenic sites. J. Viral. 48, 52-60. BURNET, F. M., MCCREA, J. R., and STONE, J. D. (1946). Modification of human red cells by virus action. I. The receptor gradient for virus action in human red cells. Bfit. /. fxp. Patho/. 27, 228-236. DANIELS, R. S., DOUGLAS, A. R., SKEHEL, J. J., WILEY, D. C., NAEVE, C. W., WEBSTER, R. G., ROGERS, G. N., and PAULSON, J. C. (1984). Antigenic analyses of influenza virus hemagglutinins with different receptor-binding specificities. Virology 138, 174-l 77. DEOM, C. M., CATON, A. J., and SCHULZE, I. T. (1986). Host cell-mediated selection of a mutant influenza A virus that has lost a complex oligosaccharide from the tip of the hemagglutinin. Proc. Nat/, Acad. Sci. USA 83, 3771-3775. FENTON, R. O., CLARK, A., and POTTER, C. W. (1981). Immunity to influenza in ferrets. XIV. Comparative immunity following infection or immunization with live or inactivated vaccine. Brit. J. fxp. Patho/. 62,297-307. KIDA, H., BROWN, L. E., and WEBSTER, R. G. (1982). Biological activity of monoclonal antibodies to operationally defined antigenic regions

VARIATION

OF

INFLUENZA

395

on the hemagglutinin molecule of A/Seal/Massachusetts/l/80 (H7N7) influenza virus. Virology 122, 38-47. KOHLER, G., and MILSTEIN, C. (1976). Derivation of specific antibodyproducing tissue culture and tumor lines by cell fusion. Eur. 1. Immunol. 6, 51 l-51 9. MADIN, S. H., and DARBY, N. B. (1958). Established kidney cell lines of normal adult bovine and ovine origin. Proc. Sot. fxp. Biol. Med. 98, 574-576. MAXAM, A. M., and GILBERT, W. (1977). A new method for sequencing DNA. Proc. Nat/. Acad. Sci. USA 74, 560-564. OXFORD, J. S., ABBO, H., CORCORAN, T., WEBSTER, R. G., SMITH, A. J., GRILLI, E., and SCHILD, G. C. (1983). Antigenic and biochemical analysis of field isolates of influenza B virus: Evidence for intraand inter-epidemic variation. 1. Gen. Viral. 64, 2367-2377. OXFORD, J. S., CORCORAN, T., KNOT, R., BATES, J., BARTOLOMEI, O., MEUOR, D., NEWMAN, R. W., ROBERTSON, J., WEBSTER, R. G., and SCHILD, G. C. (1986). Serological studies with influenza A (Hl N 1) viruses cultivated in eggs or mammalian cells. Bull. WHO, in press. PATTERSON, S., and OXFORD, I. S. (1986). Analysis of antigenic determinants on internal and external proteins of influenza virus and identification of antigenic subpopulations of virions in recent field isolates using monoclonal antibodies and immunogold labelling. Arch. Virol. 88, 189-202. ROBERTSON, J. S., NAEVE, C. W., WEBSTER, R. G., BOOTMAN, J. S., NEWMAN, R., and SCHILD, G. C. (1985). Alterations in the hemagglutinin associated with adaptation of influenza B virus to growth in eggs. Virology 143, 166-174. ROGERS, G. N., PAULSON, J. C., DANIEL% R. S., SKEHEL, J. J., WILSON, I. A., and WILEY, D. C. (1983). Single amino acid substitutions in influenza hemagglutinin change receptor binding specificity. Nature (London) 304,76-78. SCHILD, G. C.. OXFORD, J. S., DE JONG, J. C., and WEBSTER, R. G. (1983). Evidence for host-cell selection of influenza virus antigenic variants. Nature (London) 303, 706-709. SCHILD, G. C., OXFORD, I. S., and WEBSTER, R. G. (1984). Evidence for host Cell selection of antigenic variants of influenza A (Hl Nl) and B virus: Antigenic differences between the hemagglutinins of viruses grown in eggs and mammalian cells. In “The Molecular Virology and Epidemiology of Influenza” (Sir Charles H. Stuart-Harris and C. W. Potter, Eds.), pp. 163-l 74. Academic Press, Orlando/ New York/London. SIX, H. R., WEBSTER, R. G., KENDAL, A. P., GLEZEN, W. P., GRIFFIS, G., and COUCH, R. B. (1983). Antigenic analysis of H 1 N 1 viruses isolated in the Houston metropolitan area during four successive seasons. Infect. Immun. 43, 453-458. SKEHEL, 1. J., DANIELS, R. S., DOUGLAS, A. R., and WILEY, D. C. (1983). Antigenic and amino acid sequence variations in the hemagglutinin of type A influenza viruses recently isolated from human subjects. Bull. WHO 61, 671-676. STADEN, R. (1982). Automation of the computer handling of gel reading data produced by the shotgun method of DNA sequencing. Nucleic Acids Res. 10,4731-4751, WILEY, D. C., WILSON, I. A., and SKEHEL, 1. J. (1981). Structural identification of the antibody-binding sites of Hong Kong influenza hemagglutinin and their involvement in antigenic variation. Nature (London) 289, 373-378. WILSON, I. A., SKEHEL, I. J.. and WILEY, D. C. (1981). Structure of the hemagglutinin membrane glycoprotein of influenza virus at 3 A resolution. Nature (London) 289, 366-373.