﻿JOURNAL OF VIROLOGY, Feb. 1992, p. 1129-1138 Vol. 66, No. 2
0022-538X/92/021129-10$02.00/0
Copyright © 1992, American Society for Microbiology
Evolution of the H3 Influenza Virus Hemagglutinin from
Human and Nonhuman Hosts
W. J. BEAN,'* M. SCHELL,2 J. KATZ,' Y. KAWAOKA,l C. NAEVE,1 0. GORMAN,' AND R. G. WEBSTER'
Department of Virology and Molecular Biology' and Department ofBiostatistics,2 St. Jude Children's
Research Hospital, 332 North Lauderdale, P.O. Box 318, Memphis, Tennessee 38101
Received 1 July 1991/Accepted 29 October 1991
The nucleotide and amino acid sequences of 40 influenza virus hemagglutinin genes of the H3 serotype from
mammalian and avian species and 9 genes of the H4 serotype were compared, and their evolutionary
relationships were evaluated. From these relationships, the differences in the mutational characteristics of the
viral hemagglutinin in different hosts were examined and the RNA sequence changes that occurred during the
generation of the progenitor of the 1968 human pandemic strain were examined. Three major lineages were
defined: one containing only equine virus isolates; one containing only avian virus isolates; and one containing
avian, swine, and human virus isolates. The human pandemic strain of 1968 was derived from an avian virus
most similar to those isolated from ducks in Asia, and the transfer of this virus to humans probably occurred
in 1965. Since then, the human viruses have diverged from this progenitor, with the accumulation of
approximately 7.9 nucleotide and 3.4 amino acid substitutions per year. Reconstruction of the sequence of the
hypothetical ancestral strain at the avian-human transition indicated that only 6 amino acids in the mature
hemagglutinin molecule were changed during the transition between an avian virus strain and a human
pandemic strain. All of these changes are located in regions of the molecule known to affect receptor binding
and antigenicity. Unlike the human H3 influenza virus strains, the equine virus isolates have no close relatives
in other species and appear to have diverged from the avian viruses much earlier than did the human virus
strains. Mutations were estimated to have accumulated in the equine virus lineage at approximately 3.1
nucleotides and 0.8 amino acids per year. Four swine virus isolates in the analysis each appeared to have been
introduced into pigs independently, with two derived from human viruses and two from avian viruses. A
comparison of the coding and noncoding mutations in the mammalian and avian lineages showed a significantly
lower ratio of coding to total nucleotide changes in the avian viruses. Additionally, the avian virus lineages of
both the H3 and H4 serotypes, but not the mammalian virus lineages, showed significantly greater conservation
of amino acid sequence in the internal branches of the phylogenetic tree than in the terminal branches. The
small number of amino acid differences between the avian viruses and the progenitor of the 1968 pandemic
strain and the great phenotypic stability of the avian viruses suggest that strains similar to the progenitor strain
will continue to circulate in birds and will be available for reintroduction into humans.
Influenza viruses of the H3 serotype appeared in humans
and caused a major pandemic in 1968. Subsequently, it was
found that antigenically related viruses had been present in
ducks and horses for a minimum of 5 years earlier (27). It is
now known that H3 serotype viruses are widely distributed
in waterfowl and have also been associated with occasional
outbreaks of swine influenza. Serologic archaeology studies
have suggested that antigenically related viruses circulated
in humans during the 1890s, leading to the hypothesis that
this and other serotypes may periodically recycle through
the human population (28).
Previous studies with isolates of the human H3 viruses (5,
6, 14, 43) showed the progressive accumulation of mutations
in the prevalent circulating strains and correlated these
mutations with the progressive antigenic changes seen in the
human strains. Antigenic characterization of H3N2 swine
influenza viruses isolates (40, 41) suggested that these vi-
ruses had been derived from early human H3N2 viruses and
had maintained antigenic characteristics that had been lost in
later human virus isolates. Kida et al. (24, 25) analyzed a
series of Asian avian and swine viruses and found some of
the avian H3 hemagglutinin closely related to both the swine
and human H3N2 viruses. The interrelationships of the H3
*
Corresponding author.
equine influenza viruses have been studied by Daniels et al.
(10) and Kawaoka et al. (22).
We initiated this study to define the detailed evolutionary
relationships among the H3 influenza viruses, to determine
the mutational characteristics of the virus in different hosts,
and to investigate the characteristics of the virus strain
proposed to be the progenitor of the human pandemic strain.
MATERIALS AND METHODS
Virus strains and nucleic acid sequencing. The nucleic acid
sequences of 40 HA genes from influenza viruses of the H3
serotype and 9 of the H4 serotype were analyzed in this
study. The sources of the virus isolate sequences are sum-
marized in Table 1. Those not previously published were
obtained from the repository ofSt. Jude Children's Research
Hospital. Virus was grown and purified, and virion RNA was
prepared as described previously (3). Sequencing was either
directly from virion RNA as described previously (4) or from
full-length clones in the pATX vector as described by
Kawaoka et al. (22). Of the sequences taken from the
literature, 17 reported only the nucleotides coding for the
mature peptide. For these strains (Table 1), the sequence
coding for the signal peptide was obtained from virion RNA.
An aligned compilation of the 40 H3 sequences will be
provided on request.
1129
1130 BEAN ET AL.
TABLE 1. Influenza virus strains studied in this analysis
Strain designation GenBank accession no. Reference or source
A/Memphis/12/85 (H3N2)
A/Memphis/2/85 (H3N2)
A/Memphis/6/86 (H3N2)
A/USSR/3/85 (H3N2)
A/Bangkok/1/79 (H3N2)
A/England/321/77 (H3N2)
A/Swine/Ukkel/1/84 (H3N2)
A/Swine/Colorado/1/77 (H3N2)
A/Victoria/3/75 (H3N2)
A/Udorn/307/72 (H3N2)
A/Memphis/102/72 (H3N2)
A/Memphis/1/71 (H3N2)
A/NT/60/68 (H3N2)
A/Aichi/2/68 (H3N2)
A/Duck/Ukraine/1/63 (H3N8)
A/Duck/Hokkaido/5/77 (H3N2)
A/Duck/Hokkaido/9/85 (H3N8)
A/Duck/Hokkaido/10/85 (H3N8)
A/Duck/Hokkaido/33/80 (H3N8)
A/Duck/Hokkaido/8/80 (H3N8)
A/Duck/Hokkaido/7/82 (H3N8)
A/Swine/Hong Kong/126/82 (H3N2)
A/Swine/Hong Kong/81/78 (H3N2)
A/Mallard/New York/6874/78 (H3N2)
A/Duck/Alberta/78/76 (H3N8)
A/Duck/Memphis/928/74 (H3N8)
A/Duck/Hokkaido/21/82 (H3N8)
A/Equine/Uruguay/1/63 (H3N8)
A/Equine/Miami/63 (H3N8)
A/Equine/Algiers/72 (H3N8)
A/Equine/Tokyo/71 (H3N8)
A/Equine/New Market/76 (H3N8)
A/Equine/Fontainebleau/76 (H3N8)
A/Equine/France/73 (H3N8)
A/Equine/Romania/80 (H3N8)
A/Equine/Santiago/1/85 (H3N8)
A/Equine/Kentucky/2/86 (H3N8)
A/Equine/Johannesburg/86 (H3N8)
A/Equine/Tennessee/5/85 (H3N8)
A/Equine/Kentucky/1/87 (H3N8)
M21648
J02092
M73775
M73774
J02172
V02089
J02132
J02135
J02090
J02109
M16737
M16742
M16743
M16739
M16738
M16740
M19056
M19057
M73776
M73771
M73772
M16741
M24718
M24719
M24721
M24720
M24722
M24723
M73773
M24724
M24725
M24727
M24726
M24728
A/Duck/Alberta/28/76 (H4N6)
A/Chicken/Alabama/1/75 (H4N8)
A/Ruddy Turnstone/NJ/47/85 (H4N2)
A/Turkey/Minnesota/833/80 (H4N2)
A/Seal/Massachusetts/133/82 (H4N5)
A/Duck/Czechoslovakia/56 (H4N6)
A/Budgerigar/Hokkaido/1/77 (H4N6)
A/Duck/New Zealand/31/76 (H4N6)
A/Grey Teal/Australia/2/79 (H4N4)
Katz et al. (20)0
Katz et al. (20)0
Katz and Webster (21)0
Zhdanov et al. (48)
Both and Sleigh (5)
Hauptman et al. (17)
This report
This report
Min Jou et al. (29)
Naeve et al. (30)
Sleigh et al. (42)a
Newton et al. (33)
Both and Sleigh (5)
Verhoeyen et al. (43)
Fang et al. (12)
Kida et al. (24)a
Kida et al. (24)a
Kida et al. (24)0
Kida et al. (24)a
Kida et al. (24)a
Kida et al. (24)a
Kida et al. (25)0
Kida et al. (25)0
This report
This report
This report
Kida et al. (24)0
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka et al. (22)
This report
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka et al. (22)
Kawaoka and Webster (23)
Kawaoka et al. (22)
Kawaoka et al. (22)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
Donis et al. (11)
a The published sequence reported only the sequence coding for the mature polypeptide. The sequence of the signal peptide region was determined from viral
RNA as described in Materials and Methods.
Phylogenetic analysis. Phylogenetic analysis was primarily
performed by using the program PAUP (Phylogenetic Anal-
ysis Using Parsimony), version 2.4 (David L. Swofford,
Illinois Natural History Survey, Champaign, Ill.). Searches
for the most parsimonious topologies for the nucleotide and
amino acid sequences of the entire data set (49 taxa) were
done using the MULPARS and global swap options. De-
tailed studies of specified regions of the tree to determine
optimal and alternative topologies were done by using the
branch and bound algorithm with the BBSAVE options.
Phylogenetic analyses of the nucleic acid sequence in which
coding substitutions were weighted more heavily than non-
coding changes was done by appending each amino acid
sequence to its nucleotide sequence and submitting the
combined sequence to maximum parsimony analysis. This
effectively doubled the weight of each coding substitution.
Greater weights for the coding changes were obtained by
applying the WEIGHTS option of PAUP to the characters
representing the appended amino acids. Hypothetical ances-
tral sequences were generated as described by Fitch (13).
Statistical analysis. The amino acid change to nucleotide
change ratio was estimated for the internal and terminal
J. VIROL.
EVOLUTION OF THE INFLUENZA VIRUS H3 HEMAGGLUTININ 1131
branches of the phylogenetic tree for the different host
species by regressing amino acid changes on nucleotide
changes. Because the variance of amino acid changes for a
given branch is proportional to nucleotide changes (binomial
theory), the data were weighted by 1/nucleotide change,
which is a standard method for assigning weights. Differ-
ences in the mutation rates of the HAl and HA2 domains for
different lineages were assessed by using the chi-square test.
Nucleotide sequence accession numbers. The nucleotide
sequences for the strains not previously published are avail-
able from GenBank under accession numbers M73771
through M73776.
RESULTS
The deduced amino acid sequences for the entire coding
regions of the hemagglutinins of the 40 H3 influenza virus
strains and that of a hypothetical avian virus ancestor are
shown in Fig. 1. The 13 equine virus isolates are 1 amino acid
shorter than those from other species, and the alignment
used by Daniels et al. (10) is shown. Of the 566 amino acids,
351 are invariant, and 68 others are invariant in all except
one of the isolates. The most conserved of the amino acids is
tryptophan. Of the 12 tryptophans in the sequence of the
ancestral avian virus, 10 are invariant, followed by tyrosine
(16 of 20), methionine (7 of 9), histidine (8 of 11), and
cysteine (13 of 18). These five conserved amino acids are
also among the least abundant. The least conserved amino
acid is valine. Of 33 valines in the ancestral sequence, 14 are
invariant.
For the estimation of evolutionary relationships among the
virus isolates, only the nucleotide sequence of the coding
region (nucleotides 30 to 1730) was used in this analysis,
because the sequences of the noncoding 3' and 5' ends,
although highly conserved where they are known, are not
available for many of the strains studied. Insertions and
deletions of codons were recorded for this analysis to
assume that each occurred as a single mutational event
rather than as three mutations. The H4 hemagglutinin was
previously shown by Air (1) to be more closely related to the
H3 hemagglutinins than that of any of the other subtypes.
Therefore, nine H4 influenza virus gene sequences were
included in these analyses as an outgroup to provide a
hypothetical origin for the H3 lineage and to provide further
information on the genetic conservation of the avian viruses.
Their alignment with the H3 sequences is that used by Donis
et al. (11).
The shortest path connecting the 49 nucleotide sequences
required 2,817 steps. This pathway split the H3 taxa into
three major lineages: one containing only avian viruses; one
containing avian, human, and swine virus isolates; and one
containing all of the equine virus isolates. Several ap-
proaches were used to test the robustness of the topology.
Since the number of taxa was too great to allow an exhaus-
tive search of all possible topologies, alternative topologies
and shorter paths were searched for by' reanalyzing the data
after dividing the tree into major sections. The data set was
also analyzed by using the amino acid sequences. This
resulted in a tree with the same general topology as the tree
based on the nucleic acid sequences, but the presence of
several short and zero-length internal branches connecting
the avian virus isolates resulted in a large number (>50) of
equally parsimonious solutions. This problem was elimi-
nated by including the entire RNA sequence and applying
greater weight to the coding changes. This allowed the
junctions affected by amino acid changes to be studied while
preserving those supported only by silent mutations. The
coding changes supported a phylogenetic tree differing from
that supported by all nucleotide changes at two ofthe branch
points. These are labeled A and B in Fig. 2 and are detailed
in Fig. 3. One of these (node B) significantly affects the
interpreted origin of the equine lineage and is discussed
below. The other has a minor effect on the Asian avian
lineage near the origin of the human virus but does not affect
the interpretation of the avian-human virus junction. The
only other significant ambiguity was at the joining of the two
1968 human virus strains (Aichi and NT/60) with the rest of
the tree. In the topology shown, a 3-nucleotide, 0-amino-acid
branch joins these two strains with the trunk of the tree. An
equally parsimonious solution branches each separately
from the same point on the trunk. Analysis of the H3
nucleotide data using the Neighbor Joining Method (36) gave
a topology nearly identical to that obtained when all nucle-
otides were used with the maximum parsimony method. The
only difference was that Swine/Ukkel/84 branched from the
main trunk immediately before the branch containing Vic/75
and Swine/Col/77, rather than immediately after this branch.
The significance of this difference is not known.
The remarkable features of the phylogenetic tree are the
close linkage of the human virus lineage to the avian viruses,
the rapid divergence of the human viruses from their origin,
the great separation of the equine viruses from those of other
species, and the strong conservation of the amino acid
sequence in the avian lineages.
Analysis of the avian-human junction. The phylogenetic
tree indicates that the human virus of the H3N2 serotype
originated from a lineage closely related to a series of avian
virus isolates from Asia described by Kida et al. (24). To
identify the changes that occurred in this virus during its
transition from an avian pathogen to a human pandemic
strain, the hypothetical ancestral sequences of the last avian
and first human strains (Fig. 2, nodes A and C) were
determined (13) and their differences were studied in detail.
The transition required 13 nucleic acid changes, 7 of which
affect the amino acid sequence. The locations of these
changes are shown in Fig. 1. One change is not part of the
mature protein but is at a highly variable site in the signal
peptide; the others are all in the HAl peptide and in the head
of the molecule. These locations are detailed in Fig. 4, and
their significance is discussed below.
Rates of accumulation of mutations. The mutation rate of
the human sublineage was estimated by plotting the year of
isolation against the evolutionary path distances from node
A to each isolate (Fig. 5, top). Regression analysis gave
mutation rates of 7.9 nucleotides and 3.4 amino acids per
year for the coding region (1,701 nucleotides, 566 amino
acids). Two swine influenza virus isolates derived from the
human virus lineage (Swine/Col/1/77 and Swine/Ukkel/84)
were included in these calculations, but the calculated
mutation rate is not significantly affected when they are
excluded. Extrapolation of the nucleotide regression lines
gives dates of 1967 and 1965 for nodes C and A, respectively.
Similar analysis of the equine virus lineage (Fig. 5, bot-
tom) gives substitution rates of 3.1 nucleotides and 0.8 amino
acids per year. The date of origin for the lineage (node D,
Fig. 2), extrapolated from the nucleotide regression, is 1952.
The approximately linear relationship between the number
of mutations and the isolation date is valid for the topology
shown in Fig. 2 but not for the alternative topology (Fig. 3,
top right). With the alternative topology, the 1971 and 1972
isolates are farther from the origin (node B) than are the 1986
strains, and the four earliest isolates appear to form a
VOL. 66, 1992
1132 BEAN ET AL. J. VIROL.
. *.1..........20...........40......... 0*........ D
.......
..
*.100.........119
Neie12/85 A ISGAE
KG P LE N R DS EN G NK G
Nem2/85 A I G AKG P LE N R DS EN G NK
Ner,6/86 A I GEKG P LE N R DS EN G NK I
LUSSR/3/85 A ISYVTE G L N R DS EN G NE
Bangkok/79 A I GN G L N R DS EN G NK
Erotanid/ A I QVLA NG A L N R DS EN G NK
Sw/Ukket/84 LA!I VIG G G L NFNAND D NK T I
Sw/CoL/77 AlI A G L N D IN G NK
Victoria/76 A I V A G L N IN G NE
UdDrn/72 A I VLG FG L N I G N
Neni/172 A I VLG FG L N I G N L
Nei1171 ANHI VIGYG L N I G N H
NT/60/68 A I 16 G N L I N
Aichi/68 A I LG G L IN
Dk/Nok/5/77 A A I VG G Y N L P N
Dk/Hok/9/85 A I V G Y L N
DkIHokI10/85 A 1 G 1 N
Dk/Hok/33/80 A I G G L N
Dk/Hok/8/80 A I G G N E
Dk/Hok/7/82 A I G 6 N
S14VHK/126/82 A I G G E V N S
Sw/HK/81 /78 A I G GT E I N
Dk/Ukr/6.3 V A IS T G C N R N N I
Avian Ancestor NKTI IVLSYFFCLAFS DLPENDNSTATLCLGHHAVPNGTIVETITD0QIEVTNATELVQSSSTGKICNNPHRILDGIRDCTLIDALLGDPHCDVF0DETWDLFVERSKAiFSNCYPYDVPDYASLRSLVASSGTLE
Nat/NY16874/78 C S N N Y S
Dk/Atb/7B/76 L YS SN Y S D
Dk/Nsu928/74 L YS SN Y S
Dk/Hok/21/82 C C YS N Y S S
EqIVUru/63 TTI ILLT.HWVHSONIGGKN A L T Y V N N YGN I S I
Eo/Niami/63 TT ILLT.HWHN8NNTGG N A L T Y V N N Y Y N I S
EqtAtg/72 T IILLI.HWHSOI IS N A LL T T Y V EN N YE
K S I
Eca/To/71 TTI ILLT.HW.NSQI I N A LL T SY V EN N YEN S N L I
Ea/Neid4kt/79 TI ILLT.HWVYSQN ISG N A L T I Y V N N4 A YN I S I I
Ea/Fon/79 TI ILLT.HWVYSQN TSG N A L T I Y V N N YN I S I I
EqlFrance/73 TI ILLT.HNVDSOhISG N A L T I Y V N N YN I S I I
EqRina/80 TI ILLT.HWVYSONTSGCN A L T I Y V N N YN I S I I
Eo/San/85 TI ILLT.HWVYSWNTSG N A L I I YV N NV YN I1ST I I
Eqfenuky/86 TI ILLT.HWVYSQNTSG N A L I I SY V N N YN I S S I I
E/Jdh/8 TI ILLT.HWVYSONTSG N A L I I SY V N N YN I S I I
enn/Te 86 TI ILLT.HWVYSONTSG N A L I I SY V N N YN I S I I
E uEetcky/87 T ILLT.HWYSON TSG N A L I I SY V N N YN I S I I
N1285 N N S Y SVNS Y EEKA GK DE R E V P LS IL ITEK I
Nmm N N S Y SVNS YE EEKA GK DE R E V LS IL STE I
N686 N N S Y SVNS YE EYEA GE EE R E V LS I L T C I
USSR N N LS Y SVNS YEEEK GK DE REK E V LS T I LRT I
N N S Y SDNS YEEEK6 DE R E LS L I
N
N Y DNS Y E DKC E E V LS I L I
U~L NDN D S Y N N S AR E E LS I HNR
SicoL N N S DN Y a S DE 0 E EI V V LSV Y
Vic76 N N S 0 Y Q S DE E E V LS I
Udomn S D Y 0 5 E LS I
Nem72 N L D Y D S E SI
Neu71 G E S L S
NT/60 G S F LS
Aichi G I S LS
Dk577 D S
Dk985 N G V
OklO N V
0k33 F G V L
Dk880 E RG
Dk782 S
Sw126 S
SbBl N S E N4 A LS I
IDkUkr D N E A P N
Ances FITEGFTWTGVTQNCGSNACKRGPASGFFSRLNWLTKSGSTYPVLNVTMPNWONFOKLYIWGVHHPSTN0EQTNLYVQASGRVTVSTRRS0OTI IPNIGSRPWVRGQSGRISIYWTIVEPGDVLVINSNGNLIAPRGYFEN4RTGK
N4atNY A C GN A V R I
DkAtb V
DkNs 0 V 0 I
0k21 L V A
Mm 14A ~ R SRS DS NS T I NEK E N
E
-,a
14A SR SDS E ST I N E E N
Ec~Lg N4A R SSR SODS E SST I NE E L A
N
o A R SSRS DS S ST V I D E E L L. V N I A
ENI TA R GSRS DS DS T N I NEK E E A N4 V E
Ec*Fon TA RGR SDS NS T N I NEK EL E G IN V
E r TA CR505D NS T I NEK EL E C A IN
14 V
Eom TA CR505D NS T N I N E EL E IN V
TA RCG SODS NS T N I S N KEIE E 6I IN1V V
E TA R G SODS NS T N I S N KEIE E I I V L
EqO TA RCG SODS NS T N I S N KEIE E IN V L
EqT TA R G SODS NS T N I S N KEIE E IN
1 V L
Ec7 TA RCG SODS NS T N I S N EIE E IN V L
FIG. 1. Predicted amino acid sequences for the H3 influenza virus strains. The sequence of a hypothetical avian ancestral strain (node D,
Fig. 2) was reconstructed as described by Fitch (13) and used as a baseline in this figure. For the other sequences, only those that differ from
the ancestral strain are shown. The groups defined by Duck/Ukraine, the Asian avian viruses, and the human viruses are displayed above the
ancestral strain, while the smaller avian virus group, represented by three North American isolates and DkIHokI21/82, is immediately below
the ancestral sequence, followed by the equine viruses. Amino acid changes at the avian-human junction (nodes A and C, Fig. 2) are marked
with asterisks. Not indicated on the figure is the insertion of an additional asparagine at position 8 in the AlVictorial3/75 strain.
separate sublineage with an apparent mutation rate about Phenotypic change and conservation. It is clear from the
twice as fast as that of the other sublineage (4.8 versus 2.5 examination of Fig. 2 that the number of amino acid changes
nucleotide substitutions per year). Either topology requires relative to the number of nucleotide chang'es is lower among
several parallel mutations, and neither can be positively the H3 and H4 avian viruses than among the mammalian
excluded on the basis of sequence information from the viruses. Table 2 shows ratios of amino acid changes versus
currently available virus isolates. nucleotide changes for the human, equine, swine, and avian
Among the avian viruses, no consistent relationship was regions of the tree. The data are also separated into terminal
seen with respect to the date of isolation and the path lengths branches, connecting actual viral sequences with the tree,
from common nodes and thus no meaningful divergence and internal branches, connecting nodes within the tree. The
dates for other interior nodes could be calculated. ratios shown were calculated from the sums of nucleotide
EVOLUTION OF THE INFLUENZA VIRUS H3 HEMAGGLUTININ 1133
...280
.300.
V G SG T Rf
V G S T Rf
G T Rf
G SF Rf
G S
G S
S
G S
G S
G
G
V
V V
V
G S
V
V
VF
V
V
V
V
V V
V V
V V
V V
V V
V L
N V K
V K
K
K
V K
V K
V K
V E
V K
V K
V K
V K
V K
.....320......340. 360.380. 400.
I V L E
I V L E
I V L
IL V T L
l~~~~~~~~~
I I V G
I S
I R
NS
N
N
L V
R
G
K V
K
R I I
I I
RL I
I G RL I
I N I I
I N I I
I N I I
I N I I
I N I I
IR N I I
IR N I I
IR N I I
IR N I I
V
V
y
y
V Y
V Y
V Y
V Y
V Y
V Y
V Y
V Y
V Y
G
R
R
R
R
R
R
R
R
R
420 ..440 ..460 ..480 ...500.520.540.
EL K S G V U K
EL K S G V U K
EL K SG V K
EL K G V K
EL G V K
EL G V K
EL NE K
E G V K
EL G V K
EL G V K
EL G G V K
EL V
EL V
EL E V
EL K E V V S
R V V
R V V
R K
L
METKIDLWSYNADVVALENOHTIDLDSELfEKTRRQLRENA GCFKI YWCDNACIESIRNGTYDHDIYRDEALNRFQIKGVELKSGYKDWILWISfAISCFLLCVWLLG.FIIRWIRCNICI
~~~~~~~~~~CGV
V
R S
A
A
A
R
R
G
G
G
G
G
G
G
G
G
y
T G
G Y
G Y
G Y
G Y
G Y
G Y
G Y
G Y
R G Y
R
R
RS
R
I K
I K
I K
I K
I K
I K
I EK
I K
I K
I K
I K
I K
I K
FIG. 1-Continued.
and amino acid branch lengths for each branch type and
range from 0.46 for the internal human virus isolates to 0.074
for the internal branches of the H4 virus strains. The ratios
of coding to total changes of the human internal branches are
greater than those for the human terminal branches. The
opposite is the case for the equine viruses; none of the
differences among the mammalian virus branches is statisti-
cally significant. However, the ratios for the internal
branches of the avian virus lineages are significantly lower
than those for the terminal branches, and the differences
between the avian and mammalian branches are also signif-
icant.
To allow examination of the distribution of coding and
noncoding mutations on the H3 hemagglutinin polypeptide,
the hypothetical ancestral sequences of the H3 strains were
reconstructed as described by Fitch (13) and each variable
position in the RNA sequence was examined. Figure 6
shows each mutation plotted according to its position in the
protein sequence, its host lineage, and whether it is coding or
noncoding.
The paucity of coding changes in the avian virus lineage
and their concentration between amino acids 50 and 300 in
the human strains are readily apparent. The partition of the
coding and noncoding changes between HAl and HA2 in the
various species and branch types of the H3 strains was
analyzed in detail (Table 3). The coding changes for each of
the mammalian lineages show a significant deviation from a
random distribution, whereas the mutations in the avian
virus lineages do not. No coding regions were found notably
lacking in silent mutations.
VOL. 66, 1992
M1285
M285
VA86
USSR
Eno7
Sicol
Vic76
UdDrn
Nme7l
NT/60
Aichi
Dk577
Dk985
DklO
Dk33
Dk880
Dk782
Sw126
SkB1
DkUkr
Anres
MalNY
DkAtb
DkIe
Dk2l
Etru
E lg
E dM
E on
Er
E a
M1285
M285
M686
USSR
Erm7
Vic76
UdoM
Mes72
Ni2
Men7l
NT/60
Aichi
Dk577
Dk985
DklO
Dk33
Dk8B0
Dk782
Sw126
Sw61
Dkdkr
Arees
NaLNY
DkALb
DkMan
Dk2l
D210
Ecron
EcFr
Ec tan
EKY'
N
I A
D EL
EL
EL
EL
EL
EL
EL T
EL
EL
EL
EL
EL
EL
EL
EL
EL
n
SSIMRSDAPIDTCISECITPNGSIPNDKPFQNVNKITYGACPKYVKQSTLKLATGMRNVPEKQTRGLFGAIAGFIENGWEGMIDGWYGFRHQW.3EGTGQAADLKSTQAAIDQINGKLNRVIEKTNEKFHOIBCEFSEVEGitICDL
1134 BEAN ET AL.
D
131.35
1594 MllUl21285
2M 17ee9128
_Vlc76
C 30. 1
J. SW/COU77
A > 2 ;J~~~~2do.Vl72
23,
\2 22NT168
Dk"115177
_<k0/Hok/9185
13,1 20 7 3DsHokaltIHok/10195
2
k1/Hokl[/1o1338
tO3, 9 | k/H!Wl
ok/712682
v | S4, tS DltRU /" 8WlHOkl8V78~~~Hokt2WB
,01 95.13 20 aW
dlNY16B74178
| 4,7 D I llm9874
1 7 _ 1kHk2U82
167.26
iY/SO
360.155
;oh/S 15, a
TNIU 8, I ~~~~~~~~~~~~~CSIAl&175
y/8I7 23,0 RTINJ/85
82,45
|
78, la.
| 14 Dk/AbI76
82,5 33,6 -TJVoNMISO
laSIM"s/82
_ 54S4
70, 21
PBud/Hokl77
63, 6 Dk/Cz/56
102.9 _ A -r_-IGTIAus/79
15.5
I et69
9 k/NZ/70
FIG. 2. Hypothetical phylogenetic tree joining the H3 and H4 influenza A virus hemagglutinin genes. This proposed phylogeny requires
2,122 nucleotide changes and 571 amino acid changes. The number of nucleotide changes and amino acid changes are indicated on each
branch. Branches with one number have no amino acid changes. Alternative joinings for two ambiguous junctions (A and B) are detailed in
Fig. 3. The proposed first human virus is indicated as C. The hypothetical avian ancestral strain used as a baseline in Fig. 1 is indicated as
D.
DISCUSSION
FIG. 3. Alternative topologies for two ambiguous junctions of
the phylogenetic tree. Differences between the most parsimonious
trees calculated from the nucleotide sequences and slightly longer
trees requiring fewer amino acid changes are shown. (Top) Equine-
avian junction (Fig. 2, node B). The shortest tree is shown on the
right. With this topology, the four earliest equine viruses share a
common divergence from the main lineage and require 2,817 nucle-
otide changes for the 49 taxa. With the topology on the left,
Eq/Tokyo/71 and Eq/Algeria/72 branch from the main lineage before
the two 1963 isolates. This topology requires two fewer amino acid
changes but five additional nucleotide changes. The topology on the
left is used in Fig. 1 and was chosen because it provides a consistent
relationship between the date of virus isolation and distance from
the origin ofthe equine virus lineage. (Bottom) Four avian taxa near
the avian-human junction (Fig. 2, node A). The joining on the right
requires three additional nucleotide substitutions but one less amino
acid substitution. The topology on the left was used in Fig. 1.
This study was initiated to define the source and charac-
teristics of the progenitor of the hemagglutinin of the current
H3 human viruses and to provide a better understanding of
the genetic interrelationships and mutational constraints of
the virus strains in different hosts. The H3 hemagglutinin
phylogeny calculated from the sequence data clearly shows
that there is a remarkably close relationship among the first
human H3 viruses and some of the virus strains still circu-
lating in avian species. The results indicate that the human
hemagglutinin gene was very recently introduced into the
virus infecting humans and that it underwent only a few
mutations in its transition from an avian virus to a human
pandemic strain. One of these mutations (position 226) has
been previously implicated in host specificity (30). Of the
others, three (positions 62, 144, and 193) are among the most
variable in the molecule, each having three or four different
amino acids at the site within the human lineage. These
locations have been shown to be affected by antigenic drift
or host range selection (5, 6, 20, 21, 44). One of the sites
(position 193), which changes from Asn to Ser during the
transition, later reverts to the ancestral form. The changes at
these positions may not have been part of the adaptation of
the avian virus to humans but may have been the early stages
of antigenic drift that occurred before the virus had been
detected.
The calculated date of 1965 for the introduction of the H3
gene into a human virus is made on the basis of assumptions
that the mutation rate ofthe virus has been constant, that the
J. VIROL.
33.7
E41AW72
20.9
EqIT0171
Eq/Uns/63
A a
EqiMk"6
NW79
EqlFonf79
B 47.14 EqIFF173
10. Eq/RMVSO
1.1
4
EqIS4
Eq/N
EqIJ
EVOLUTION OF THE INFLUENZA VIRUS H3 HEMAGGLUTININ
I
193 Asn-Ser
-) ,, 144 Ala-Gly
---81 Asp-Asn
< VJi~~ -- - - 62 Arg-lle
N -- 92 Asn-Lys
FIG. 4. Amino acid changes in the hemagglutinin proposed to
have been associated with the early adaptation of the avian hemag-
glutinin to humans. The figure shows the human H3 hemagglutinin
structure, as determined by Wilson et al. (46). The locations of the
six amino acid changes required between nodes A and C of Fig. 2 are
indicated. The position number is followed by the amino acids
present in the avian and human forms, respectively.
introduction occurred at node A on the phylogenetic tree,
and that there were no intermediate hosts. The assumption
that the introduction occurred at node A rather than at node
C (Fig. 2) is based on the high ratio of coding to noncoding
changes in the branch linking nodes A and C which is
characteristic of the virus in humans rather than in birds.
The evolutionary record provides no evidence for an
intermediate host between the avian and human sublineages,
but it cannot be ruled out that one may have existed, and the
possibility of swine intermediaries in the generation of
human influenza virus strains has been previously consid-
ered (15, 19, 26, 31, 32, 37-39, 45). Of the four swine virus
isolates included in this analysis, two were introduced into
swine from the human virus lineage and two are recent
introductions from avian viruses. This finding and previous
evidence for the transmission of virus between birds and
swine (19, 34), as well as documented transmissions of
influenza virus from swine to humans (18, 35), leaves open
the possibility that a swine intermediate could have been
involved. The ratios of the coding changes to mutations in
the four swine virus isolates included in this study were
similar to those of the human strains. Thus, the data are
consistent with a transfer of the hemagglutinin from the
avian reservoir into swine at node C and then into humans at
node A. Regardless of the initial mammalian host, if the
amino acid substitutions during this period were selected by
antibody pressure, the virus is likely to have been the
Human
Nucleotides
150 Amino Acids
_ 100
0-
to
o50
U
._a
0
S
E 0
0
0
0 100
0
z
50
190
1950
7.9 Substitutionslyear
3.4 Substitutionslyear
1960 1970 1980 1990
Year of Isolation
FIG. 5. Rates of accumulation of nucleotide and amino acid
changes in the human and equine viruses. Rates were calculated by
the regression of the date of isolation and the total branch distance
from nodes A and B for the human and equine virus isolates,
respectively (Fig. 2). The amino acid and nucleotide distances are
plotted as open and closed circles, respectively.
predominant strain in a population, sufficiently large to have
maintained the virus but limited enough to have escaped
detection.
In contrast to the progressive changes of both the nucle-
otide and amino acid sequences of the mammalian virus
lineages, the avian viruses show far less variation and no
clear relationship between the position on the phylogenetic
tree and the date of isolation. Additionally, most of the
coding changes in the avian lineages have occurred in the
terminal branches, whereas in the mammalian lineages the
terminal and internal branches have similar ratios of coding
and noncoding changes. This fundamental difference in the
TABLE 2. Ratios of coding changes to total nucleotide changes
in the terminal and internal branches of the avian and
mammalian evolutionary lineages
No. of Amino acid Nucleotide Ratio
Branch typea branches changes changes (AAC/NC)
(AAC) (NC)
Avian (H3), int. 12 39 423 0.092b C
Avian (H3), ter. 11 82 334 0.246c
Avian (H4), int. 7 28 380 0.074bc
Avian (H4), ter. 9 80 338 0.234c
Human, int. 13 77 168 0.458
Human, ter. 12 43 120 0.358
Equine, int. 11 46 132 0.348
Equine, ter. 13 59 143 0.413
Swine, ter. 4 55 126 0.436
a Phylogenetic tree branches (Fig. 2). ter., terminal branches connecting
virus isolates to the tree; int., internal branches connecting nodes on the tree.
b Significantly less than corresponding terminal branches (P = 0.01).
' Significantly less than corresponding mammalian branches (P = 0.01).
Equine 0
Nucleotides 3.1 Substitutionslyear
- Amino Acids 0.8 Substitutionslyear
* 0-
0
0
- 0
-
0
0
VOL. 66, 1992 1135
1136 BEAN ET AL.
1-
50-
100-
150-
200-
250-
>
450-
500- -
550-
Human
Ter. Int.
Avian
Ter. Int.
*,4
I
i
-1
.8
I
A
I
1
m2
I
I
I
a
P-- I
J
a
I
a
I
i
-i
I
I
-3
I
I
I
i
r I
Equine
Ter. Int.
-1 I
J. VIROL.
Swine Av-Eq
Linlc
ii
a
11
a
1:
4:
I
I
I
aF
FIG. 6. Distribution of coding and noncoding mutations of the
phylogenetic tree for the H3 hemagglutinin. Separate plots were
constructed for the terminal and internal branches for human, avian,
and equine lineages, for terminal branches for swine, and for the link
between the avian and equine lineages. The numbers on the vertical
axis represent amino acid positions 1 to 556 in the hemagglutinin
molecule, with the initiating methionine as 1. Dashes on the left and
right of the vertical axes represent noncoding and coding mutations,
respectively.
TABLE 3. Comparison of the occurrence of mutations in the
HAl and HA2 domains of different host lineages
No. of mutations
Branch typea P valueb
HAl HA2
Human, TN 44 22 0.09
Human, TC 32 6 0.0004
Human, IN 58 23 0.004
Human, IC 51 10 <0.0001
Avian, TN 146 93 0.08
Avian, TC 53 28 0.09
Avian, IN 174 116 0.12
Avian, IC 20 8 0.11
Equine, TN 47 24 0.09
Equine, TC 41 12 0.001
Equine, IN 44 39 0.44
Equine, IC 35 10 0.003
Swine, TN 34 27 0.84
Swine, TC 42 9 <0.0001
Avian-equine, IN 137 95 0.39
Avian-equine, IC 32 8 0.002
a Branch type (from Fig. 6). TN, terminal branches, noncoding mutations;
TC, terminal branches, coding mutations; IN, internal branches, noncoding
mutations; IC, internal branches, coding mutations.
b The P values were determined by the chi-square test. Significant results
indicate differences in the occurrence of mutations in the two domains.
evolution of the avian viruses suggests that their long-term
survival favors those that have maintained the original
phenotype. If survival favors those that have not changed,
then virus populations in environments that undergo rela-
tively few replication cycles would be more likely to yield
progeny that do not have deleterious mutations. Those
replicating in other environments or mutants in the original
population might have a temporary selective advantage in a
particular host or environment, but the accumulation of
mutations in these subpopulations would be deleterious in
other circumstances. Thus, the original population (perhaps
often in a very small minority) would have a selective
advantage as hosts or environmental conditions change.
The striking difference in the topologies of the H3 and H4
avian lineages in comparison with the human virus lineage is
apparently due to a heavy positive selective pressure on
viruses replicating in humans that is not seen when replicat-
ing in birds. Fitch et al. (14) have studied the unusual
"cactuslike" topology of the human influenza A viruses and
have shown that the proportion of amino acid changes
affecting antigenic regions of the hemagglutinin (HA1) is
greater on the main trunk than on the side branches of the
tree. This provided convincing evidence for positive Darwin-
ian selection on the HAl molecule, mediated by immune
pressure. They calculated the age of the nonsurviving side
branches (the distance from the main lineage trunk to the
branch tips) to be 1.6 years. The present study, using a very
different data set and the entire coding sequence, gives a
very similar value of 1.5 years. The increased mutation rate
seen in the hemagglutinin when transferred from an avian to
a mammalian host is paralleled by smaller increases in the
mutation rates of other genes (9, 14, 16).
The topology of the avian virus lineages resembles that
described for the human influenza C virus (7, 8, 47), with
multiple cocirculating lineages and little correlation between
tree position and date of isolation. In both cases this may
reflect a long-established equilibrium between the virus and
the host. The influenza B viruses show a regular relationship
between tree position and isolation date but a slower rate of
i
i
I
z
a
!8
:1
a
-1
I
EVOLUTION OF THE INFLUENZA VIRUS H3 HEMAGGLUTININ 1137
change than the human influenza A strains and cocirculating
branches that survive much longer than the short side
branches of the influenza A viruses (14, 47). Air et al. (2)
have analyzed the proportion of silent and nonsilent muta-
tions in the human A and B viruses and have proposed that
the evolution of the B viruses is not primarily driven by
immune selection.
In the present study, the only internal avian branches with
a large proportion of coding changes on the phylogenetic tree
are the long links connecting the H3 equine viruses and the
H4's with the rest of the taxa. However, these lengths and
their connection points to each other must be interpreted
with caution. The large number of silent mutations, particu-
larly in the link to the H4 viruses, make it likely that the total
number of mutations is underestimated, since any of them
could have mutated multiple times. Lacking evolutionary
intermediates and with no information on what selective
pressures or time periods were involved in the separation of
the HA subtypes, we cannot estimate the actual lengths of
these branches with any certainty.
The stability of the avian strains suggests that the virus has
reached an adaptive optimum in birds that has not been
achieved in humans. There are several possible explanations
for this difference. One obvious possibility is that because
this gene was only recently introduced into the human virus,
it has not had time to reach an optimal configuration.
Another possibility is that when the virus is in humans there
is sufficient flexibility in the interaction of the hemagglutinin
with the host that no single configuration is optimal and that
mutations selected by antibody may not significantly affect
other properties or functions of the protein. This implies that
in an immunologically naive population, the viruses derived
by continued immunological selection would not be at a
selective disadvantage if placed in competition with the
original strain. A third possibility, and perhaps the most
likely, is that the H3 hemagglutinin was already in its optimal
configuration after its initial adaptation as a human patho-
gen. Subsequent mutations in response to antibody pres-
sures in the human population, while essential for the
continued survival of the virus lineage, would put the virus
at a disadvantage ifforced to compete with the original strain
in an immunologically naive population. This question re-
mains to be tested directly but is consistent with the ob-
served periodic replacement of influenza virus strains in the
human population with another serotype and the hypothe-
sized recycling ofinfluenza virus strains (28). If the recycling
of influenza virus serotypes is to occur, it requires a mech-
anism for maintaining the virus while it is not circulating in
humans, and the avian virus reservoir clearly provides one.
The highly conserved phenotype of the H3 hemagglutinin in
birds suggests that strains very similar to the progenitor of
the 1968 pandemic will continue to circulate in birds and will
be available for reintroduction into mammalian hosts in the
future.
ACKNOWLEDGMENTS
We thank Raphael Onwuzuruigbo and Evelyn Stigger for techni-
cal assistance.
This work was supported by Public Health Service grants Al-
20591, AI-08831, AI-29680, AI-29599, and AI-27497 from the Na-
tional Institute of Allergy and Infectious Diseases, by Cancer Center
Support Grant (CORE) CA-21765, and by the American Lebanese
Syrian Associated Charities.
REFERENCES
1. Air, G. M. 1981. Sequence relationships among the hemagglu-
tinin genes of 12 subtypes of influenza A viruses. Proc. Natl.
Acad. Sci. USA 78:7639-7643.
2. Air, G. M., A. J. Gibbs, W. G. Laver, and R. G. Webster. 1990.
Evolutionary changes in influenza B are not primarily governed
by antibody selection. Proc. Natl. Acad. Sci. USA 87:3884-
3888.
3. Bean, W. J., G. Sriram, and R. G. Webster. 1980. Electropho-
retic analysis of iodine-labeled influenza virus RNA segments.
Anal. Biochem. 102:228-232.
4. Bean, W. J., S. C. Threlkeld, and R. G. Webster. 1989. Biologic
potential of amantadine-resistant influenza A virus in an avian
model. J. Infect. Dis. 159:1050-1056.
5. Both, G. W., and M. J. Sleigh. 1981. Conservation and variation
in the hemagglutinins of Hong Kong subtype influenza viruses
during antigenic drift. J. Virol. 39:663-672.
6. Both, G. W., M. J. Sleigh, N. Cox, and A. P. Kendal. 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. Virol. 48:52-60.
7. Buonagurio, D. A., S. Nakada, W. M. Fitch, and P. Palese. 1986.
Epidemiology of influenza C virus in man: multiple evolutionary
lineages and low rate of change. Virology 153:12-21.
8. Buonagurio, D. A., S. Nakada, U. Desselberger, M. Krystal, and
P. Palese. 1985. Noncumulative sequence changes in the hem-
agglutinin gene of influenza C virus isolates. Virology 146:221-
232.
9. Buonagurio, D. A., S. Nakada, J. D. Parvin, M. Krystal, P.
Palese, and W. M. Fitch. 1986. Evolution of human influenza A
viruses over 50 years: rapid, uniform rate of change in NS gene.
Science 232:980-982.
10. Daniels, R. S., J. J. Skehel, and D. C. Wiley. 1985. Amino acid
sequences of haemagglutinins of influenza viruses of the H3
subtype isolated from horses. J. Gen. Virol. 66:457-464.
11. Donis, R. O., W. J. Bean, Y. Kawaoka, and R. G. Webster. 1989.
Distinct lineages of influenza virus H4 hemagglutinin genes in
different regions of the world. Virology 169:408-417.
12. Fang, R., W. Min Jou, D. Huylebroeck, R. Devos, and W. Fiers.
1981. Complete structure of A/Duck/Urkraine/63 influenza hem-
agglutinin gene: animal virus as progenitor of human H3 Hong
Kong 1968 influenza hemagglutinin. Cell 25:315-323.
13. Fitch, W. M. 1971. Toward defining the course of evolution:
minimum change for a specific tree topology. Syst. Zool.
20:406-416.
14. Fitch, W. M., J. M. E. Leiter, L. Xingqiang, and P. Palese. 1991.
Positive Darwinian evolution in human influenza viruses. Proc.
Natl. Acad. Sci. USA 88:4270-4274.
15. Gorman, 0. T., W. J. Bean, Y. Kawaoka, I. Donatelli, Y. Gou,
and R. G. Webster. 1990. Evolution of the influenza A virus
nucleoprotein genes: implications for the origins of HlNl hu-
man and classical swine viruses. J. Virol. 65:3704-3714.
16. Gorman, 0. T., W. J. Bean, Y. Kawaoka, and R. G. Webster.
1990. Evolution of the nucleoprotein gene of influenza A virus.
J. Virol. 64:1487-1497.
17. Hauptman, R., L. D. Clarke, R. C. Mountford, H. Bachmayer,
and J. W. Almond. 1983. Nucleotide sequence of the haemag-
glutinin gene of influenza virus A/England/321M7. J. Gen. Virol.
64:215-220.
18. Hinshaw, V. S., W. J. Bean, R. G. Webster, and B. C. Easterday.
1978. The prevalence of influenza viruses in swine and the
antigenic and genetic relatedness of influenza viruses from man
and swine. Virology 84:51-62.
19. Hinshaw, V. S., R. G. Webster, W. J. Bean, J. C. Downie, and
D. A. Senne. 1983. Swine influenza-like viruses in turkeys: a
potential source of virus for humans? Science 220:206-208.
20. Katz, J. M., C. W. Naeve, and R. G. Webster. 1987. Host
cell-mediated variation in H3N2 influenza viruses. Virology
156:386-395.
21. Katz, J. M., and R. G. Webster. 1988. Antigenic and structural
characterization of multiple subpopulations of H3N2 influenza
virus from an individual. Virology 165:446-456.
22. Kawaoka, Y., W. J. Bean, and R. G. Webster. 1989. Evolution
VOL. 66, 1992
1138 BEAN ET AL.
of the hemagglutinin of equine H3 influenza viruses. Virology
169:283-292.
23. Kawaoka, Y., and R. G. Webster. 1989. Origin of the hemagglu-
tinin on A/Equine/Johannesburg/86 (H3N8): the first known
equine influenza outbreak in South Africa. Arch. Virol. 106:
159-164.
24. Kida, H., Y. Kawaoka, C. W. Naeve, and R. G. Webster. 1987.
Antigenic and genetic conservation of H3 influenza in wild
ducks. Virology 159:109-119.
25. Kida, H., K. F. Shortridge, and R. G. Webster. 1988. Origin of
the hemagglutinin gene of H3N2 influenza viruses from pigs in
China. Virology 162:160-166.
26. Kundin, W. D. 1970. Hong Kong A2 influenza virus infection
among swine during a human epidemic in Taiwan. Nature
(London) 228:857.
27. Laver, W. G., and R. G. Webster. 1973. Studies on the origin of
pandemic influenza. III. Evidence implicating duck and equine
influenza viruses as possible progenitors of the Hong Kong
strain of human influenza. Virology 51:383-391.
28. Masurel, N., and W. M. Marine. 1973. Recycling of Asian and
Hong Kong influenza A virus hemagglutinins in man. Am. J.
Epidemiol. 97:44 49.
29. Min Jou, W., M. Verhoeyen, R. Devos, et al. 1980. Complete
structure of the hemagglutinin gene from the human influenza
A/Victoria/3/75 (H3N2) strain as determined from cloned DNA.
Cell 19:683-696.
30. Naeve, C. W., V. S. Hinshaw, and R. G. Webster. 1984.
Mutations in the hemagglutinin receptor-binding site can change
the biological properties of an influenza virus. J. Virol. 51:567-
569.
31. Nakajima, K., E. Nabusawa, and S. Nakajima. 1984. Genetic
relates between A/Swine/Iowa/15/30 and human influenza vi-
ruses. Virology 139:194-198.
32. Nerome, K., M. Ishida, A. Oya, and K. Oda. 1982. The possible
origin of HlNl (HswlNl) virus in the pig population of Japan
and antigenic analysis of isolates. J. Gen. Virol. 62:171-175.
33. Newton, S. E., G. M. Air, R. G. Webster, and W. G. Laver.
1983. Sequence of the hemagglutinin gene of influenza virus
A/Memphis/1/71 and previously uncharacterized monoclonal
antibody-derived variants. Virology 128:495-501.
34. Pensaert, M., K. Ottis, J. Vandeputte, M. M. Kaplan, and P. A.
Bachman. 1981. Evidence for the natural transmission from wild
ducks to swine and its potential importance for man. Bull
W.H.O. 59:75-78.
35. Rota, P. A., E. P. Rocha, and M. W. Harmon. 1989. Laboratory
characterization of a swine influenza virus isolated from a fatal
case of human influenza. J. Clin. Microbiol. 27:1413-1416.
36. Saitu, N., and M. Nei. 1987. The neighborjoining method: a new
method for reconstructing phylogenetic trees. Mol. Biol. Evol.
4:406-425.
37. Scholtissek, C., H. Burger, 0. Kistner, and K. F. Shortridge.
1985. The nucleoprotein as a possible major factor in determin-
ing host specificity of influenza H3N2 viruses. Virology 147:
287-294.
38. Scholtissek, C., W. Rohde, V. Von Hoynigen, and R. Rott. 1978.
On the origin of the human influenza virus subtypes H2N2 and
H3N2. Virology 87:13-28.
39. Schultz, U., W. M. Fitch, S. Ludwig, J. Mandler, and C.
Scholtissek. 1991. Evolution of pig influenza viruses. Virology
183:61-73.
40. Shortridge, K. F., A. Cherry, and A. P. Kendal. 1979. Further
studies on the antigenic properties of H3N2 strains of influenza
A isolated from pigs in Southeast Asia. J. Gen. Virol. 44:251-
254.
41. Shortridge, K. F., R. G. Webster, W. K. Butterfield, and C. H.
Campbell. 1977. Persistence of Hong Kong influenza virus
variants in pigs. Science 196:1454-1455.
42. Sleigh, M. J., G. W. Both, P. A. Underwood, and V. J. Bender.
1981. Antigenic drift in the hemagglutinin of Hong Kong influ-
enza subtype: correlation of amino acid changes with changes in
viral antigenicity. J. Virol. 37:845-853.
43. Verhoeyen, M., R. Fang, W. Min Jou, R. Devos, D. Huylebroeck,
E. Saman, and W. Fiers. 1980. Antigenic drift between the
haemagglutinin of the Hong Kong influenza strains A/Aichi/2/68
and A/Victoria/3/75. Nature (London) 286:7771-7776.
44. Wang, M., J. M. Katz, and R. G. Webster. 1989. Extensive
heterogeneity in the hemagglutinin of egg-grown influenza vi-
ruses from different patients. Virology 171:275-279.
45. Webster, R. G., and W. G. Laver. 1975. Antigenic variation of
influenza viruses, p. 296-314. In E. D. Kilbourne (ed.), The
influenza viruses and influenza. Academic Press, Inc., New
York.
46. Wilson, I. A., J. J. Skehel, and D. C. Wiley. 1981. The hemag-
glutinin membrane glycoprotein of influenza virus: structure at 3
A resolution. Nature (London) 289:366-373.
47. Yamashita, M., M. Krystal, W. Fitch, and P. Palese. 1988.
Influenza B virus evolution: cocirculating lineages and compar-
ison of evolutionary pattern with those of influenza A and C
viruses. Virology 163:112-122.
48. Zhdanov, V. M., N. A. Petrov, A. A. Grinev, M. A. Yakhno,
V. A. Ishachenko, Y. A. Gorbunov, I. A. Vtorushina, S. V.
Netesov, S. K. Vaselenko, and L. S. Sandakhchiev. 1989. Primary
structure of influenza A viruses (H3N2) isolated in USSR, 1985.
Vopr. Virusol. 2:155-160.
J. VIROL.
