﻿Minimal molecular constraints for respiratory
droplet transmission of an avian­human
H9N2 influenza A virus
Erin M. Sorrell1
, Hongquan Wan1,2
, Yonas Araya, Haichen Song3
, and Daniel R. Perez4
Department of Veterinary Medicine, University of Maryland, College Park, and Virginia­Maryland Regional College of Veterinary Medicine, 8075
Greenmead Drive, College Park, MD 20742
Edited by Peter Palese, Mount Sinai School of Medicine, New York, NY, and approved March 17, 2009 (received for review January 27, 2009)
Pandemic influenza requires interspecies transmission of an influ-
enza virus with a novel hemagglutinin (HA) subtytpe that can
adapt to its new host through either reassortment or point muta-
tions and transmit by aerosolized respiratory droplets. Two pre-
vious pandemics of 1957 and 1968 resulted from the reassortment
of low pathogenic avian viruses and human subtypes of that
period; however, conditions leading to a pandemic virus are still
poorly understood. Given the endemic situation of avian H9N2
influenza with human-like receptor specificity in Eurasia and its
occasional transmission to humans and pigs, we wanted to deter-
mine whether an avian­human H9N2 reassortant could gain re-
spiratory transmission in a mammalian animal model, the ferret.
Here we show that following adaptation in the ferret, a reassor-
tant virus carrying the surface proteins of an avian H9N2 in a
human H3N2 backbone can transmit efficiently via respiratory
droplets, creating a clinical infection similar to human influenza
infections. Minimal changes at the protein level were found in this
virus capable of respiratory droplet transmission. A reassortant
virus expressing only the HA and neuraminidase (NA) of the
ferret-adapted virus was able to account for the transmissibility,
suggesting that currently circulating avian H9N2 viruses require
little adaptation in mammals following acquisition of all human
virus internal genes through reassortment. Hemagglutinin inhibi-
tion (HI) analysis showed changes in the antigenic profile of the
virus, which carries profound implications for vaccine seed stock
preparation against avian H9N2 influenza. This report illustrates
that aerosolized respiratory transmission is not exclusive to current
human H1, H2, and H3 influenza subtypes.
aerosol  ferrets  contact  pandemic  preparedness
H5, H7, and H9 avian influenza subtypes top the World Health
Organization's (WHO) list with the greatest pandemic poten-
tial. A transition from avian-like 2,3-linked sialic acid (SA2,3)
receptors to human-like 2,6-linked sialic acid (SA2,6) receptors
appears to be a crucial step for avian influenza viruses to replicate
efficiently and transmit in humans (1). An increasing number of
contemporary avian H9N2 viruses contain leucine (L) at position
226 in the hemagglutinin (HA) receptor-binding site (RBS), sup-
porting the preferential binding to SA2,6 receptors and the ability
to replicate efficiently in human respiratory epithelial cells and in
the ferret model, an in vivo model which closely resembles human
airway epithelium and clinical infection (2­5). Since the mid-1990's,
H9N2 influenza viruses have become endemic in poultry through-
out Eurasia and have occasionally transmitted to humans and pigs
(6­8). In addition to possessing human virus-like receptor speci-
ficity, avian H9N2 viruses induce typical human flu-like illness,
which can easily go unreported, and therefore have the opportunity
to circulate, reassort, and improve transmissibility. Seroepidemio-
logical studies in Asia suggest that the incidence of human H9N2
infections could be more prevalent than what has been reported and
possible human-to-human transmission cannot be completely ex-
cluded (9­11). These direct infections with avian H9N2 confirm
that interspecies transmission of H9N2 from avian species to
mammalian hosts occurs and it is not uncommon. Reassortment
between the current human epidemic strain and an avian virus of
a different subtype is postulated to generate the next pandemic
strain. Given the receptor specificity of avian H9N2 viruses and
their repeated introduction into humans, as recent as December
2008 (Vietnam Partnership on Avian and Human Influenza
(PAHI) http://www.avianinfluenza.org.vn/), the opportunity for
their reassortment and/or adaptation for human-to-human trans-
mission is ever present. However the question remains what is
missing for the H9N2 virus to transmit from human-to-human and
possibly lead to the next pandemic.
In our previous study (4), we showed that human virus-like
receptor specificity, specifically leucine (L) at position 226 in the
HA RBS, is critical for direct transmission of avian H9N2 viruses
in ferrets. Creation of an H9N2 avian­human reassortant virus led
to increased replication, direct transmission, and expanded tissue
tropism in ferrets compared to the parental avian H9N2 virus. The
reassortant, 2WF10:6M98, contained the surface genes [HA and
neuraminidase (NA)] of A/guinea fowl/Hong Kong/WF10/99
(H9N2) [WF10] and the internal genes (PB2, PB1, PA, NP, M, NS)
of A/Memphis/14/98 (H3N2) [M98](4). This reassortant however,
lacked the ability to transmit via respiratory droplets despite clinical
signs including high titers in nasal washes and sneezing, indicating
that additional traits are needed. The transmission modes postu-
lated for natural influenza A infections include large droplets, direct
contact, and aerosols with aerosol transmission having obvious
implications for pandemic influenza (12). The ferret is an ideal
model for this study as aerosolization is the main mode of influenza
A transmission in this species (13­16). As a result, we began
adapting this avian­human reassortant, 2WF10:6M98 in ferrets
and after 10 passages achieved respiratory droplet transmission.
Here we show for the first time an avian­human H9N2 reassortant
that can transmit efficiently in respiratory droplets. We have
identified key changes in the surface proteins that are critical for
respiratory droplet transmission and also play important roles in
antigenic variation. Our studies provide valuable information for
pandemic preparedness against H9N2 strains.
Author contributions: D.R.P. designed research; E.M.S., H.W., Y.A., and H.S. performed
research; E.M.S., H.W., H.S., and D.R.P. analyzed data; and E.M.S., H.W., and D.R.P. wrote
the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
Data deposition: The nucleotide sequences of the RCP10 virus, A/ferret/Maryland/P10-
UMD/08 (H9N2), have been deposited at NCBI's GenBank (accession nos. CY036274­
CY036281).
1E.M.S. and H.W. contributed equally to this work.
2Present address: Molecular, Virology, and Vaccines Branch, Influenza Division, Centers for
Disease Control and Prevention, 1600 Clifton Road, Atlanta, GA 30333.
3Present address: Synbiotics Corporation, 8075 Greenmead Drive, College Park, MD 20742.
4To whom correspondence should be addressed. E-mail: dperez1@umd.edu.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0900877106/DCSupplemental.
www.pnas.orgcgidoi10.1073pnas.0900877106 PNAS  May 5, 2009  vol. 106  no. 18  7565­7570
MICROBIOLOGY
Results
Respiratory Droplet Transmission of a H9N2 Avian­Human Reassor-
tant. The generation of a H9N2 avian­human reassortant,
2WF10:6M98, containing the surface genes of A/guinea fowl/Hong
Kong/WF10/99 (H9N2) [WF10] and the internal genes of A/Mem-
phis/14/98 (H3N2) [M98] led to increased replication, direct trans-
mission, and tissue tropism when compared to the parental WF10
virus. Clinical signs displayed were similar to those observed during
infection with the full human M98 H3N2 virus and included high
viral titers in nasal washes and sneezing, yet no transmission to
respiratory droplet contacts occurred (4). To determine the key
components necessary for efficient respiratory droplet transmission
we began adapting the 2WF10:6M98 H9N2 virus in ferrets. Ferrets
were infected intranasally (i.n.) with 106
tissue culture infectious
dose 50 (TCID50) of 2WF10:6M98 (passage 1); nasal washes were
collected 3 days postinfection (p.i.), pooled, and used as the dose for
the following passage of ferrets. Respiratory droplet contact ferrets
were introduced at passages 1 and 2; however, no transmission was
observed. After 9 passages of nasal wash, we arbitrarily tested the
transmissibility of this virus during the 10th passage, herein referred
to as P10. Within 3 days postcontact (p.c.), direct contact ferrets
were shedding virus and were able to transmit to respiratory droplet
contact ferrets by days 4 and 6 p.c. (Fig. 1 A and B). All ferrets,
including respiratory droplet contacts, shed virus up to 6­7 days and
displayed clinical signs, including sneezing and fever, similar to that
of a human virus infection (4) and showed high antibody titers to
the homologous virus (Tables 1 and 2). The transmission phenotype
of the P10 virus was confirmed in additional groups of ferrets (Fig.
1 C and D), consistently resulting in efficient respiratory droplet
transmission and clinical signs. It is necessary to note that in the
second round of experiments, 1 of the 2 direct contact ferrets died
on day 6 p.c. (Fig. 1C); however, postmortem examination was
inconclusive. Our adaptation study shows that our H9N2 avian­
human virus is able to sustain efficient, reproducible respiratory
droplet transmission in ferrets causing an infection similar in
duration and clinical signs to typical human H3N2 strains. These
results suggest that current H9N2 viruses circulating in poultry
require little adaptation in mammals, following reassortment and
acquisition of human internal genes, to cause respiratory droplet
transmission.
Consistent Isolation of Ferret-Adapted P10 H9N2 Virus in Lung Tissue.
The consistency of obtaining respiratory droplet transmission in
multiple rounds of transmission led us to compare tissue tropism of
the P10 virus to the parental 2WF10:6M98 and WF10 viruses (4).
Ferrets were also mock infected with PBS as a negative control.
Tissues were collected on day 5 p.i., homogenized, and virus
titrations performed as previously described (4). While
2WF10:6M98 was able to replicate and expand tissue tropism
compared to WF10, the P10 virus shows over 1.5 log10 higher viral
Fig. 1. Respiratory droplet transmission of H9N2 avi-
an­human reassortant viruses. Ferrets were inocu-
lated with 106 TCID50 of P10 ferret-adapted H9N2 virus
(A and C) or 2RCP10:6M98 reassortant virus (E). Direct
contact ferrets (A, C, and E) and respiratory droplet
contact ferrets for P10 (B and D) and 2RCP10:6M98 (F)
were introduced at 24 h p.i. and nasal washes were
collected daily. Black and white bars represent individ-
ual ferrets. In C, day 6 p.c., the direct contact from the
group represented by the black bars died, as noted by
an asterisk in the bar graph. Titers are expressed as
log10 values of TCID50/mL with the limit of detection at
0.699 log10TCID50/mL.
Table 1. Clinical signs, virus replication, and seroconversion
associated with H9N2 reassortant viruses in infected ferrets
Infected ferrets
Virus
Weight Loss
(%)*
Sneezing
(day of onset)
Serum
(HI titer)
2WF10:6M98 5.1  0.85 2/2 (2, 2) 2560, 2560
P10 4.01  1.2 4/4 (3, 5, 7) 2560, 2560, 2560, 2560
2RCP10:6M98 4.67  1.7 3/4 (5, 6) 2560, 2560, 2560, 2560
RCP10 (A189, G192) 5.0  2.48 4/4 (5, 6) 2560, 2560, 2560, 1280
RCP10 (T189, R192)  3.69  1.43 4/4 (5, 6) 2560, 2560, 1280, 1280
2WF10:6RCP10 1.9  1.0 4/4 (5, 6) 1280, 1280, 2560, 1280
*Average body weight loss is shown as average  standard deviation.
At 2 weeks p.i. convalescent sera was collected and used with the homolo-
gous virus in HI assays to detect anti-H9 antibodies.
Two independent experiments with 2 infected, 2 direct, and 2 respiratory
droplet ferrets each.
7566  www.pnas.orgcgidoi10.1073pnas.0900877106 Sorrell et al.
titers than 2WF10:6M98 (Fig. 2). We also isolated virus from the
brain of the P10 ferrets, suggesting that in addition to improving its
transmissibility phenotype, this virus has the potential to become
more virulent. However, we must note that this study focuses largely
on the molecular features that alter the transmission phenotype of
an H9N2 virus in ferrets. The molecular markers that modulate
virulence of this virus in ferrets are beyond the scope of the present
report and are currently being evaluated.
Minor Sequence Changes Observed During Ferret Respiratory Droplet
Adaptation. Viruses collected from the nasal washes of respiratory
droplet contacts, A/ferret/Maryland/P10-UMD/08 (H9N2)
[RCP10], were directly sequenced to determine the molecular
changes supporting respiratory droplet transmission. Sequence
analysis of nasal washes collected on days 5 and 8 p.c., from 4
independent respiratory droplet contacts, revealed the same 5
amino acid changes from 2WF10:6M98 to the RCP10 virus, indi-
cating their selection during respiratory droplet transmission. Three
amino acid changes were found on the surface proteins while 2 were
found on the internal proteins. Two changes occurred on the HA,
one on the HA1 portion of the molecule at position 189 (H3
numbering) within antigenic site B and in close proximity to the
RBS (Fig. 3). This amino acid change from threonine (T) to alanine
(A) has been documented before (17) and is also found in naturally
occurring isolates (18). However, the combination of key amino
acid residues at the RBS found in the RCP10 ferret-adapted virus;
i.e., histidine (H) 183, A189, glutamic (E) 190, and L226 has yet to
be identified in nature (Fig. 3D). The available human H9N2
sequences from the NCBI database that have yet to show sustained
human-to-human transmission, contain H183 or asparagine (N)
183, T189, E190, and L226. The only major difference in these
viruses and the RCP10 viruses is at position 189. The second change
is located on the HA2 at position 192 (H3 numbering), a change
from glycine (G) to arginine (R), 3 amino acids away from the
transmembrane region of the HA2. Unfortunately, this amino acid
change lies within a region that has not been resolved by crystal-
lography and therefore cannot be mapped structurally. The change
in the NA at position 28, isoleucine (I) to valine (V), is located in
the transmembrane domain. This domain has been reported to
participate in virus assembly and/or shedding (19). The 2 remaining
changes, L to I and H to tyrosine (Y), at positions 374 and 110 of
PB2 and M1, respectively, map to regions of unassigned functions
within these 2 proteins.
To establish whether the amino acid changes observed in RCP10
occurred before passage 10, we sequenced the P10 inoculum virus
[P10] (nasal wash from passage 9 ferrets used to infect passage 10),
P9, and P8 inoculum viruses (nasal wash from passages 8 and 7
ferrets used to infect passages 9 and 8, respectively) and nasal wash
from respiratory droplet contacts from the second transmission
study of P10 [RCP102] (Fig. 1 C and D). Sequence analysis of
passages 8­10 revealed that changes observed in the PB2 and NA
occurred either before or at passage 8 with the M1 having a mixed
population at P8 selecting for Y at P9. Interestingly the 2 changes
observed on the HA were not present until RCP10 and nasal
washes collected from RCP102 respiratory droplet contacts re-
vealed the same 2 changes in the HA gene, implying these changes
were selected during adaptation and are perhaps necessary for
respiratory droplet transmission (Table 3). The sequence analysis
shows minor changes are necessary to support respiratory droplet
transmission, one of which alters the RBS of the HA and most likely
results in the observed transmission phenotype.
Adaptive Mutations on RCP10 Surface Proteins Support Respiratory
Droplet Transmission. A majority of the adaptive amino acid changes
occurred before passage 10, with the exception of changes found on
the HA. Therefore we wanted to determine whether the amino acid
changes on the surface proteins alone are sufficient for respiratory
droplet transmission in the background of the M98 backbone. Using
reverse genetics, we created a reassortant virus, 2RCP10:6M98,
which contains the HA and NA genes from the ferret-adapted
RCP10 virus and the internal genes from the human M98 virus. We
found that the changes in the surface proteins alone are indeed
sufficient for respiratory droplet transmission (Fig. 1 E and F) with
direct contacts shedding virus days 4 and 5 p.c. and transmission to
respiratory droplet contacts on the same day with similar titers to
P10. Clinical signs were also similar to those observed during the
P10 infection, highlighting the role of the surface protein changes
on transmissibility (Tables 1 and 2). Respiratory droplet transmis-
sion of 2RCP10:6M98 was confirmed in a second, independent
study in which 1 out of 2 respiratory droplet contacts became
positive for virus shedding (Table 4). Although the M98 back-
Table 2. Clinical signs, virus replication, and seroconversion associated with H9N2 reassortant viruses in direct contact and
respiratory-droplet contact ferrets
Direct contacts Respiratory-droplet contacts
Virus
Weight loss
(%)*
Sneezing
(day of onset)
Serum
(HI titer)
Weight loss
(%)*
Sneezing
(day of onset)
Serum
(HI titer)
2WF10:6M98 1.65  0.50 2/2 (4, 5) 1280, 2560 ND 0/2 10, 10
P10 5.36  0.1 4/4 (5, 7) 2560, 2560, 2560, 2560 7.91  1.98 4/4 (7, 8, 9) 2560, 2560 1280, 2560
2RCP10:6M98 2.79  1.43 4/4 (7, 9) 1280, 1280, 1280, 1280 2.07  0.59 4/4 (6, 7, 8) 1280, 1280, 640, 640
RCP10 (A189, G192) 1.67  0.82 4/4 (5, 7) 1280, 1280, 2560, 2560 ND 0/4 10, 10, 10, 10
RCP10 (T189, R192) 8.65  5.16 4/4 (6, 7) 1280, 2560, 2560, 2560 ND 0/4 10, 10, 10, 10
2WF10: 6RCP10 2.3  1.4 4/4 (6, 7) 1280, 2560, 1280, 1280 1.2  0.4 0/4 10, 40, 640, 40
*Average body weight loss is shown as average  standard deviation.
Homologous virus was used in HI assays to detect anti-H9 antibodies (sera collected at 2 weeks p.c.).
Two independent experiments with 2 infected, 2 direct, and 2 respiratory droplet ferrets each. ND, not determined, because no viral replication occurred in ferrets.
Fig. 2. Consistent isolation of P10 H9N2 virus in lung tissue. Two ferrets were
infected with the ferret-adapted P10 virus or mock infected with PBS. Data are
compared to the 2WF10:6M98 and WF10 viruses published in ref. 4. Tissues
were collected at 5 dpi. *Note only 1 of 2 ferret lungs were positive for virus
in the 2WF10:6M98 group. Titers are expressed as log10 values of TCID50/mL
with the limit of detection at 0.699 log10TCID50/mL. OB, olfactory bulb; NT,
nasal turbinate.
Sorrell et al. PNAS  May 5, 2009  vol. 106  no. 18  7567
MICROBIOLOGY
bone is likely to play a role in transmission, our study indicates
that the adaptive changes in PB2 and M1 in the RCP10 virus are
not essential for the respiratory droplet transmission phenotype.
At most, 3 amino acid changes on the surface proteins of the
avian H9N2 support respiratory droplet transmission in the M98
backbone.
Because HA is the major determinant in the transmission of
pandemic influenza, and is key for respiratory droplet transmission
of RCP10, we wanted to determine whether both changes in the
RCP10 HA are required for respiratory transmission. We used
site-directed mutagenesis to create RCP10 HA-mutant viruses that
carry 1 of the 2 adaptive HA changes. RCP10 (A189, G192)
contains the adaptive change of alanine at HA1 189 and the avian
glycine at HA2 192 while RCP10 (T189, R192) contains the avian
threonine at HA1 189 and adaptive arginine at HA2 192. Our
transmission studies suggest that both mutations in HA are neces-
sary for respiratory droplet transmission of the avian­human H9N2
reassortant viruses (Fig. 4 A­D). Both mutant viruses replicate
efficiently and transmit to direct contacts within 5 to 6 days p.c.
inducing weight loss and sneezing (Tables 1 and 2); however,
transmission to respiratory droplet contacts in neither nasal wash
nor serum was detected. Interestingly, we could predict that a
change in HA1, in close proximity to the RBS (Fig. 3), would be
necessary for respiratory droplet transmission; however, we could
not anticipate that a change in the HA2 portion of the molecule
would have an impact on transmission. Furthermore, because
infection with neither RCP10 (A189, G192) nor RCP10 (T189,
R192) resulted in quick selection of strains with respiratory droplet
transmission, we must conclude that both the T189A and G192R
mutations arose as aleatory mutations during adaptation in the
absence of selective immune pressure. Perhaps multiple rounds of
infection would be required before a dominant population con-
taining A189 and R192 can emerge from either the RCP10 (A189,
G192) or RCP10 (T189, R192) viruses. These studies highlight the
complexities associated with transmissibility of influenza viruses
and emphasize the need for in vivo studies, like those shown here,
to better understand mechanisms of influenza transmission.
To determine the role the adaptive mutations in the internal
proteins of RCP10 play in respiratory droplet transmission, we
rescued the reassortant H9N2 virus encoding the internal genes of
RCP10 and the unadaptive HA and NA of WF10, 2WF10:6RCP10.
We found that the virus was able to replicate and transmit effi-
ciently to direct contact ferrets; however, respiratory droplet trans-
mission was observed in only 1 of 4 respiratory droplet contacts,
which shed titers roughly 2 logs lower than RCP10 and
Fig. 3. Adaptive mutations in the H9 HA surface protein necessary for
respiratory droplet transmission. (A) Cartoon representation of the H9 HA
monomer as described by Ha et al., (18) binding the Ltsc 2,6 sialic acid analog
(orange and red lines) in the RBS. (B) Magnification of the globular head of the
HA showing stick representations (in green) of key amino acids in the RBS:
N183, G228, L226, T189, and V190, binding to 2,6 sialic acid (SIA, red lines).
Numbers correspond to amino acid positions based on the H3 HA numbering
system. (C) H9 HA RBS with amino acids corresponding to the WF10 HA
wild-type sequence, which differs from the published crystal structure at 2
positions: H183 (dark blue stick) and E190 (red stick). T at position 189 is
represented as a bright green stick. (D) H9 HA RBS with amino acids corre-
sponding to the RCP10 HA sequence, which differs from the WF10 HA se-
quence at A189, represented as an olive green stick. Structures generated
using MacPymol (DeLano Scientific).
Table 3. Sequence analysis of avian­human H9N2 viruses
obtained through adaptation in ferrets
Gene Origin
Amino acid
position Parent P8 P9 P10 RCP10 RCP102
PB2 Human 374 L I I I I I
PB1 Human No changes* ND ND
PA Human No changes ND ND
HA Avian HA1 189 T T T T A A
HA2 192 G G G/R G/R R R
NP Human No changes ND ND
NA Avian 28 I V V V V V
M1 Human 110 H H/Y Y Y Y Y
M2 Human No changes ND ND
NS1 Human No changes ND ND
NEP Human No changes ND ND
*No amino acid changes detected between the parent and either the RCP10
or the RCP102 viruses.
ND, sequencing not done.
Bold and italicized letters denotes more prominent residue at particular
amino acid position based on electropherograms of sequencing profiles.
Table 4. Summary of reassortant viruses tested for replication
and transmission in ferrets
Virus Replication
Transmission
Direct Respiratory droplet
P10* 4/4 4/4 4/4
2RCP10:6M98 4/4 4/4 3/4
RCP10 (A189, G192) 4/4 4/4 0/4
RCP10 (T189, R192) 4/4 4/4 0/4
2WF10:6RCP10 4/4 3/4 1/4
*Two separate studies of 2 infected, 2 direct, and 2 respiratory droplet
contacts each.
Minimal shedding for 1 of the 3 positive direct contacts.
7568  www.pnas.orgcgidoi10.1073pnas.0900877106 Sorrell et al.
2RCP10:6M98 respiratory droplet contacts (Fig. 4 E and F, Table
2). Sequence analysis directly from the respiratory droplet contact's
nasal wash indicated no adaptive changes on the HA. This result is
consistent with the notion that adaptive mutations on the surface
proteins are essential for efficient respiratory droplet transmission
of the avian­human H9N2 reassortant virus. However we should
note that the adaptive changes in PB2 and M1 do play a role in the
transmission phenotype noted by the 1 positive respiratory droplet
contact in Fig. 4F and when comparing the viral titers and length
of shedding in respiratory droplet contacts from RCP10 and
2RCP10:6M98. The internal genes have been implicated in trans-
mission not only within avian species (20­22) but also from avian
to mammalian species (23, 24). A complete set of viruses tested and
their transmission phenotype are listed (Table 4 and supporting
information (SI) Fig. S1 and Table S1).
Adaptation and Respiratory Droplet Transmission Leads to Changes in
Hemagglutinin Inhibition (HI) Profile, Implications for Pandemic Pre-
paredness. Our results suggest that one of the important determi-
nants for respiratory droplet transmission is located in close prox-
imity to the RBS overlapping a major antigenic site of the HA
molecule (site B, Fig. 3); this led us to resolve the HI profiles for the
reassortant viruses tested. Interestingly we found that the RCP10
virus displays a different antigenic profile from the parental WF10
virus. The HI titers to the WF10 virus are greatly reduced if serum
antibodies raised in response to the RCP10 are used instead of those
against the parental WF10 virus (Table 5). The RCP10 serum also
reacted inefficiently against other H9N2 viruses in HI assays. The
opposite is also true: the HI titers to the RCP10 virus are greatly
reduced if serum antibodies raised in response to the WF10 virus
are used. More importantly, HI titers using anti-RCP10 antiserum
were similar for the RCP10 and the RCP10 (A189, G192) viruses,
implicating amino acid 189 in antigenicity and in agreement with its
position in the tip of the globular head of HA1 (Fig. 3). It has been
speculated that given natural conditions, immune pressure can
select for variants with altered host specificity and an ability to
escape host immunity, which are key factors in the evolution of
avian H9N2 viruses. This RCP10 virus, in the absence of immune
pressure, created an antigenically variant HA in the ferret. The
findings above carry huge implications for vaccine stocks for
pandemic preparedness; highlighting the potential discrepancy in
antibody protection from the avian field isolate (chosen to prepare
the vaccine stock) versus the antigenic makeup of the virus that
gains respiratory droplet (or human-to-human) transmissibility.
This study also highlights the inherent limitations in the selection of
vaccine seed stocks from current avian H9N2 strains, which resem-
bles the WF10 virus in HA sequence. It will be important to
determine whether the changes seen in the HA of RCP10, namely
amino acid 189 in the HA1, can confer transmissibility in additional
H9 HAs and other avian subtypes and whether this should be
Fig. 4. Transmission phenotype supported through
both T189A and G192R changes on the HA. Ferrets
were inoculated intranasally (i.n.) with 106 TCID50 of
either RCP10 (A189, G192) (A), RCP10 (T189, R192) (C),
or 2WF10:6RCP10 virus (E). Twenty-four hours later,
direct contact (A, C, and E) and respiratory droplet
contact ferrets (B, D, and F) were introduced and nasal
washes collected daily. Black and white bars represent
individual ferrets. Titers are expressed as log10 values
of TCID50/mL with the limit of detection at 0.699
log10TCID50/mL.
Table 5. HI profiles show antigenic differences among H9N2
wild-type and ferret-adapted strains
Ferret
sera*/virus WF10 RCP10
RCP10
(A189, G192)
RCP10
(T189, R192)
WF10 5120, 5120 320, 640 640, 640 1280, 2560
RCP10 1280, 640 5120, 5120 5120, 2560 1280, 640
RCP10 (A189, G192) 1280, 640 5120, 2560 5120, 5120 1280, 640
RCP10 (T189, R192) 2560, 2560 1280, 640 640, 1280 5120, 2560
M98 10, 10 10, 10 10, 10 10, 10
Dk/Y280 80, 80 80, 80 ND§ ND
Ch/SF3 160, 160 320, 640 ND ND
*Sera collected at 2 weeks p.i. was used in HI assays against homologous and
heterologous viruses.
Dk/Y280 corresponds to influenza A/duck/Hong Kong/Y280/97 (H9N2).
Ch/SF3 corresponds to influenza A/chicken/Hong Kong/SF3/99 (H9N2).
§ND, not done.
Sorrell et al. PNAS  May 5, 2009  vol. 106  no. 18  7569
MICROBIOLOGY
considered a critical antigenic site for vaccine candidates. It is
interesting to note that residue 189 has been implicated not only in
H9 escape mutants but also in escape mutants of the highly
pathogenic H5 and pandemic H2 viruses (17, 25­27). However, it
must be noted that selection of alanine at position 189 in this study
occurred in the absence of preexisting immune pressure in the
ferrets.
Discussion
The threat of avian H9N2 strains, and for that matter any avian
influenza subtype, becoming a pandemic virus is ever-present.
However the key mechanism, human-to-human transmission, is an
obstacle yet to be achieved and a process we cannot predict. Insight
into the mechanism behind efficient human-to-human transmission
can aid in surveillance, countermeasures, vaccine production, and
quick reaction/response to outbreaks. Previous studies have com-
pared avian influenza viruses with early pandemic strains, partic-
ularly H1 and H3 strains and studied the particular ``adaptive
mutations'' that lead to respiratory droplet transmission using
different mammalian and avian animal models (28­31). These and
other studies have confirmed the importance of SA2,6 receptor
specificity for sustained transmission in humans (and ferrets) and
defined position 226 in the RBS of HA as a key component in
influenza host range (32, 33). We have recently shown that L226 in
the RBS of the HA of H9 viruses also plays a crucial role in
replication in ferrets; however, these viruses have yet to gain the
ability to transmit by respiratory droplets regardless of high viral
titers and sneezing (4).
Our unique study describes respiratory droplet transmission of an
avian­human H9N2 influenza virus in ferrets and pinpoints the
minimal changes necessary for respiratory droplet transmission in
this model. It is important to note that this particular strain has yet
to establish itself in a mammalian host; however, after only 10
passages of nasal washes we were able to establish infection and
sustain respiratory droplet transmission that was reproducible in
multiple studies. This adaptation resulted in only 5 amino acid
changes in the entire genome implying that little is needed for
currently circulating avian H9N2 viruses to transmit human-to-
human following reassortment with a human strain. Studies to
identify the minimal changes necessary indicated the 3 changes in
the surface HA and NA as key point mutations essential for
respiratory droplet transmission. More importantly, we identified
and located a change that dramatically alters the antigenicity of the
virus, bringing to light the inherent limitations in the selection of
vaccine seed stocks for avian H9N2 viruses and the possible
inefficiency regarding the seed stock selection of other avian
influenza strains. Whether these changes can affect transmission
phenotypes of additional avian H9N2 strains and possibly other
influenza subtypes, most notably H5 and H7, is to be determined.
However as we have mentioned, changes at position 189 (T189A)
have been highlighted in H5 and H2 escape mutants. Our studies
show that respiratory droplet transmission in mammals is not an
exclusive property of the few virus subtypes that have caused human
pandemics (namely H1, H2, and H3 influenza viruses). Other virus
subtypes, like H9N2, can also overcome the natural barriers that
prevent them from transmitting in a similar manner, provided the
ideal host and environmental conditions. It is critical to determine
these conditions to develop countermeasures for the impending
pandemic whether it is H9, H5, or any other subtype.
Materials and Methods
Viruses and Cells. VirusesweregrownandtitratedinMadin-DarbyCaninekidney
(MDCK) cells, as described previously (4). Live viruses were handled under a
biosafety level-3 containment. The viruses collected from P8, P9, P10, and
respiratory droplet contacts, RCP10 and RCP102, were sequenced directly from
the nasal washes without any passages in embryonated chicken eggs or cell
culture. Please see online supporting information for a complete description on
Materials and Methods.
ACKNOWLEDGMENTS. We are indebted to Ivan Gomez-Osorio for his excel-
lent laboratory techniques and animal handling assistance. We thank Andrea
Ferrero for her laboratory managerial skills, Danielle Hickman for editing the
manuscript, and Dr. Robert Webster for supplying the wild-type viruses used
in this study. We are also indebted to Drs. Bin Lu, Rosemary Broome, and the
team at MedImmune Inc. for their training in ferret work. This research was
possible through funding by the Centers for Disease Control and Prevention­
Health and Human Services Grant (1U01CI000355), National Institute of Al-
lergy and Infectious Diseases­National Institutes of Health (NIAID-NIH) Grant
(R01AI052155), Cooperative State Research, Education, and Extension Servic-
es­U.S. Department of Agriculture Grant (2005­05523), and NIAID-NIH Con-
tract (HHSN266200700010C).
1. Horimoto T, Kawaoka Y (2001) Pandemic threat posed by avian influenza A viruses. Clin
Microbiol Rev 14(1):129­149.
2. Matrosovich MN, Krauss S, Webster RG (2001) H9N2 influenza A viruses from poultry
in Asia have human virus-like receptor specificity. Virology 281(2):156­162.
3. Wan H, Perez DR (2007) Amino acid 226 in the hemagglutinin of H9N2 influenza viruses
determines cell tropism and replication in human airway epithelial cells. J Virol
81(10):5181­5191.
4. Wan H, et al. (2008) Replication and transmission of H9N2 influenza viruses in ferrets:
Evaluation of pandemic potential. PLoS ONE 3(8):e2923.
5. van Riel D, et al. (2006) H5N1 virus attachment to lower respiratory tract. Science
312(5772):399.
6. Maines TR, et al. (2008) Pathogenesis of emerging avian influenza viruses in mammals
and the host innate immune response. Immunol Rev 225:68­84.
7. Peiris JS, et al. (2001) Cocirculation of avian H9N2 and contemporary ``human'' H3N2
influenza A viruses in pigs in southeastern China: Potential for genetic reassortment?
J Virol 75(20):9679­9686.
8. Peiris M, et al. (1999) Human infection with influenza H9N2. Lancet 354(9182):916­917.
9. Jia N, et al. (2008) Increased sensitivity for detecting avian influenza-specific antibodies
by a modified hemagglutination inhibition assay using horse erythrocytes. J Virol
Methods 153(1):43­48.
10. Guo Y, Li J, Cheng X (1999) Discovery of men infected by avian influenza A (H9N2) virus.
Zhonghua Shi Yan He Lin Chuang Bing Du Xue Za Zhi 13(2):105­108.
11. Butt KM, et al. (2005) Human infection with an avian H9N2 influenza A virus in Hong
Kong in 2003. J Clin Microbiol 43(11):5760­5767.
12. Tellier R (2006) Review of aerosol transmission of influenza A virus. Emerg Infect Dis
12(11):1657­1662.
13. Haff RF, Schriver PW, Stewart RC (1966) Pathogenesis of influenza in ferrets: Nasal
manifestations of disease. Br J Exp Pathol 47(5):435­444.
14. Herlocher ML, et al. (2001) Ferrets as a transmission model for influenza: Sequence
changes in HA1 of type A (H3N2) virus. J Infect Dis 184(5):542­546.
15. Renegar KB (1992) Influenza virus infections and immunity: A review of human and
animal models. Lab Anim Sci 42(3):222­232.
16. Smith H, Sweet C (1988) Lessons for human influenza from pathogenicity studies with
ferrets. Rev Infect Dis 10(1):56­75.
17. Kaverin NV, et al. (2004) Structural differences among hemagglutinins of influenza A
virus subtypes are reflected in their antigenic architecture: Analysis of H9 escape
mutants. J Virol 78(1):240­249.
18. Ha Y, Stevens DJ, Skehel JJ, Wiley DC (2002) H5 avian and H9 swine influenza virus
haemagglutinin structures: Possible origin of influenza subtypes. EMBO J 21(5):865­875.
19. Barman S, et al. (2004) Role of transmembrane domain and cytoplasmic tail amino acid
sequences of influenza a virus neuraminidase in raft association and virus budding.
J Virol 78(10):5258­5269.
20. Sorrell EM, Perez DR (2007) Adaptation of influenza A/Mallard/Potsdam/178­4/83
H2N2 virus in Japanese quail leads to infection and transmission in chickens. Avian Dis
51(Suppl 1):264­268.
21. Makarova NV, Ozaki H, Kida H, Webster RG, Perez DR (2003) Replication and trans-
mission of influenza viruses in Japanese quail. Virology 310(1):8­15.
22. Hossain MJ, Hickman D, Perez DR (2008) Evidence of expanded host range and
mammalian-associated genetic changes in a duck H9N2 influenza virus following
adaptation in quail and chickens. PLoS ONE 3(9):e3170.
23. Steel J, Lowen AC, Mubareka S, Palese P (2009) Transmission of influenza virus in a
mammalian host is increased by PB2 amino acids 627K or 627E/701N. PLoS Pathog
5(1):e1000252.
24. Van Hoeven N, et al. (2009) Human HA and polymerase subunit PB2 proteins confer
transmission of an avian influenza virus through the air. Proc Natl Acad Sci USA
106(9):3366­3371.
25. Tsuchiya E, et al. (2001) Antigenic structure of the haemagglutinin of human influenza
A/H2N2 virus. J Gen Virol 82(Pt 10):2475­2484.
26. Kaverin NV, et al. (2002) Structure of antigenic sites on the haemagglutinin molecule
of H5 avian influenza virus and phenotypic variation of escape mutants. J Gen Virol
83(Pt 10):2497­2505.
27. Ilyushina N, Rudneva I, Gambaryan A, Bovin N, Kaverin N (2004) Monoclonal antibodies
differentially affect the interaction between the hemagglutinin of H9 influenza virus
escape mutants and sialic receptors. Virology 329(1):33­39.
28. Tumpey TM, et al. (2007) A two-amino acid change in the hemagglutinin of the 1918
influenza virus abolishes transmission. Science 315(5812):655­659.
29. Taubenberger JK, Morens DM, Fauci AS (2007) The next influenza pandemic: Can it be
predicted? J Am Med Assoc 297(18):2025­2027.
30. Garcia-Sastre A, Whitley RJ (2006) Lessons learned from reconstructing the 1918
influenza pandemic. J Infect Dis 194(Suppl 2):S127­S132.
31. Naeve CW, Webster RG, Hinshaw VS (1983) Phenotypic variation in influenza virus
reassortants with identical gene constellations. Virology 128(2):331­340.
32. Matrosovich M, et al. (2000) Early alterations of the receptor-binding properties of H1,
H2, and H3 avian influenza virus hemagglutinins after their introduction into mam-
mals. J Virol 74(18):8502­8512.
33. Stevens J, et al. (2006) Glycan microarray analysis of the hemagglutinins from modern
and pandemic influenza viruses reveals different receptor specificities. J Mol Biol
355(5):1143­1155.
7570  www.pnas.orgcgidoi10.1073pnas.0900877106 Sorrell et al.
