INTRODUCTION
In March 2013, three fatal human cases of infection with influenza A virus (H7N9)
were reported in China (1). More cases were rapidly recognized in China; although
no cases were reported in the summer of 2013, the number of cases started accumulating
rapidly again in the fall of that same year (2). Common clinical symptoms in patients
with H7N9 infection are fever and cough with lymphocytopenia; the disease progresses
to an acute respiratory distress syndrome, often with a fatal outcome (3). Perimortem
lung biopsies show diffuse alveolar damage, with infiltration of polymorphonuclear
cells, formation of hyaline membranes, and fibroproliferative changes (4).
Although influenza A viruses of the H7 subtype have caused human infections in the
past (reviewed in reference 5), most notably during an outbreak of highly pathogenic
avian influenza virus H7N7 in the Netherlands in 2003 that resulted in 89 human cases
(6), this is the first time that an H7 subtype virus has caused so many severe cases.
The H7N9 virus is a reassortant consisting of Eurasian wild bird H7 and N9 viruses
with H9N2 viruses circulating in poultry in China (1, 7, 8). An amino acid substitution
in HA, Q226L, present in some H7N9 strains, increases the binding of HA of these viruses
to α2,6-linked sialic acids (9). However, this single amino acid change does not seem
to be sufficient for efficient human-to-human transmission.
The newly emerged H7N9 virus poses a significant pandemic threat, due to the rapid
accumulation of cases in several distinct regions of China and the ability of the
H7N9 viruses to be transmitted, albeit inefficiently, via respiratory droplets or
aerosols in the ferret model (10
–
13). In order to allow a better estimation of the effect of a potential H7N9 pandemic,
we studied the A/Anhui/1/2013 strain of this virus in the cynomolgus macaque model.
This model was chosen because it most closely reflects the human physiology and the
development of pneumonia, cytokine, and chemokine responses (14), and the pattern
of attachment of H7N9 virus to respiratory tissues of macaques was recently shown
to be more similar to that in humans than that in other frequently used animal models
(15). Upon inoculation with influenza virus A/Anhui/1/2013, cynomolgus macaques developed
transient, moderate disease with virus replication in the upper and lower respiratory
tracts. The emerging H7N9 influenza virus was more pathogenic than seasonal influenza
A virus and most isolates of the pandemic H1N1 virus but not as pathogenic as the
1918 Spanish influenza virus or highly pathogenic avian influenza (HPAI) H5N1 virus
in cynomolgus macaques.
RESULTS
Clinical signs in cynomolgus macaques inoculated with A/Anhui/1/2013.
Eight 5-year-old cynomolgus macaques (4 male and 4 female) were inoculated with 7
× 106 50% tissue culture infectious doses (TCID50) of A/Anhui/1/2013 (H7N9) via a
combination of ocular, oral, intranasal, and intratracheal inoculation as described
previously (16). Animals started to show signs of disease at 1 day postinoculation
(dpi). Clinical disease peaked at 3 and 4 dpi (Fig. 1A) and was moderate. Six out
of 8 animals showed obvious respiratory signs, such as increased respiration rates,
abdominal breathing, and coughing (see Table S1 in the supplemental material). Nasal
discharge and cough were noted in only one animal. Hematologic and blood chemical
analyses did not reveal any abnormalities during the course of infection (data not
shown).
FIG 1
Clinical scores, virus shedding and lung pathology in cynomolgus macaques after infection
with influenza virus A/Anhui/1/2013. (A) Eight cynomolgus macaques were inoculated
with 7 × 106 TCID50 of influenza virus A/Anhui/1/2013 and were observed twice daily
for clinical signs of disease and scored using a previously described clinical scoring
system (58). The average clinical score for all animals per time point is given. Clinical
scores for individual animals can be found in Table S1 in the supplemental material.
(B) Nose (green symbols) and throat (purple symbols) swabs and bronchoalveolar lavage
fluid (blue symbols) were collected at 1, 2, 3, 4, and 6 days after inoculation. Virus
titers in these samples were determined by titration in MDCK cells. (C) The percentage
of area of each lung lobe affected by lesions was determined ventrally and dorsally
in animals euthanized at 3 (blue symbols) and 6 (red symbols) dpi. The average percent
gross pathology of dorsal and ventral measurements was calculated for each lung lobe
of each animal. Of note, 4 animals were euthanized at 3 days postinoculation, and
thus data in panels A and B from 4 dpi onwards are the averages for 4 animals. Symbols
in panels B and C represent individual animals. Closed circles, H7N9-1; closed squares,
H7N9-2; closed triangles, H7N9-3; closed inverted triangles, H7N9-4; open circles,
H7N9-5; open squares, H7N9-6; open triangles, H7N9-7; open inverted triangles, H7N9-8.
Radiographic changes in lungs of animals inoculated with A/Anhui/1/2013.
At 1, 2, 3, 4, and 6 dpi, ventral-dorsal and lateral chest X-rays were taken to monitor
the development of pneumonia. Radiographic changes in the lungs of inoculated animals
were observed starting 2 and 3 dpi and were observed in all inoculated animals to
various degrees (see Table S2 in the supplemental material). The interstitial infiltration
was observed first in the lower right lung lobe and, in individual animals, spread
over time to the right middle, left lower, left middle, and right upper lung lobes
and developed into severe diffuse interstitial infiltration (Fig. 2; also, see Table S2
in the supplemental material).
FIG 2
Radiographic changes in lungs of cynomolgus macaques inoculated with influenza virus
A/Anhui/1/2013. Ventral-dorsal X rays were obtained from macaque H7N9-5 before inoculation
(day 0; baseline) and at 1, 3, 4 and 6 days after inoculation with 7 × 106 TCID50
of influenza virus A/Anhui/1/2013. A detailed analysis of radiographs taken from all
eight infected macaques during infection is given in Table S2 in the supplemental
material.
Virus shedding in cynomolgus macaques inoculated with A/Anhui/1/2013.
Clinical exams were performed at 0, 1, 2, 3, 4, and 6 dpi, and nasal, oropharyngeal,
ocular, and rectal swabs were collected. Swabs were initially analyzed for the presence
of viral RNA by real-time reverse transcription-PCR (RT-PCR). Because of the large
number of swabs in which viral RNA was detected, all nasal, oropharyngeal, and ocular
swabs were titrated on MDCK cells; rectal swabs were not titrated, as only 3 rectal
swabs were positive by PCR. Oropharyngeal swabs were positive in virus titration by
1 dpi and remained positive in all animals until the end of the experiment at 6 dpi
(Fig. 1B). Not all nasal swabs were positive by virus titration; most of the virus
shedding via the nose occurred in animals H7N9-7 and H7N9-8 (Fig. 1B). Despite the
ability of influenza A viruses of the H7 subtype to cause conjunctivitis, ocular swabs
were only sporadically positive by virus titration: only the ocular swabs collected
from H7N9-1 at 1, 2, and 3 dpi, the ocular swab collected from H7N9-4 at 1 dpi, and
the ocular swab collected from H7N9-8 at 3 dpi were positive, with virus titers between
100.8 and 102.8 TCID50/ml (data not shown).
During clinical exams, bronchalveolar lavages (BAL) were performed, and the samples
were analyzed for the presence of infectious virus. Virus could be isolated from the
BAL fluid of all animals at 1 dpi; BAL fluid remained positive throughout the experiment,
with significant amounts of virus still detected at 6 dpi, ranging from 3.2 × 102
TCID50/ml to 1.4 × 104 TCID50/ml (Fig. 1B).
Gross lung pathology in cynomolgus macaques inoculated with A/Anhui/1/2013.
Upon necropsy of 4 animals at 3 dpi, gross lesions were observed in the lungs of all
animals. There was variation in the area of the lung affected between animals, but
at least two lobes showed gross lesions in all animals, varying from 5% to 100% of
tissue affected (Fig. 1C). In line with our observation on X-rays, the right lung
lobes were more severely affected than the left lobes; this is likely a result of
the intratracheal inoculation and the anatomy of the lung (17). By 6 dpi, the area
of the lung displaying gross lesions had increased, with 100% of all three right lung
lobes being affected in 2 of 4 animals (Fig. 1C).
Virus titers in tissues of cynomolgus macaques inoculated with A/Anhui/1/2013.
For each animal, virus titers were determined in tissue samples collected from all
6 lung lobes and 2 lung lesions. In line with our X-ray and gross pathology observations,
at 3 dpi virus could be detected in all three lobes of the right lung but not in all
three lung lobes of the left lung in all animals (Fig. 3A). Virus was present in the
collected lung lesions, but unexpectedly, virus titers were not higher in lung lesions
than in the collected lung lobe samples, indicating widespread virus replication throughout
the lower respiratory tract. By 6 dpi, the virus titers in the lung lobes and lung
lesions had decreased compared to those at 3 dpi, although this decrease was not statistically
significant; the number of animals with positive virus titration also decreased (Fig. 3A).
FIG 3
Virus replication and histopathological changes in cynomolgus macaques inoculated
with influenza virus A/Anhui/1/2013. Cynomolgus macaques were inoculated with 7 ×
106 TCID50 of influenza virus A/Anhui/1/2013; 4 animals were euthanized at 3 and 6 days
postinoculation, and tissue samples were collected. (A and B) Virus titers in the
indicated tissues collected at 3 and 6 days postinoculation were determined by titration
in MDCK cells. Geometric mean titers were calculated; error bars represent standard
deviations. Numbers above bars indicate the number of animals in which virus titration
was positive (out of 4 animals per time point). R, right; L, left; LN, lymph node.
(C and D) Histopathological changes in lungs of cynomolgus macaques inoculated with
influenza virus A/Anhui/1/2013 at 3 (C) and 6 (D) days postinoculation. Lung tissue
was collected and stained with hematoxylin and eosin. In severe cases of infection
at 3 dpi, lesions were characterized by edema, alveolar fibrin (black asterisk), and
hyaline membrane formation (arrow). At 6 dpi, edema, organizing fibrin (white asterisk),
and type II pneumocyte hyperplasia (arrowheads) were observed. Magnification (C and
D), ×400.
In other tissues of the respiratory tract, i.e., nasal turbinates, oropharynx, trachea,
and right and left bronchus, virus titers were generally higher than in the lungs
at 3 dpi and 6 dpi (Fig. 3B). Interestingly, although virus titers in the lung lobes
and lung lesions decreased between 3 and 6 dpi, virus titers in the other respiratory
tract tissues did not. Virus could also be detected in the tonsils and mediastinal
lymph nodes of infected animals at 3 and 6 dpi (Fig. 3B). Virus could be detected
in the conjunctiva of only one animal at 3 and 6 dpi.
The remaining tissues that were collected at 3 and 6 dpi (i.e., heart, liver, spleen,
kidney, stomach, jejunum, ileum, transverse colon, and brain) were analyzed for the
presence of viral RNA by real-time RT-PCR. Except for the liver, where vRNA was detected
in 6 out of 8 animals, vRNA was detected sporadically in all tissues except in heart
(see Table S3 in the supplemental material). Since cycle threshold values (C
T
) in almost all of these samples were higher than the level at which virus titration
is usually successful (C
T
< 32), virus titration was not attempted on these samples.
Histopathology of respiratory tissues of cynomolgus macaques inoculated with A/Anhui/1/2013.
At 3 dpi, histopathology of the lungs was similar for all 4 animals necropsied and
was characterized as mild to marked, acute, bronchointerstitial pneumonia. The pneumonia
was characterized microscopically as mild to marked thickening of alveolar septa by
fibrin, edema, neutrophils, and macrophages, and the alveoli contained small to large
amounts of these same inflammatory components (Fig. 3C). Multifocally, hyaline membranes
were observed. The lumens of terminal bronchioles frequently contained fibrin, edema,
hyaline membranes, neutrophils, and macrophages, with multifocal necrosis and loss
of lining epithelium. Inflammation and necrosis of bronchial submucosal glands were
frequently noted, with mild, subacute periglandular inflammation or more severe changes
that ranged from neutrophils and macrophages within ductular lumens to necrosis of
acinar and ductular epithelium, occasionally affecting the entire gland. Larger bronchioles
and bronchi were generally much less severely affected than terminal airways, with
intact, viable lining epithelium and only occasional mild, neutrophilic luminal exudate.
At 6 dpi, histopathology of the lungs was characterized as mild to marked, subacute
to chronic, bronchointerstitial pneumonia. Microscopic changes were again characterized
by the presence of neutrophils, macrophages, fibrin, and edema within alveolar septa,
alveoli, and terminal bronchioles (Fig. 3D). Additionally, type II pneumocyte hyperplasia
was observed in extensive portions of each lung lobe, and large clumps of alveolar
fibrin frequently engulfed neutrophils and macrophages and were lined and infiltrated
by fibroblasts.
Immunohistochemistry (IHC) of sections of lung demonstrated low to moderate numbers
of antigen-positive alveolar type I and type II pneumocytes, macrophages, and epithelium
lining bronchioles, bronchi, and bronchial submucosal glands (Fig. 4). The numbers
of positively stained alveolar pneumocytes and bronchial and bronchiolar submucosal
gland epithelial cells were similar at 3 and 6 dpi. The number of positively stained
pulmonary macrophages was increased at 6 dpi compared with 3 dpi. Positive staining
in alveolar and submucosal macrophages was cytoplasmic, most likely indicating active
phagocytosis of virus rather than replication of virus in these cells (Fig. 4H). Antigen-positive
macrophages were also consistently noted within mediastinal lymph nodes, with increased
numbers of these cells being noted at day 6 dpi. Low numbers of positively stained
macrophages were noted in pharyngeal tonsils, oropharynges, tracheas, and extrapulmonary
bronchi.
FIG 4
Immunohistochemical analysis of lungs of cynomolgus macaques inoculated with influenza
virus A/Anhui/1/2013. Cynomolgus macaques were inoculated with 7 × 106 TCID50 of influenza
virus A/Anhui/1/2013; 4 animals were euthanized at 3 (A to D) and 6 (E to J) days
postinoculation, and lung samples were collected. Matching tissue sections were stained
with hematoxylin and eosin (H&E) or studied by immunohistochemistry (IHC) using an
anti-NP monoclonal antibody (IHC; visible as red-brown staining). At 3 dpi, mild to
marked thickening of alveolar septa by fibrin, edema, neutrophils, and macrophages
was observed (A), and virus antigen was present in type I and type II (black arrows)
pneumocytes (B). Inflammation and necrosis of bronchial submucosal glands were frequently
noted (C), with virus antigen being present in submucosal gland (open arrowhead) and
bronchial epithelium (white arrow). Infection of the bronchial lining epithelium was
more pronounced by 6 dpi (E and F). Cytoplasmic staining of alveolar macrophages indicates
active phagocytosis on 6 dpi (H; black arrowheads). (J) Virus replication in bronchial
submucosal gland epithelium on 6 dpi. Magnification, ×400.
There was more variation between individual animals in histopathological lesions in
the remaining tissues. Mild, subacute conjunctivitis was noted histologically in one
of eight animals at 6 dpi, and low numbers of macrophages positive for influenza A
virus antigen were present in the submucosa from that animal; the viral load in the
eye swab collected from this animal (H7N9-7) at 6 dpi was still high (104 TCID50 equivalents/ml),
likely indicating active virus replication. Influenza A virus-positive cells were
noted in the epithelium of the nasal turbinates with minimal inflammation in 3 of
4 animals at 3 dpi (Fig. 5A and B). Although no microscopic abnormalities of the tonsil
were noted, there were low to moderate numbers of positively stained tonsillar macrophages
in all eight animals, with an increase in the amount of antigen detected between 3
and 6 dpi. Focal, subacute pharyngitis with ulceration was noted in one animal at
3 dpi; rare pharyngeal epithelial cells were positive according to IHC but not at
the site of the ulcer (Fig. 5C and D). In one animal at 3 dpi and one at 6 dpi, influenza
A virus antigen could be detected in respiratory epithelial cells from the nasopharynx.
In other tissue sections of the pharynx, there were few macrophages that were positive
by IHC. Tracheal inflammation affecting the mucosal lining and/or submucosal glands
was noted in all macaques, with ulceration of lining epithelium occurring in one animal
on 3 dpi and one on 6 dpi. Few epithelial cells and macrophages stained positive in
IHC (Fig. 5E and F). Minimal to moderate subacute inflammation was noted in at least
one of the extrapulmonary bronchi from each animal, and in one animal there was an
extensive area of marked inflammation with ulceration. Upon IHC staining of the same
sections of bronchi, few positively stained epithelial cells and/or macrophages were
noted (Fig. 5G and H).
FIG 5
Extrapulmonary histopathological changes in cynomolgus macaques inoculated with influenza
virus A/Anhui/1/2013. Cynomolgus macaques were euthanized at 3 (A and B) and 6 (C
to H) days postinoculation, and tissue was collected and stained with hematoxylin
and eosin (H&E) or immunohistochemistry (IHC) using an anti-NP monoclonal antibody
(IHC; visible as red-brown staining). In the rostral nasal cavity (A and B) and nasopharynx
(C and D), influenza virus A/Anhui/1/2013 mainly replicated in the lining epithelium.
In the trachea (E and F), cell debris (arrowhead) with immunopositive staining is
visible in a submucosal gland. In the bronchus (G and H), bronchial lumen exudate
is visible, with virus replication in the epithelium lining the bronchus. Magnification,
×400.
Systemic cytokine and chemokine profiles in infected macaques.
Serum levels of 23 cytokines and chemokines were determined in samples obtained at
0, 1, 2, 3, 4, and 6 dpi. Transient differences were observed in granulocyte colony-stimulating
factor (G-CSF), gamma interferon (IFN-γ), interleukin 1 receptor antagonist (IL-1RA),
IL-2, IL-6, IL-8, IL-10, IL-15, monocyte chemoattractant protein 1 (MCP-1), macrophage
inflammatory protein 1α (MIP-1α), MIP-1β, tumor necrosis factor alpha (TNF-α), and
vascular endothelial growth factor (VEGF) levels. At 1 dpi, there was a statistically
significant increase in levels of G-CSF, IL-1RA, IL-6, IL-10, IL-15, MCP-1, and TNF-α
(Fig. 6); levels of IFN-γ, IL-2, IL-8, MIP-1α, MIP-1β, and VEGF increased at that
time point, but increases were not statistically significant (Fig. 6; also data not
shown). By 6 dpi, all levels of cytokines and chemokines had returned to baseline.
FIG 6
Cytokine and chemokine levels in serum samples of cynomolgus macaques infected with
influenza virus A/Anhui/1/2013. Serum samples were collected 0, 1, 2, 3, 4, and 6 days
postinoculation, and chemokine and cytokine levels were analyzed using the nonhuman
primate cytokine Milliplex 23-plex map kit (Millipore). Only cytokines and chemokines
with detectable levels that fluctuated during the course of the experiment are shown.
Each symbol indicates one animal. ●, H7N9-1; ■, H7N9-2; ▲, H7N9-3; ▼, H7N9-4; ♦, H7N9-5;
○, H7N9-6; □, H7N9-7; Δ, H7N9-8. Bars at the tops of the graphs indicate statistically
significant differences between time points (P < 0.05; one-way ANOVA with Bonferroni’s
posttest).
Alterations in host gene expression upon infection with A/Anhui/1/2013.
To elucidate global host responses specifically associated with sites of virus-induced
airway injury in influenza virus A/Anhui/1/2013-infected macaques, we used microarrays
to assess transcriptional profiles induced in lung lesions compared to the adjacent
lung tissue. Welch’s t test (P < 0.05; change > 1.5-fold) identified 802 differentially
expressed genes (DEG) in the lesions compared to lung samples from the same animals
at 3 dpi (Fig. 7A; also, see Table S4 in the supplemental material); these 802 genes
were differentially expressed in all animals. Of those, 429 were upregulated and 372
were downregulated in lesions compared to samples from the right lower lung lobe.
Using Ingenuity pathway analysis (IPA), we assessed the functional significance of
this molecular signature and constructed a network of critical molecules based on
direct interactions in the IPA knowledge base (IPKB) (Fig. 7B). As expected, there
are numerous transcripts upregulated from functional categories previously observed
in cynomolgus macaque models of influenza infection (18
–
21), including known mediators of antiviral immunity, such as pattern recognition
receptors and downstream signaling molecules (TLR4 and MYD88), interferons and interferon-stimulated
genes (IFNL3), interferon-regulatory factors (IRF1, IRF5, and JAK3), and proinflammatory
cytokines and inflammatory mediators (IL-6, NLR family, NLRP3, TNFRSF1B, TNFRSF6B,
and TNFSF8). We also identified a number of upregulated genes associated with leukocyte
migration and differentiation (CXCL10, CXCL11, SELE, IL4R, IL-18, and CSF2RA). This
suggests that molecules that recruit infiltrating effector leukocytes are increased
at the site of lesions and is in line with the observed influx of neutrophils and
macrophages observed microscopically in the lungs. Moreover, hyaluronic acid synthase
2 (HAS2), a molecule known to be associated with lung injury, was strongly upregulated
in the lung lesion samples.
FIG 7
Transcriptional signatures associated with influenza virus A/Anhui/1/2013 infection
in cynomolgus macaques. Cynomolgus macaques were inoculated with 7 × 106 TCID50 of
influenza virus A/Anhui/1/2013; 4 animals were euthanized 3 days postinoculation,
and tissue samples were collected from the right lower lung lobes and from lung lesions.
(A) Heat map showing log2 expression ratios of 802 DEG in lung lesions relative to
the right lower lung lobe in individual animals, as determined by Welch’s t test (P
< 0.05, fold change ≥1.5), and grouped by hierarchical clustering. (B) Molecular interaction
network built using DEG shown in panel A. Solid lines show direct molecular interactions.
Red molecules are those upregulated in lesions relative to lung, while green molecules
are downregulated. Text on the outside of the network indicates functional categories
in which molecules in this part of the network cluster.
We also observed several functional categories of molecules that were downregulated
in lesions relative to the adjacent lung tissue (Fig. 7B). These were associated predominantly
with lipid metabolism and adipogenesis regulated by peroxisome proliferator-activated
receptor alpha (PPARα). We also observed downregulation of growth factors, particularly
those related to insulin signaling and regulation of glucose levels (INSR and IGFBP5).
By 6 dpi, 154 DEG were identified in lung tissue versus lung lesions of animals euthanized
at 6 dpi (see Fig. S1 and Table S4 in the supplemental material). Due to substantial
variability between animals, we were not able to determine any significantly enriched
functional categories at this time point.
Next, the Upstream Analysis function in IPA was used to identify drugs predicted to
act as significant upstream regulators of the key DEG identified at 3 dpi. Drugs that
are predicted to inhibit pathological host responses were selected based on the activation
z score. Negative activation z scores are predicted to cause opposite or inhibitory
effects on significant genes associated with pathology in the lesions. These findings
were confirmed using Connectivity Map (CMap), a resource allowing comparison with
transcriptional signatures from multiple cell types treated with a library of chemical
and genetic perturbagens (22), by determining if the connectivity score was also negative,
indicating that the compound would have inhibitory effects on transcriptional signatures
associated with lesion formation. We identified ten compounds in IPA (Table 1), four
of which were perturbagens listed in CMap. We identified two compounds that met our
criteria in IPA and CMap, rosiglitazone and simvastatin, predicted to have inhibitory
effects on pathological host responses associated with lesions in influenza virus
A/Anhui/1/2013-infected animals (Table 1). These molecules are both involved in regulating
lipid biosynthesis and metabolism. Rosiglitazone is an FDA-approved diabetes drug
that modulates PPAR activity and insulin sensitivity and has been shown to affect
RNA virus replication (23
–
25), virus-induced inflammation (26, 27), and lung inflammation (28
–
30). Simvastatin is a FDA-approved statin which lowers blood cholesterol by inhibiting
3-hydroxy-3-methyl-glutaryl coenzyme A (HMG-CoA) reductase. It has been shown to reduce
lung inflammation in mouse models of airway injury (31
–
34) and bacterial infections (35
–
37). However, simvastatin was not shown to have a significant effect on influenza
A virus pathogenesis in mice (38
–
40). Besides rosiglitazone and simvastatin, mainly immunomodulatory drugs were predicted
to have inhibitory effects on pathological host responses to A/Anhui/1/2013 infection.
Table 1
Top ten drugs and chemicals predicted by Ingenuity pathway analysis to inhibit lung
lesion-specific host responses after infection with influenza virus A/Anhui/1/2013
Drug
Drug class
IPA activation z score
IPA P value of overlap
CMap enrichment score
FDA-approved formulation
Rosiglitazone
Thiazolidinedione (PPAR antagonist; insulin sensitizer)
−2.358
8.41 × 10−6
−0.146
Yes (Avandia)
Dexamethasone
Glucocorticoid steroid (anti-inflammatory)
−2.234
0.000125
0.321
Yes (generic)
Tyrphostin AG-1478
Tyrosine kinase inhibitor (EGFR signaling inhibitor)
−2.425
0.00125
0.754
No
Simvastatin
Statin (HMG-CoA reductase inhibitor)
−2.096
0.00226
−0.535
Yes (generic)
PD98059
Kinase inhibitor (MEK inhibitor)
−2.838
0.00228
NA
No
Infliximab
Monoclonal antibody (anti-TNF-α)
−2.236
0.00424
NA
Yes (Remicade)
U0126
Kinase inhibitor (MEK inhibitor)
−2.837
0.0045
NA
No
Oleic acid
Monounsaturated fatty acid
−2.4
0.00556
NA
No
Pyrrolidine dithiocarbamate
Cell cycle inhibitor, nitric oxide synthase inhibitor, metal chelator
−2.169
0.013
NA
No
Curcumin
Antioxidant (anti-inflammatory)
−2.123
0.0703
NA
No
The ability of rosiglitazone to inhibit A/Anhui/1/2013 replication was tested in MDCK
cells treated with different concentrations of this drug after infection with A/Anhui/1/2013
at a multiplicity of infection of 0.001. Twenty-four hours after addition of 100 µM
rosiglitazone, but not at lower concentrations, a statistically significant (P = 0.0314;
two-tailed unpaired t test) 14-fold reduction in virus titers in the supernatant of
treated, infected cells compared to mock-treated cells was observed (see Fig. S2A
in the supplemental material); a small cytotoxic effect of the drug was also noticed
at this time point in an MTS [3-(4,5-dimethyl-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium,
inner salt] assay (see Fig. S2B in the supplemental material). At 72 h after treatment,
the inhibitory effect of rosiglitazone on A/Anhui/1/2013 replication could no longer
be detected (data not shown).
DISCUSSION
The recent emergence of H7N9 influenza A virus in humans in China has shifted the
focus away from HPAI H5N1 to H7N9 as one of the main pandemic threats. Although much
attention has been drawn to the ability of this virus to be transmitted between ferrets,
the pathogenicity of H7N9 influenza A virus in ferrets does not match the severity
of disease observed in human cases (10
–
13). The physiology of the cynomolgus macaque lung is more similar to that of humans,
and cytokine and chemokine responses to infection in macaques are similar to those
in humans (14). Moreover, it was shown recently that the pattern of attachment of
H7N9 influenza A virus to respiratory tissues of cynomolgus macaques is more similar
to the attachment pattern in humans than that in other animal models frequently used
for influenza A virus pathogenesis studies, including the ferret (15). Therefore,
we studied the H7N9 infection of cynomolgus macaques in detail.
In agreement with the observed abundant attachment of the H7N9 influenza A virus to
the human upper and lower respiratory tracts (41), and the replication of the H7N9
virus in ex vivo cultures of the human upper as well as lower respiratory tract (42),
H7N9 virus replicated well in the upper and lower respiratory tracts of cynomolgus
macaques, as indicated by virus titers in nasal turbinates, oronasopharynges, tracheas,
bronchi, and lung tissue samples, reflecting the previously described receptor distribution
of influenza virus H7N9 in the macaque respiratory tract (15). Surprisingly, the virus
titers in nasal turbinates, oronasopharynges, tracheas, and bronchi did not decrease
between 3 and 6 dpi, unlike the virus titers in the lung samples, which did decrease.
This extended virus replication in the upper respiratory tract could result in prolonged
virus shedding and thus an increased risk of virus transmission.
The histopathology of the lungs in fatal human H7N9 cases was similar to, albeit more
severe than, that observed in cynomolgus macaques, with diffuse alveolar damage, infiltration
of polymorphonuclear cells, formation of hyaline membranes, and, later after onset
of symptoms, pneumocyte hyperplasia and fibroproliferative changes (4). However, clinically
and histopathologically, H7N9 infection was not as severe in cynomolgus macaques as
it has been described in humans, where H7N9 has a high case fatality rate. Although
viral characteristics could cause differences in disease severity between humans and
macaques, another possible explanation for the discrepancy in disease severity could
be the high percentage of human cases with underlying medical complications (3).
Since some of the H7N9 influenza virus strains isolated from human cases have an R292K
substitution in NA that renders them partially resistant to treatment with neuraminidase
inhibitors (12, 43
–
45), it is important to identify drugs that either directly inhibit virus replication
in the host or reduce the severity of disease and the level of lung injury after infection.
By analyzing the gene expression profiles in lung lesions compared to adjacent infected,
nonlesional lung tissue, we were able to predict drugs reported to act as upstream
regulators of some of these genes that may thus play a role in the development of
lung lesions during H7N9 infection. One of the predicted drugs, rosiglitazone, modestly
reduced replication of influenza virus A/Anhui/1/2013 in vitro. A similar data analysis
resulted in the identification of a compound that reduced replication of the Middle
East respiratory syndrome coronavirus in vitro (46); however, the validity of this
approach has not yet been tested in vivo. Thus, this analysis of the gene expression
profile hints at new avenues of treatment to explore in in vivo models, rather than
revealing novel treatments directly applicable in the clinic.
Although it is difficult to compare studies investigating the pathogenicity of different
influenza A viruses in the macaque model, because of different inoculation routes
and doses used and because sampling schemes do not overlap between studies, the H7N9
infection in macaques in our study was similar to that described previously (12).
In both studies, animals developed transient clinical signs, with virus replication
in the upper as well as lower respiratory tract and similar histopathological lesions
in the lower respiratory tract. More importantly, H7N9 infection seemed to be clinically
more severe than most infections with isolates of pandemic H1N1 (16); virus shedding
from the throat was higher in H7N9-infected animals, and a larger area of the lung
was affected with gross lesions (47). Inoculation of cynomolgus macaques with seasonal
H3N2 influenza A virus resulted in infection with mild or even asymptomatic disease
(48, 49). Cynomolgus macaques inoculated via the same route and with the same dose
of 1918 Spanish influenza virus developed lethal disease (50), and HPAI H5N1 influenza
A virus inoculated into macaques at a 40-fold-lower dose caused more severe disease
than H7N9 influenza A virus in this study (51). Thus, the emerging H7N9 influenza
virus is more pathogenic than seasonal influenza A virus and most isolates of the
pandemic H1N1 virus but not as pathogenic as the 1918 Spanish influenza virus and
HPAI H5N1 virus in cynomolgus macaques. However, the pathogenicity of the H7N9 virus
may decrease if the virus adapts further to solely using α2,6-linked sialic acids
as the receptor for entry, as pandemic influenza viruses to date have done (52
–
55). Exclusive attachment to α2,6-linked sialic acids would most likely result in
a shift to replication mainly in the upper respiratory tract of humans, likely resulting
in less severe disease, as has been described for the 2009 pandemic H1N1 virus (56)
and upon adaptation of HPAI H5N1 virus to efficient transmission via respiratory droplets
or aerosols (57).
MATERIALS AND METHODS
Ethics statement.
All animal experiments were approved by the Institutional Animal Care and Use Committee
of the Rocky Mountain Laboratories and performed following the guidelines of the Association
for Assessment and Accreditation of Laboratory Animal Care, International (AAALAC),
by certified staff in an AAALAC-approved facility.
Cells.
Madin-Darby canine kidney (MDCK) cells were cultured in Eagle’s modified essential
medium (EMEM) (Gibco) supplemented with 10% fetal calf serum (FCS), 50 IU/ml penicillin,
50 µg/ml streptomycin, 2 mM glutamine, 0.75 mg/ml sodium bicarbonate, and nonessential
amino acids.
Virus.
A/Anhui/1/2013 (passage E2/E1) was obtained from the Centers for Disease Control in
Atlanta, GA, and passaged once in MDCK cells.
Animal study and sample collection.
Eight cynomolgus macaques (4 males, 4 females; age, 5 years; 4 to 6 kg) were inoculated
with 7 × 106 TCID50 of A/Anhui/1/2013 via a combination of intratracheal (4 × 106
TCID50; 4 ml), intranasal (1 × 106 TCID50; 500 µl/nostril), oral (1 × 106 TCID50;
1 ml), and ocular (1 × 106 TCID50; 20 µl/eye) routes. The animals were observed twice
daily for clinical signs of disease and scored using a previously described clinical
scoring system (58). At 1, 2, 3, 4, and 6 days postinoculation, clinical exams were
performed on anesthetized animals, and lateral X rays were taken and analyzed by a
veterinarian. Nasal, oral, urogenital, and rectal swabs were collected in 1 ml Dulbecco’s
modified essential medium (DMEM) with 50 U/ml penicillin and 50 µg/ml streptomycin;
bronchoalveolar lavages (BAL) were performed using 10 ml sterile saline solution;
blood was collected for hematology, blood chemistry analysis, and peripheral blood
mononuclear cell (PBMC) isolation. The total white blood cell count, lymphocyte, platelet,
reticulocyte, and red blood cell counts, hemoglobin and hematocrit values, mean cell
volume, mean corpuscular volume, and mean corpuscular hemoglobin concentrations were
determined from EDTA-containing blood with the Hemavet 950FS+ hemoglobin analyzer
(Drew Scientific). PBMC were isolated by centrifugation over a Histopaque gradient
(Sigma) as per the manufacturer’s recommendation. At 3 and 6 dpi, 4 macaques were
euthanized, and samples of the conjunctivas, right and left eyes, nasal turbinates,
tonsils, oronasopharynges, tracheas, right and left bronchi, all six lung lobes, lung
lesions, mediastinal lymph nodes, hearts, livers, spleens, kidneys, stomachs, jejuna,
ilea, transverse colons, olfactory bulbs, cerebella, brain stems, and bone marrow
were collected. The percentage of the lung surface area affected by gross lung lesions
was quantitated for each lung lobe, ventrally and dorsally, by a board-certified veterinary
pathologist at the time of necropsy.
Histopathology and immunohistochemistry.
Histopathology and immunohistochemistry were performed on macaque tissues. After fixation
for 7 days in 10% neutral buffered formalin and embedding in paraffin, tissue sections
were stained with hematoxylin and eosin (H&E). To detect influenza A virus antigen,
immunohistochemistry was performed using an anti-NP monoclonal HB-65 antibody (EVL,
the Netherlands) as a primary antibody.
RNA extraction.
RNA was extracted from swabs, BAL fluid, and whole-blood samples using the QIAamp
viral RNA kit (Qiagen) according to the manufacturer’s instructions. Tissues were
stored at −80°C until further processing; tissue samples (30 mg) were homogenized
in RLT buffer, and RNA was extracted using the RNeasy kit (Qiagen).
Quantitative real-time RT-PCR.
A one-step real-time RT-PCR targeted at the matrix gene of influenza A virus was performed
using the Quantifast probe kit (Qiagen) according to instructions of the manufacturer
using the primers and probe described in reference 59.
Virus titrations.
Viruses were titrated by endpoint dilution in MDCK cells. MDCK cells were inoculated
with tenfold serial dilutions of culture supernatants. One hour after inoculation,
cells were washed with phosphate-buffered saline (PBS) and supplemented with infection
medium (EMEM supplemented with 50 IU/ml penicillin, 50 µg/ml streptomycin, 2 mM glutamine,
0.75 mg/ml sodium bicarbonate, nonessential amino acids, and 5 µg/ml trypsin). Three days
after inoculation, the supernatants of infected cell cultures were tested for agglutination
activity using turkey red blood cells as an indicator of infection of the cells. Infectious
titers were calculated from 5 replicates by the Spearman-Karber method (60).
Serum cytokine and chemokine analysis.
Serum samples for analysis of cytokine/chemokine levels were inactivated with gamma
radiation (5 megarads) according to standard operating procedures. Concentrations
of granulocyte colony-stimulating factor, granulocyte-macrophage colony-stimulating
factor, IFN-γ, IL-1β, IL-1 receptor antagonist, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10,
IL-12/23 (p40), IL-13, IL-15, IL-17, MCP-1, MIP-1α, MIP-1β, soluble CD40 ligand (sCD40L),
transforming growth factor α, TNF-α, vascular endothelial growth factor (VEGF), and
IL-18 were measured on a Bio-Plex 200 instrument (Bio-Rad) using the nonhuman primate
cytokine Milliplex 23-plex map kit (Millipore) according to the manufacturer’s instructions.
Microarray data and functional analysis.
RNA was extracted using Qiagen micro-RNeasy spin columns per the manufacturer’s protocol.
Low-yield samples were concentrated using the RNA Clean and Concentrator kit (Zymo
Research, Irvine, CA). As multiple lesions were collected from each animal, equal
masses of RNA from these samples (3 or 4 samples per animal) were pooled to investigate
comprehensive signatures from multiple pathological sites within the same individual.
Probe labeling was carried out using the Agilent low-input processing protocol and
hybridized to Agilent rhesus macaque 8×60K microarrays (Agilent Technologies, Santa
Clara, CA) using the manufacturer’s one-color analysis protocol. For comparisons of
differentially expressed genes (DEG) in infected lungs and lung lesions, raw array
data were uploaded to Genedata Analyst 7.6 (Genedata Inc., San Francisco, CA). Data
were normalized using the quantile normalization method, and the log2 ratio expression
was calculated relative to the mean probe values of the 4 right lower lung lobe samples
per time point (lesion versus lung comparisons). Statistically significant DEG were
identified using Welch’s t test (P < 0.05; fold change, ≥1.5). Hierarchical clustering
of DEG was performed by the unweighted average method (unweighted pair group with
arithmetic mean [UPGMA]) using Spotfire DecisionSite 9.1.1 (Tibco, Somerville, MA).
Analysis of functional enrichment was performed using Ingenuity Pathway Analysis (IPA)
software (Ingenuity Systems, Redwood City, CA), and upstream drug efficacy predictions
were made using both the Upstream Analysis function of IPA and Connectivity Map 02
(Broad Institute, Cambridge, MA). Drugs predicted to inhibit pathological host responses
were selected based on the activation z score. Negative activation z scores are predicted
to cause opposite or inhibitory effects on significant genes associated with pathology
in the lesions, and in this study we sought compounds with activation z scores less
than −2 and a P value less than 0.05. These findings were confirmed using Connectivity
Map (CMap) (22). CMap assigns enrichment scores ranging between −1 and 1, and we accepted
compounds with CMap enrichment scores less than −0.1. These are compounds inducing
transcriptional profiles that are negatively connected with the gene expression signature
associated with lung pathology, confirming that these drugs may induce a transcriptional
profile that is inhibitory to tissue damage and lesion formation.
Antiviral assay.
Confluent MDCK cells in 24-well culture plates were infected in triplicate with influenza
virus A/Anhui/1/2013 at a multiplicity of infection of 0.001. After 1 h at 37°C, cells
were washed once with PBS, and infection medium containing rosiglitazone (0 to 100 µM)
was added to the cells. Cells were incubated for 24 h; supernatant was then harvested,
stored at −80°C for subsequent virus titration, and replaced with fresh infection
medium containing rosiglitazone. Supernatant was again collected at 72 h after infection
and stored at −80°C for subsequent virus titration. To determine a potential cytotoxic
effect of rosiglitazone, MDCK cells were simultaneously plated in 96-well culture
plates and treated with rosiglitazone (0 to 100 µM). After 24 h incubation, cytotoxicity
was tested using the CellTiter 96 AQueous one-solution cell proliferation assay (MTS)
(Promega) according to the manufacturer’s instructions.
Microarray data accession number.
Raw microarray data have been deposited in NCBI’s Gene Expression Omnibus database
(GSE48976) and are also available to the public at http://viromics.washington.edu.
SUPPLEMENTAL MATERIAL
Figure S1
Transcriptional signatures associated with influenza virus A/Anhui/1/2013 infection
in cynomolgus macaques. Cynomolgus macaques were inoculated with 7 × 106 TCID50 of
influenza virus A/Anhui/1/2013; 4 animals were euthanized 6 days postinoculation,
and tissue samples were collected from the right lower lung lobes and from lung lesions.
The heat map shows log2 expression ratios of 154 DEG in lung lesions relative to the
right lower lobe of the lung from individual animals, as determined by Welch’s t test
(P < 0.05, fold change ≥1.5), and grouped by hierarchical clustering. Download
Figure S1, TIF file, 0.1 MB
Figure S2
Effect of rosiglitazone on replication of influenza virus A/Anhui/1/2013 in MDCK cells.
(A) MDCK cells were infected with influenza virus A/Anhui/1/2013 at a MOI of 0.001
and subsequently treated with different concentrations of rosiglitazone. At 24 h postinfection,
supernatants were harvested, and virus titers in the supernatants were determined
by endpoint titration on MDCK cells. Geometric mean virus titers are indicated; error
bars indicate standard deviations, and asterisks indicate a statistically significant
effect (P < 0.05; two-tailed unpaired t test). (B) The cytotoxic effect of the different
rosiglitazone concentrations on MDCK cells was tested using an MTS assay. The average
OD, measured at 405 nm, calculated from three individual wells is indicated; error
bars indicate standard deviations. Download
Figure S2, TIF file, 0.1 MB
Table S1
Clinical scores of A/Anhui/1/2013-inoculated cynomolgus macaques. Animals were observed
twice daily, and clinical parameters were recorded.
Table S1, DOCX file, 0.1 MB.
Table S2
Summary of radiographic findings for A/Anhui/1/2013-inoculated cynomolgus macaques.
The quality and location of interstitial infiltration observed from ventral-dorsal
and lateral X rays are indicated.
Table S2, DOCX file, 0.1 MB.
Table S3
Viral RNA in tissues from cynomolgus macaques infected with influenza virus A/Anhui/1/2013.
Numbers are C
T
values obtained in real-time RT-PCR.
Table S3, DOCX file, 0.1 MB.
Table S4
Overview of the differentially expressed genes shown in Fig. 7 and in Fig. S1 in the
supplemental material, comparing samples from the right lower lung lobe or lung lesions
collected at 3 and 6 days postinoculation to those from uninfected controls.
Table S4, PDF file, 0.2 MB.