crossmark

Constitutively Expressed IFITM3 Protein in Human Endothelial Cells

Poses an Early Infection Block to Human Influenza Viruses
Xiangjie Sun, Hui Zeng, Amrita Kumar, Jessica A. Belser, Taronna R. Maines, Terrence M. Tumpey
Immunology and Pathogenesis Branch, Influenza Division, National Center for Immunization and Respiratory Diseases, Centers for Disease Control and Prevention,
Atlanta, Georgia, USA

ABSTRACT
A role for pulmonary endothelial cells in the orchestration of cytokine production and leukocyte recruitment during influenza
virus infection, leading to severe lung damage, has been recently identified. As the mechanistic pathway for this ability is not
fully known, we extended previous studies on influenza virus tropism in cultured human pulmonary endothelial cells. We found
that a subset of avian influenza viruses, including potentially pandemic H5N1, H7N9, and H9N2 viruses, could infect human
pulmonary endothelial cells (HULEC) with high efficiency compared to human H1N1 or H3N2 viruses. In HULEC, human influ-
enza viruses were capable of binding to host cellular receptors, becoming internalized and initiating hemifusion but failing to
uncoat the viral nucleocapsid and to replicate in host nuclei. Unlike numerous cell types, including epithelial cells, we found that
pulmonary endothelial cells constitutively express a high level of the restriction protein IFITM3 in endosomal compartments.
IFITM3 knockdown by small interfering RNA (siRNA) could partially rescue H1N1 virus infection in HULEC, suggesting
IFITM3 proteins were involved in blocking human influenza virus infection in endothelial cells. In contrast, selected avian influ-
enza viruses were able to escape IFITM3 restriction in endothelial cells, possibly by fusing in early endosomes at higher pH or by
other, unknown mechanisms. Collectively, our study demonstrates that the human pulmonary endothelium possesses intrinsic
immunity to human influenza viruses, in part due to the constitutive expression of IFITM3 proteins. Notably, certain avian in-
fluenza viruses have evolved to escape this restriction, possibly contributing to virus-induced pneumonia and severe lung disease
in humans.

IMPORTANCE
Avian influenza viruses, including H5N1 and H7N9, have been associated with severe respiratory disease and fatal outcomes in
humans. Although acute respiratory distress syndrome (ARDS) and progressive pulmonary endothelial damage are known to be
present during severe human infections, the role of pulmonary endothelial cells in the pathogenesis of avian influenza virus in-
fections is largely unknown. By comparing human seasonal influenza strains to avian influenza viruses, we provide greater in-
sight into the interaction of influenza virus with human pulmonary endothelial cells. We show that human influenza virus infec-
tion is blocked during the early stages of virus entry, which is likely due to the relatively high expression of the host antiviral
factors IFITMs (interferon-induced transmembrane proteins) located in membrane-bound compartments inside cells. Overall,
this study provides a mechanism by which human endothelial cells limit replication of human influenza virus strains, whereas
avian influenza viruses overcome these restriction factors in this cell type.

I nfluenza A viruses are important respiratory pathogens in hu-

mans and are responsible for approximately 250,000 to 500,000
fatal cases of influenza during annual epidemics worldwide (1).
Occasionally, influenza A viruses of novel strains or subtypes
against which the general human population has no preexisting
immunity emerge and cause severe pandemics, as was demon-
strated in 1918, 1957, 1968, and, most recently, in 2009 (2). Mean-
while, certain influenza A viruses of avian origin are capable of
crossing host species barriers, resulting in sporadic infection in
humans. Among these viruses, highly pathogenic avian influenza
(HPAI) H5N1 viruses cause the highest mortality rate in humans,
approximately 60% based on WHO reports (3). While exhibiting
the 1918 and 2009 H1N1 viruses) or recently isolated HPAI H5N1
viruses possess the ability to replicate in human lower respiratory
tract tissues and induce exacerbated innate immune responses
(6–9). This is demonstrated by early recruitment of inflammatory
leukocytes to the lung and excessive cytokine production, ulti-
mately leading to acute respiratory distress syndrome (ARDS) and
high mortality rates (10, 11). While the molecular mechanisms of
severe illness caused by influenza virus infection have not been
completely uncovered, it is believed that aberrant proinflamma-
Received 29 June 2016 Accepted 27 September 2016

reduced mortality in humans, low-pathogenicity avian influenza Accepted manuscript posted online 5 October 2016

(LPAI) viruses of the H7N9 subtype have also been associated with
severe disease, with over 700 reported cases since their initial de-
tection in humans in 2013 (4, 5).
Human influenza A viruses primarily target epithelial cells in
the upper respiratory tract due to their abundant expression of
␣-2,6-linked sialic acids, the preferred receptors for human influ-
enza viruses (1). However, pandemic influenza viruses (including
Citation Sun X, Zeng H, Kumar A, Belser JA, Maines TR, Tumpey TM. 2016.
Constitutively expressed IFITM3 protein in human endothelial cells poses an early
infection block to human influenza viruses. J Virol 90:11157–11167.
doi:10.1128/JVI.01254-16.
Editor: B. Williams, Hudson Institute of Medical Research
Address correspondence to Terrence M. Tumpey, tft9@cdc.gov.
Copyright © 2016, American Society for Microbiology. All Rights Reserved.

December 2016 Volume 90 Number 24 Journal of Virology jvi.asm.org 11157

Sun et al.
tory cytokine production and the resulting damage to the epithe-
lial-endothelial barrier of the pulmonary alveolus play an impor-
tant role in the development of severe disease (12). Recently, it has
been revealed that pulmonary endothelial cells are central orches-
trators of cytokine production and leukocyte recruitment in mice
inoculated with the 2009 pandemic H1N1 virus (13). The work
suggests that despite not representing a primary site for influenza
virus replication, pulmonary endothelial cells contribute to the
severity of the infection (13). Moreover, in vitro studies have
shown that influenza virus infection can upregulate the expression
of several endothelial adhesion molecules (14, 15), which may
facilitate extravasation of neutrophils and macrophages into the
alveoli. The persistent influx of such inflammatory cells can lead to
damage of the epithelial-endothelial barrier by releasing reactive
oxygen species, cytokines, and neutrophil extracellular traps (16).
Additionally, pulmonary endothelial cells are susceptible to HPAI
H5N1 virus infection in vitro in an envelope-dependent manner
and express high levels of proinflammatory cytokines upon infec-
tion, whereas most human influenza viruses display only limited
infectivity under these conditions (17–19). However, the exact
molecular mechanism governing how selected highly pathogenic
H5N1 viruses, but not human influenza viruses, possess the ability
to replicate and induce excessive cytokine production is still
largely unknown.
IFITMs (interferon-induced transmembrane proteins) were
first identified as type I and type II interferon (IFN)-induced pro-
teins in 1984 (20) and comprise a family of small, 10- to 15-kDa
proteins. In humans, there are four functional IFITM genes: the
IFITM1, IFITM2, IFITM3, and IFITM5 genes, with IFITM4 being
a pseudogene (21). Although previous studies on IFITMs mainly
focused on their roles in embryonic development, their functions
as host antiviral factors were only recently discovered by RNA
interference genomic screening for host factors involved in in-
fluenza virus infection (22). It was subsequently revealed that
IFITMs can restrict the early stages of replication of a wide variety
of viruses, including influenza virus (22), flavivirus (dengue and
West Nile viruses) (23, 24), filovirus (Marburg virus and Ebola
viruses) (23), and coronavirus (23). Among all the IFITM mem-
bers present in humans, IFITM3 has shown to be the most potent
antiviral factor in restricting influenza virus infection, as IFITM3
knockout mice displayed enhanced morbidity and mortality asso-
ciated with seasonal or 2009 pandemic H1N1 influenza virus in-
fection (25, 26). Furthermore, the single-nucleotide polymor-
phism (SNP) (12252-C) in the IFITM3 gene, which results in
decreased IFITM3 protein expression, has been linked with a
higher risk of hospitalization among individuals infected with sea-
sonal or 2009 pandemic H1N1 virus, as well as the novel H7N9
virus (26, 27).
In this study, we investigated the mechanism by which human
endothelial cells limit replication of human influenza viruses and
how avian influenza viruses overcome these restriction factors in
this cell type. Our results show that human influenza virus infec-
tion is blocked during the early stages of virus entry, precisely,
after hemifusion, likely due to the relatively high constitutive ex-
pression of the host antiviral factor IFITM3 located in endosomal
and lysosomal compartments. Conversely, certain avian influenza
viruses may circumvent this restriction by fusing at a higher pH in
early endosomes or by other, unknown mechanisms. Overall, our
study suggests human hosts are able to restrict influenza virus
infection in pulmonary endothelial cells partially by constitutively
11158 jvi.asm.org Journal of Virology
expressing high levels of IFITM proteins before interferon induc-
tion.
MATERIALS AND METHODS
Cells and viruses. Primary human lung blood microvascular endothelial
cells (HMVEC) (Lonza, Walkersville, MD), immortalized human lung
microvascular endothelial cells (HULEC), and human umbilical vein en-
dothelial cells (HUVEC) (obtained from the Scientific Resources Pro-
gram, CDC, Atlanta, GA) were cultured in endothelial cell basal medium
2 (EBM-2) (Lonza) with supplements as previously described (17). Hu-
man brain vascular endothelial cells (HBVEC), kindly provided by Mo-
nique Stins, Johns Hopkins University, were cultured in RPMI 1640 me-
dium supplemented with 10% fetal bovine serum (FBS), 10% Nu-Serum
(Fisher Scientific), 2 mM L-glutamine, 1 mM sodium pyruvate, 1⫻ min-
imal essential medium (MEM) nonessential amino acids, MEM vitamins,
and penicillin-streptomycin (100 U/ml). Human bronchial epithelium
(Calu-3) cells were cultured in Eagle’s MEM supplemented with 10% FBS,
2 mM L-glutamine, 1 mM sodium pyruvate, and nonessential amino ac-
ids. Human lung epithelial A549 cells and Madin-Darby canine kidney
(MDCK) cells were grown in Dulbecco’s modified Eagle medium
(DMEM) supplemented with 10% FBS. Primary human bronchial epithe-
lial (NHBE) cells were grown in serum-free and hormone-supplemented
bronchial epithelial growth medium according to the manufacturer’s in-
structions (Lonza). The A549 cells stably expressing IFITM3 or the vector
pQCXIP (Clontech) were originally developed and kindly provided by
Gregory B. Melikyan and Mariana Marin, Emory University, and were
maintained with 1.5 mg/ml of puromycin.
Influenza viruses A/Puerto Rico/8/1934 (PR8) (H1N1), A/Vietnam/
1203/2004 (VN/04) (H5N1), A/Anhui/1/2013 (Anhui/1) (H7N9), A/
chicken/Vietnam/NCVD-1156/2011 (CK/11) (H9N2), A/Netherlands/
219/2003 (NL/03) (H7N7), A/shoveler/Egypt/00215-NAMRU3/2007
(Shov/07) (H7N9), A/Brisbane/59/2007 (Bris/07) (seasonal H1N1), A/
Panama/2007/1999 (Pan/99) (seasonal H3N2), and 2009 pandemic
A/Mexico/4482/2009 (Mex/09) (H1N1) were grown in the allantoic cav-
ities of 10-day-old embryonated hen’s eggs for 24 to 48 h at 33°C to 37°C.
A/Brisbane/59/2007 (Bris/07) (seasonal H1N1) and 2009 pandemic
A/Mexico/4482/2009 (Mex/09) (H1N1) were propagated in MDCK cells
as described previously (28). Allantoic fluid or cell culture supernatant
was clarified by centrifugation, aliquoted, and stored at ⫺70°C; virus titers
were determined by standard plaque assay using MDCK cells. All research
with HPAI H5, H7, and H9 subtype viruses was conducted under bio-
safety level 3 (BSL3) containment, including enhancements required by
the U.S. Department of Agriculture and the National Select Agent Pro-
gram.
The reassortant viruses bearing HA and NA genes from VN/VN/04
(H5N1) or Anhui/1 (H7N9) virus and internal genes from the PR8 donor
virus (VN:PR8 and Anhui:PR8) (kindly provided by Li-Mei Chen, Influ-
enza Division, CDC) were propagated in the allantoic cavities of 10-day-
old embryonated hen’s eggs for 48 h at 35°C under BSL2⫹ laboratory
conditions.
Virus purification. PR8 viruses were propagated in the allantoic cav-
ities of 10-day-old embryonated hen’s eggs and then concentrated and
purified by equilibrium density centrifugation through a 30 to 60% linear
sucrose gradient as previously described (29). The concentrations of pu-
rified virus were determined using a Bio-Rad protein assay kit (Bio-Rad
Laboratories, Hercules, CA) and then diluted to 2 mg/ml in phosphate-
buffered saline (PBS) for virus labeling.
Virus infection. HULEC or A549 cells were incubated with influenza
virus at multiplicities of infection (MOI) ranging from 1 to 5 in viral
infection medium (base culture medium supplemented with 0.3% bovine
serum albumin [BSA]) for 1 h, followed by washing with viral infection
medium, and cultured with fresh infection medium at 37°C in a 5% CO2
atmosphere for 8 h (time of peak expression) before being fixed with 4%
paraformaldehyde for 20 min. Viral infectivity was quantified based on
the percentage of nucleoprotein (NP)-positive cells by indirect immuno-
December 2016 Volume 90 Number 24

fluorescence microscopy (counting at least 200 cells per infection) using a
mouse monoclonal anti-influenza A virus NP clone A1 and A3 blend
(Millipore, Billerica, MA).
Acid-induced fusion assay at the cell plasma membrane (acid bypass
assay). Influenza viruses at an MOI of 5 were bound to A549 cells and
HULEC on ice for 1 h, washed twice with prechilled viral infection me-
dium to remove any unbound virus particles, and then incubated with
prewarmed fusion buffer (20 mM HEPES, 2 mM CaCl2, 150 mM NaCl, 20
mM citric acid monohydrate-sodium citrate tribasic dehydrate, pH 5.0)
for 2 min at 37°C to induce virus fusion at the cell surface. Following acid
treatment, the cells were washed twice with prechilled viral infection me-
dium and then incubated for 8 h with viral infection medium containing
30 mM NH4Cl to prevent viral infection via endocytic pathways. Virus
infection mediated by viral fusion at the cell surface was quantified based
on the percentage of NP-positive cells, using immunofluorescence mi-
croscopy.
Virus internalization assay. Concentrated and purified influenza PR8
viruses were labeled with biotin and used in a virus internalization assay as
described previously (30). Briefly, 1 ml of purified virus (2 mg/ml) was
incubated with 10 mM Sulfo-NHS-SS-biotin (Pierce) for 2 h at 4°C, fol-
lowed by ultracentrifugation to remove any unincorporated biotin. The
labeled viruses were resuspended in PBS at a concentration of 1 mg/ml
and filtered through a 0.22-␮m filter before use. During the virus inter-
nalization assay, 10 ␮l of biotin-labeled virus was incubated with cells in
viral infection medium for 1 h at 4°C, followed by washing with ice-cold
viral infection medium to remove any unbound virus particles. To induce
internalization, the virus-bound cells were shifted to 37°C for 1 h in viral
infection medium. To distinguish cell surface-bound virus particles from
internalized virions, cells were incubated with excess quantities of uncon-
jugated streptavidin (2 mg/ml; Thermo Fisher Scientific Inc., Rockford,
IL) on ice after internalization to mask biotins from surface-bound vi-
ruses, after which the cells were fixed and permeabilized with 0.5% Triton
X-100 and visualized by immunofluorescence microscopy by staining
with Alexa 488-conjugated streptavidin (Thermo Fisher Scientific Inc.).
Virus fusion assay. Concentrated and purified PR8 virus at 100 ␮g/ml
was labeled with R18 and SP-DiOC18 (Thermo Fisher Scientific Inc.) at
final concentrations of 0.4 ␮M and 0.2 ␮M, as described previously (31,
32). Ten microliters of labeled PR8 viruses was bound to cells grown on
24-well plates on ice for 1 h, followed by washing three times with ice-cold
viral infection medium, before the plates were transferred to a 37°C incu-
bator with 0.5% CO2 for 1.5 h to induce viral internalization and fusion.
The cells were fixed with 4% paraformaldehyde (PFA) and stained with
DAPI (4=,6-diamidino-2-phenylindole) before being scanned with an
LSM 710 Zeiss inverted confocal microscope at wavelengths of 510 to 525
nm (green) and 575 to 640 nm (red) simultaneously. Viral fusion in en-
dosomes is indicated by a fluorescence color shift from the red to the green
channel.
Detection of influenza virus nucleocapsid uncoating. PR8 viruses at
an MOI of 40 were bound to HULEC and A549 cells on ice for 1 h in viral
infection medium, after which the virus-bound cells were incubated at
37°C for 3.5 h following washing with viral infection medium. The cells
were fixed, permeabilized, and stained with the anti-M1 monoclonal an-
tibody HB64 (ATCC, Manassas, VA) by indirect immunofluorescence
microscopy.
siRNA transfection. HULEC and A549 cells grown in 24-well plates
were transfected with a mixture of IFITM3 small interfering RNA (siRNA)
(catalog number 284737) or negative-control siRNA (catalog number
AM4611) and Lipofectamine 2000 (Thermo Fisher Scientific Inc.) follow-
ing the manufacturer’s recommended transfection procedures as de-
scribed previously (33); the knockdown effect on the IFITM3 protein
expression level was confirmed by Western blotting and by immunofluo-
rescence microscopy with rabbit anti-human IFITM3 antibody (catalog
number PA5-11274; Thermo Fisher Scientific Inc.) at 72 h posttransfec-
tion.
December 2016 Volume 90 Number 24 Journal of Virology
Influenza Virus Infection in Pulmonary Endothelial Cells
Immunofluorescence microscopy and confocal imaging. Images
were captured with a Zeiss Axioskop 2 fluorescence microscope with a
20⫻ or 40⫻ objective lens or an LSM 710 Zeiss inverted confocal micro-
scope with a 40⫻, 63⫻, or 100⫻ oil objective lens and processed using
Zen 2010 (Zeiss) and Adobe Photoshop (Adobe Inc.). The colocalization
coefficient was analyzed with Zen 2010 and calculated based on the for-
mula described by Manders et al. (34) from an average of at least 10
individual cells. For colocalization studies, the following reagents were
used: rabbit anti-EEA1 antibody (Cell Signaling), CellLight Rab7-green
fluorescent protein (GFP) (ThermoFisher Scientific), rabbit anti-human
CD107a (LAMP-1) and anti-human CD63 antibodies (BD Biosciences),
and mouse anti-EEA1 and LAMP-1 from Santa Cruz Biotechnology Inc.
Western blotting. Human endothelial cells and epithelial cells from
either cell lines or primary culture were grown to confluence in T-75
flasks. The cells were trypsinized and resuspended at 2 ⫻ 106/ml before
being spun down and lysed in 2⫻ Laemmli sample buffer (Bio-Rad) con-
taining 5% ␤-mercaptoethanol. The samples were boiled at 95°C for 5
min before being loaded into a 4 to 15% Mini-Protean Tris-glycine pre-
cast gel (Bio-Rad). Rabbit anti-IFITM3 (ThermoFisher Scientific) and
anti-␤-actin (Sigma-Aldrich) antibodies were used to detect IFITM3 and
actin expression in cell lysates, respectively.
Influenza virus pH optimum of fusion. Analysis of the pH threshold
of fusion was evaluated by virus-induced hemolysis assay as described by
Shelton et al. (35). Briefly, viruses at 128 HA units (HAU) per 50 ␮l or PBS
mock control was mixed with 50 ␮l of 1% (vol/vol) turkey red blood cells
and incubated at 4°C in a 96-well plate for 1 h; then, the mixtures were
pelleted and resuspended in 100 ␮l of fusion buffer (20 mM HEPES, 2 mM
CaCl2, 150 mM NaCl, 20 mM citric acid monohydrate-sodium citrate
tribasic dehydrate) with various pH values (from 4.8 to 7.4) at 37°C for 1
h to trigger hemolysis mediated by viral fusion activity. The released he-
moglobin was measured as the optical density at 405 nm (OD405). The OD
value from mock PBS control at each pH value was subtracted to calculate
hemolysis. The maximal hemolysis at a given pH was normalized to 100%,
and the hemolysis at other pH values was expressed as a percentage of the
maximal hemolysis.
RESULTS
A group of selected avian influenza viruses exhibited extended
tropism in human pulmonary endothelial cells. It has been doc-
umented previously that human influenza viruses of the H1N1
and H3N2 subtypes possess low infectivity in human pulmonary
endothelial cells, whereas HPAI H5N1 viruses are able to infect
and replicate efficiently in these cells (17–19). In order to further
examine the tropism of influenza viruses in pulmonary endothe-
lial cells, we infected HULEC with human or avian influenza vi-
ruses for 8 h; viral infectivity was measured by NP expression and
examined by immunofluorescence microscopy. Different sub-
types were used for HULEC infection, and parallel viral infections
in human lung epithelial A549 cells, which are highly susceptible
to human and avian influenza viruses, were included. The amount
of virus corresponding to an MOI of 1 to 5, which produced 80 to
95% infectivity in A549 cells, was chosen to infect HULEC, and
virus infectivity in both cell types was determined (Fig. 1A). To
directly compare viral infectivity among different viruses, viral
infection efficiency was expressed as the ratio of the percentage of
NP-positive HULEC versus that of NP-positive permissive A549
cells. As shown in Fig. 1B, human influenza viruses of H1 and H3
subtypes exhibited very limited viral infection rates in HULEC
with an infection ratio below 0.1 compared to viral infection rates
in A549 cells. However, unlike human influenza viruses, the avian
influenza viruses, VN/04 (HPAI H5N1), Anhui/1 (LPAI H7N9),
and CK/11 (LPAI H9N2), could efficiently infect HULEC at an
infection ratio of 0.65 or greater. Interestingly, the H7 viruses
jvi.asm.org 11159
Sun et al.

FIG 1 Influenza virus infectivity in HULEC. Influenza viruses at an MOI of 1 to 5, i.e., seasonal PR8 (H1N1), Bris/07 (H1N1), Pan/99 (H3N2) and 2009
pandemic Mex/09 (H1N1), VN/04 (H5N1), A/Anhui/1 (H7N9), CK/11 (H9N2), NL/03 (H7N7), Shov/07 (H7N9), and Anhui:PR8 and VN:PR8, were used to
infect HULEC and A549 cells for 8 h, and viral infectivity was quantified based on the percentage of NP-positive cells by immunofluorescence microscopy
(counting at least 200 cells per infection). (A) Viral infectivity in HULEC and A549 cells. (B) Ratios of viral infectivity in HULEC versus A549 cells. The error bars
represent the standard deviations (SD) of the mean from three independent experiments. Statistical analysis was performed to compare the ratio of viral
infectivity in HULEC versus A549 cells between human and avian influenza viruses by an unpaired t test with Welch’s correction.

NL/03 (HPAI H7N7) and Shov/07 (LPAI H7N9), the latter of
which shares high sequence homology with the HA and NA genes
from Anhui/1 virus, showed only intermediate infectivity in
HULEC, as their infectivity in these cells was approximately 35 to
40% of that observed in A549 cells (Fig. 1).
To better understand which viral genes are involved in the
extended tropism of avian influenza viruses in HULEC, recombi-
nant PR8 viruses harboring the HA and NA genes from Anhui/1
or VN/04 (lacking the polybasic cleavage site) were tested in the
viral infection assay. Both recombinant influenza viruses showed
enhanced viral infectivity in HULEC compared to PR8 virus, al-
beit at slightly lower infection rates than wild-type (wt) Anhui/1
or VN/04 virus (Fig. 1). This suggests that the HA and NA genes
represent the main determinants of influenza virus tropism in
HULEC. Taken together, our results show that, unlike human
H1N1 and H3N2 viruses, avian influenza viruses of the H5, H7,
and H9 subtypes have gained extended cellular tropism in human
pulmonary endothelial cells in an apparently viral envelope pro-
tein-dependent manner.
Human influenza viruses can infect endothelial cells by fus-
ing at the cell surface. To elucidate the stage of human influenza
virus infection that is blocked in endothelial cells, we first set out
to determine whether virus infectivity could be rescued by forcing
human influenza viruses to fuse at the endothelial cell surface. For
this purpose, viruses were incubated with endothelial cells on ice
for 1 h to allow virus binding, and then the cells were exposed to
pH 5.0 buffer to force fusion between surface-bound influenza
viruses and the host plasma membrane. In this assay approach, if
viruses are able to bind and fuse at the cell surface, the viral nu-
cleocapsid will be directly released into the cytosol instead of being
transported through endocytic pathways, as occurs during normal
viral entry. At 8 h after inducing fusion at the cell surface, viral
infectivity in the presence of 30 mM NH4Cl, which was used to
block viral entry via the endocytic pathway, was quantified based
on NP staining. As a control, viral infection in permissive A549
cells was used to demonstrate that PR8 (H1N1) and Pan/99
(H3N2) viruses could achieve approximately 75% and 95% infec-
tivity, respectively, by inducing fusion at the cell surface upon
exposure to pH 5.0, but not pH 7.4, conditions (Fig. 2). Interest-
ingly, although PR8 and Pan/99 viruses showed only very low
infectivity in HULEC by the normal viral entry route, the infection
rates of both viruses in HULEC reached up to 95% upon inducing
fusion at the cell surface, comparable to their infectivity in A549
cells under the same conditions (Fig. 2). These findings indicate
that human influenza viruses can bind to endothelial cells and fuse
at the cell surface in low-pH environments at efficiencies compa-
rable to those in permissive A549 cells, thus providing the means
for the viral genome to enter the cytoplasm and initiate replica-
tion. Our results from this acid bypass assay confirmed that the
infection block to human influenza virus in HULEC occurs down-
stream of viral binding but upstream of viral replication in the
host nucleus.
Human influenza viruses can become internalized and initi-
ate hemifusion in endosomes but fail to uncoat during entry
into endothelial cells. During influenza virus entry into host cells,
the virus first binds to cellular receptors and then becomes inter-
nalized by endocytosis into endosomal compartments in which
the viral membrane fuses with the endosomal membrane to re-
lease the viral contents into the cytosol and initiate replication.
The fusion between the influenza virus membrane and host endo-
somal membrane starts with lipid mixing (hemifusion), which is
then followed by fusion pore formation (36). After we determined
that the block in human influenza virus infection in endothelial
cells occurs downstream of viral binding but upstream of viral
replication in the nucleus, we next performed stepwise entry as-
says to determine the exact step at which the infection is inhibited.
First, we examined whether human influenza virus can bind and
be internalized by endocytosis in HULEC. For this purpose, we
used biotin-labeled PR8 virus in a virus binding and internaliza-
tion assay as described previously (30). As shown in Fig. 3A (left
column), PR8 virus was able to bind to both HULEC and A549
cells by staining with Alexa 488-streptavidin following virus incu-
bation with cells at 4°C for 1 h. This further confirms that the virus
has no defect in binding to HULEC compared to A549 cells. As
expected, the signal from cell surface-bound virions could be
blocked by preincubation with excessive amounts of unconju-
gated streptavidin (Fig. 3A, second column from left). Upon in-
ducing internalization at 37°C for 1 h, internalized viruses stained
with Alexa 488-streptavidin following unconjugated streptavidin
preincubation were visible in both HULEC and A549 cells (Fig.
3A, right column), suggesting that PR8 virus can become internal-
ized into HULEC with an efficiency similar to that in A549 cells.

11160 jvi.asm.org Journal of Virology December 2016 Volume 90 Number 24

 Influenza Virus Infection in Pulmonary Endothelial Cells

FIG 2 Human influenza virus infection in HULEC induced by fusing at the cell surface (acid bypass infection). Cell surface-bound PR8 H1N1 and Panama
H3N2 viruses at an MOI of 5 were exposed to fusion buffer (pH 5.0 or 7.4) at 37°C for 2 min, after which the cells were cultured for 8 h in the presence of 30 mM
NH4Cl to block virus fusion in the endocytic pathway. (Left) Immunofluorescence microscopy (20⫻ objective lens). Cells were stained with anti-NP antibody
(green) and DAPI (blue) for nuclei. (Right) Viral infectivity expressed as the mean percentage of NP-positive cells from three independent experiments. The error
bars represent SD. Two-way analysis of variance (ANOVA) was done to compare viral infectivity at pH 5.0 and at pH 7.4, ***, P ⬍ 0.001.

As PR8 virus demonstrated efficient binding and internaliza-
tion into both HULEC and A549 cells, we next investigated if PR8
virus could fuse in the endosomes of endothelial cells. Following
previously established dual-wavelength imaging methods to mon-
itor influenza virus fusion (31), we labeled purified PR8 virus with
dual lipophilic dyes, R18 (red) and SP-DiOC18 (green). In this
assay, lipid mixing between the viral membrane and host endo-
somal membrane correlates with the fluorescence color shift from
red to green due to the release of self-quenching DiOC18 and
dissolution of fluorescent resonance energy transfer (FRET) from
DiOC18 to R18 in the labeled viruses. As shown in Fig. 3B (top
row), upon binding at 4°C, cell-bound viruses could be visualized
only in the red channel (R18) in HULEC. Upon inducing inter-
nalization for 1.5 h, the green fluorescence signal was increased,
indicating that PR8 virus can initiate hemifusion (lipid mixing) in
HULEC (Fig. 3B, middle row). As a negative fusion control, virus
internalization was induced in the presence of 30 mM NH4Cl to
inhibit endosomal acidification; as a result, internalized viruses
could no longer fuse in HULEC, as demonstrated by the lack of
green signal in Fig. 3B (bottom row). PR8 virus fusion in A549
cells was included as a positive control, with results obtained in
this cell line generally similar to those observed for HULEC (data
not shown).
Following confirmation of successful viral lipid mixing in en-
dosomes of endothelial cells, we next monitored uncoating of the
nucleocapsid by M1 protein staining with HB64 antibody in the
presence of 1 mM cycloheximide to prevent synthesis of new viral
proteins (37). The HB64 antibody recognizes the epitope in M1
protein, which becomes more accessible once the M1 protein is
dispersed into the cytosol (38). In control A549 cells at 3.5 h
postinternalization, PR8 M1 protein displayed bright dispersed
cytoplasmic signals, indicating that viruses had undergone un-
coating events, releasing M1 proteins into the cytosol (Fig. 3C). In
contrast, in HULEC, the M1 protein exhibited a punctate staining
pattern, suggesting the M1 protein was confined in vacuole-like
compartments and failed to be released into the cytosol. In order
to determine the location of M1 proteins in HULEC, we per-
formed colocalization immunostaining with an early endosome
marker (EEA1) and a late endosome/lysosome marker (LAMP-1).
The overlap coefficient of M1 with EEA1 was 0.34, and it was 0.73
for M1 with LAMP-1 (Fig. 3C), demonstrating that PR8 virus M1
protein was mainly trapped in late endosome/lysosome compart-

December 2016 Volume 90 Number 24 Journal of Virology jvi.asm.org 11161

Sun et al.
FIG 3 PR8 virus binding, internalization, fusion, and uncoating in HULEC and
A549 cells. (A) Biotin-labeled PR8 virus binding and internalization. Cell surface-
bound PR8 virus following binding at 4°C for 1 h was visualized by Alexa-488
streptavidin (strep) (left column); the signal from surface-bound PR8 viruses
could be blocked by preincubation with unconjugated streptavidin (1 mg/ml)
(second column from left). Upon inducing internalization at 37°C for 1 h, the cells
were incubated with unconjugated streptavidin to block the signal from surface-
bound viruses, and then the cells were fixed and permeabilized before being
stained with Alexa-488 streptavidin for internalized PR8 virus visualization (right
column). Both cell surface-bound and internalized viruses were shown by Alexa-
488 streptavidin staining without unconjugated preincubatioin (third column
from left). The images were taken under a 63⫻ oil objective lens. (B) Virus-medi-
ated lipid mixing in endosomes. R18 (red) and SP-DiOC18 (green) dual-color-
labeled PR8 viruses induced internalization at 37°C for 1.5 h; the lipid mixing
(hemifusion) between the viral membrane and the host endosomal membrane
was indicated by increased intensity of green fluorescence due to the release of
self-quenching DiOC18 and dissolution of FRET. The images were taken under a
63⫻ oil objective lens. (C) (Top) PR8 virus uncoating. PR8 virus M1 proteins were
visualized by immunofluorescence using anti-M1 antibody (HB64) at 3.5 h
postinternalization and DAPI (blue) for nuclei (40⫻ objective lens). (Bottom)
The PR8 virus M1 proteins (green) in HULEC were colocalized with the early
endosome marker EEA1 (red) and the lysosomal marker LAMP-1 (red), and the
overlap coefficients (ranging from 0 to 1, with 1 representing perfectly colocalized
11162 jvi.asm.org Journal of Virology
ments during entry into HULEC. Taken together, these results
show that the blocking of human influenza virus infection in
HULEC is at a posthemifusion step; as a result, virus can no longer
uncoat nucleocapsid and release the viral genome into the cytosol.
The virus replication restriction factor IFITM3 is constitu-
tively expressed in human endothelial cells but at a significantly
lower level in epithelial cells. The infection block in HULEC at a
posthemifusion stage prompted us to investigate whether the
IFITM3 protein, which has recently been suggested to block
influenza virus infection at the same stage (39), plays a similar role
in influenza virus infection in endothelial cells. IFITM proteins
are generally expressed in cells at a basal level but can be signifi-
cantly induced by type I and type II IFNs (20). We first examined
by Western blotting whether human endothelial cells constitu-
tively express IFITM3 without interferon induction. As shown
in Fig. 4A, cell lysates from endothelial cells, including HULEC,
HMVEC, and HUVEC, showed high levels of endogenous IFITM3
protein expression; HBVEC expressed a much lower level of IFITM3
protein. Compared to endothelial cells, human lung epithelial
cells, including A549 and Calu-3 cells, failed to display IFITM3
protein expression under the same detection conditions, and
NHBE cells had only marginal IFITM3 protein expression com-
pared to that in HULEC.
Using immunofluorescence microscopy, we next examined
the locations of endogenous IFITM3 proteins in HULEC. In this
cell type, IFITM3 showed a punctate staining pattern close to nu-
clear regions. In a colocalization study, we found that the overlap
coefficient between IFITM3 and the early endosome marker EEA1
was 0.76, and it was 0.67 between IFITM3 and the late endosome/
lysosome marker LAMP-1, indicating endogenous IFITM3 is
mainly located in endosomal compartments (Fig. 4B). Taken to-
gether, our results show that, unlike most respiratory epithelial
cells, human pulmonary endothelial cells constitutively express
high levels of IFITM3 in endosomes, which may potentially re-
strict influenza virus infection.
IFITM3 knockdown by siRNA in HULEC partially enhances
human influenza virus infection. To determine whether consti-
tutively expressed IFITM3 proteins are involved in the restriction
of influenza virus infection in HULEC, we used silencing RNA
(siRNA) to knock down IFITM3 expression in HULEC prior to
virus infection. As shown in Fig. 5A, 100 pmol of IFITM3 siRNA
could significantly downregulate IFITM3 expression, as demon-
strated by Western blotting (Fig. 5A). Following viral infection
with PR8 virus, IFITM3 siRNA-transfected cells showed signifi-
cantly enhanced viral infectivity (P ⬍ 0.001), with approximately
12-fold (21% versus 1.7%) and 8-fold (47% versus 5.8%) in-
creases compared to cells transfected with negative-control siRNA
at MOI of 0.25 and 1, respectively (Fig. 5B). However, knocking
down IFITM3 expression in HULEC had a less significant effect
on Anhui:PR8, as Anhui:PR8 virus infectivity increased from 28%
to 35% at an MOI of 0.25 and increased from 46% to 56% at an
MOI of 1 (Fig. 5B). The different restriction effects of endoge-
nously expressed IFITM3 in HULEC on human and avian influ-
enza viruses seem to be cell type specific. Thus, IFITM3 proteins
pixels) of the green and red signals were calculated with the program Zen from
Zeiss and are marked in the right corners of the images. The images were taken
with an LSM 710 Zeiss inverted confocal microscope under a 40⫻ objective lens
with digital zoom 2. The scale bars represent 10 ␮m.
December 2016 Volume 90 Number 24
 Influenza Virus Infection in Pulmonary Endothelial Cells

FIG 4 Constitutive IFITM3 expression in human endothelial cells. (A) IFITM3 expression in endothelial cell lysates (HULEC, HMVEC, HUVEC, and HBVEC)
and epithelial cell lysates (Calu-3, A549, and NHBE cells) was detected by anti-IFITM3 antibody in a Western blot assay; actin expression monitored by
anti-␤-actin antibody blotting was used as an internal loading control. Actin and IFITM3 protein expression levels in the Western blot were quantified with a
ChemiDoc MP system with Image Lab software. The relative IFITM3 protein expression was plotted as the mean ratio of the intensity of the IFITM3 signal to that
of the actin internal control from three independent experiments. (B) Localization of IFITM3 in HULEC. The endogenous IFITM3 proteins (red) in HULEC
were colocalized with the early endosome marker EEA1 (green) and the lysosomal marker LAMP-1 (green), and the overlap coefficients of the green and red
signals were calculated with the program Zen from Zeiss and marked in the right corners of the images. DAPI (blue) was used to stain nuclei. 63⫻ oil objective
lenses were used in microscopy.

overexpressed in A549 cells exhibited similar degrees of inhi-
bition of all the influenza viruses we tested (Fig. 5C), which was
consistent with the results of a previous study (22). Our results
demonstrate that constitutively expressed IFITM3 proteins are
able to selectively restrict human influenza virus infection in
endothelial cells but have a less potent effect on certain avian
influenza viruses.
A higher virus fusion threshold is not associated with escape
from IFITM3 restriction. Thus far, we have shown that human
influenza virus entry into endothelial cells is blocked at a pos-
themifusion step, partially due to the constitutive expression of
IFITM3. However, certain avian influenza viruses, including
HPAI H5N1, LPAI H9N2, and novel LPAI H7N9 viruses, show
significantly higher infection rates than H1N1 and H3N2 human

December 2016 Volume 90 Number 24 Journal of Virology jvi.asm.org 11163

Sun et al.
FIG 5 Constitutively expressed IFITM3 proteins restrict PR8 virus infection
in HULEC. (A) Different amounts (12.5 to 100 pmol) of IFITM3 siRNA and
negative-control siRNA were transfected in HULEC, and the expression of
IFITM3 was examined at 72 h posttransfection by Western blotting with anti-
IFITM3 antibody; anti-actin antibody was included as an internal loading
control. (B) HULEC transfected with 100 pmol of IFITM3 siRNA for 72 h were
infected with PR8 or Anhui:PR8 virus at MOI of 0.25 and 1 for 8 h before viral
infectivity was quantified based on NP staining by immunofluorescence mi-
croscopy. The graph represents the mean infectivities of three independent
experiments, with the error bars showing SD. Two-way ANOVA statistical
analysis was performed to compare viral infectivities between IFITM3 siRNA
and negative-control siRNA groups; ***, P ⬍ 0.001; *, P ⬍ 0.05. (C) A549 cells
stably expressing IFITM3 or vector pQCXIP were infected with the indicated
influenza viruses at an MOI of approximately 1, and viral infectivity at 8 h
postinfection was quantified by NP-positive cells in immunofluorescence mi-
croscopy. Shown are the means of three independent experiments with SD.
Two-way ANOVA statistical analysis was performed to compare viral infectiv-
ities in IFITM3 and vector-expressing A549 cells.
influenza viruses in endothelial cells. One hypothesis for how
avian influenza viruses circumvent this infection block is that
avian influenza viruses can fuse in early endosomes at relatively
high pH values before traveling to late endosomes, where IFITM3
proteins have been suggested to be located and to restrict influ-
11164 jvi.asm.org Journal of Virology
FIG 6 pH fusion thresholds for influenza viruses. Influenza viruses at 128
HAU were incubated with 1% turkey erythrocytes at 4°C for 1 h before fusion
was induced with 100 ␮l of fusion buffer at pHs ranging from 4.8 to 7.4. The
hemolysis was measured as the OD405 from duplicate samples, and the maxi-
mal hemolysis was normalized to 100%. The hemolysis efficiencies at various
pHs were expressed as percentages of the maximal hemolysis. The graph rep-
resents the average hemolysis efficiencies from three independent experi-
ments. The pH thresholds for fusion (the pH range at which 50% of maximal
hemolysis was achieved) are shown in the table. Viral infectivities in HULEC
were classified as high, low, or intermediate based on the viral infection rates
shown in Fig. 1.
enza virus entry. To test this possibility, we used a virus-mediated
hemolysis assay to compare viral fusion pHs for selected human
and avian influenza viruses. As shown in Fig. 6, human H1N1
influenza viruses, including PR8, Bris/09, and Mex/09, preferen-
tially fused at a relatively low pH (⬍5.3). In contrast, the viruses
that displayed the highest infectivity in HULEC (Anhui/13
[H7N9] and VN/04 [H5N1] viruses) had a fusion threshold of
approximately pH 5.8 (Fig. 6). The H7 viruses NL/03 (H7N7) and
Shov/07 (H7N9), which showed intermediate infection rates in
HULEC, possessed slightly lower pH fusion thresholds (pH 5.5)
than the Anhui/13 (H7N9) and VN/04 (H5N1) viruses. However,
human Pan/99 (H3N2) virus shared a similar pH fusion profile
with VN/04 virus despite the different infection rates of these two
viruses in HULEC (Fig. 1 and 5). Similarly, CK/11 (H9N2) virus
had a pH fusion range similar to that of human H1N1 viruses but
was able to infect HULEC with much higher efficiency than H1N1
virus. Our results suggest that although having a relatively high
fusion threshold might prompt some avian influenza viruses, like
Anhui/1 and VN/04, to fuse early before reaching late endosomes
to avoid IFITM3 restriction, other avian influenza viruses, such as
CK/11 virus, may explore other, unknown routes to escape this
restriction in endothelial cells.
December 2016 Volume 90 Number 24

DISCUSSION
Human pulmonary cells constitute approximately 30% of the to-
tal cell population in the lung, providing the lining for the network
of capillaries in the alveolus (40). Recently, pulmonary endothe-
lium dysfunction caused by cytokines released from alveolar epi-
thelial cells and leukocytes has been suggested to play an impor-
tant role in virus-induced lung damage in severe cases of influenza
virus infection (41). Additionally, given the close proximity of the
pulmonary endothelium and alveolar epithelium, endothelial
cells exposed to influenza virus particles released from infected
neighboring epithelial cells may lead to cell death, increased en-
dothelial permeability, and vascular destruction, further contrib-
uting to epithelial-endothelial barrier damage (41). However, pre-
vious studies in vitro have shown that pulmonary endothelial cells
are not susceptible to infection with most human influenza vi-
ruses, as only HPAI H5N1 viruses were able to replicate efficiently
and induce excessive production of proinflammatory cytokines
(17, 19). Here, we extend previous work on influenza virus tro-
pism in human pulmonary endothelial cells by identifying cellular
and viral factors that contribute to viral infection in this cell type.
We showed that human pulmonary endothelial HULEC are less
permissive to human influenza virus infection than a subset of
avian influenza viruses, including highly pathogenic H5N1 (VN/
04), LPAI H7N9, and H9N2 viruses. Despite limited expression of
␣-2,6-linked sialic acid residues (17), the receptor for human in-
fluenza viruses, these viruses are still able to bind to endothelial
cells with efficiency comparable to that of binding on epithelial
cells. This was shown not only by the biotin-labeled-virus binding
assay (Fig. 3A), but also by an acid bypass infection assay (Fig. 2).
We further showed that human PR8 virus could be internalized
and successfully initiate hemifusion in endothelial cells but could
not uncoat and release the viral genome into the cytosol to initiate
subsequent replication processes. Although human influenza vi-
rus infection is blocked at the early stages of viral entry in endo-
thelial cells, this does not necessarily mean that the pulmonary
endothelium cannot be activated through direct contact with hu-
man influenza viruses. It has been previously reported that a UV-
inactivated, replication-deficient human seasonal H3N2 virus was
capable of causing a significant reduction in endothelial permea-
bility by inducing degradation of the tight-junction protein Clau-
din 5. This was accomplished without causing cytotoxic effects on
the endothelium, and binding of the virus alone was not sufficient
to induce permeability changes (14). We postulate that host pat-
tern recognition receptors (PRR) residing in host endocytic path-
ways, such as Toll-like receptor 7 (TLR7) or TLR8 in endosomes,
might be involved in recognizing viral components and activating
signal transductions in endothelial cells in a viral-replication-in-
dependent manner (42). Despite the fact that avian influenza vi-
ruses, such as HPAI H5N1 and LPAI H7N9, were able to replicate
efficiently in endothelial cells and to induce strong cytokine pro-
duction in vitro, further studies are needed to elucidate their exact
role in viral pathogenesis in vivo, as postmortem analysis of pa-
tients who succumbed to H5N1 infection rarely showed endothe-
liotropism (6).
In this study, we revealed that, unlike human bronchial epithe-
lial Calu-3 and alveolar basal epithelium-derived A549 cells, hu-
man primary (HMVEC) or transformed (HULEC) pulmonary
endothelial cells, as well as HUVEC, constitutively express high
levels of IFITM3 without IFN induction, whereas NHBE cells
December 2016 Volume 90 Number 24 Journal of Virology
Influenza Virus Infection in Pulmonary Endothelial Cells
showed only intermediate levels of IFITM3 expression. Although
comprehensive immunohistochemical staining for IFITM3 ex-
pression in human lung tissues is still lacking, various levels of
endogenous IFITM3 expression in mouse respiratory epithelial
cells and pulmonary endothelial cells have been revealed (25).
This suggests that endogenous expression of IFITM3 may be tis-
sue or cell type specific; accordingly, higher endogenous expres-
sion of IFITM3 might provide a certain degree of advantage in
protecting the host from influenza virus infection prior to inter-
feron production. Although the antiviral activity of IFITM3 has
been well established, the exact mechanism by which IFITM3 re-
stricts viral infection is not yet fully understood. Early evidence
suggests that IFITM3 restricts viral-membrane hemifusion, pos-
sibly by affecting the membrane’s molecular order and fluidity
(43) or by modulating cholesterol homeostasis in the endosome
by interacting with VAPA (vesicle-associated membrane protein-
associated protein A), resulting in inhibition of viral fusion (44).
However, it was subsequently found that IFITM3 restricts viral
infection by blocking the formation of fusion pores following vi-
rus-endosome hemifusion, possibly by modulating endosomal-
membrane rigidity (39). Furthermore, IFITM3 subcellular local-
izations, which can be modulated by protein posttranslational
modification, has proven to be important for its antiviral activi-
ties, and depending on the cell types and stimulation status,
IFITM3 may localize differently in cells (45). In unstimulated
WI-38 human primary fibroblasts, the majority of IFITM3 resides
in the ER and becomes distributed in a vesicular pattern through-
out the cell upon IFN exposure (22). When overexpressed in A549
cells, IFITM3 predominantly localizes in enlarged compartments
shared with markers of endosomes and lysosomes (23). In our
study, we showed that the constitutively expressed IFITM3 in
HULEC mainly localized in endosomal and lysosomal compart-
ments, as indicated by their colocalization with the early endo-
some and late endosome/lysosome markers EEA1 and LAMP-1,
respectively, suggesting that endogenous IFITM3 proteins in
HULEC already reside in the compartments involved in the influ-
enza virus entry pathway and might be intrinsically primed for
blocking influenza virus infection. However, we did not observe
IFITM3-enriched large compartments, as previously revealed in
overexpressed or IFN-induced A549 cells; whether this reflects a
difference between cell types or between endogenously expressed
versus overexpressed IFITM3 requires further investigation (39).
In our study, we demonstrated that PR8 viral infection in HULEC
was blocked after a hemifusion step. This suggests a potential role
for endosome-residing constitutively expressed IFITM3 proteins
in this block, based on our findings that knocking down IFITM3
expression with siRNA significantly improved PR8 virus infectiv-
ity in HULEC. However, the improved infectivity in HULEC upon
IFITM3 siRNA treatment is still lower than that in A549 cells,
suggesting a role for other restriction factors in endothelial cells.
Moreover, a subset of avian influenza viruses from the H5, H7,
and H9 subtypes seemed to be able to partially escape this IFITM3
restriction block in endothelial cells, as IFITM3 siRNA knock-
down had only a moderate effect on improving the infectivity of a
recombinant PR8 virus bearing the Anhui/1 H7N9 HA and NA.
Future studies exploring whether interferon-induced IFITM3 in
pulmonary endothelial cells, which may have different subcellular
locations, has an effect on avian influenza virus infection similar
to that of constitutively expressed IFITM3 are warranted.
Most influenza viruses have a pH fusion range of 5.0 to 5.5,
jvi.asm.org 11165
Sun et al.
which matches the pH in the late endosome (46). However, cer-
tain influenza viruses, such as the novel H7N9 virus Anhui/1, have
a higher pH threshold (pH 5.8) for fusion; such viruses may be
able to fuse in early endosomes, which have pH values ranging
from 5.4 to 6.2, depending on the cell type (47, 48). IFITM3 is
mostly located in late endosomes or lysosomes when overex-
pressed or upon stimulation by IFN (23, 49) and has less inhibi-
tory effects on viruses, such as vesicular stomatitis virus (VSV),
that preferentially fuse at early endosomes or late endosome/mul-
tivesicular bodies (50); therefore, IFITM3 might be less potent in
restricting influenza viruses, which can fuse in early endosomes.
In our study, we did observe that VN/04 H5N1 and Anhui/13
H7N9 viruses, which showed efficient viral replication in HULEC,
had a higher pH threshold (pH ⬎ 5.5) for fusion, indicating that
the viruses could potentially escape IFITM3 restriction by fusing
early, before reaching late endosomes. However, CK/11 H9N2
virus was able to achieve high infectivity in HULEC despite having
a low pH fusion threshold, suggesting that certain influenza vi-
ruses may have developed different strategies to counteract
IFITM3 restriction in endothelial cells.
Altogether, we demonstrated that human influenza virus in-
fection in human pulmonary endothelial cells is less efficient than
that of selected strains of H5, H7, and H9 avian influenza viruses,
partially due to the high constitutive expression of IFITM3 in late
endosomes and lysosomes, which may block the procession from
hemifusion to fusion pore formation during the viral entry pro-
cess. In comparison, certain avian influenza viruses were able to
escape the restriction in endothelial cells by fusing early in the
endocytic pathway at a higher pH value or by other, unknown
mechanisms. However, we cannot rule out the possibility that
other, unidentified restriction factors may exist in human pulmo-
nary endothelial cells that confer their resistance to influenza virus
infection. An additional factor to be mentioned is the binding
preference of the laboratory-adapted human influenza PR8 virus,
with ␣-2,6- and ␣-2,3-linked sialic acid dual receptor binding
specificity. This virus was able to bind endothelial cells (followed
by internalization of the virus), but whether PR8 viruses are trans-
ported into the same endocytic compartments as avian influenza
viruses, which preferentially bind to ␣-2,3-linked sialic acids, is
unknown. Previously, it had been suggested that HPAI H5N1 in-
fluenza viruses with different receptor binding specificities might
be sensed or recognized differently in human monocyte-derived
macrophages and dendritic cells, subsequently leading to distinct
signaling cascades (51). Whether influenza virus receptor speci-
ficity is involved in human pulmonary endothelial cell restriction
to human influenza viruses requires further study.
ACKNOWLEDGMENTS
The findings and conclusions in this report are ours and do not necessarily
represent the official position of the Centers for Disease Control and Pre-
vention.
X. Sun was supported by the Oak Ridge Institute for Science and Ed-
ucation.
FUNDING INFORMATION
This work, including the efforts of Xiangjie Sun, was funded by HHS |
Centers for Disease Control and Prevention (CDC).
REFERENCES
1. Palese PSM. 2007. Orthomyxoviridae: the viruses and their replication, p
1647–1689 In Knipe DM, Howley PM, Griffin DE, Lamb RA, Martin MA,
11166 jvi.asm.org Journal of Virology
Roizman B, Straus SE (ed), Fields virology, 5th ed. Lippincott Williams &
Wilkins, Philadelphia, PA.
2. Kuiken T, Riteau B, Fouchier RA, Rimmelzwaan GF. 2012. Pathogenesis
of influenza virus infections: the good, the bad and the ugly. Curr Opin
Virol 2:276 –286. http://dx.doi.org/10.1016/j.coviro.2012.02.013.
3. Poovorawan Y, Pyungporn S, Prachayangprecha S, Makkoch J. 2013.
Global alert to avian influenza virus infection: from H5N1 to H7N9. Pat-
hog Glob Health 107:217–223. http://dx.doi.org/10.1179/2047773213Y
.0000000103.
4. Liu Q, Liu DY, Yang ZQ. 2013. Characteristics of human infection with
avian influenza viruses and development of new antiviral agents. Acta
Pharmacol Sin 34:1257–1269. http://dx.doi.org/10.1038/aps.2013.121.
5. WHO. 2016. Monthly risk assessment summary: influenza at the human-
animal interface. http://www.who.int/influenza/human_animal_interface
/HAI_Risk_Assessment/en. Accessed 16 May 2016.
6. Gu J, Xie Z, Gao Z, Liu J, Korteweg C, Ye J, Lau LT, Lu J, Gao Z, Zhang
B, McNutt MA, Lu M, Anderson VM, Gong E, Yu AC, Lipkin WI. 2007.
H5N1 infection of the respiratory tract and beyond: a molecular pathology
study. Lancet 370:1137–1145. http://dx.doi.org/10.1016/S0140-6736(07)
61515-3.
7. Perrone LA, Tumpey TM. 2007. Reconstruction of the 1918 pandemic
influenza virus: how revealing the molecular secrets of the virus responsi-
ble for the worst pandemic in recorded history can guide our response to
future influenza pandemics. Infect Disord Drug Targets 7:294 –303. http:
//dx.doi.org/10.2174/187152607783018772.
8. Yeh E, Luo RF, Dyner L, Hong DK, Banaei N, Baron EJ, Pinsky BA.
2010. Preferential lower respiratory tract infection in swine-origin 2009
A(H1N1) influenza. Clin Infect Dis 50:391–394. http://dx.doi.org/10
.1086/649875.
9. Gao R, Bhatnagar J, Blau DM, Greer P, Rollin DC, Denison AM,
Deleon-Carnes M, Shieh WJ, Sambhara S, Tumpey TM, Patel M, Liu L,
Paddock C, Drew C, Shu Y, Katz JM, Zaki SR. 2013. Cytokine and
chemokine profiles in lung tissues from fatal cases of 2009 pandemic in-
fluenza A (H1N1): role of the host immune response in pathogenesis. Am
J Pathol 183:1258 –1268. http://dx.doi.org/10.1016/j.ajpath.2013.06.023.
10. Herold S, Becker C, Ridge KM, Budinger GR. 2015. Influenza virus-
induced lung injury: pathogenesis and implications for treatment. Eur
Respir J 45:1463–1478. http://dx.doi.org/10.1183/09031936.00186214.
11. Fukuyama S, Kawaoka Y. 2011. The pathogenesis of influenza virus
infections: the contributions of virus and host factors. Curr Opin Immu-
nol 23:481– 486. http://dx.doi.org/10.1016/j.coi.2011.07.016.
12. Short KR, Kroeze EJ, Fouchier RA, Kuiken T. 2014. Pathogenesis of
influenza-induced acute respiratory distress syndrome. Lancet Infect Dis
14:57– 69. http://dx.doi.org/10.1016/S1473-3099(13)70286-X.
13. Teijaro JR, Walsh KB, Cahalan S, Fremgen DM, Roberts E, Scott F,
Martinborough E, Peach R, Oldstone MB, Rosen H. 2011. Endothelial
cells are central orchestrators of cytokine amplification during influenza
virus infection. Cell 146:980 –991. http://dx.doi.org/10.1016/j.cell.2011
.08.015.
14. Armstrong SM, Wang C, Tigdi J, Si X, Dumpit C, Charles S, Gamage
A, Moraes TJ, Lee WL. 2012. Influenza infects lung microvascular endo-
thelium leading to microvascular leak: role of apoptosis and claudin-5.
PLoS One 7:e47323. http://dx.doi.org/10.1371/journal.pone.0047323.
15. Hiyoshi M, Indalao IL, Yano M, Yamane K, Takahashi E, Kido H. 2015.
Influenza A virus infection of vascular endothelial cells induces GSK-
3beta-mediated beta-catenin degradation in adherens junctions, with a
resultant increase in membrane permeability. Arch Virol 160:225–234.
http://dx.doi.org/10.1007/s00705-014-2270-5.
16. Narasaraju T, Yang E, Samy RP, Ng HH, Poh WP, Liew AA, Phoon
MC, van Rooijen N, Chow VT. 2011. Excessive neutrophils and neutro-
phil extracellular traps contribute to acute lung injury of influenza pneu-
monitis. Am J Pathol 179:199 –210. http://dx.doi.org/10.1016/j.ajpath
.2011.03.013.
17. Zeng H, Pappas C, Belser JA, Houser KV, Zhong W, Wadford DA,
Stevens T, Balczon R, Katz JM, Tumpey TM. 2012. Human pulmonary
microvascular endothelial cells support productive replication of highly
pathogenic avian influenza viruses: possible involvement in the pathogen-
esis of human H5N1 virus infection. J Virol 86:667– 678. http://dx.doi.org
/10.1128/JVI.06348-11.
18. Chan MC, Chan RW, Yu WC, Ho CC, Chui WH, Lo CK, Yuen KM,
Guan YI, Nicholls JM, Peiris JS. 2009. Influenza H5N1 virus infection of
polarized human alveolar epithelial cells and lung microvascular endothe-
lial cells. Respir Res 10:102. http://dx.doi.org/10.1186/1465-9921-10-102.
December 2016 Volume 90 Number 24

19. Ocana-Macchi M, Bel M, Guzylack-Piriou L, Ruggli N, Liniger M,
McCullough KC, Sakoda Y, Isoda N, Matrosovich M, Summerfield A.
2009. Hemagglutinin-dependent tropism of H5N1 avian influenza virus
for human endothelial cells. J Virol 83:12947–12955. http://dx.doi.org/10
.1128/JVI.00468-09.
20. Friedman RL, Manly SP, McMahon M, Kerr IM, Stark GR. 1984.
Transcriptional and posttranscriptional regulation of interferon-induced
gene expression in human cells. Cell 38:745–755. http://dx.doi.org/10
.1016/0092-8674(84)90270-8.
21. Hickford D, Frankenberg S, Shaw G, Renfree MB. 2012. Evolution of
vertebrate interferon inducible transmembrane proteins. BMC Genomics
13:155. http://dx.doi.org/10.1186/1471-2164-13-155.
22. Brass AL, Huang IC, Benita Y, John SP, Krishnan MN, Feeley EM, Ryan
BJ, Weyer JL, van der Weyden L, Fikrig E, Adams DJ, Xavier RJ, Farzan
M, Elledge SJ. 2009. The IFITM proteins mediate cellular resistance to
influenza A H1N1 virus, West Nile virus, and dengue virus. Cell 139:1243–
1254. http://dx.doi.org/10.1016/j.cell.2009.12.017.
23. Huang IC, Bailey CC, Weyer JL, Radoshitzky SR, Becker MM, Chiang
JJ, Brass AL, Ahmed AA, Chi X, Dong L, Longobardi LE, Boltz D, Kuhn
JH, Elledge SJ, Bavari S, Denison MR, Choe H, Farzan M. 2011. Distinct
patterns of IFITM-mediated restriction of filoviruses, SARS coronavirus,
and influenza A virus. PLoS Pathog 7:e1001258. http://dx.doi.org/10.1371
/journal.ppat.1001258.
24. Jiang D, Weidner JM, Qing M, Pan XB, Guo H, Xu C, Zhang X, Birk
A, Chang J, Shi PY, Block TM, Guo JT. 2010. Identification of five
interferon-induced cellular proteins that inhibit west Nile virus and den-
gue virus infections. J Virol 84:8332– 8341. http://dx.doi.org/10.1128/JVI
.02199-09.
25. Bailey CC, Huang IC, Kam C, Farzan M. 2012. Ifitm3 limits the severity
of acute influenza in mice. PLoS Pathog 8:e1002909. http://dx.doi.org/10
.1371/journal.ppat.1002909.
26. Everitt AR, Clare S, Pertel T, John SP, Wash RS, Smith SE, Chin CR,
Feeley EM, Sims JS, Adams DJ, Wise HM, Kane L, Goulding D, Digard
P, Anttila V, Baillie JK, Walsh TS, Hume DA, Palotie A, Xue Y,
Colonna V, Tyler-Smith C, Dunning J, Gordon SB, Gen II, Investiga-
tors M, Smyth RL, Openshaw PJ, Dougan G, Brass AL, Kellam P. 2012.
IFITM3 restricts the morbidity and mortality associated with influenza.
Nature 484:519 –523. http://dx.doi.org/10.1038/nature10921.
27. Wang Z, Zhang A, Wan Y, Liu X, Qiu C, Xi X, Ren Y, Wang J, Dong
Y, Bao M, Li L, Zhou M, Yuan S, Sun J, Zhu Z, Chen L, Li Q, Zhang
Z, Zhang X, Lu S, Doherty PC, Kedzierska K, Xu J. 2014. Early hyper-
cytokinemia is associated with interferon-induced transmembrane pro-
tein-3 dysfunction and predictive of fatal H7N9 infection. Proc Natl Acad
Sci U S A 111:769 –774. http://dx.doi.org/10.1073/pnas.1321748111.
28. Zeng H, Pappas C, Katz JM, Tumpey TM. 2011. The 2009 pandemic
H1N1 and triple-reassortant swine H1N1 influenza viruses replicate effi-
ciently but elicit an attenuated inflammatory response in polarized human
bronchial epithelial cells. J Virol 85:686 – 696. http://dx.doi.org/10.1128
/JVI.01568-10.
29. Sun X, Cao W, Pappas C, Liu F, Katz JM, Tumpey TM. 2014. Effect of
receptor binding specificity on the immunogenicity and protective effi-
cacy of influenza virus A H1 vaccines. Virology 464-465:156 –165. http:
//dx.doi.org/10.1016/j.virol.2014.07.004.
30. Sun X, Whittaker GR. 2007. Role of the actin cytoskeleton during influ-
enza virus internalization into polarized epithelial cells. Cell Microbiol
9:1672–1682. http://dx.doi.org/10.1111/j.1462-5822.2007.00900.x.
31. Sakai T, Ohuchi M, Imai M, Mizuno T, Kawasaki K, Kuroda K,
Yamashina S. 2006. Dual wavelength imaging allows analysis of mem-
brane fusion of influenza virus inside cells. J Virol 80:2013–2018. http://dx
.doi.org/10.1128/JVI.80.4.2013-2018.2006.
32. Stauffer S, Feng Y, Nebioglu F, Heilig R, Picotti P, Helenius A. 2014.
Stepwise priming by acidic pH and a high K⫹ concentration is required
for efficient uncoating of influenza A virus cores after penetration. J Virol
88:13029 –13046. http://dx.doi.org/10.1128/JVI.01430-14.
33. Sun X, Yau VK, Briggs BJ, Whittaker GR. 2005. Role of clathrin-
December 2016 Volume 90 Number 24 Journal of Virology
Influenza Virus Infection in Pulmonary Endothelial Cells
mediated endocytosis during vesicular stomatitis virus entry into host
cells. Virology 338:53– 60. http://dx.doi.org/10.1016/j.virol.2005.05.006.
34. Manders EMM, Verbeek FJ, Aten JA. 1993. Measurement of co-
localization of objects in dual-colour confocal images. J Microsc 169:375–
382. http://dx.doi.org/10.1111/j.1365-2818.1993.tb03313.x.
35. Shelton H, Roberts KL, Molesti E, Temperton N, Barclay WS. 2013.
Mutations in haemagglutinin that affect receptor binding and pH stability
increase replication of a PR8 influenza virus with H5 HA in the upper
respiratory tract of ferrets and may contribute to transmissibility. J Gen
Virol 94:1220 –1229. http://dx.doi.org/10.1099/vir.0.050526-0.
36. Razinkov VI, Melikyan GB, Cohen FS. 1999. Hemifusion between cells
expressing hemagglutinin of influenza virus and planar membranes can pre-
cede the formation of fusion pores that subsequently fully enlarge. Biophys J
77:3144 –3151. http://dx.doi.org/10.1016/S0006-3495(99)77144-4.
37. Banerjee I, Yamauchi Y, Helenius A, Horvath P. 2013. High-content
analysis of sequential events during the early phase of influenza A virus
infection. PLoS One 8:e68450. http://dx.doi.org/10.1371/journal.pone
.0068450.
38. Martin K, Helenius A. 1991. Transport of incoming influenza virus nu-
cleocapsids into the nucleus. J Virol 65:232–244.
39. Desai TM, Marin M, Chin CR, Savidis G, Brass AL, Melikyan GB. 2014.
IFITM3 restricts influenza A virus entry by blocking the formation of
fusion pores following virus-endosome hemifusion. PLoS Pathog 10:
e1004048. http://dx.doi.org/10.1371/journal.ppat.1004048.
40. Piantadosi CA, Schwartz DA. 2004. The acute respiratory distress syn-
drome. Ann Intern Med 141:460 – 470. http://dx.doi.org/10.7326/0003
-4819-141-6-200409210-00012.
41. Armstrong SM, Darwish I, Lee WL. 2013. Endothelial activation and
dysfunction in the pathogenesis of influenza A virus infection. Virulence
4:537–542. http://dx.doi.org/10.4161/viru.25779.
42. Ichinohe T, Pang IK, Iwasaki A. 2010. Influenza virus activates inflam-
masomes via its intracellular M2 ion channel. Nat Immunol 11:404 – 410.
http://dx.doi.org/10.1038/ni.1861.
43. Li K, Markosyan RM, Zheng YM, Golfetto O, Bungart B, Li M, Ding S,
He Y, Liang C, Lee JC, Gratton E, Cohen FS, Liu SL. 2013. IFITM
proteins restrict viral membrane hemifusion. PLoS Pathog 9:e1003124.
http://dx.doi.org/10.1371/journal.ppat.1003124.
44. Amini-Bavil-Olyaee S, Choi YJ, Lee JH, Shi M, Huang IC, Farzan M,
Jung JU. 2013. The antiviral effector IFITM3 disrupts intracellular cho-
lesterol homeostasis to block viral entry. Cell Host Microbe 13:452– 464.
http://dx.doi.org/10.1016/j.chom.2013.03.006.
45. Chesarino NM, McMichael TM, Yount JS. 2014. Regulation of the traffick-
ing and antiviral activity of IFITM3 by post-translational modifications. Fu-
ture Microbiol 9:1151–1163. http://dx.doi.org/10.2217/fmb.14.65.
46. Sorkin A, Von Zastrow M. 2002. Signal transduction and endocytosis:
close encounters of many kinds. Nat Rev Mol Cell Biol 3:600 – 614. http:
//dx.doi.org/10.1038/nrm883.
47. Murphy RF, Powers S, Cantor CR. 1984. Endosome pH measured in
single cells by dual fluorescence flow cytometry: rapid acidification of
insulin to pH 6. J Cell Biol 98:1757–1762. http://dx.doi.org/10.1083/jcb.98
.5.1757.
48. Rybak SL, Murphy RF. 1998. Primary cell cultures from murine kidney
and heart differ in endosomal pH. J Cell Physiol 176:216 –222.
49. Feeley EM, Sims JS, John SP, Chin CR, Pertel T, Chen LM, Gaiha GD,
Ryan BJ, Donis RO, Elledge SJ, Brass AL. 2011. IFITM3 inhibits influ-
enza A virus infection by preventing cytosolic entry. PLoS Pathog
7:e1002337. http://dx.doi.org/10.1371/journal.ppat.1002337.
50. Weidner JM, Jiang D, Pan XB, Chang J, Block TM, Guo JT. 2010.
Interferon-induced cell membrane proteins, IFITM3 and tetherin, inhibit
vesicular stomatitis virus infection via distinct mechanisms. J Virol 84:
12646 –12657. http://dx.doi.org/10.1128/JVI.01328-10.
51. Ramos I, Bernal-Rubio D, Durham N, Belicha-Villanueva A, Lowen
AC, Steel J, Fernandez-Sesma A. 2011. Effects of receptor binding spec-
ificity of avian influenza virus on the human innate immune response. J
Virol 85:4421– 4431. http://dx.doi.org/10.1128/JVI.02356-10.
jvi.asm.org 11167