Project 2 Host and viral determinants of human clinical outcome in influenza infection

lynxfatkidneyedΔίκτυα και Επικοινωνίες

26 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

175 εμφανίσεις

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

Project 2


Host and viral

determinants of human clinical outcome in influenza infection

The severity of disease during an influenza infection is determined by both the extent of viral replication and
the nature of the host response. These two factors are mutually dependent. For
example
, large amounts

of

virus can drive an extreme host respons
e that results in significant immunopathology
,

including extensive
tissue damage and poor clinical outcome. In contrast, a measured and well
-
timed host response can limit viral
replication.

The goal of this project is to
defin
e
mechanistic features

of the
host response

that
determine

severe and mild
disease.


Recent studies have shown that the molecular networks activated by influenza virus involve thousands of
genes and a vastly larger number of interactions between gene products
(1

4)
. Though the effects of infe
ction
are broad,
our recent work has shown that
diverse influenza viruses induce a conserved host “influenza
response” in infected epithelial cells
, the output of which contains approximately 300 highly induced genes
.
While the core “influenza response” in

airway epithelial cells is qualitatively similar across a wide range of
infections, both the magnitude and kinetics of subsets of the response over the first 24 hours following infection
vary in a strain
-
specific manner that correlates with disease outcom
e.
This observation suggests that the
severity of infection may be attributable to quantitative differences in a smal
l

number of critical virus
-
host interactions.

In particular, our data demonstrate that in some infections, the host response is significant
ly
delayed relative to viral growth through unknown mechanisms that are independent of the predominant
influenza
immune suppressive protein NS1.

T
he challenge
is to filter the

large number of genes and pathways that are activated by infection

and

identify

the much smaller number of proximal virus
-
host interactions that

most strongly contribute to the outcome of
infection, as defined by control of viral replication and non
-
pathogenic immune responses (clinical recovery).

We will attack this problem using t
wo complimentary screening strategies to identify these interactions: one
that exploits the quantitative variations we have observed in the host response to viruses of varying
pathogenicity and one that exploits correlations between human genetic variation

and disease severity.
Importantly, our preliminary data indicate that the critical variations in these virus
-
host interactions that
correlate with outcome are kinetic, requiring a kinetic network model as a framework for interpreting these
data. As critic
al interactions are deciphered, these will be included in updated versions of the model. New
global measures of the host response will be combined with prior data and reinterpreted through the model,
likely suggesting a reprioritization of candidates for t
he next cycle of screening and adding candidates including
those whose transcription is not altered by infection.


Aim 1: Determine the contribution of influenza response network components to host
-
defense in
human airway epithelial cells.

We will systemat
ically assess the functional importance of host response
components by perturbing their expression in airway epithelial cells. The initial set of candidates has been
selected based on our preliminary studies and targets for subsequent experimental cycles w
ill be determined
based on refinement of the kinetic model developed in Aim 3.

Aim 2: Determine the relative contribution of
human polymorphisms in influenza response network
components

to defense against severe disease in humans.

We will analyze the impac
t of

host
polymorphisms in candidate genes by quantifying the attributable risk of genetic variation to the severity
of
disease in a case
-
control study. This investigation will be facilitated by our access to a large number of banked
samples from clinical
studies in which we participate.

Aim 3:

Determine the mechanism by which each gene identified as a critical node in Aims 1 or 2
mediates host
-
defense to influenza.

We will use the tools of systems biology to determine the biological
mechanisms governing th
e function of each gene identified in Aims 1 and 2 and integrate these data into a
kinetic model of influenza infection.







Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

A. SIGNIFICANCE


Influenza A infection persists as a major economic and health burden worldwide. In the United States, it is
estimated that 20% of the population is infected with seasonal influenza annually, leading to ~200,000
hospitalizations and nearly 40,000
deaths

(5

7)
. The economic burden of seasonal influenza infection in the
United States is estimated to exceed $80 billion annually
(5, 8)
.

The severity of disease during an influenza infection is determined b
y both the extent of viral replication
and the nature of the host response. These two factors are mutually dependent. For
example
, large amounts

of

virus can drive an extreme host response that results in significant immunopathology
,

including extensive
ti
ssue damage and poor clinical outcome.
By

contrast, a measured and well
-
timed host response can limit viral
replication.

Understanding how the host and the virus interact at the molecular level to confer these very
different outcomes is fundamental to unde
rstanding and

controlling
disease

cause by influenza infection
.

Recent studies have shown that the molecular networks activated by influenza virus involve thousands of
genes and a vastly larger number of interactions between gene products
(1

4)
.

These studies
, combining
technologies such as transcriptomics, proteomics, and RNAi screens, have defined the relevant “parts lists” of
major players of the anti
-
influenza host response and the host factors necessary for viral replication. The

challenge
is

to filter the

large number of genes and pathways that are activated by infection

and

identify the
much smaller number of proximal virus
-
host interactions that

most strongly contribute to clinical outcome.

This project addres
ses this challenge using two c
omplementary screens: an
in vitro

study in human airway
epithelial cells and a genetic association study facilitated by our access to materials from multiple clinical
studies in which we participate.
As described below, our preliminary data suggest
that wh
ile the
core host
-
response is qualitatively similar across a wide range of infections, both the
magnitude and kinetics of
subsets of the response

vary in a manner that correlates with disease outcome.

While previous systems
biology studies of influenza hav
e developed static network models of the host response that provide a critical
foundation, these in themselves are not sufficient for predictive modeling of the kinetic differences in infection
that we observe.

T
his proposal will integrate data
from both o
f these screens
with kinetic modeling appro
aches
to
generate testable hypotheses for the mechanisms controlling influenza pathogenesis.

B. INNOVATION

This project integrates
cutting
-
edge technologies in

immunology, human genetics, systems biology, and
comp
utational modeling in order to decipher aspects of influenza pathogenicity in humans that cannot be
understood
by applying any of these approaches in isolation
.
The starting points for this project are 1) access
to unique sets of clinical data and samples
and

2)
a model derived from
extensive transcriptional analyses of
the host response to a panel of viruses that have different transmission phenotypes, replication efficiencies,
and disease outcomes.


Although studies in mice have been useful for understanding many aspects of influenza infection,
influenza
is not a natural mouse pathogen and therefore these studies

often employ high doses

of virus
, strains that are
attenuated by laboratory passage, and
artificial routes of inoculation that do not always
adequately model
human
infection
.
Laboratory mice are

usually inbred, rarely vaccinated, and often only one strain of a
standardized age is studied

(9, 10)
. Our grant addresses these problems by directly examin
ing the huma
n

response to influenza. W
e will comprehensively examine
the
role of candidate molecules identified in our
preliminary data on the anti
-
viral response in human airway epithelial cells. In parallel, we will determine the
relevance of these molecules in medi
ating severe disease in humans using genetic association studies of three
clinical cohorts. Importantly, unlike many genetic associations studies where the

DNA variant associated with
the disease is only a marker
, we will functionally characterize each mol
ecule using banked samples from these
studies that

contain
epithelial cells,

neutrophils, monocytes and antigen
-
specific CD4 and CD8 T cells
. We
have developed protocols to immortalize these primary cells, providing a unique
resource
of
cells derived from
individuals carrying the causative polymorphism

that

will facilitate these studies.


Although traditional network models
incorporating protein signaling pathways, transcriptional responses
,

and antiviral effectors

have been useful for identifying critical components, they are not adequate to fully
describe the host response to infection.

Based on
our intensive time sampling of
primary cells
infected with a
panel of viruses possessing a wide range of pathogeni
ci
ties
,
we determined that
while
the qualitative character
of the host response

the genes induced

is

virtually

identical across all infections,
the kinetics of induction
were dramatically different and correlated with overall

pathogenicity
.
These experiments sug
gested that the
ability of the virus to delay the host response is
a
crit
ical determinant of disease outcome.
Because these
Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

temporal effects are not captured by traditional static network models, in this proposal we will

develop

a kinetic
modeling framew
or
k

to interpret

these data

and

guide mechanistic studies

utilizing a panel
of systems
-
based
strategies, including advanced proteomic and transcriptional analysis
.

In summary,
the

project has several innovative features that include
:

1.
Access to primary cells from
clinically, virologically and immunologically characterized individuals and protocols
to transform primary cells
that combined
will allow a characterization of human variation novel in a study of infectious diseases.
2
.

Discove
ry and characterization of common human immune deficiencies.

These
polymorphic humans

offer
advantages similar to the characterization of

knockout mice generated with gene deletion

strategies and yet
avoid the limitations of mouse models

(11

18)
.
3
.
The generation
of a

comprehensive
, kinetic host response

model that will be refined throughout the course of the study

that

will allow us to assign a quantitative
contribution to each element of the host response network we thoroughly investigate. By analogy to
“attribut
able risk” calculations standard in epidemiological studies, we will prioritize the role of each factor and
natural polymorphisms that control its expression or function, calculating the contribution each makes to the
overall network and by extension, to t
he burden of human disease. Critically, this analysis will summarize what
has been learned about the biological importance of any given protein and will also reveal how much of the
burden of disease and network function we have yet to capture.
4
.

Technolog
ically leading edge transcriptional
and proteomic analysis, including advanced analytical approaches developed within our group
.



C. BACKGROUND AND PRELIMINARY STUDIES

C.1 Host response networks following influenza infection

In order to focus on those
components of the host response most associated with severe disease we
measured the transcriptional response in primary respiratory human epithelial cells following infection with a
panel of viruses that are associated with different clinical outcomes. We

chose epithelial cells for these
experiments because they are the major cell type where influenza viruses complete their life cycle. We tested
a panel of 28 viruses, of which we focus on three viruses
: a representative of the

2009 H1N1 pandemic,
A/Cal
ifornia/04/2009 (
CA
), a highly fit human virus;

a representative of the Eurasian swine lineage,
A/Swine/Italy/13901
-
2/1995 (
IT
)
, capable of replicating in humans but otherwise poorly fit;

and the previously
circulating human seasonal H1N1 virus A/Brisbane/
59/2007 (
BR
)
, based on epidemiological studies
considered less fit and capable of causing disease than the CA 2009 pandemic virus
.
These were selected
because, while sharing some genetic relationships, they induce different levels of disease, with CA induc
ing
the greatest burden of disease worldwide, followed by BR and IT. It is important to note that, CA caused a
much higher rate of mortality in young adults than
the circulating

seasonal viruses (including BR)
(19)
. BR is a
human seasonal strain that induces a strong host response. In contrast, infection of airway epithelial cells with
CA induces a delayed host response despite having a somewhat accel
er
ated growth rate.
IT share
s

non
-
overlapping genetic elements with CA, induce
s

a quantitatively distinct host response from BR and CA, and
do
es

not cause transmissible disease in humans. Looking at three time points in the first 24 hours (12, 16, 24
hr),

2
95 unique transcripts were differentially expressed by more than 2
-
fold relative to mock infection by at
least one virus (p < 0.05). Despite spanning a wide range of virulence in humans,
a
pplying the Bayesian
Information Criterion to a simple k
-
means clust
ering of the scaled temporal expression data indicates that the
295 differentially expressed genes are optimally grouped into just two clusters, one containing 271 transcripts
strongly up
-
regulated by infection and a second containing 24 genes whose expres
sion is either uniformly
down
-
regulated or inconsistent between strains.
Thus, despite diverse genetic compositions, these
three
influenza

viruses induced qualitatively identical host transcriptional responses during the first 24 hours of
infecti
on in prim
ary epithelial cells. This finding was consistent across our panel of 28 viruses.

However, despite this qualitative similarity in the induced host response network

the “parts” of each
virus’s induced network were virtually identical

the magnitude of induct
ion of each component was distinct.
This was determined by a high
-
throughput real time PCR analysis of a subset of genes over an 8
-
24 h
ou
r time
course. W
e scaled the induction of each gene under each condition by the maximum induction observed under
all co
nditions and computed the ratio of this quantity to the expression of the influenza M
-
gene at that time
point. For a
subset
of genes up
-
regulated by at least two
-
fold over the first 24 hours, the IT and
BR
strains
induced significantly greater expression t
han the
“stealthy”
CA strain
. That is, there was a lower induction of

a
set of

host response genes per unit of viral M gene CA strain co
mpared to the

IT and BR viruses. It takes
“more M gene”
of
CA to get the equivalent host response as IT.

Furthermore, when the set of genes
differentially regulated by IT were analyzed using a promoter scanning

approach, binding sites for NF
κB were
found to be enriched
(Fig. 1)
. In contrast, binding sites for IRF transcription factors were enriched in genes
Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

induced equally by all viruses and those induced more strongly by IT.
These kinetic variations can be used to
construct models of the host response, identifying critical regulatory branches. Critically, we determined using a
reverse genetics approach that
this phenotype is not
mediated by the NS gene segment alone.

The finding that the alterations in
the host response
were more likely to be kinetic, rather than qualitative
(inducing a distinct genetic module), led us to the
hypothesis that the same mechanisms that allow viruses
to evade the induction of host recognition might also allow
viruses to esca
pe the effector mechanisms associated
with the host response. Mechanistically,
viral
RNA is the
most potent
stimulus

innate immune responses
and

is
also the target of many innate immune effectors
. Most of
these effectors are also strongly upregulated by se
creted
type I interferons inducing these defense mechanisms in
naive cells in the vicinity of an infection.
We tested
our

hypothesi
zed link between induction and evasion of
effector mechanisms by measuring the sensitivity of
“noisy” and “stealth” strains o
f virus to pre
-
treatment of
target cells with IFN
-
α

which we quantified by determining
the concentration

necessary to reduce the growth rate of
each strain by 50% (
the IC
50
) after 1 hour of pretreatment.
This assay essentially bypasses the need for viral r
ecognition and therefore is exclusively testing a given
viruses susceptibility to host viral killing mechanisms
.
Mirroring their degree of stealthy behavior, we found that
C
A had the highest IC
50

value
(
26.1 units/mL
) while

the IC
50
s for
BR and IT
were mor
e than four
-
fold lower (
6
units/mL
)
.
These data provide an initial confirmation of the hypothesis that viruses that evade innate
recognition are also less susceptible to innate effector mechanisms.

Taken together, our findings suggest a model
in which a

network of viral protein interactions that controls
accessibility to immune triggers and antiviral targets (likely the viral RNA) interact
s

with
the

host response
network consisting of immune detector and effectors
that interact with these same triggers a
nd targets
. In this
model, it seems likely that immune detection and antiviral activity may
be directly

linked
. Furthermore, our
temporal data suggest that the kinetics of the response differ between infections suggesting that identification
of components
of the host
-
response that play dual roles of sensor and effector might best be identified and
understood in terms of a kinetic model.



O
ne protein with this potential that was highlighted in our analysis was the Myxovirus resistance protein A
(MxA).
This
protein has been a
focus of influenza antiviral effector studies as several lines of evidence
su
gg
est

that it plays a role in host
defense
(20

25)
.
Although murine studies have

highlighted the importance of this
protein, the human ortholog exhibits distinct sub
-
cellular localization from its murine counterpart and the role of
MxA in human cells remains poorly characterized.

MxA is a dynamin
-
like GTPase. As an antiviral effector
, it has been shown that MxA physically sequesters
components of the viral polymerase complex (the nucleocapsid and PB2) to prevent viral replication
(26
)
. The
precise mechanism of MxA function is unclear, though one mechanism proposes that a supramolecular
assembly of MxA ring
-
shaped oligomers constricts around the nucelocapsid upon GTP hydrolysis
(25)
.

MxA is strongly induced by both influenza infection and type I interferons and most studies have
hypothesized that it
functions primarily as a direct anti
-
viral effector. In experiments designed to d
etermine the
kinetics of MxA function, we noted that MxA is present in unstimulated human epithelial cells and localized to
the viral endosome at very early time points following influenza infection, suggesting that MxA also act as a
sensor, upstream of th
e IFN pathway

(
Fig. 2
)
. To assess the role of MxA in the Type I IFN response, we
generated
human airway epithelial
cells with MxA
expression
knock
ed
down via shRNA transduction

(Mx KD)
.
To confirm its role during an influenza infection,
we infected either
Mx KD or control cells with BR at an

MOI
of
1 and measured production of infectious particles by TCID50.As expected,
viral titer was
significantly higher
(three
-
fold) in Mx

KD cells relative to controls
(
Fig.

2B
). To examine the role of Mx in the interferon
-
mediated
pathway, Mx KD or control cells were infected with 1 MOI
CA

or
BR

over a 6 h time course, and expression
levels of IFN
-
α/β and other interferon
-
stimulated ge
nes were measured using qRT
-
PCR
.


Figure 1
:
Expression of selected genes in human airway
epithelial cells 24 hours following

infection with each of
the three

strains described in the text. The set of genes
whose induction was greater than 2
-
fold could be divided
into two classes:
A)

Genes whose expression in cells
infected with the Eurasian swine strain (I
T
) was more
than twice as high as the mean of the other three
infections, and
B)

Genes whose induction across all
strains did not differ by more than two fold. Promoter
scanning analysis demonstrated that while binding sites
for IRF
-
family transcription fac
tors were enriched in both
sets relative to a bac
kground of expressed but uninduc
ed
genes, binding sites for NF
κ
B were enriched in the set of
genes more strongly induced by infection with the
Eurasian swine strain

(IT)
.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

We observe
d a dramatic reduction of IFN induction in the
CA
-
infected Mx KD cells, and a smaller but still
statistically significant reduction in the
BR
-
infected Mx KD cells. For both strains, the expression of interferon
-
stimulated genes was reduced in the Mx KDs. T
he lack of IFN
-
mediated response in
the

KD cells at very early
time points
suggests

t
hat the loss of Mx
A

compromises the

activation of the IFN response pathway

(Fig. 2BD)
.
Upon phosphorylation, the transcription factor IRF
-
3 will translocate to the nucleus

and upregulate IFN
-
α/β and
other ISGs at early time points after viral infection

(27)
. To investigate whether the transcriptional changes we
observed

correlated with IRF
-
3 activation, we used con
focal microscopy to examine IRF
3 phosphorylation and
nuclear localization after 0.1 MOI
CA

or
BR

infection

and found dramatic reductions in IRF3 localization in the
nucleus
(Fig. 2C)
.

Thus, we have discovered

a novel role for MxA in viral sensing may be mechanistically related to its
descr
ibed role as an anti
-
viral effector. MxA appears to be acting as a feed
-
forward regulator

enhancing the
activation of the Type I IFN pathway that will further elevate its transcription.
We propose to pursue the precise
function of this protein in the antiv
iral detection network in aims 2 and 3.
The potential for host response
molecules to play
multiple, complementary
roles enforces the need
for a mathematical
analysis that can capture
these effects and a
biological approach that is
able to disentangle this
diversity
. A similar
approach will be used for
additional candidate host
response genes identified
through computational
analysis as described
below in the
Approach.

C.2 Host genetic
polymorphisms as
modulators of
susceptibility and
pathogenesis

Host genet
ic
polymorphisms play a
critical role in a number of
infectious diseases and
ample evidence suggests
influenza infection may be
similarly influenced
genetically.
Genetic
studies can be particularly
useful in resolving the
relevant function of a
gene candid
ate. For
example, ISG15 has been studied for many years as part of the antiviral response with little evidence of its
functional contribution, until a genetic study revealed its primary role in protection from
Mycobacterium
tuberculosis
(28)
.
We have shown dramatic differences in susceptibility to highly pathogenic influenza infection
across inbred strains of mice and have mapped the loci

corresponding to some of this variability

(29, 30)
. A
recent report describes a splice
-
acceptor mutation in the
Ifitm3

gene correlated with severe outcomes following
influenza infection

(31)
. Other human genetic associations have been found

(32, 33)
,

but given the importance
of influenza disease, this area is relatively understudied.


Figure 2:

MxA functions upstream of the antiviral response.
(A)

Confocal microscopy of
endosomal markers Rab5 and
CD63 in influenza
-
infected cells 45 minutes after infection.
Nuclear marker

DAPI is in blue. Note the perinculear localization of MxA.
(B)

MxA shRNA
kno
ckdown (blue) results in enhanced viral titers compared to scrambled shRNA (black), but
reduced host response gene induction as early as 30 minutes after infection.
(C)
IRF3 nuclear
translocation is suppressed in cells with MxA knockdown.
(D)
Heat map summ
arizing host
response induction over the first six hours of infection in ells infected with influenza. Yellow
indicates upregulation, blue no change.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

Figure. 3: Lipopeptide Induced
Whole Blood IL6 Production Is
Regulated by TLR1 Polymorphism
T1805G
. Whole blood cytokine
assays were performed and IL
-
6
responses were analyzed by TLR1
genotypes designated as T
(genotypes TT or GT, black circles)
or G (genotype 1805GG, open
circles). The median level is
depicted by a bar. * = P<0.05 by
Mann
-
Whitney U test.




We have extensive experience characterizing

the role of

human

polymorphisms in many infe
ctious
diseases

and we present selected examples of our work below to demonstrate our approach.

C.2A. Human TLR Polymorph
isms Regulate Cytokine Secretion:

Over the past 12 years, we have
examined the TLR pathway for common polymorphisms that regulate the innate immune response and are
associated with susceptibility to infection.
We and others previously discovered and charac
terized TLR1, TLR5,
and TLR6 non
-
synonymous coding region polymorphisms which regulate IL
-
6 secretion in monocytes after
ligand stimulation
(11, 15, 17, 18, 34

36)
.
In 2003, we discovered a TLR5 stop codon polymorphism in TLR5
that is common (10% of the population), abrogates flagellin
-
induced cytokine secretion in PBMCs, acts in a
dominant fashion, and is
associated with susceptibility to Legionnaires’ Disease
(17)
. After this discovery, we
extended our studies to other TLRs including a polymorphism in
TLR1 (I602S or T1805G) that regulates the
immune response to tri
-
acylated lipopeptides
(16, 37)
.

These polymorphisms genetically define TLR1, TLR5,
and TLR6
-
deficient individuals
.

As one example of our data,
we examined
whether TLR1 T1805G regulated

signaling in primary cells from
individuals with different genotypes. In a whole blood cytokine assay,
individuals with the T allele (TT/GT genotypes) produced substantially more
IL
-
6 in response to PAM3 (the TLR1 ligand) stimulation in comparison to
thos
e without the T allele (GG genotype)
(Fig. 3A)
. As controls, stimulation
with PAM2 (TLR2/6 ligand), zymosan (TLR2/6 ligand), and LPS (TLR4
ligand) did not show any difference in IL
-
6 when stratified by TLR1
genotype
(Fig. 3A and B and data not shown)
. On a
verage, individuals
with the GG genotype produced IL
-
6 levels that were approximately 5
-
12%
of those with the TT genotype.
Together, these luciferase and whole blood
cytokine results suggest that the transmembrane domain SNP
TLR1
T1805G
is

a critical regul
ator of lipopeptide
-
induced signaling.

More
recently, we extended this work and discovered that TLR1/6 deficiency is
associated with altered in vivo T
-
cell responses after BCG vaccination
(35)
.

C.2B
.
Isolation and immortalization of primary cells from nasal wash
cultures
:
We
developed
procedures to isolate viable cells from nasal
washes of influenza infected patients as part of the St. Jude based FLU09
-
cohort study. An average of 10 million viable cells were recovered from
infected patients from a single wash, with the cellular composition
distributed 30% epithelial, 30% lymphocytes and 40% monocytes and neutrophils. To demonstrate that these
cells were
representative of the site of infection inflammatory mileu, we stained for the presence of influenza
-
specific T cells.
Figure 4

shows a representative image of tetramer staining from a day 7 infected sample,
indicating that the cells recovered represent th
e local immune response to influenza infection. While we
recover abundant numbers of these cells (in some patients, up to 100 million viable cells) in order to perform
extensive experiments, we next developed
an immortalization protocol allowing the
long t
erm
characterization
of diverse cell types. As a proof of principle, we focused on Type I epithelial cells. Type I alveolar epithelial
cells are a replicative niche for influenza
in vivo,

yet their response to infection is not fully understood. To better
c
haracterize their cellular responses, we created an
immortalized primary murine lung epithelial type I cell line
(LET1) using a retrovirus expressing SV40 large T antigen.
These cells support spreading influenza infection in the absence
of exogenous protea
se, and thus permit simultaneous analysis
of viral replication dynamics and host cell responses. LET1 cells
can be productively infected with influenza, and exhibit
expression of an antiviral transcriptional program and robust
cytokine secretion. We charac
terized influenza virus replication
dynamics and host responses of lung type I epithelial cells and
identified the capacity of epithelial cell
-
derived type I interferon
to establish an effective antiviral state in the absence of a
“professional” immune res
ponse. This response was

Figure 4:
Flow cytometry staining of
flu
-
specific T
cells from a nasal wash. Nasal wash cells from an
infected HLA*A0201 individual were stained with
CD3, CD8 and HLA
-
A*0201 Influenza
-
M1
(GILGFVFTL) tetramer demonstrating the
specificity of the cell types recovered.


Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

investigated using gain and loss of function approaches, revealing specific modules of antiviral effectors
responsible for controlling infection within a population of lung type I epithelial cells. Together, our results
indicate th
at the type I epithelial cell can play a major role in restricting influenza infection without contribution
from the hematopoietic compartment. This approach will be applied in Aim 2 to primary samples obtained from
human clinical studies in order to valid
ate the role of specific polymorphisms.
We will apply a similar protocol to
generate transformed cells derived from nasal wash samples.
These cells will be used to confirm and
functionally characterize the genetic polymorphisms detected in aim 2
.

C.3 Systems analysis of an antiviral
response

A key component of systems approaches to viral pathogenesis is to understand the gene regulatory
networks that respond to
infection
. These studies can reveal critical molecular networks and their dynamics as
ke
y regulators and ultimately points of intervention. We have considerable experience in this area, examining
regulatory networks in various contexts using a wide range of systems approaches and technologies

(38

50)

.
As an example of the approach most relevant to this proposal, we present a recently completed a study that
illustrates how a systems analysis can reveal both the structure of a network and

its

dynamic properties
(51)
.
We analyzed ~30 microarrays of mouse bone marrow macrophages (BMMs) exposed to polyinosinic
-
polycytidylic acid (PIC), a
surrogate for double
-
stranded RNA (dsRNA) viruses that stimulates the interferon
response
,

and controls. Gene expression data were clustered using the K
-
means and hierarchical clustering
algorithms and combined with transcription factor binding site (TFBS)

motif scanning algorithms to infer a
network of associations between transcription factors and target genes that were activated and enriched in
specific clusters. These data led us to hypothesize that FOXO3 acts as a repressor of the IRF and STAT
transcri
ption factors, master regulators of the IFN
-
I pathways.


In order to gain insight into the dynamics of this regulation,

we constructed a kinetic model of the
FOXO3/IRF7/IFB
-
I gene regulatory circuit. The kinetic model consists of ODEs describing FOXO3, IRF
7 and
IFB
-
I activities in the following form: d
x
i
/d
t

=
k
in
xi
f
xi



k
out
xi
x
i
, where
x
i

is the activity of the
i
-
th component of the
system (e.g. FOXO3, IRF7 and IFB
-
I);
k
in
xi

and
k
out
xi

are the maximum influx and efflux rates of the
i
-
th
component of the system, respectively; and
f
xi

is the fractional activity function for the influx of the
i
-
th
component of the system.
f
xi

depends on the activities of other components of the system including external
stimuli and is represented by a ge
neralized Hill function
(52)

or a rational polynomial. These fractional activity
functions were chosen to have the simplest possible form that is cons
istent with the experimentally
reconstructed FOXO3/IRF7/IFB
-
I gene regulatory circuit.

To systematically investigate how perturbations/changes in the model parameters impact the system we
performed parameter sensitivity analysis. Specifically, we quantitat
ively investigated how variation
in

FOXO3
activity affects the basal activity of IRF7 and potentially prevents leakiness of IRF7
-
induced genes in the
absence of a viral infection by limiting the transcription of
Irf7
.
The model predicts that the response t
ime of the
system is directly affected by FOXO3 activity, suggesting the
hypothesis that FOXO3 prevents spurious
system activation by dampening the IRF7/IFN
-
I
-
induced positive feedbacks. Both of these findings are key
features of the complex molecular netw
ork interactions that give rise to precise coordination of the antiviral
responses which must be tightly regulated to defend rapidly against infection while minimizing inflammatory
damage. The model predictions are in accordance the experimental data for g
ene expression and protein
activity levels in macrophages isolated from WT and
Foxo3
-
null mice. Moreover, the mathematical kinetic
model allows one to quantitatively understand how the antiviral response depends on the spectrum of potential
activities of t
he system components (e.g. FOXO3, IRF7 and IFB
-
I)
, an understanding that would be difficult to
intuit from the data alone.

D. APPROACH

Systems
-
guided selection of target candidates

Based on
the

preliminary data

present above and in Project 1
, we have established:

1.

That while the core outputs of the transcriptional program activated in primary human airway epithelial
cells during the first 24 hours following infection are
induced

similarly by
strains of varying
pathogenicity
,
subsets of the
re
sponse are distinguishable by their magnitude and kinetics.

2.

That t
hese divergent transcriptional responses
are

decoupled from viral
load
, indicating that they are
not attributable to a trivial difference in viral life cycle.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

3.

A comprehensive network of host

proteins that interact with NS1
,
the predominant modulator of the
innate immune response. Critically this was accomplished using replication
-
competent affinity
-
tagged
influenza viruses combined with global proteomics approaches (See Project 1) strongly su
ggesting that
the interactions are not artifacts of
over
expression or
in vitro

reconstitutions typical of such studies.

4.

That n
umerous interactions in the host
-
NS1 network are strain dependent.

The roles
in host
-
defense against influenza
of most of the
se ge
nes

are currently unknown.
Although the
function of a small number of these genes in controlling infection has been defined, it remains unclear whether
some have multiple anti
-
viral functions and whether they are equally effective against all strains.
In t
his project,
we will employ a systems level approach to determine the role of a prioritized subset of these genes/proteins in
regulating anti
-
influenza host
-
defense. Although defining the role of all of these components is well beyond the
scope of this pro
gram, the approach described below provides a roadmap for this long
-
term goal.

W
e hypothesize that the genes
which mediate host
-
virus interactions leading
to
severe infection are those
that are differentially regulated by viral strains of varying
pathogenicity or genes whose products directly
interact with influenza proteins. We have therefore p
rioritized
an

initial set of candidates for
study
base
d on the
following crit
eria:

1)

Degree to which induction of the gene is strain specific.

2)

Degree to which

induction is decoupled from viral load.

3)

Basal expression level of an induced gene, suggesting that it could act as a sensor.

4)

Measured interaction with any NS1 variant.

5)

Degree to

which interaction with NS1 is strain
-
dependent.

Based on these five criteria, we have selected an

initial list of 20 genes to focus our analysis in the first year
listed here with an associated criterion (some factors qualify under multiple criteria) (Table 1).

D.1 Specific Aim 1: Determine the relative contribution of influenza response network compo
nents to
host
-
defense in human airway epithelial cells.

We will determine the impact of each candidate gene on influenza replication in airway epithelial cells using the
following experimental approach.

D.1A shRNA knockdown:
We will use lentiviral shRNA v
ectors to knockdown each gene in A549 cells and
validate knockdown by Western blot if appropriate antibodies are available. If antibodies are not available, a
tagged version of the protein will be expressed to confirm specificity of knockdown. Knockdown ce
lls will be
infected with a reference strain of human
inf
luenza (A/Brisbane/59/2007) for which

we have
previously
generated extensive host response dat
a
.
We will measure the transcriptional response at 4, 8,12, 16, and 24
hours following infection by multi
plexed real
-
time PCR. We will initially use a panel of genes representing
outputs of the NFkB, IRF3, and IRF7 pathways, as well as the influenza M gene, and adjust this panel in
subsequent experimental cycles, as the kinetic response model is refined to in
clude additional pathways. We
will also measure the rate of viral protein production
by
staining for influenza NP and production of infectious
viral particles in the supernatant by TCID50.

D.1B Overexpression:

In a complementary set of experiments, we wil
l overexpress each target protein using
lentiviral vectors and repeat the analysis described in D.1A. These experiments will be performed in an A549
-
clone that we have generated with stable knockdown of MAVS/IPS
-
1 in order to eliminate the impact of the
ho
st response which is MAVS
-
dependent in this cell line.

These data will inform what targets cause the greatest perturbations in the network and have the largest
consequences for the influenza life cycle
.
We
expect to discover novel functions for many of
th
ese components, as demonstrated in the
preliminary data by the MxA results
.
Approximately 20 targets will be assessed each
year for the first four years of the program. The
selection of targets in subsequent years will be
done based on the new information
learned
from the analysis of the network in previous
years, leading to an iterative
improvement

in
the resolution of our model.

Criterion 1

Criterion 2

Criterion 3

Criterion 4

Criterion 5

IFIT2

OAS1

MxA

HNRNPH3

Smu1

IFI44

IFI35

IFIT3

HNRNPD

SF3B2

IFI44L

ISG20


DHX9

XRN2

IFI6



DDX21

PCBP1

IFIT5




GIGYF2

KLF4




PCID2

Table 1.
Year 1 candidate list.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

D.1.C Strain
-
specific response:

In addition

to analyzing these targets against the human seasonal BR virus, we
will also invest
igate the response to a panel of viruses of interest, including circulating human viruses,
emerging viruses (such as the vH3N2 viruses), “highly pathogenic” virus that pose a great risk (e.g. H5N1,
H7N7) and specific mutant or “swap” viruses generated in P
roject 1. As Project 1 identifies the necessary
adaptations for maximal human fitness, many of these mutations will likely have consequences on host
response pathways. We can immediately adapt them into our panel for study and explore the consequences of
e
ach mutation based on our defined network models. As new viruses emerge in nature, our status as a Center
of Excellence for Influenza Research and Surveillance and a World Health Organization Collaborating Center
gives us immediate access to new and emergi
ng strains. The high throughput nature of these analyses allows
for a significant number of viruses to be tested in each experimental system.

Genes identified in these screening experiments to play a role in controlling influenza replication or altering
th
e host response to infection will be examined in more detail in Aim 3 in order to integrate them into the kinetic
model.

D.2 Specific Aim 2: Determine the relative contribution of
human polymorphisms in influenza response
network components

to defense agai
nst severe disease in humans

Association

of genetic polymorphisms with severe disease:
We will use a case
-
control candidate gene
association study to examine whether polymorphisms are associated with influenza disease severity. For
genes with polymorphisms associated with influenza, we will perform detailed functional studies to discover the

causal SNP that regulates gene function. We will then assess the cellular mechanism of how these gene
variants regulate the response to influenza infection.

D.2
A. Clinical Enrollment of Influenza cohorts:

We have access to samples from three influenza co
horts.
Cases will have severe influenza
-
associated illness defined as requiring hospitalization and meeting the WHO
definition of severe acute respiratory infection (SARI) associated with

influenza. The WHO definition contains

the following inclusion crite
ria: history of fever, cough or sore throat, dyspnea, and need for hospitalization (all
must be present).
We have extensive epidemiologic data on all of the cohorts including
the presence of high
-
risk conditions,
such as

diabetes mellitus, cancer, end
-
stag
e liver or renal disease,
pregnancy and
immunosuppressive conditions
such as

HIV

or corticosteroid use
. The
re are no exclusion criteria for enrollment
due to a plan to analyze associations with and without identifiable risk factors.
The controls will have
documented influenza infection characterized as influenza
-
like illness (ILI) with the following criteria: history of
fever and cough or sore throat. In all cohorts we have the full case history and the viral titers or RT
-
PCR result
from each subject.

The three sites for clinical enrollment are: 1) FLU09, a CEIRS
-
network program based at St. Jude and Le
Bonheur Children’s Hospital (open to enrollment since 2009, approved through 2014). This site is in Memphis,
TN, in an urban setting.
Due to the smaller

sample size, this cohort will primarily be used for functional studies
of nasal wash epithelial cells and other cell types.
2) SHIVERS, CDC
-
funded and based at ESR in New
Zealand, on which Richard Webby and Paul Thomas are co
-
investigators.
This is the la
rgest cohort and will be
the primary sample set used for the discovery phase of the genetic studies.
3) La Red, an NIH
-
funded Mexico
City based biobank of specimens collected from respiratory illness.
This cohort will be used for validating
findings from t
he SHIVERS studys.
See
Table
2
. To our knowledge, th
e combination of these three sites
constitute

the largest influenza cohort ever assembled for this type of genetic analysis, though there may be
unpublished efforts of which we are not aware.

D.2
B. SNP s
election
:

We will select non
-
synonymous coding region, promoter region, and haplotype
-
tagging
polymorphisms from several sources. First, we will query the 1000 Genomes (http://www.1000genomes.org/)
and HapMap (
http://hapmap.ncbi.nlm.nih.gov/
) databases for known polymorphisms in populations
represented in our cohort. Second, we will amplify and sequence the coding and promoter regions in 100
individuals. Although most SNPs will be shared between
the countries
where the samples are obtained

and
populations in the public databases, we will use this resequencing strategy to ensure full coverage of the
genes
.

This strategy may be particularly important for the SHIVERS cohort since New Zealand populations
(Maori and

Pacific Peoples) are not represented in the 1000 Genomes of HapMap databases.
The sequence
will be aligned and analyzed with the programs PHRED/PHRAP and CONSED to determine the frequency of
known and novel polymorphisms
(53)
. We will then genotype the polymorphisms in our cases and c
ontrols
.
If
polymorphisms are not in Hardy Weinberg Equilibrium (HWE) in population controls, we will not assess them
further (due to presumed technical genotyping error).

D.2
C. Analysis of clinical phenotypes
:


Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

T
able 2
: Influenza disease cohort characteristics

Cohort name

# of cases: Flu+, SARI

(Collected and
predic
ted by Fall 2014
)

# of
controls:
Flu+, ILI

Ethnicity of cohort

SHIVERS

870

700

Maori

19%

Pacific Peoples

32%

Asians

10%

Europeans

39%

La Red

250

200

Mestizo

98%

FLU09

44

141

Black,
non
-
Hispanic

75%

White, non
-
Hispanic

24%

Other

1%


i
. Discovery Phase
.
In our
primary analysis,
we will examine
whether polymorphisms are
associated with the severity of
influenza infection by comparing
the
genotype

frequency in cases
and controls with a
genotypic
model in a discovery cohort of
500 cases of severe influenza
and 500 controls with mild
disease. A
n unadjusted

P value <
0.05 will be considered significant
in the discovery phase.
We will prioritize analys
is of SNPs that are significant after adjustment for multiple
comparisons which will be

performed with a Bonferroni correction based on the number of effective
independent tests accounting for LD structure among SNPs in the same gene, as determined by the
program
SNPSpDlite (http://gump.qimr.edu.au/general/daleN/SNPSpDlite/) using the estimate proposed by Li and Ji
(54)
.

ii. Population Admixture
:
Populati
on stratification (presence of multiple “hidden population structures”), may
confound observed genetic associations.
We will examine and correct for population stratification by several
methods.
First, the discovery phase will be performed in the SHIVERS s
tudy, which is the largest of our
cohorts. To avoid further population admixture problems, we will only examine the dominant ethnic groups in
the discovery phase (European Ancestry 39% and Pacific Peoples 32% within SHIVERS). Second,
we will
perform subgro
up analyses stratified by ethnicity
.

Third
, we will adjust for ethnicity in logistic regression
analysis with the use of self
-
identified ethnicity as a variable.
Fourth
, we will genotype a panel of ancestry
informative markers (AIMs) which are designed to distinguish admixture in populations that otherwise appear
homogenous
(55, 56)
.

These “control SNPs” are used to assess an individual’s genetic ancestry and then
ancestry
-
adjusted genoty
pe and phenotype are used in logistic regression analysis to compute association
statistics.


iii. Validation
: We will use a two
-
stage study design and validate SNPs from Stage 1 in a 2
nd

valida
tion cohort
of the same size. The initial validation strategy will be with the same populations (European and Pacific
Peoples ancestry) in the SHIVERS study. We will then attempt to
validate findings in the remaining SHIVERS populations (Maori and
Asian) a
nd in the La Red cohort from Mexico. We will also examine
whether the association of SNPs with influenza severity is altered
when adjusted for identifiable clinical risk factors (e.g. co
-
morbid
conditions, immunosuppressed states) in a logistic regression
model.

iv. Power Calculation:
For a power calculation, we examined a
variety of scenarios of minor allele frequen
cy and odds ratios (OR)
(Table 3
, calculated with CaTS Power Calculator)

(60)
. Depending
on the SNP frequency, we will have sufficient power to detect an
OR of 1.75 given our sample size, and will be able to identify weak
effects with an OR 1.5, but with less significance when the allele
frequency is low.
Based on these power calcu
lations, we will be able to identify modest or strong associations
between polymorphisms and susceptibility to severe influenza infection.

We have used all of these statistical
techniques with other genetic analyses in our previous studies

(25, 29, 38, 40,

61).

v. Fine Mapping:
These initial findings will uncover genetic markers of influenza severity rather than causal
SNPs. We will perform additional sequencing and analysis of the genes associated with influenza severity to
fine map these regions and disco
ver which
S
NPs are most strongly associated with influenza severity AND
regulate function in cellular assays outlined below
.


D.2D Molecular characterization of polymorphisms
:

After identifying the polymorphisms that are most
strongly associated with influ
enza susceptibility, we will determine the molecular mechanism of how select
variants regulate cellular function. For the analyses below, we will use primary cells and immortalized cells
generated as described
in C.2B
.

Synonymous polymorphisms & transcrip
tional regulation:
We will first determine whether the polymorphisms
are associated with altered total or splice variant mRNA levels in epithelial cells. We will
utilize our established
Table 3. Power to genetic association
effect at

=0.05



Contrasting cases (N=500) and controls
(N=500)


Power to detect:

MAF

OR=1.5

OR=1.75

OR=2.0

OR=2.5

0.05


0.68


0.99


0.99



0.99

0.10


0.91


0.99


0.99


0.99

0.20


0.99


0.99


0.99


0.99

0.30


0.99


0.99


0.99


0.99

MAF=minor allele frequency; OR=odds ratio.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

Fluidigm panels

to measure mRNA levels
in individuals
from our enrolle
d cohorts. We will complement this
strategy in
New Zealand and Mexico
with parallel work in Seattle where we have a
large, established
repository of healthy individuals

and in Memphis where we have a cohort of individuals with influenza
. We can
screen th
es
e

cohort
s

for the relevant polymorphisms already identified in the

SHIVERS

cohort, and then have
the option to pursue additional assays that are more readily performed in Seattle

or Memphis
. We will isolate
epithelial cells from nasal wash obtained from these individuals who are homozygous or heterozygous for each
variant of interest and
examine whether any polymorphisms are associated with altered mRNA levels or its
splice variants.


If as
sociations with mRNA levels are found, then we will use bio
-
informatic approaches to assess whether
the polymorphisms are present in predicted transcription factor binding sites or splice donor/acceptor sites. We
will then test this hypothesis by cloning t
he region of genomic DNA spanning the polymorphism and the
predicted transcription factor binding site. We will insert this DNA fragment upstream of a luciferase reporter
plasmid (pGL4), transfect it into different cell lines (HEK293, THP1, etc) and compar
e luciferase activity of the
polymorphism and wild type variants. If differences are observed in luciferase activity, then we will truncate the
DNA segment to find the minimal essential element that regulates signaling. If a specific transcription factor i
s
known to bind to this DNA fragment, we will test this possibility with an electrophoretic mobility shift assay
(EMSA) with nuclear extracts.
If successful, these experiments will identify a transcription factor that
differentially binds the polymorphic v
ariant and regulates mRNA transcription.

Non
-
synonymous Coding Region Polymorphism:

If no association of the polymorphism with transcriptional
regulation is found, then the SNPs may be in linkage disequilibrium with a non
-
synonymous SNP that alters
gene fu
nction. To test this hypothesis, we will sequence the coding region of 50 individuals with a
polymorphism associated with the outcome (either clinical cases or those with a specific
cellular

response). If a
non
-
synonymous polymorphism is found, we will com
pare the WT and variant function in reconstitution assays.
These studies will enable us to determine whether a non
-
synonymous polymorphism alters the function of the
gene
.

D.3 Specific Aim 3
Determine the mechanism by which each gene identified as a criti
cal node in Aims 1
or 2 mediates host
-
defense to influenza.

We will use multiple approaches to determine the mechanistic role of each gene identified in Aims 1 and 2 in
regulating the anti
-
influenza response program defined above (Section C.1). Importantly
, our focus in this
Project is on identifying those aspects of the host
-
response that make the most significant contribution to
severe infections.
We have constructed a simple kinetic model of the host
-
response that parameterizes several
critical features
the response that we have identified in our preliminary work or are suggested from the
literature as correlating with pathogenicity (
Figure 5
).

The details of the model are described in D.3A. We then
illustrate how we will use the model to guide our experi
mental approach, using MxA as an example and refine
the model based on the outcome of these experiments (Section D.3B). Finally, we outline our approach to
defining the role of each gene identified above in immune cells.

D.3A

Model design.

Based on
our preliminary data and data from the literature (for review,
see

(57)
)
, we
have
construct
ed

a kinetic model
that represents multiple
aspects of

the
host
-
virus

interaction
.
These include active evasion or
suppression of the RNA sensing, differential activation of the IRF and
NF
κ
B arms of the transcription
al response, and the auto
-
regulation of
these pathways
.

We have chosen to include only the NF
κ
B and IRF3/7
in the initial model because our preliminary data suggestin
g
strain
-
specific differential responses of these regulators and their targets.
Each

of th
ese regulators
, and related family members, has

individually,
been the subject of kinetic modeling

(58

63)

. The model also includes
the primary mechanisms by which influenza avoids the intracellular
detection. While this model is basic, it is non
-
trivial and it can guide in
contextualizing the host factor in the viral response. We can measure
inputs (viral RNA), activation of some key components of the path
ways
(NFkB, IRF3, IRF7
), and the output (transcriptional response). By
examining these eleme
nts of the network upon disruption or knock
-
down of specific host factors, we can readily place components within

Figure 5.

General kinetic model
framework for activation of innate
immune responses by influenza.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

context of this network. Factors that can explain the differential kinetic responses we have already observed in
a mechanistic way will be exa
mined with high priority.

The

first version of the

kinetic model consists of
differential equations

describing
six dynamical variables
(
a
i
) or

gener
ic

activities

of the system
(e.g.
flu virus (V)
,
RIG
-
I
,
IPS
-
I
,
IRF3/7
,
NFκB

and
the pool of the target
genes

(T)
) in the following form: d
a
i
/d
t

=
F
a
i



G
ai
a
i
, where
F
a
i

and

G
ai

are

the fractional activity function
s

for the
influx

and efflux, respectively,

of the
i
-
th component of the system.
F
a
i

and

G
ai

depend on the activities of other
system components and
will be formally described as

generalized Hill
functions

(52)
.

This framework also allows for the introduction of independent time
-
delays for each br
anch which in future
refinements will
appropriately model different characteristic times of the
host
-
virus

interaction system

processes
(e.g. signal transduction, gene expression changes, virus replication etc.).
In addition, it

will
help us
to distinguish

between the
evasion
strategies of different influenza virus strains and

guide further
experiment
ation

to
mechanistically understand
host
-
virus

interactions, which will be necessary to place genetic
polymorphisms in the proper context
.

Specifically, the pr
oposed kinetic model will help us to quantitatively
understand which host
-
virus interactions, such as the
differential activation of the IRF and NF
κ
B arms of the
transcriptional response
, the virus interference with the upstream signaling pathway, or their combination are
critical to
the “stealthy” and “noisy”
pheno
types

and the dysregulation in observed in severely ill patients.

D.3B

Define the role of
each gene

in regulating the
anti
-
i
nfluenza response

We will illustrate our methods in this sub
-
aim using MxA as an example. The approach outlined below will be
applied to other genes of interest as they are identified.

D.3B
.1 Impact on transcriptional response:
O
ur preliminary data have r
evealed a previously unknown role for
M
xA in controlling host

responses to influenza. Because shRNA knockdown of MxA results in suppression of
numerous critical anti
-
viral genes including type I interferons, we
will start by using RNA
-
seq to identify the f
ull
compendium of genes that are affected by MxA knockdown in A549 cells responding to infection with a virus
that induces a strong host response (BR) and a virus that induces a delayed response (CA).

We will begin
by
clustering

the expression data, group
ing genes with the largest differences in their
temporal expression profiles between MxA KD and untreated cells (
See
Core A
-

Modeling). We will then apply
promoter
-
scanning algorithms to identify transcription factors that may be mediating the host respon
se
downstream of MxA. We will prioritize the list of enriched transcription factors in each cluster by level of
expression, magnitude of enrichment, and annotated role in immune responses. We will then use chromatin
-
immunoprecipitation (ChIP) to identify t
hose transcription factors whose binding to the promoters of
representative genes from the cluster is altered by influenza infection and MxA knockdown. In our initial model,
viral sensing occurs through RIG
-
I and drives the transcriptional response via IRF
3/7 and NF
κ
B. The
experiments described here will extend the transcriptional regulatory portion of this model by adding MxA as a
sensor and by adding additional transcription factors and their associated upstream activators.

D.3B
.2 Impact on signaling res
ponse:

We will also determine the impact of MxA knockdown on signaling
responses to influenza infection by comparing the kinetics of activation of several key components of the IRF3
and NFkB pathways in MxA knockdown and control cells infected with the BR
and CA strains. Flux through the
NF
κ
B pathway will be measured by I
κ
B
α

degradation and translocation of NF
κ
Bp50 and p65 subunits to the
nucleus. Similarly, activity through the IRF3/7 axis will be quantified by measuring their phosphorylation and
rate of t
ranslocation to the nucleus. Each of these activities will be incorporated into the model
.


D.3B
.3 Functional interpretation:
Although our preliminary data presented above are consistent with a role for
MxA in direct detection of viral RNA and subsequent s
ignaling, an alternative hypothesis is that the observed
transcriptional repression arises from an increased abundance of viral NS1 protein, a known suppressor of
type I interferon activation. In this scenario, the increase in the suppressive effect of NS1

dominates over the
presumed increase in RIG
-
I activation driven by viral RNA.

In order to distinguish these alternatives, we have implemented
a simplified instance
the modeling
framework described above and constructed two preliminary models, one with and

one without RNA sensing
and subsequent positive feedb
ack by MxA (see Figure 6
). In addition to MxA, these models include replication
of viral RNA, sensing of viral RNA by RIG
-
I, production of influenza NS1 and its attenuation of RIG
-
I signaling,

transcrip
tion of host effectors and their destruction of viral RNA. (
Figure 6
). While this model vastly simplifies
the underlying biology, it is an appropriate starting point for building intuition about the behavior of the system
and illuminating the most promisin
g experimental approaches. First, both models were adjusted so that they
produced approximately the same output in the unperturbed (WT) state. Then, we simulated knockdown of
Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

MxA and compared the accumulation of viral RNA and transcription of host effector
s. MxA knockdown
increased viral load in both models, in concordance with the preliminary data described above (Figure 2B),
However, the reduction in transcription of host effectors was significantly stronger in the model that included
MxA
-
RNA sensing and
positive feedback supporting this hypothesized novel role for the protein.

Although the concordance of the RNA
-
sensing model with the preliminary data is suggestive, both the
model and the data are currently insufficient to define this possible role for M
xA. This simple model does
however serve to guide experiments
. A
dditional measurements such as the abundance of NS1 and MxA over
the course of infection can help to constrain each model and ultimately aid in discriminating them.

Furthermore, this modeling approach is applicable to any of the effectors that we are studying and can be
extended to include additional parameters that are experimentall
y accessible such as concentrations of
both

RNA and protein for virus and host, the impact of viral proteins on the rate of translation, and multiple host
effectors with differing potencies. As described in the Technology Core, we have access to several
te
chnologies ideally suited to simultaneous, high
-
resolution temporal quantification of multiple of RNA and
protein abundance such as multiplex real
-
time PCR
(Fluidigm) and SRM mass spectrometry and we will use
both of these approaches to constrain these mor
e
complex models.

D.3C

Define the role of
each gene

in immune cells.

Most of the genes in our candidate list above are
expressed in many cell types in addition to epithelial cells
including macrophages and T cells. One of the
advantages our approach is that we have included a
screen for disease in humans
(Specific Aim 2)
an
d are
not limited to detecting defects in epithelial cell
responses. While we will focus the analysis described
abov
e on epithelial cells,
it is possible we will find a
significant genetic association, but the functional
consequences of the association will not manifest, or not
manifest strongly in this compartment. In those cases, we
will proceed to use the assays above, along with cell
spec
ific assays, to test the functional consequences in
monocytes and
macrophages

and
CD4 and CD8 T

cells
. Dr. Hawn is an expert in monocyte and
macrophage biology, while Drs. Thomas and Doherty
have extensive experience in the functional
characterization of T

cells
(64

70)
. These cell types ar
e
present in abundance in the nasal washes we’ve
processed and are also easily isolated from PBMCs.
Our
hypothesis is that the epithelial cell is the critical cell type
for dissecting the host response network during influenza
infection, but in the case wh
ere there is not an epithelial
phenotype but a confirmed, statistically significant SNP,
we will maximize that information by identifying the appropriate cell for
further
investigation
. Briefly, for
monocytes and macrophages we will infect cells with influ
enza and examine cytokine production and viral
replication. If the gene of interest has a known ligand (such as an innate immune receptor ligand), then we will
also examine whether the SNP regulates the immune response to stimulation with the ligand. For
T cells, we
will stimulate with polyclonal stimulators and look at cytokine production and lytic activity. If autologous APCs
are available, we will perform influenza
-
specific intracellular cytokine assays and CTL assays. The
transcriptional response will
be measured using a T
-
cell specific Fluidigm panel.

If a role is discovered for the gene in immune cells, the same modeling formalism described above, is
applicable to signaling in these cells and these models will be developed as appropriate in later yea
rs of the
program. As described in Section C.3, we have previously applied these approaches to model macrophage
responses to viral RNA
(51)
.


Figure 6
.

Alternative
models for the role of MxA in the
host response to influenza.

Two alternative kinetic
models are shown
A)

with and
B)

without MxA sensing of
influenza virus and subsequent signaling. The transcription
of influenza viral RNA and host effectors are shown in
the
lower panels indicating the effect of MxA knockdown in
each model (dotted lines).

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

Referenc
es

1.

Karlas A, Machuy N, Shin Y, Pleissner K
-
P, Artarini A, Heuer D, Becker D, et al. 2010. Genome
-
wide
RNAi screen identifies human host factors crucial for influenza virus replication.
Nature

463(7282):
818

822.

2.

Nakaya HI, Wrammert J, Lee EK, Racioppi L, Marie
-
Kunze S, Haining WN, Means AR, et al. 2011.
Systems biology of vaccination for seasonal influenza in humans.
Nat Immunol

advance online
publication:

3.

Brass AL, Huang I
-
C, Benita Y, John SP, K
rishnan MN, Feeley EM, Ryan BJ, et al. 2009. The IFITM
proteins mediate cellular resistance to influenza A H1N1 virus, West Nile virus, and dengue virus.
Cell

139(7):1243

1254.

4.

Shapira SD, Gat
-
Viks I, Shum BOV, Dricot A, de Grace MM, Wu L, Gupta PB, et

al. 2009. A physical and
regulatory map of host
-
influenza interactions reveals pathways in H1N1 infection.
Cell

139(7):1255

1267.

5.

Meltzer MI, Cox NJ, and Fukuda K. 1999. The economic impact of pandemic influenza in the United
States: priorities for in
tervention.
Emerging Infect. Dis.

5(5):659

671.

6.

Bhat N, Wright JG, Broder KR, Murray EL, Greenberg ME, Glover MJ, Likos AM, et al. 2005. Influenza
-
associated deaths among children in the United States, 2003
-
2004.
N Engl J Med

353(24):2559

2567.

7.

201
2. Prevention and Control of Influenza with Vaccines: Recommendations of the Advisory Committee
on Immunization Practices (ACIP)
-

United States, 2012
-
13 Influenza Season.
MMWR Morb. Mortal.
Wkly. Rep.

61:613

618.

8.

Molinari N
-
AM, Ortega
-
Sanchez IR, Mess
onnier ML, Thompson WW, Wortley PM, Weintraub E, and
Bridges CB. 2007. The annual impact of seasonal influenza in the US: measuring disease burden and
costs.
Vaccine

25(27):5086

5096.

9.

Casanova JL and Abel L. 2004. The human model: a genetic dissection
of immunity to infection in natural
conditions.
Nat Rev Immunol

4(1):55

66.

10.

Quintana
-
Murci L, Alcais A, Abel L, and Casanova JL. 2007. Immunology in natura: clinical,
epidemiological and evolutionary genetics of infectious diseases.
Nat Immunol

8(11):
1165

71.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

11.

Hawn TR, Berrington WR, Smith IA, Uematsu S, Akira S, Aderem A, Smith KD, and Skerrett SJ. 2007.
Altered Inflammatory Responses in TLR5
-
Deficient Mice Infected with Legionella pneumophila.
J
Immunol

179(10):6981

7.

12.

Cooke GS, Campbell SJ,

Sillah J, Gustafson P, Bah B, Sirugo G, Bennett S, et al. 2006. Polymorphism
within the interferon
-
gamma/receptor complex is associated with pulmonary tuberculosis.
Am J Respir
Crit Care Med

174(3):339

43.

13.

Hawn TR, Dunstan SJ, Thwaites GE, Simmons CP
, Thuong NT, Lan NT, Quy HT, et al. 2006. A
polymorphism in toll
-
interleukin 1 receptor domain containing adaptor protein is associated with
susceptibility to meningeal tuberculosis.
J Infect Dis

194(8):1127

34.

14.

Bochud PY, Hawn TR, Siddiqui MR, Saunde
rson P, Abraham I, Tadesse A, Janer M, Zhao LP, Kaplan G,
and Aderem A. 2007. Toll
-
like Receptor 2 Polymorphisms Are Associated with Reversal Reaction in
Leprosy.
J. Inf. Dis.

In press.:

15.

Dunstan SJ, Hawn TR, Hue NT, Parry CP, Ho VA, Vinh H, Diep TS, e
t al. 2005. Host susceptibility and
clinical outcomes in toll
-
like receptor 5
-
deficient patients with typhoid fever in Vietnam.
J Infect Dis

191(7):1068

71.

16.

Hawn TR, Misch EA, Dunstan SJ, Thwaites GE, Lan NT, Quy HT, Chau TT, et al. 2007. A common
hum
an TLR1 polymorphism regulates the innate immune response to lipopeptides.
Eur J Immunol

37(8):2280

9.

17.

Hawn TR, Verbon A, Lettinga KD, Zhao LP, Li SS, Laws RJ, Skerrett SJ, et al. 2003. A common dominant
TLR5 stop codon polymorphism abolishes flagelli
n signaling and is associated with susceptibility to
Legionnaires’ Disease.
J. Exp Med

198:1563

72.

18.

Misch EA, Macdonald M, Ranjit C, Sapkota BR, Wells RD, Siddiqui MR, Kaplan G, and Hawn TR. 2008.
Human TLR1 Deficiency Is Associated with Impaired Mycobacterial Signaling and Protection from
Leprosy Reversal Reaction.
PLoS Negl Trop Dis

2(5):e231.

19.

Vib
oud C, Miller M, Olson D, Osterholm M, and Simonsen L. 2010. Preliminary Estimates of Mortality and
Years of Life Lost Associated with the 2009 A/H1N1 Pandemic in the US and Comparison with Past
Influenza Seasons.
PLoS Curr
:RRN1153.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

20.

Pavlovic J, Haller

O, and Staeheli P. 1992. Human and mouse Mx proteins inhibit different steps of the
influenza virus multiplication cycle.
Journal of virology

66(4):2564.

21.

Turan K. 2004. Nuclear MxA proteins form a complex with influenza virus NP and inhibit the trans
cription
of the engineered influenza virus genome.
Nucleic Acids Research

32(2):643

652.

22.

Dittmann J, Stertz S, Grimm D, Steel J, GarcÃ-a
-
Sastre A, Haller O, and Kochs G. 2008. Influenza A virus
strains differ in sensitivity to the antiviral action of
Mx
-
GTPase.
J Virol

82(7):3624

3631.

23.

Salomon R, Staeheli P, Kochs G, Yen H
-
L, Franks J, Rehg JE, Webster RG, and Hoffmann E. 2007. Mx1
gene protects mice against the highly lethal human H5N1 influenza virus.
Cell Cycle

6(19):2417

2421.

24.

Haller O, Frese M, and Kochs G. 1998. Mx proteins: mediators of innate resistance to RNA viruses.
Rev
Sci Tech

17(1):220

30.

25.

Gao S, von der Malsburg A, Dick A, Faelber K, Schröder GF, Haller O, Kochs G, and Daumke O. 2011.
Structure of myxovirus resis
tance protein a reveals intra
-

and intermolecular domain interactions required
for the antiviral function.
Immunity

35(4):514

525.

26.

Sadler AJ and Williams BR. 2008. Interferon
-
inducible antiviral effectors.
Nature Reviews Immunology

8(7):559

568.

27.

Grandvaux N, Servant MJ, tenOever B, Sen GC, Balachandran S, Barber GN, Lin R, and Hiscott J. 2002.
Transcriptional profiling of interferon regulatory factor 3 target genes: direct involvement in the regulation
of interferon
-
stimulated genes.
J. Virol.

76(
11):5532

5539.

28.

Bogunovic D, Byun M, Durfee LA, Abhyankar A, Sanal O, Mansouri D, Salem S, et al. 2012.
Mycobacterial disease and impaired IFN
-
γ immunity in humans with inherited ISG15 deficiency.
Science

337(6102):1684

1688.

29.

Boon ACM, Finkelstein

D, Zheng M, Liao G, Allard J, Klumpp K, Webster R, Peltz G, and Webby RJ.
September. H5N1 Influenza Virus Pathogenesis in Genetically Diverse Mice Is Mediated at the Level of
Viral Load.
mBio

2(5):

30.

Boon ACM, deBeauchamp J, Hollmann A, Luke J, Kotb M,

Rowe S, Finkelstein D, et al. 2009. Host
genetic variation affects resistance to infection with a highly pathogenic H5N1 influenza A virus in mice.
J.
Virol

83(20):10417

10426.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

31.

Everitt AR, Clare S, Pertel T, John SP, Wash RS, Smith SE, Chin CR, et al
. 2012. IFITM3 restricts the
morbidity and mortality associated with influenza.
Nature

484(7395):519

523.

32.

Zhou J, To KK
-
W, Dong H, Cheng Z
-
S, Lau CC
-
Y, Poon VKM, Fan Y
-
H, et al. 2012. A functional variation
in CD55 increases the severity of 2009 pande
mic H1N1 influenza A virus infection.
J. Infect. Dis.

206(4):495

503.

33.

Zúñiga J, Buendía
-
Roldán I, Zhao Y, Jiménez L, Torres D, Romo J, Ramírez G, et al. 2012. Genetic
variants associated with severe pneumonia in A/H1N1 influenza infection.
Eur. Respir
. J.

39(3):604

610.

34.

Johnson CM, Lyle EA, Omueti KO, Stepensky VA, Yegin O, Alpsoy E, Hamann L, Schumann RR, and
Tapping RI. 2007. Cutting Edge: A Common Polymorphism Impairs Cell Surface Trafficking and
Functional Responses of TLR1 but Protects agains
t Leprosy.
J Immunol

178(12):7520

7524.

35.

Randhawa AK, Shey MS, Keyser A, Peixoto B, Wells RD, de Kock M, Lerumo L, et al. 2011. Association
of human TLR1 and TLR6 deficiency with altered immune responses to BCG vaccination in South African
infants.
PLo
S Pathog.

7(8):e1002174.

36.

Wurfel MM, Gordon AC, Holden TD, Radella F, Strout J, Kajikawa O, Ruzinski JT, et al. 2008. Toll
-
like
Receptor 1 Polymorphisms Affect Innate Immune Responses and Outcomes in Sepsis.
Am J Respir Crit
Care Med
:

37.

Thuong NT, D
unstan SJ, Chau TT, Thorsson V, Simmons CP, Quyen NT, Thwaites GE, et al. 2008.
Identification of tuberculosis susceptibility genes with human macrophage gene expression profiles.
PLoS
Pathog

4(12):e1000229.

38.

Hwang D, Rust AG, Ramsey S, Smith JJ, Lesli
e DM, Weston AD, de Atauri P, et al. 2005. A data
integration methodology for systems biology.
Proc. Natl. Acad. Sci. U.S.A.

102(48):17296

17301.

39.

Saleem RA, Knoblach B, Mast FD, Smith JJ, Boyle J, Dobson CM, Long
-
O’Donnell R, Rachubinski RA,
and Aitch
ison JD. 2008. Genome
-
wide analysis of signaling networks regulating fatty acid
-
induced gene
expression and organelle biogenesis.
J. Cell Biol.

181(2):281

292.

40.

Saleem RA, Long
-
O’Donnell R, Dilworth DJ, Armstrong AM, Jamakhandi AP, Wan Y, Knijnenburg T
A, et
al. 2010. Genome
-
wide analysis of effectors of peroxisome biogenesis.
PLoS ONE

5(8):e11953.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

41.

Saleem RA, Rogers RS, Ratushny AV, Dilworth DJ, Shannon PT, Shteynberg D, Wan Y, et al. 2010.
Integrated phosphoproteomics analysis of a signaling networ
k governing nutrient response and
peroxisome induction.
Mol. Cell Proteomics

9(9):2076

2088.

42.

Smith JJ, Marelli M, Christmas RH, Vizeacoumar FJ, Dilworth DJ, Ideker T, Galitski T, Dimitrov K,
Rachubinski RA, and Aitchison JD. 2002. Transcriptome profil
ing to identify genes involved in peroxisome
assembly and function.
J. Cell Biol.

158(2):259

271.

43.

Smith JJ, Ramsey SA, Marelli M, Marzolf B, Hwang D, Saleem RA, Rachubinski RA, and Aitchison JD.
2007. Transcriptional responses to fatty acid are coordi
nated by combinatorial control.
Mol. Syst. Biol.

3:115.

44.

Smith JJ, Sydorskyy Y, Marelli M, Hwang D, Bolouri H, Rachubinski RA, and Aitchison JD. 2006.
Expression and functional profiling reveal distinct gene classes involved in fatty acid metabolism.
M
ol.
Syst. Biol.

2:2006.0009.

45.

Ratushny AV, Shmulevich I, and Aitchison JD. 2011. Trade
-
off between responsiveness and noise
suppression in biomolecular system responses to environmental cues.
PLoS Comput. Biol.

7(6):e1002091.

46.

Ratushny AV, Ramsey S
A, Roda O, Wan Y, Smith JJ, and Aitchison JD. 2008. Control of transcriptional
variability by overlapping feed
-
forward regulatory motifs.
Biophys. J.

95(8):3715

3723.

47.

Ratushny AV, Ramsey SA, and Aitchison JD. 2011. Mathematical modeling of biomolecula
r network
dynamics.
Methods Mol. Biol.

781:415

433.

48.

Shmulevich I and Aitchison JD. 2009. Deterministic and stochastic models of genetic regulatory networks.
Meth. Enzymol.

467:335

356.

49.

Ramsey SA, Smith JJ, Orrell D, Marelli M, Petersen TW, de Ata
uri P, Bolouri H, and Aitchison JD. 2006.
Dual feedback loops in the GAL regulon suppress cellular heterogeneity in yeast.
Nat. Genet.

38(9):1082

1087.

50.

Ratushny AV, Saleem RA, Sitko K, Ramsey SA, and Aitchison JD. 2012. Asymmetric positive feedback
lo
ops reliably control biological responses.
Mol. Syst. Biol.

8:577.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

51.

Litvak V, Ratushny AV, Lampano AE, Schmitz F, Huang AC, Raman A, Rust AG, Bergthaler A, Aitchison
JD, and Aderem A. 2012. A FOXO3
-
IRF7 gene regulatory circuit limits inflammatory seque
lae of antiviral
responses.
Nature
:

52.

Likhoshvai V and Ratushny A. 2007. Generalized hill function method for modeling molecular processes.
J Bioinform Comput Biol

5(2B):521

531.

53.

Gordon D, Abajian C, and Green P. 1998. Consed: a graphical tool for sequence finishing.
Genome Res

8(3):195

202.

54.

Li J and Ji L. 2005. Adjusting multiple testing in multilocus analyses using the eigenvalues of a correlation
matrix.
Heredity

95(3):221

7.

55.

Seldin MF, Pasaniuc B, and Price AL. 2011. New approaches to disease mapping in admixed
populations.
Nat. Rev. Genet.

12(8):523

528.

56.

Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, and Reich D. 2006. Principal
components analysis
corrects for stratification in genome
-
wide association studies.
Nat Genet

38(8):904

9.

57.

Ramos HJ and Gale M Jr. 2011. RIG
-
I like receptors and their signaling crosstalk in the regulation of
antiviral immunity.
Curr Opin Virol

1(3):167

176.

58.

Hoffman
n A. 2002. The Ikappa B
-
NF
-
kappa B Signaling Module: Temporal Control and Selective Gene
Activation.
Science

298(5596):1241

1245.

59.

Nelson DE, Ihekwaba AEC, Elliott M, Johnson JR, Gibney CA, Foreman BE, Nelson G, et al. 2004.
Oscillations in NF
-
kappaB s
ignaling control the dynamics of gene expression.
Science

306(5696):704

708.

60.

Ashall L, Horton CA, Nelson DE, Paszek P, Harper CV, Sillitoe K, Ryan S, et al. 2009. Pulsatile
Stimulation Determines Timing and Specificity of NF
-
κB
-
Dependent Transcription
.
Science

324(5924):242

246.

61.

Litvak V, Ramsey SA, Rust AG, Zak DE, Kennedy KA, Lampano AE, Nykter M, Shmulevich I, and
Aderem A. 2009. Function of C/EBPδ in a regulatory circuit that discriminates between transient and
persistent TLR4
-
induced signals.

Nature Immunology

10(4):437

443.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

62.

Rand U, Rinas M, Schwerk J, Nöhren G, Linnes M, Kröger A, Flossdorf M, et al. 2012. Multi
-
layered
stochasticity and paracrine signal propagation shape the type
-
I interferon response.
Molecular Systems
Biology

8(1):

63
.

Maiwald T, Schneider A, Busch H, Sahle S, Gretz N, Weiss TS, Kummer U, and Klingmüller U. 2010.
Combining theoretical analysis and experimental data generation reveals IRF9 as a crucial factor for
accelerating interferon α
-
induced early antiviral signal
ling.
FEBS Journal

277(22):4741

4754.

64.

Dash P, McClaren JL, Oguin TH, Rothwell W, Todd B, Morris MY, Becksfort J, et al. 2011. Paired
analysis of TCRα and TCRβ chains at the single
-
cell level in mice.
J. Clin. Invest

121(1):288

295.

65.

Thomas PG, Bro
wn SA, Morris MY, Yue W, So J, Reynolds C, Webby RJ, and Doherty PC. 2010.
Physiological numbers of CD4+ T cells generate weak recall responses following influenza virus
challenge.
J. Immunol

184(4):1721

1727.

66.

La Gruta NL, Rothwell WT, Cukalac T, Swan NG, Valkenburg SA, Kedzierska K, Thomas PG, Doherty
PC, and Turner SJ. 2010. Primary CTL response magnitude in mice is determined by the extent of naive
T cell recruitment and subsequent clonal expansion.
J. Clin.
Invest

120(6):1885

1894.

67.

Rutigliano JA, Morris MY, Yue W, Keating R, Webby RJ, Thomas PG, and Doherty PC. 2010. Protective
memory responses are modulated by priming events prior to challenge.
J Virol

84(2):1047

1056.

68.

Wang GC, Dash P, McCullers JA
, Doherty PC, and Thomas PG. 2012. T
-
cell receptor αβ diversity
inversely correlates with antibody levels in human cytomegalovirus infection.
Sci Transl Med

In press:

69.

Doherty PC, Turner SJ, Webby RG, and Thomas PG. 2006. Influenza and the challenge fo
r immunology.
Nat Immunol

7(5):449

55.

70.

Turner SJ, La Gruta NL, Kedzierska K, Thomas PG, and Doherty PC. 2009. Functional implications of T
cell receptor diversity.
Curr. Opin. Immunol

21(3):286

290.

71.

Hawn TR, Berrington WR, Smith IA, Uematsu S, Ak
ira S, Aderem A, Smith KD, and Skerrett SJ. Nov 15.
Altered Inflammatory Responses in TLR5
-
Deficient Mice Infected with Legionella pneumophila.
J
Immunol

179(10):6981

7.

Program Director/Principal Investigator (Last, First,
Middle):

Aitchison, John

PHS 398/2590 (Rev. 06/09
)

Page


Continuation Format Page

72.

Hayashi F, Smith KD, Ozinsky A, Hawn TR, Yi EC, Goodlett DR, Eng JK, Akira S, Und
erhill DM, and
Aderem A. 2001. The innate immune response to bacterial flagellin is mediated by Toll
-
like receptor 5.
Nature

410(6832):1099

103.

73.

Sorensen TI, Nielsen GG, Andersen PK, and Teasdale TW. Mar 24. Genetic and environmental
influences on pre
mature death in adult adoptees.
N Engl J Med

318(12):727

32.

74.

Berrington WR and Hawn TR. 2007. Mycobacterium tuberculosis, macrophages, and the innate immune
response: does common variation matter?
Immunological Reviews

219(1):167

186.

75.

Casanova JL

and Abel L. 2002. Genetic dissection of immunity to mycobacteria: the human model.
Annu
Rev Immunol

20:581

620.

76.

Cooke GS and Hill AV. 2001. Genetics of susceptibility to human infectious disease.
Nat Rev Genet

2(12):967

77.

77.

Misch EA and Hawn TR.

Mar. Toll
-
like receptor polymorphisms and susceptibility to human disease.
Clin
Sci (Lond)

114(5):347

60.