Microbiota regulates immune defense against respiratory tract influenza A virus infection.

Abstract

Although commensal bacteria are crucial in maintaining immune homeostasis of the intestine, the role of commensal bacteria in immune responses at other mucosal surfaces remains less clear. Here, we show that commensal microbiota composition critically regulates the generation of virus-specific CD4 and CD8 T cells and antibody responses following respiratory influenza virus infection. By using various antibiotic treatments, we found that neomycin-sensitive bacteria are associated with the induction of productive immune responses in the lung. Local or distal injection of Toll-like receptor (TLR) ligands could rescue the immune impairment in the antibiotic-treated mice. Intact microbiota provided signals leading to the expression of mRNA for pro-IL-1β and pro-IL-18 at steady state. Following influenza virus infection, inflammasome activation led to migration of dendritic cells (DCs) from the lung to the draining lymph node and T-cell priming. Our results reveal the importance of commensal microbiota in regulating immunity in the respiratory mucosa through the proper activation of inflammasomes.

Full text

Page 1

Microbiota regulates immune defense against
respiratory tract influenza A virus infection
Takeshi Ichinohea,b,1, Iris K. Panga,1, Yosuke Kumamotoa, David R. Peaperc, John H. Hoa, Thomas S. Murrayc,d,
and Akiko Iwasakia,2
aDepartment of Immunobiology,dDepartment of Pediatrics, andcLaboratory Medicine, Yale University School of Medicine, New Haven, CT 06520;
andbDepartment of Virology, Faculty of Medicine, Kyushu University, Fukuoka 812-8582, Japan
Edited by Dan R. Littman, New York University Medical Center, New York, NY, and approved February 7, 2011 (received for review December 23, 2010)
Although commensal bacteria are crucial in maintaining immune
homeostasis of the intestine, the role of commensal bacteria in
immune responses at other mucosal surfaces remains less clear.
Here, we show that commensal microbiota composition critically
regulates the generation of virus-specific CD4 and CD8 T cells and
antibody responses following respiratory influenza virus infection.
By using various antibiotic treatments, we found that neomycin-
sensitive bacteria are associated with the induction of productive
immune responses in the lung. Local or distal injection of Toll-like
receptor (TLR) ligands could rescue the immune impairment in the
antibiotic-treated mice. Intact microbiota provided signals leading
to the expression of mRNA for pro–IL-1β and pro–IL-18 at steady
state.Followinginfluenzavirusinfection,inflammasomeactivation
ledtomigrationofdendriticcells(DCs)fromthelungtothedraining
lymphnodeandT-cellpriming.Ourresultsrevealtheimportanceof
commensal microbiota in regulating immunity in the respiratory
mucosa through the proper activation of inflammasomes.
mucosal immunity|NLRP3|caspase-1|adaptive immunity
V
tality in both humans and animals worldwide. Influenza virus
is the major source of severe viral respiratory infections in adults,
causing annual epidemics that result in severe morbidity and
mortality involving 3 to 5 million people annually. The constant
threat of the emergence of a novel influenza subtype engenders
an even greater risk to society, as the recent pandemic with the
swine flu (1) clearly demonstrated. Innate recognition of in-
fluenza virus through pattern recognition receptors (PRRs) plays
a central role in the generation of adaptive immune responses.
Recent studies highlighted the importance of the NOD-like re-
ceptor (NLR) activation of inflammasomes in antiviral defense
(2). Work from our laboratory and others demonstrated that
inflammasome activation and the downstream cytokines (3) play
a key role in innate (4, 5) and adaptive (6) immune defense
against influenza virus infection in vivo. Using a sublethal dose
(10 pfu) of A/PR8 strain, we demonstrated that mice deficient in
caspase-1 or apoptosis-associated speck-like protein containing
a caspase recruitment domain (ASC) have diminished T- and B-
cell responses and ultimately succumb to viral infection. Two
other groups, using high lethal doses of influenza virus challenge,
demonstrated that in the absence of the NLRP3 inflammasome,
mice succumb to high doses of A/PR8 infection due to innate
immune defects (4, 5). However, how inflammasomes are acti-
vated to elicit adaptive immunity following respiratory influenza
infection in vivo remains unclear.
Commensal bacteria are essential in shaping intestinal im-
mune responses in both health and disease (7, 8). Germ-free
mice have underdeveloped gut-associated lymphoid tissues in-
cluding Peyer’s patches, isolated lymphoid follicles, and mesen-
teric lymph nodes (9). Commensal bacteria are sensed by the
innate pattern-recognition receptors to maintain the homeostasis
of intestinal epithelial cell turnover and integrity (10). Gut
commensal microbiota also support intestinal immune homeo-
stasis by regulating Tregs (11–13) and Th17 cells (14–17). In
some instances, specific bacteria have been associated with im-
munological outcomes. Commensal bacteria, particularly the
iral respiratory infections cause severe morbidity and mor-
segmented filamentous bacteria, promote Th17 development in
the intestine (14, 15). In addition to their beneficial effects in the
intestine, recent studies highlight the importance of gut bacterial
composition in a number of pathological conditions including
diabetes (18) and obesity (19–22). Inflammatory bowel disease
can develop as a result of the emergence of harmful intestinal
bacteria (23). However, it remains unclear whether there is a role
for microbiota in shaping the immune inductive function at a
nonintestinal mucosal surface such as the lung.
Here, we examine the role of commensal bacteria in the ini-
tiation of adaptive immunity after respiratory infection with in-
fluenza virus. We demonstrate the requirement for intact com-
mensal bacterial community in the establishment of Th1, CTL,
and IgA responses to respiratory influenza virus infection. No-
tably, we found that neomycin-sensitive bacteria contributed to
immunocompetence in the lung. This was in part mediated by
providing signals for robust priming of pro–IL-1β and pro–IL-18
expression at steady state. Thus, our data reveal a key role for
commensal bacteria in controlling adaptive immunity against a
respiratory virus infection.
Results
Immune Responses to Respiratory Influenza Virus Infection Are
Diminished by Antibiotic Treatment. To determine the impor-
tance of commensal microbiota in immune responses within the
respiratory tract, mice were subjected to a 4-wk oral adminis-
tration of antibiotic combination, vancomycin, neomycin, met-
ronidazole, and ampicillin (V/N/M/A) (10, 11, 24–26). This
treatment resulted in significant changes in the composition of
culturable commensal bacteria (Fig. S1). Antibiotic-treated mice
were then infected intranasally with a sublethal dose (10 pfu) of
A/PR8 influenza virus. Mice were kept on the V/N/M/A regimen
for the entire duration of the experiments. Two weeks later, vi-
rion-specific Ig levels and T-cell responses were measured. In-
fluenza virus-specific antibody titers (Fig. 1A) and CD4 T-cell
responses (Fig. 1B) were significantly reduced in the antibiotic-
treated group. Both cytokine secretion (Fig. 1 B and C) and the
frequency of influenza virus-specific CTLs as determined by
tetramer staining (Fig. 1D) were diminished in the antibiotic-
treated mice. As a consequence, viral titer in the lung remained
significantly elevated in the antibiotic-treated mice at day 9
postinfection (Fig. 1E). These data indicated that commensal
microbiota composition profoundly affects the adaptive immune
responses to respiratory influenza virus infection and that anti-
biotic treatment predisposes the mice to high viral replication in
the lung.
Author contributions: T.I., I.K.P., Y.K., D.R.P., T.S.M., and A.I. designed research; T.I., I.K.P.,
Y.K., D.R.P., J.H.H., and T.S.M. performed research; T.I., I.K.P., Y.K., D.R.P., J.H.H., T.S.M.,
and A.I. analyzed data; and T.I., I.K.P., D.R.P., T.S.M., and A.I. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1T.I. and I.K.P. contributed equally to this work.
2To whom correspondence should be addressed. E-mail: akiko.iwasaki@yale.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1019378108/-/DCSupplemental.
5354–5359
| PNAS
| March 29, 2011
| vol. 108
| no. 13 www.pnas.org/cgi/doi/10.1073/pnas.1019378108

Page 2

Selective Role of Commensal Microbiota in Immune Responses to
Influenza Virus. To examine whether the antibiotic treatment
resulted in general immunodeficiency, we immunized these
mice with ovalbumin (OVA) in complete Freund’s adjuvant
(CFA) in the footpad. Unlike lung infection with influenza virus
(Fig. 1), immunization with ovalbumin in CFA led to normal Ig
(Fig. S2A) and T-cell responses (Fig. S2B) in antibiotic-treated
mice. Adaptive immunity following respiratory infection with
influenza virus requires the inflammasome activation (6). This
led us to consider the possibility that commensal microbiota may
regulate immune responses that require (i) the activation of the
inflammasomes, and/or (ii) respiratory infection. To address
these possibilities, we took advantage of the fact that immunity
to genital infection with herpes simplex virus type 2 (HSV-2)
does not require inflammasome activation (27). Thus, mice
treated with antibiotics were infected intranasally with HSV-2. In
contrast to flu infection, we observed WT levels of CD4 and CD8
T-cell activation in response to intranasal infection with HSV-2
in antibiotic-treated mice (Fig. S2 C and D). Next, we examined
whether intact commensal microbiota is required for immune
response to a bacterial respiratory pathogen. We chose to study
Legionella pneumophila, as it is a natural respiratory pathogen
and is resistant to the antibiotics used in this study (28, 29). Upon
intranasal infection with L. pneumophila, both CD4 and CD8 T-
cell responses remained intact in antibiotic-treated mice (Fig. S2
E and F). Moreover, similar immune responses were generated
in caspase-1–deficient mice, indicating that neither inflamma-
somes nor intact commensal bacteria is required for adaptive
immune responses to L. pneumophila. These data suggested that
the requirement for intact commensal bacteria is restricted to
inflammasome-dependent priming and not a general character-
istic unique to the respiratory mucosa or to viral infections.
Treg Suppression Is Not Responsible for Commensal Microbiota-
Dependent Immunity Against Influenza Virus. Several possible mech-
anisms could explain how commensal microbiota might support
the generation of adaptive immunity to influenza virus in the lung
(Fig. S3). First, commensal microbiota might contribute to inflam-
masome activation. Second, commensal bacteria-derived products
couldprovideligandsforTLRsorotherPRRsand enhancestimu-
lation of adaptive immunity induced by influenza virus infection.
Third, because commensal bacteria-associated DNA could block
the conversionofTregswithinthegutmucosa(11),antibiotictreat-
ment could result in an enhanced Treg-mediated suppression of
adaptive immunity to influenza virus in the lung.
To test the third hypothesis, we measured the percentages of
Foxp3+Tregs in various tissues of water-fed vs. antibiotic-treated
mice. As reported previously (11), Treg numbers were increased
in the Peyer’s patches but not in the spleen of antibiotic-treated
mice (Fig. S4A). In addition, in lung and in the mediastinal LNs,
we detected higher percentages of Foxp3+CD4+T cells in an-
tibiotic-treated mice (Fig. S4A). Thus, we examined whether
immunedeficiency intheantibiotic-treated micecouldberescued
by Treg depletion. To this end, Tregs were depleted by antibody
against CD25 a few days before influenza virus infection in anti-
biotic-treated animals (Fig. S4B). This procedure has been shown
previously to deplete Tregs transiently, during which critical
suppressive function of the Tregs is mediated, resulting in en-
hancement of effector T-cell responses to various antigens (30,
31). Treg depletion was confirmed by flow cytometry (Fig. S4C).
Our results showed that neither antibody (Fig. S4D) nor T-cell
responses to influenza virus (Fig. S4E) were recovered by de-
pletion of Tregs in antibiotic-treated mice. These data indicated
that the antibiotic-mediated impairment in antiinfluenza immu-
nity is not sufficiently explained by the effects of commensal
bacteria on suppression of Treg conversion.
Rectal TLR Stimulation Restores Immune Response to Influenza Virus
Infection in Antibiotic-Treated Mice. Even though the colon con-
tains by far the highest number of commensal bacteria in the
body, upper respiratory mucosa is also inhabited by numerous
bacterial species. Thus, to examine whether a TLR agonist can re-
store immune responses to influenza virus infection in antibiotic-
treated mice, and to determine whether the location of such TLR
signals dictates the outcome of immune responses to influenza virus
infection, we inoculated LPS either intranasally (local) (Fig. 2 A–D)
or intrarectally (distal) (Fig. 2 E–H) at the time of influenza virus
challenge in antibiotic-treated mice. We picked rectal inoculation
to mimic the effect of the high levels of commensal bacteria pre-
sentinthecolon.Remarkably, bothantibodyandT-cellresponses
were completely restored by a single inoculation of LPS either
locally or distally (Fig. 2). Next, we tested whether other TLR
agonists exert similar effects. We found that rectal inoculation of
CpG (TLR9 agonist), Poly I:C (TLR3 agonist), and to a lesser
extent, peptidoglycan (TLR2 agonist), could restore immunity to
influenza virus in the lung (Fig. 2I). These data indicated that
a TLR stimulus is sufficient to restore immune responses in anti-
biotic-treatedmiceandfurthersuggestedthatsignalscomingfrom
distal commensal bacterial products (colon) may be sufficient to
support immune priming in the lung.
Neomycin-Sensitive Commensal Bacteria Are Required for Immune
Responses to Influenza Virus Infection. To better understand the
bacterial classes responsible for endowing immune functions in
the respiratory tract, we treated mice with individual antibiotics.
Mice were treated with either combination of all four antibiotics
(Abt) or individual antibiotics, ampicillin (Amp), vancomycin
(Vanco),neomycin(Neo),ormetronidazole(Metro).Remarkably,
Fig. 1.
munity to influenza virus infection. C57BL/6 mice were given
ampicillin (1 g/L), vancomycin (500 mg/L), neomycin sulfate
(1 g/L), and metronidazole (1 g/L) in drinking water for 4 wk
before PR8 virus infection (10 pfu per mouse). Two weeks
later, serum and nasal wash were collected and Ag-specific
antibody titers were measured (A), and T-cells were isolated
from spleen and restimulated with flu virion or NP peptide
for 72 h, and IFN-γ production from CD4 T cells (B) and CD8 T
cells (C) was measured. (D) Lymphocytes were collected from
the lung of infected animals at 14 d postinfection and stained
with flu-specific tetramer. (E) The lung washes of flu-infected
mice were harvested at 9 d postinfection, and viral titers were
measured by plaque assay. *P < 0.05 and ***P < 0.001 vs.
water-fed group. Data represent the mean ± SD. Similar
results were obtained from three separate experiments.
Antibiotic-treated mice fail to induce acquired im-
Ichinohe et al.PNAS
| March 29, 2011
| vol. 108
| no. 13
| 5355
IMMUNOLOGY

Page 3

whereas treatment with ampicillin, vancomycin, or metronidazole
alone had only a mild and variable effect on CTL responses after
9 or 14 d of influenza infection, neomycin treatment alone re-
capitulated the effects of the combination antibiotic treatment on
the suppression of CTL responses (Fig. 3A). Metronidazole treat-
ment still maintained CD8 T-cellresponses inthelung. Incontrast,
oral neomycin treatment abolished CD8 T-cell responses in the
lung. To determine the bacterial species that are affected by the
various antibiotic treatments, stool and nasal wash samples from
these mice were subjected to microbiological analyses. Culturable
bacteria were enumerated and identified by a combination of bio-
chemical analyses and 16S ribosomal sequencing (Fig. 3 B–E). In
the antibiotic-untreated mice and in mice treated with oral van-
comycin or metronidazole, Lactobacillus spp. (Gram positive)
dominated the intestinal and nasal tracts. It has been reported that
orally delivered neomycin is not absorbed through the gastroin-
testinal tract, thus selectively targeting aerobic bacteria in the in-
testinal lumen but not elsewhere (32). Consistently, oral neomycin
treatment resulted in a significant decrease in the intestinal (Fig.
3B), but not nasal (Fig. 3C), bacterial load. Moreover, the stools
of neomycin (aminoglycoside)-treated mice were dominated by
Sphingomonas spp., which are known to be resistant to amino-
glycoside antibiotics (33) (Fig. 3D). Oral ampicillin treatment di-
minished Gram-positive bacteria in both the intestine and in the
nasal mucosa. Instead, both colonic and nasal mucosae were
dominated by an outgrowth of Enterobacter spp., which are mostly
ampicillin resistant (34) (Fig. 3 D and E). Collectively, these data
indicated that oral treatment with neomycin impaired immune
responses to respiratory influenza virus infection, which was asso-
ciated with depletion of Gram-positive bacteria in the gut but not
the nasal tract.
Commensal Bacterial Supply Signal 1 for IL-1β and IL-18 Secretion.
Next, we asked how commensal bacteria are required for optimal
respiratory immune responses against influenza virus infection.
Because intact commensal microbiota was required for adaptive
immune responses to inflammasome-dependent (flu), but not
inflammasome-independent pathogen (HSV-2, Legionella) (Fig.
S2), we examined whether commensal microbiota might provide
signals necessary for inflammasome-dependent cytokine secre-
tion. Two distinct stimuli are necessary for the processing and
secretion of IL-1β; the first stimulus (signal 1) to induce tran-
scription and translation of pro–IL-1β; and the second (signal 2),
to activate caspase-1 that processes IL-1β into mature form (35).
The priming signal (signal 1), often provided through TLRs,
leads to transcriptional activation of the genes encoding pro–IL-
1β, pro–IL-18, and NLRP3 (36, 37). The second signal is medi-
ated through the activation of the inflammasome complex,
leading to the proteolytic activation of caspase-1. To test whether
commensal microbiota might supply signals necessary for IL-1β
and IL-18 secretion, we measured the levels of secreted IL-1β in
the bronchoalveolar lavage (BAL) of mice infected with in-
fluenza that had been pretreated with the combination anti-
biotics. Notably, antibiotic treatment of WT mice impaired
mRNA expression of pro–IL-1β, pro–IL-18, and NLRP3 and
reduced secretion of mature IL-1β protein in the BAL to the
levels of ASC-deficient mice following influenza infection (Fig.
S5). These data indicated that intact commensal microbiota
provides signals that support the expression of pro–IL-1β and
pro–IL-18 (signal 1) even at the steady state. Upon influenza
infection, virus-inflicted damage, including the activity of M2 ion
channel (signal 2) (38), leads to the activation of inflammasomes
and release of mature forms of IL-1β and IL-18.
Antibiotic Treatment Impairs DC Homeostasis and Migration by
Reducing Priming Signals for Inflammasome-Dependent Cytokines.
Because T-cell priming upon influenza virus infection is medi-
ated by DCs in the mediastinal LNs (39), we examined the an-
tigen presenting capacity of mLN DCs in antibiotic-treated mice.
Mice were infected intranasally with a recombinant influenza
virus encoding for the GP33-41 epitope of lymphocytic chorio-
meningitis virus (6). At steady state, the number of non-DCs in
the lung, mLN, spleen, blood, and bone marrow were compa-
rable in antibiotic-treated vs. -untreated mice. In contrast, gross
lymph node size (Fig. 4A), cellularity (Fig. 4B), and DC num-
bers (Fig. 4C) in the mLN were considerably reduced in anti-
biotic-treated influenza virus-infected mice. Further, freshly
isolated mLN DCs from antibiotic-treated mice, in the absence
of exogenous peptide, induced no proliferation (Fig. 4D) or
differentiation (Fig. 4E) of P14 CD8 TCR Tg T cells specific for
the GP33-41 epitope presented on H-2Db. This defect was due
to impairment of viral antigen peptide presentation by the mLN
Fig. 2.
fluenza virus infection in antibiotic-treated mice. C57BL/6 mice were given
antibiotics in drinking water for 4 wk before 10 pfu of PR8 viral infection
with or without 2 μg of LPS injected intranasally (A–D) or intrarectally (E–H).
Two weeks later, serum was collected and Ag-specific antibody titers were
measured (A and E), and T cells were isolated from spleen and restimulated
with flu virion (B and F) or NP peptide (C, D, G, and H) for 72 h, and IFN-γ
production from CD4 T cells (B, C, F, and G) and CD8 T cells (D and H) was
measured. (I) Water-fed and antibiotic-treated mice were infected in-
tranasally with 10 pfu of PR8 virus with or without intrarectal injection of
LPS (5 μg), CpG2216 (50 μg), peptidoglycan (20 μg), or Poly (I:C) (50 μg). Total
numbers of influenza virus-specific CD8 T cells in the lung are shown. Data
represent the mean ± SD, and are representative of at least three inde-
pendent experiments (A–H) or are pooled from two independent experi-
ments (I). *P < 0.05; **P < 0.01; ***P < 0.001.
Local and distal TLR stimulation restores immune response to in-
5356
| www.pnas.org/cgi/doi/10.1073/pnas.1019378108Ichinohe et al.

Page 4

DCs from antibiotic-treated mice, because the same DCs were
able to differentiate P14 naïve CD8 T cells after exogenous
GP33 peptide addition (Fig. 4F).
We hypothesized that the reduction in antigen-presenting
DCs in the antibiotic-treated mice is due to (i) reduction in the
number of respiratory DCs, and/or (ii) the failure of DCs to
migrate from the lung to the mLN. To determine the possible
effect of antibiotics on respiratory DCs, flow cytometric analyses
of the phenotype and number of lung DCs were carried out.
Of note, we observed a significant reduction in the number of
CD103+DCs in the lung and mLNs of antibiotic-treated mice
at the steady state (Fig. 4G). Respiratory tract CD103+DCs are
responsible for priming CD8 T-cell responses early after in-
fluenza virus infection (39, 40). In addition, expression levels
of CD86, CD80, and MHC class II were lower in DCs from
antibiotic-treated mice (Fig. 4H). Next, to determine the mi-
gration of respiratory tract DCs into the mLN upon influenza
virus infection, mice were first inoculated with carboxy-
fluorescein succinimidyl ester (CFSE) intranasally and lung-mi-
grant DCs were enumerated after 18 h post-infection in the mLN
(Fig. 4I) (41). These data showed that commensal bacteria are
required for lung DCs to migrate to the mLN. Consequently, P14
naïve CD8 T cells specific for the GP33 peptide failed to pro-
liferate in the mLNs at 5 d after infection with either 10 pfu or
100 pfu of influenza virus in vivo (Fig. S6). Next, to test whether
such a defect in lung DCs in the antibiotic-treated mice corre-
lates with their failure to prime for inflammasome-dependent
cytokines, we examined the ability of DCs to migrate from the
lung to the mLN in caspase-1–deficient mice’s response to flu
infection. Migration of lung DCs to the mLN was severely im-
paired in the absence of caspase-1 (Fig. 4J). These data indicated
that inflammasome activation, which is diminished by antibiotic
treatment (Fig. S5), is required for migration of respiratory DCs
to the mLNs. Finally, we examined whether the impaired im-
mune responses observed in antibiotic-treated mice was directly
linked to the diminished DC migration. To this end, we tested
whether the TLR rescue of the influenza response in antibiotic-
treated mice (Fig. 2) was colinked to DC migration from the lung
to the mLN. Notably, LPS inoculation (intranasal or intrarectal)
completely restored DC migration into the mLN after flu in-
fection in antibiotic-treated mice (Fig. 4K). These data collec-
tively indicated that alteration in the microbiota by antibiotics
leads to the failure in the activation of inflammasome-dependent
Fig. 3.
immune responses to respiratory influenza virus infection. C57BL/6 mice
were each given single or four combinatorial antibiotics in drinking water
for 4 wk before PR8 virus infection (10 pfu per mouse). Lymphocytes were
collected from the lung of infected animals at 9 d or 14 d postinfection and
stained with flu-specific tetramer (A). Bacterial load in the stool (B) and nasal
wash (C) from antibiotic-treated mice (n = 3 per condition) were measured.
Bacterial compositions in the stool (D) and nasal wash (E) from single anti-
biotic-treated mice are depicted. Purple and yellow tones denote Gram-
positive and Gram-negative bacteria, respectively. Similar results were
obtained from two to three separate experiments. *P < 0.05; **P < 0.01.
Effect of single antibiotic treatment on bacterial colonization and
Fig. 4.
cell responses in antibiotic-treated mice. C57BL/6 mice were given antibiotics
in drinking water for 4 wk before intranasal infection with 1,000 pfu of PR8-
GP33 viruses. (A–C) Three days later, CD11c+DCs were isolated from the
mediastinal LN. (D–F) Naïve p14 tg CD8 T cells (2 × 105cells per well) were
cocultured with different numbers of CD11c+DCs isolated from mLN of
infected animals with (F) or without GP33 peptide (D and E) for 72 h. Splenic
DCs from infected animals were used as negative control (E). IFN-γ pro-
duction (E and F) and the number of CD8 T cells (D) were measured. The
number of CD103+DCs (G) and phenotype of total DC population (H) in the
lung and mLN were measured in antibiotic-treated mice without influenza
infection. (I and J) Water-fed, antibiotic-treated, and caspase-1–deficient
mice were inoculated intranasally with CFSE. Six hours later, mice were
infected with 1,000 pfu of PR8 viruses. Eighteen hours after infection, me-
diastinal LNs were collected. The numbers of CFSE+CD11c+DCs are shown. (K)
LPS inoculation (intranasal or intrarectal) restored DC migration to the mLN
following intranasal influenza virus infection. Data represent the mean ±
SD. Similar results were obtained from three separate experiments.
Respiratory tract DCs fail to migrate to the draining LN and prime T-
Ichinohe et al.PNAS
| March 29, 2011
| vol. 108
| no. 13
| 5357
IMMUNOLOGY

Page 5

cytokine release and to the impairment of respiratory tract DC
homeostasis and their migration to the draining LN. Such a de-
fect in DC migration results in diminished T-cell responses fol-
lowing influenza infection.
Discussion
Commensal bacteria provide the mammalian hosts with essential
functions well beyond the digestive system. Although their im-
portance in the maintenance of immune homeostasis in the gut
mucosa is widely appreciated (7), our study revealed that the
commensal microbiota could regulate immune responses in the
respiratory mucosa. We demonstrated that oral antibiotic treat-
ments resulted in defective CD4 T-, CD8 T-, and B-cell immunity
following intranasal infection with influenza virus. This was
not due to global or local immune deficiencies in these mice, as
antibiotic treatments did not impair immune responses against
protein antigen injected with CFA or to respiratory infection
with HSV-2 or Legionella pneumophila. Commensal bacteria-
mediatedsupport ofthe immune response against flu infection was
also independent of their effect on Tregs, but was sufficiently re-
stored by distal (rectal) inoculation of TLR agonists. In antibiotic-
treated mice, synthesis of pro–IL-1β, pro–IL-18, and NLRP3 was
impaired even at the steady state. Antibiotic treatment led to im-
paired respiratory DC distribution and activation status at steady
state and reduced DC migration from the lung to the draining
lymph nodes. Acute injection of LPS was able to overcome de-
fective inflammasome activation and DC migration in antibiotic-
treated mice. Of note, IL-1β is known to play a key role in the
migration of Langerhans cells from the skin to the draining lymph
node in a contact hypersensitivity model (42), suggesting that this
cytokine may be a common inducer of DC mobilization.
Our data demonstrate that not all commensal bacteria could
contribute equally to the immunocompetence in the lung. Oral
neomycin, but not ampicillin, vancomycin, or metronidazole
treatment consistently abolished CTL responses to influenza vi-
rus. Microbiological examination of the stool and nasal secretions
from mice treated with a single antibiotic revealed that oral
treatments with vancomycin or metronidazole made minimal
changes to the composition or density of culturable endogenous
microbiota, which consisted mainly of Lactobacillus spp. In con-
trast,oralneomycintreatment,whichmainlytargetedgut-resident
bacteria, ablated the immune response to respiratory influenza
infection. Interestingly, neomycin depleted all culturable Gram-
positive bacteria in the colon, while maintaining those in the nasal
tract. Oral ampicillin treatment had a systemic effect on de-
pleting most of the Gram-positive bacteria, with an overgrowth of
Enterobacter spp. in both the colonic and nasal mucosae. These
data collectively indicated that commensal microbiotaresponsible
for conferring an immunogenic environment in the lungare either
the gut-resident Gram-positive bacteria and/or bacteria in the
nasal tract. Of note, neomycin treatment resulted in a shift from
mostly Lactobacillus spp. to Corynebacterium spp. in the nasal
mucosa. It is intriguing that certain bacterial species are in-
sufficient (Sphingomonas spp. in the gut and Corynebacterium spp.
in the nasal tract) to promote immunocompetence against in-
fluenza virus infection. Perhaps these bacteria fail to prime basal
levels of pro–IL-1β and pro–IL-18 expression. In the human in-
testine, anaerobic bacteria outnumber aerobic bacteria by a factor
of 100–1,000 (43). The phyla Firmicutes (Gram positive) and
Bacteroidetes (anaerobic Gram negative) are predominant in the
human colon (44), whereas aerobes (facultative anaerobes), such
as Escherichia, Enterobacter, Enterococcus, Klebsiella, Lactobacil-
lus, and Proteus, remain a small minority (43). Interestingly, a sys-
tematic metagenomic analysis of total microbiota communities
affected by similar oral antibiotic treatments in mice revealed that
intestinal mucosa-associated Lactobacillus spp. are significantly
reduced(45).However,futurestudiesareneededtodeterminethe
bacteria responsible for conferring immunocompetence against
influenza infection.
Our data revealed a link between commensal microbiota
community and inflammasome-dependent cytokine activation.
Currently, the precise mechanism by which gut commensal bac-
teria supports inflammasome activation in the lung is unknown.
We speculate that products of a select group of commensal bac-
teria could trigger a variety of pattern recognition receptors to
stimulate leukocytes either locally or systemically. Factors re-
leased by such leukocytes can support steady-state production of
pro–IL-1β, pro–IL-18, and NLR proteins, thereby priming signal
1 for inflammasome-dependent cytokine activation. This idea is
supported by our results that injection of TLR ligands can restore
immune responses to influenza virus in antibiotic-treated mice.
Commensal microbiota providing signals for PRRs has been
shown by a recent study showing that peptidoglycan translocated
from the gut microbiota to the systemic circulation is sensed by
Nod1, resulting in enhanced systemic innate immunity mediated
by neutrophils (46). We speculate that microbiome may be re-
sponsible for providing signal 1 for inflammasome-dependent
cytokine activation in vivo under circumstances in which signal 2
(such as uric acid and alum) alone are capable of eliciting robust
caspase-1–dependent IL-1β and IL-18 production.
In summary, our findings substantially expand the contribution
of the microbiota in maintaining the immunological status of the
host animal beyond the intestinal mucosa. Because antibiotic use
is prevalent in the treatment of respiratory infections, our results
imply a possible deleterious effect of such treatment in initiating
proper immune responses to influenza virus. Conversely, it will
be important to determine whether probiotic therapy can be
explored for immune-stimulating effects during the flu season.
Materials and Methods
Antibiotic Treatment. Mice were treated for 4 wk with ampicillin (1 g/L),
vancomycin (500 mg/L), neomycin sulfate (1 g/L), and metronidazole (1 g/L) in
drinking water as previously described (10, 11, 24–26). For some experiments,
individual antibiotics (at the indicated concentration) were given in drinking
water. Antibiotic-containing water was changed twice a week.
Virus Infection in Vivo. A/PR8 virus (H1N1) and recombinant PR8 virus
expressing the lymphocytic choriomeningitis virus (LCMV) glycoprotein epi-
tope GP33-41 (PR8-GP33; a gift from S. Kaech, Yale University, New Haven,
CT) used for all experiments was grown in allantoic cavities from 10- to 11-d-
old fertile chicken eggs for 2 d at 35 °C. Viral titer was quantified by a
standard plaque assay using Madin-Darby canine kidney cells and viral stock
was stored at −80 °C. For intranasal infection, mice were fully anesthetized
by i.p. injection of ketamine and xylazine and then infected by intranasal
application of 20 μL of virus suspension (10–1,000 pfu of influenza virus or
106pfu of HSV-2 in PBS). This procedure leads to upper and lower respiratory
tract infection.
Nasal and Rectal Inoculation of TLR Agonist. For intranasal inoculation, water-
fed and antibiotic-treated mice were infected by intranasal application of
20 μL of virus suspension including 10 pfu of influenza virus and 2 μg of LPS.
For intrarectal inoculation, water-fed and antibiotic-treated mice were in-
oculated by intrarectal application of 50 μL of PBS containing ultrapure LPS
from Escherichia coli (InvivoGen; 5 μg), CpG2216 (Tri-Link Biotech; 50 μg),
peptidoglycan from Bacillus subtilis (Fluka, now Sigma-Aldrich; 20 μg), or
Poly (I:C) (InvivoGen; 50 μg) at the same time with intranasal infection with
10 pfu of influenza virus.
ACKNOWLEDGMENTS. We thank S. Yu and B. Yordy for technical support
and S. Shin and R. Medzhitov for helpful advice. This study was supported
by grants from the National Institutes of Health (NIH) (AI054359, AI062428,
and AI064705). T.I. was a recipient of the Japan Society for the Promotion
of Science Postdoctoral Fellow for Research Abroad. A.I. is a recipient of
the Burroughs Wellcome Investigators in Pathogenesis of Infectious Disease.
I.K.P. was supported by a NIH National Research Service Award (T32AI07019)
from the Interdisciplinary Immunology Training Program in Yale University,
Department of Immunobiology.
1. Neumann G, Noda T, Kawaoka Y (2009) Emergence and pandemic potential of swine-
origin H1N1 influenza virus. Nature 459:931–939.
2. Fitzgerald KA (2010) NLR-containing inflammasomes: Central mediators of host
defense and inflammation. Eur J Immunol 40:595–598.
5358
| www.pnas.org/cgi/doi/10.1073/pnas.1019378108 Ichinohe et al.