Cell, Vol. 122, 107–118, July 15, 2005, Copyright ©2005 by Elsevier Inc. DOI 10.1016/j.cell.2005.05.007
An Immunomodulatory Molecule
of Symbiotic Bacteria Directs
Maturation of the Host Immune System
Sarkis K. Mazmanian,1,3,* Cui Hua Liu,1,2
Arthur O. Tzianabos,1,3and Dennis L. Kasper1,3,*
Department of Medicine
Brigham and Women’s Hospital
Harvard Medical School
Boston, Massachusetts 02115
2Institute of Molecular Biology
Southern Medical University
3Department of Microbiology and Molecular Genetics
Harvard Medical School
Boston, Massachusetts 02115
2001). The magnitude of this interaction between com-
mensal bacteria and mammals must predictably exert
fundamental influences on the physiology of both. The
most impressive feature of this relationship may be that
the host not only tolerates but has evolved to require
colonization by commensal microorganisms for its own
development and health.
Autochthonous (indigenous) bacteria in the mamma-
lian gut have long been appreciated for potential bene-
fits to the host: provision of essential nutrients, me-
tabolism of indigestible compounds, defense against
colonization by opportunistic pathogens, and contribu-
tions to the development of the intestinal architecture
(Hooper et al., 2000; Hooper et al., 2002). How and, more
importantly, why does the immunocompetent gut envi-
ronment allow the presence of multitudinous foreign
organisms? Researchers have proposed that certain
commensal bacteria have evolved to aid in the host’s
health; several organisms are being studied for pro-
biotic (beneficial) potential (Guarner and Malagelada,
2003; Rastall, 2004). The “hygiene hypothesis” sug-
gests that the appropriate bacterial constitution of the
human microflora is a factor in protection from allergy
and asthma (Umetsu et al., 2002; Von Hertzen and
Haahtela, 2004). Investigations have shown that the in-
teractions of commensal bacteria with Toll-like recep-
tors are critical for intestinal homeostasis (Rakoff-
Nahoum et al., 2004). The intimate relationships between
commensal microorganisms and the host immune sys-
tem are increasingly evident (Macpherson and Harris,
2004; Noverr and Huffnagle, 2004).
The mammalian immune system is a dynamic and
remarkable organ. In recognizing, responding, and
adapting to countless foreign and self molecules, the
immune system is central to processes of health and
disease. CD4+T cells, a major component of the im-
mune system, are required for vital aspects of proper
immune function, from reactions to infectious agents to
control of autoimmune reactions and cancers (Janeway
et al., 2001). Effector CD4+T cells are of two general
subtypes, T helper 1 (TH1) and T helper 2 (TH2), each
carrying out distinct and opposing activities. The proper
balance between TH1 and TH2 immunologic responses
is critical to overall human and animal health (Neurath
et al., 2002; Sheikh and Strachan, 2004). A role for com-
mensal bacteria in establishing this equilibrium has
been postulated (Bowman and Holt, 2001; Rook and
Brunet, 2002). We investigated the molecular contribu-
tions of specific autochthonous organisms to the cellu-
lar development of the host immune system.
Bacteroides fragilis is a ubiquitous and important
Gram-negative anaerobe that colonizes the mammalian
lower gastrointestinal tract. Bacteroides species are
among the earliest-colonizing and most numerically
prominent constituents of the gut microflora (Kononen
et al., 1992). Although capsular polysaccharides are
common in many bacterial species, B. fragilis elabo-
rates an unprecedented eight distinct surface polysac-
charides (Krinos et al., 2001). Several of these polymers
have a novel zwitterionic structure, with both positive
The mammalian gastrointestinal tract harbors a com-
plex ecosystem consisting of countless bacteria in
homeostasis with the host immune system. Shaped
by evolution, this partnership has potential for symbi-
otic benefit. However, the identities of bacterial mole-
cules mediating symbiosis remain undefined. Here
we show that, during colonization of animals with the
ubiquitous gut microorganism Bacteroides fragilis, a
bacterial polysaccharide (PSA) directs the cellular
and physical maturation of the developing immune
system. Comparison with germ-free animals reveals
that the immunomodulatory activities of PSA during
B. fragilis colonization include correcting systemic T
cell deficiencies and TH1/TH2 imbalances and directing
lymphoid organogenesis. A PSA mutant of B. fragilis
does not restore these immunologic functions. PSA
presented by intestinal dendritic cells activates CD4+
T cells and elicits appropriate cytokine production.
These findings provide a molecular basis for host-
bacterial symbiosis and reveal the archetypal mole-
cule of commensal bacteria that mediates develop-
ment of the host immune system.
Immediately after a sterile birth, mammals are initiated
into an organized and lifelong process of colonization
by foreign organisms. Shaped by eons of evolution,
some host-bacterial associations have developed into
prosperous relationships creating diverse environments.
No better example exists in biology than the astound-
ing numbers of bacteria harbored by the lower gastroin-
testinal tract of mammals (Hooper et al., 1998). By
young adulthood, humans and other mammals are host
to w1012viable bacteria per gram of colonic content,
consisting of 500–1000 microbial species and outnum-
bering host cells by 100-fold (Hooper and Gordon,
and negative charges in each repeating unit (Tzianabos
et al., 1993). Zwitterionic polysaccharides (ZPSs) are
unique T cell-dependent antigens that specifically me-
diate the proliferation of CD4+T cells in vitro (Brubaker
et al., 1999; Tzianabos and Kasper, 2002). Adoptive
transfer experiments show that responses to polysac-
charide A (PSA), the most immunodominant ZPS of
B. fragilis, are conferred by CD4+T cells, not by B cells
or other T cells (Tzianabos et al., 1999). We have de-
scribed the novel internalization and processing of PSA
within endosomes of antigen-presenting cells (APCs)
(Cobb et al., 2004). Subsequent presentation of pro-
cessed polysaccharide by major histocompatibility com-
plex class II (MHC II) molecules activates CD4+T cells
and represents a previously undescribed pathway of
antigen presentation. Thus, ZPSs appear to have evolved
novel biological activities shaped by coevolution with
the host immune system.
Herein we show that monocolonization of germ-free
animals with B. fragilis is sufficient to correct several
immunologic defects found in the absence of a bacte-
rial microflora. The organism’s immunomodulatory ac-
tivity requires production of PSA, which mediates host
immune-system development through specific cellular
and molecular interactions. The significance of PSA’s
role in immune homeostasis lies in its ability to mediate
establishment of a proper TH1/TH2 balance for the host,
a fundamental aspect of healthy immunologic function.
B. fragilis PSA is the first identified member of a novel
class of molecules, referred to here as “symbiosis fac-
tors,” that mediate the beneficial relationship between
bacteria and mammalian hosts during mutualism.
colonization in animals with a naive immune system.
We chose the model microorganism B. fragilis because
of its prominence in the normal microbial gut flora and
its production of known immunomodulatory molecules
(Kononen et al., 1992; Tzianabos and Kasper, 2002). In
the absence of competing bacterial species, germ-free
mice monoassociated with B. fragilis strain NCTC 9343
are readily colonized to high levels (>1010cfu/g of fe-
ces; Figure 1B). Flow cytometry of splenic lymphocytes
from these mice shows a nearly complete restoration
of CD4+T cells to conventional proportions (Figures 1A
and 1C). Thus, B. fragilis monocolonization is sufficient
to correct CD4+T cell deficiency in spleens of germ-
free mice. No other bacterial species alone has been
shown to correct lymphoid defects in germ-free ani-
mals (Cash and Hooper, 2005).
The Immunomodulatory Effects of B. fragilis
Require Production of PSA
At least 2 of the 8 capsular polysaccharides of B. frag-
ilis are ZPSs, a unique class of bacterial molecules with
immunomodulatory properties (Tzianabos and Kasper,
2002). We wondered whether PSA, the most immuno-
dominant and highly conserved ZPS, plays a role in
splenic T cell expansion during B. fragilis commensal-
ism. We used PSA-deficient B. fragilis ?PSA to mono-
colonize germ-free mice (Coyne et al., 2001). The level
of intestinal colonization by the mutant is indistinguish-
able from that by the isogenic parent strain, as as-
sessed by fecal cfu counts (Figure 1B). Examination of
splenic lymphocyte populations from mice colonized
with B. fragilis lacking PSA but expressing all other an-
tigens produced by this organism reveals an inability to
correct CD4+T cell deficiencies in germ-free mice (Fig-
ure 1C). In pooled experiments (n = 4), the average pro-
portions of CD4+T cells were: conventional, 17.82% ±
2.1%; B. fragilis, 18.05% ± 1.9%; B. fragilis ?PSA,
10.95% ± 2.3%; and germ-free, 11.15% ± 1.5%. The
effects were specific to CD4+T cells, as the proportions
of CD8+T cells and CD19+B cells from splenic lympho-
cytes (Figure 1D) are indistinguishable between con-
ventional and either monocolonized or germ-free mice,
as previously observed (Pereira et al., 1986). Together,
these results show that B. fragilis colonizing the gut of
germ-free mice requires PSA production to correct host
systemic CD4+T cell deficiencies during commensalism.
Monocolonization of Germ-free Animals
with B. fragilis Results in CD4+T Cell Expansion
We investigated the effects of bacterial colonization on
immune maturation in animals, exploring the role of the
microbial flora in systemic T cell development. We used
germ-free mice: animals born and raised in sterile isola-
tors devoid of microbes. Initially, spleens were har-
vested from both conventionally colonized and germ-
free mice and were analyzed for total CD4+T cells by
flow cytometry (FC). All groups of mice had similar sple-
nic total lymphocyte counts (average: 1 × 108). Consis-
tent with seminal observations of a positive immuno-
logic role of autochthonous bacteria (Dobber et al.,
1992), the lymphocyte population purified from spleens
of conventional SPF (specific-pathogen-free) mice with
a diverse gut microflora contains a greater proportion
of CD4+T cells than that of germ-free mice (Figure
1A). Previous studies have documented the beneficial
role of commensal bacteria in intestinal development
(Hooper, 2004). The observed alteration in CD4+T cell
proportions of splenic lymphocytes highlights the pro-
found effects of bacterial colonization of the gut on the
systemic immune response.
SPF mice harbor a diverse and complex microbial
flora. To stringently investigate the influence of specific
bacterial constituents of the gut flora on the host im-
mune system, we needed to colonize germ-free mice
with a single bacterial species. This approach allows
“real-time” measurements of responses to bacterial
PSA Production by B. fragilis Directs
Commensal bacteria have long been appreciated for
their positive impact on development of gut-associated
lymphoid tissues (GALT, including Peyer’s patches) and
intraepithelial lymphocytes (IELs) and production of
mucosal IgA (Hooper, 2004). We studied whether cellu-
lar immune maturation after bacterial colonization was
also manifested in the morphological and ultrastruc-
tural development of peripheral lymphoid tissues. Germ-
free animals have recently been reported to display de-
fects in splenic structural development (Macpherson
and Harris, 2004). We examined histological sections of
spleens from germ-free mice colonized with wild-type
B. fragilis or B. fragilis ?PSA. Spleens from mice mono-
colonized with B. fragilis appear normal, with well-
formed lymphocyte zones appearing as defined folli-
Gut Bacterial Molecule Directs Immune Development
Figure 1. Cellular and Physical Immune Maturation in Germ-free Mice Requires PSA Production during Intestinal B. fragilis Colonization
(A) FC analysis of α-CD4-stained splenic lymphocytes from conventionally colonized (CNV) and germ-free (GF) mice reveals depletion of
CD4+T cells in the absence of colonizing microflora in the GI tract. Results are representative of four experiments with pools of 3 to 5 mice.
(B) Quantitation of fecal cfu during monocolonization of GF mice with wild-type B. fragilis or an isogenic mutant deficient in PSA production
(B. fragilis ?PSA) reveals equivalent counts of viable bacteria from fecal pellets. Inset: Immunoblot of bacterial extracts with α-PSA after SDS-
PAGE separation and electrotransfer to PVDF membranes shows the lack of PSA expression by B. fragilis ?PSA.
(C) FC of α-CD4-stained splenic lymphocytes from GF mice shows that intestinal monocolonization of mice with B. fragilis results in comple-
mentation of CD4+T cells. In the presence of every other antigen produced during colonization with B. fragilis ?PSA, the absence of PSA
results in no increase in CD4+T cell counts. Results are representative of >10 experiments with single or pools of 3 to 5 mice.
(D) FC of splenic lymphocytes from CNV and GF mice colonized with wild-type B. fragilis or B. fragilis ?PSA reveals no significant differences
in CD19+B cells (upper panels) or CD8+T cells (lower panels) in the CD4−fraction. Results are representative of four experiments with pools
of 3 to 5 mice.
(E) H&E-stained sections of spleens from CNV mice (right) and GF mice monocolonized with wild-type B. fragilis (left) and B. fragilis ?PSA
(center). White pulp containing lymphocytes appears as darker-staining follicular structures (arrows). The lack of large, well-defined follicles
in mice colonized with B. fragilis ?PSA is a measure of T cell depletion and reflects developmental defects in organogenesis. All images were
taken at the same magnification.
cles (white pulp) similar to those in conventional mice
with a complete gut flora (Figure 1E). The interspersed
red pulp is densely packed with red blood cells and
neutrophils. Spleens from germ-free mice colonized
with B. fragilis ?PSA show gross anatomical depletion
of the lymphocyte zones similar to that in uncolonized
germ-free mice (Macpherson and Harris, 2004). Folli-
cles are smaller, less defined, and more fragmented
than in germ-free mice colonized with wild-type B. frag-
ilis or conventional mice (Figure 1E). The overall size
and shape of spleens from all groups are comparable.
This finding suggests a role for PSA specifically in lym-
phocyte development. Thus, changes in CD4+T cell
expansion that are mediated by PSA produced by in-
testinal bacteria are consistent with the correction of
physical and developmental defects in secondary lym-
phoid tissues. This observation reflects the importance
of the beneficial relationship between commensal mi-
croorganisms and host physiology.
Purified PSA Is Sufficient to Expand T Cells
in Germ-free and Conventional Animals
The inability of a PSA-deficient mutant to cause T cell
expansion shows that the activity of this immunomodu-
This treatment recapitulates intestinal exposure of ani-
mals to PSA during colonization. Oral treatment led to
a specific increase of splenic CD4+T cells in conven-
tional mice (Figure 2B), showing that PSA’s effect is not
exclusive to germ-free animals. CD8+T cell and CD19+
B cell ratios were unaffected by PSA (Figure 2C) as for-
merly shown (Figure 1D). We found that specific re-
cognition of purified PSA by host immune components
in the intestines results in splenic CD4+T cell ex-
PSA Is Specifically Recognized by Dendritic Cells,
with Consequent Cell Activation
All CD4+T cell reactions require instruction to T cells
by APCs (Kidd, 2003; Kapsenberg, 2003). PSA is recog-
nized by APCs and then presented to T cells in vitro
(Kalka-Moll et al., 2002). The cellular mechanism of PSA
recognition in the intestine and subsequent signaling to
expand splenic T cells is unknown. To characterize the
APC responsible for PSA effects in animals, we fed
mice fluorescently labeled PSA by gavage and then re-
covered cells from mesenteric lymph nodes (MLNs). As
shown by flow cytometry, PSA specifically associates
with CD11c+dendritic cells (DCs; Figure 3A), and not
with CD4+T cells or CD19+B cells (Figure 3A, data not
shown) in the MLNs. The notion of in vivo DC recogni-
tion of PSA appears appropriate, as DCs are the only
APCs known to sample luminal contents from the intes-
tine and migrate to lymph nodes to initiate T cell reac-
tions (Rescigno et al., 2001; Mowat, 2003). Consistent
with this hypothesis, no PSA is recovered from spleens
of orally treated mice (data not shown) despite splenic
T cell expansion. In addition, confocal microscopy re-
veals that PSA is internalized by primary-cultured
bone-marrow-derived DCs (BMDCs) and is subse-
quently displayed on the cell surface (Figure 3B). Thus
it appears that DCs sample intestinal PSA by antigen
uptake and migrate only as far as the local lymph
nodes. These results agree with those of Macpherson
and Uhr (2004), who observed that commensal bacteria
are internalized by intestinal DCs that migrate to
We next assessed whether the association of PSA
with BMDCs leads to cell activation and maturation, as
only mature DCs can activate T cells. PSA-mediated
maturation of DCs is measured by upregulation of MHC
II among CD11c+cells from 22% to 43% after PSA
treatment (Figure 3C), a process required for efficient
antigen display to the receptor on CD4+T cells (Ban-
chereau and Steinman, 1998; Thery and Amigorena,
2001). Treatment also increases expression of the co-
stimulatory molecules CD80 (B7.1) and CD86 (B7.2) in
a dose-dependent manner (data not shown and Figure
3D, respectively), a result further showing that PSA in-
teracts with and induces maturation of DCs. These
findings are consistent with the fact that DCs, not B
cells, mediate intestinal antigen presentation to T cells
after conventionalization of germ-free animals (Yama-
naka et al., 2003).
Figure 2. Purified PSA Treatment Is Sufficient for Expansion of
(A) FC of α-CD4-stained splenic lymphocytes from GF mice treated
intraperitoneally with PSA or PBS reveals PSA-dependent restora-
tion of CD4+T cells. Results are representative of two experiments
with pools of four mice.
(B) FC shows that oral treatment of conventional C57Bl/6 and
BALB/c mice with purified PSA (PSA) results in an increase in CD4+
T cell proportion among splenic lymphocytes over controls (PBS).
Results are representative of four experiments with pools of 3 to
(C) FC shows that oral treatment of conventional mice with purified
PSA does not affect proportions of CD8+T cells or CD19+B cells
among the CD4−splenic lymphocyte population.
latory molecule is required during B. fragilis coloniza-
tion. Using chromatography to purify PSA extracted
from the surface of B. fragilis, we investigated whether
PSA alone is sufficient to counter the CD4+T cell de-
fects in germ-free mice. Purity was assessed by various
methods, including H1-NMR, spectroscopy, and gel
electrophoresis. The preparation was devoid of con-
tamination by protein, nucleic acid, and endotoxin
(LPS). Uncolonized germ-free mice were treated intra-
peritoneally with purified PSA and then assessed for
splenic T cell expansion. We found that purified PSA
restores CD4+T cell proportions among splenic lym-
phocytes in germ-free mice to conventional levels (Fig-
To assess whether the route of administration to mice
of various backgrounds affects PSA’s immunomodula-
tory properties, both conventionally colonized C57Bl/6
and BALB/c mice received purified PSA intragastrically.
PSA Induces T Cell Proliferation When
Presented by DCs In Vitro
To further investigate cellular and molecular events un-
derlying immune responses to PSA, we developed an
Gut Bacterial Molecule Directs Immune Development
Figure 3. PSA Is Specifically Recognized by DCs in the GI Tract and In Vitro, with Consequent Cell Activation
(A) Oral treatment of mice with Alexa-594-labeled PSA results in antigen uptake by CD11c+DCs from MLNs. FC of CD11c+gated cells (boxed
left panel, middle) analyzed for the presence of PSA (horizontal axis of right panels) shows colocalization of PSA with DCs. FC of isolated
MLNs reveals no PSA associated with CD4+T cells from the same lymph nodes (boxed left panel, bottom). Isotype control is shown in the
top left panel. Gray histograms represent unstained control, and thick black lines denote Alexa-594 signal from PSA on the horizontal axis.
No PSA is detectable in splenic tissues (data not shown).
(B) Confocal microscopy of α-CD11c-labeled (green) and 24 hr PSA-treated (red) BMDCs in culture illustrates antigen in endosomes and
surface display (arrowheads). Central image is the xy plane of a medial z section; upper and side panels are assembled z stacks.
(C) FC of BMDCs cultured for 24 hr with PSA shows activation by upregulation of MHC II (horizontal axis) among CD11c+cells (vertical axis).
(D) FC of CD11c+BMDCs cultured for 24 hr with PSA shows activation through upregulation of the costimulatory molecule CD86 (B7.2).
in vitro coculture system using primary cells to assess
the ability of purified PSA to induce T cell proliferation
and cytokine expression. As shown above, PSA treat-
ment leads to CD4+T cell expansion in mice. Incuba-
tion of PSA with CD11c+BMDCs and naive splenic
CD4+T cells leads to a dose-dependent increase in T
cell proliferation (Figure 4A). Neutralization of the posi-
tive charge by chemical modification of PSA (N-acet-
ylated [NAc] PSA) results in no incorporation of
[3H]thymidine—a marker for cell replication. This ex-
tends to DCs our observation that the zwitterionic
structure of PSA is critical for biological activity (Kalka-
Moll et al., 2002; Tzianabos and Kasper, 2002). Purified
E. coli LPS does not induce T cell proliferation in mice
(Figure 4A; LPS). PSA-mediated CD4+T cell prolifera-
tion requires both DCs and T cells; either cell type alone
results in no incorporation of radiolabel (Figure 4A). To-
gether, these results show that DCs can direct T cell
proliferation in response to purified PSA in vitro.
PSA Induces T Cell Cytokine Production
in Dendritic-Cell Cocultures
The two subtypes of effector CD4+T cells, TH1 and TH2,
are defined by expression of the cytokines interferon γ
rable to that associated with several known potent IFNγ
inducers (α-CD3, LPS, and staphylococcal enterotoxin
A [SEA]) and requires both DCs and T cells (Figure 4B).
Specificity is evidenced by the lack of TH1 cytokine pro-
duction after NAc-PSA treatment (Figure 4B).
TH1 cytokine production suppresses TH2 responses;
conversely, TH2 cytokine expression inhibits TH1 re-
sponses. Normal immune responses require a con-
trolled balance of these opposing signals. Examination
of IL-4 expression in response to PSA treatment reveals
no cytokine production by purified CD4+T cells (Figure
4C; 100 ?g/ml PSA). α-CD3 and the superantigen SEA
are potent stimulators of both classes of cytokine (Fig-
ure 4C). As TH2 cytokine production is a “default path-
way” in many systems (Kidd, 2003; Amsen et al., 2004)
and TH1 cytokine production is antagonistic to TH2 ex-
pression, the specific stimulation of IFNγ by PSA in vitro
may provide a mechanism for establishing commensal-
mediated homeostasis of the host immune system by
balancing TH1/TH2 responses.
TH1 Cytokine Production in Response to PSA
Treatment Requires Signaling through
the IL-12/Stat4 Pathway and MHC II Expression
A unique immunologic molecule, PSA is the only carbo-
hydrate studied to date that is internalized into APCs
and displayed by MHC II to T cells (Cobb et al., 2004)—
a process previously reserved for protein antigens. We
further characterized the molecular pathway for PSA-
induced TH1 cytokine production. Many of the molecu-
lar signaling events involving T helper cytokine expres-
sion are well characterized. The major pathway of IFNγ
upregulation and TH1 cell differentiation involves DC
secretion of IL-12, which binds to the IL-12 receptor on
T cells and signals to activate the TH1-specific tran-
scription factor Stat4 (Trinchieri, 2003). PSA stimulation
of DC-T cell cocultures elicits a dose-dependent in-
crease in IL-12 production (Figure 5A). DCs alone stim-
ulated with PSA secrete IL-12 (Figure 5A, DC 100 ?g/
ml PSA), although at levels lower than those found for
CD4+T cells. This is the first bacterial polysaccharide
shown to signal IL-12 secretion by APCs, as previously
shown for classical protein antigens (Macatonia et al.,
1995). NAc-PSA, lacking the essential positive-charge
motif, does not stimulate cytokine production. To deter-
mine whether IL-12 is required for TH1 cell differentia-
tion, we measured IFNγ expression after PSA treatment
in the presence of increasing concentrations of neutral-
izing antibody to IL-12 (Heufler et al., 1996). IL-12 neu-
tralization abolishes PSA-mediated IFNγ secretion by
CD4+T cells in vitro (Figure 5B). Ablation of IL-12 sig-
naling does not affect α-CD3-mediated IFNγ expres-
sion, as this signal is APC independent, acting directly
on T cells. To determine whether IL-12 signaling that
results in TH1 lineage differentiation involves the Stat4
transcription factor, we incubated DCs from wild-type
mice with CD4+T cells from stat4 knockout mice and
measured IFNγ expression in response to PSA treat-
ment. The absence of Stat4 greatly reduces TH1 cyto-
kine production (Figure 5C); thus, PSA specifically in-
duces DCs to signal T cell differentiation through Stat4
To investigate whether antigen presentation of PSA
Figure 4. PSA Induces CD4+T Cell Proliferation and TH1 Cytokine
Production In Vitro
(A) CD4+T cell proliferation by [3H]thymidine incorporation in-
creases in response to irradiated BMDCs (1 × 106) incubated with
PSA. Treatment of cocultures with NAc-PSA results in no increase
in cell proliferation. LPS treatment or DCs or T cells alone do not
support CD4+T cell proliferation. Results are representative of
three experiments. Error bars represent the ± SD from triplicate
samples of a single experiment.
(B) PSA treatment stimulates IFNγ in DC-T cell cocultures, as mea-
sured by ELISA of culture supernatants after 48 hr of treatment.
NAc-PSA treatment has no effect. Treatment with α-CD3, LPS, and
SEA, all known stimulators of TH1 cytokine expression, results in
IFNγ expression. DCs or T cells alone treated with PSA do not sup-
port cytokine expression.
(C) PSA treatment does not stimulate expression of IL-4 in DC-T
cell cocultures. Treatment with α-CD3 and SEA, known stimulators
of TH2 cytokine expression, results in IL-4 production.
(IFNγ) and interleukin 4 (IL-4), respectively (Janeway et
al., 2001). As shown above, PSA induces CD4+T cell
expansion in B. fragilis-colonized mice and in vitro. To
further characterize the effects of PSA-mediated T cell
activation, we assessed cytokine profiles using purified
cellular components. Coculture of DCs and CD4+T
cells in the presence of PSA yields dose-dependent
upexpression of the TH1 cytokine IFNγ (Figure 4B). The
level of IFNγ production associated with PSA is compa-
Gut Bacterial Molecule Directs Immune Development
Figure 5. PSA Signals through the IL-12/Stat4 Pathway to Mediate TH1 Cytokine Production, which Requires Presentation by MHC II
(A) PSA stimulates expression of IL-12, the TH1 determining signal, in DC-T cell cocultures (1 × 106of each cell type). NAc-PSA treatment
has no effect. α-CD3 and SEA serve as controls for IL-12 expression by BMDCs. Results are representative of two experiments.
(B) IL-12 is required for PSA-mediated IFNγ production. IFNγ expression is abolished in DC-T cell cocultures treated with PSA (100 ?g/ml)
in the presence of neutralizing antibody to IL-12. Neutralization of IL-12 does not inhibit α-CD3-mediated IFNγ expression, which is IL-12
(C) PSA signals through Stat4 to induce IFNγ secretion by T cells. IFNγ expression from DC-T cell cocultures treated with PSA (100 ?g/ml) is
reduced when CD4+T cells are purified from spleens of stat4 knockout rather than wild-type mice (Figure 4B).
(D) MHC II expressed on DCs is necessary for PSA-mediated IFNγ expression by CD4+T cells. IFNγ production by DC-T cell cocultures
treated with PSA (100 ?g/ml) is reduced when BMDCs are from MHC II knockout mice. NAc-PSA (100 ?g/ml) treatment of wild-type DCs
(MHC II+/+) serves as the control for PSA-specific IFNγ expression.
is required for cytokine signaling, we purified DCs from
MHC II-deficient mice and treated DC-T cell cocultures
with PSA or NAc-PSA. The level of IFNγ expression is
significantly higher in wild-type (MHC II+/+) mice treated
with PSA than in MHC II knockout mice (MHC II−/−),
which express amounts similar to a NAc-PSA control
(Figure 5D). Together, these results demonstrate that
TH1 cytokine responses to PSA require MHC II expres-
sion by APCs and involve signaling through the IL-12/
Stat4 pathway to induce T cell activation and proper
man neonatal (precolonization) cytokine profile (Kirja-
vainen and Gibson, 1999; Prescott et al., 1998; Adkins,
2000; Kidd, 2003). This “default” TH2 bias in the ab-
sence of bacterial colonization again highlights the
profound contributions of the microflora to immune de-
velopment and provides a model to test the effects of
symbiotic bacteria on the establishment of appropriate
host cytokine production.
Mice colonized with wild-type B. fragilis alone display
a level of IL-4 production similar to that in conventional
mice with a complex microflora (Figure 6A); this sim-
ilarity shows the organism’s sufficiency to correct sys-
temic immune defects. Moreover, mice colonized with
B. fragilis ?PSA produce TH2 cytokines at elevated
levels similar to those in germ-free mice (Figure 6A).
Thus, the expression of a single bacterial antigen al-
lows B. fragilis to correct the IL-4 cytokine imbalance
found in uncolonized animals.
Examination of IFNγ production by purified splenic
CD4+T cells reveals that germ-free mice, which are
TH2-skewed, are deficient in production of this proto-
typical TH1 marker when compared to conventional
mice (Figure 6B). Colonization with wild-type B. fragilis
alone is sufficient to correct the defect in IFNγ expres-
sion in germ-free mice, with levels nearly as high as
those in conventional mice (Figure 6B). Lack of PSA
PSA Is Required for Appropriate CD4+T Helper
Cytokine Production during Colonization
A proper TH1/TH2 balance is critical for human and ani-
mal health; over- or underproduction of either response
is associated with immunologic disorders. We investi-
gated the effects of PSA on TH1/TH2 cytokine re-
sponses in colonized animals, again using germ-free
mice. CD4+T cells from mouse spleens were purified
and tested by ELISA for cytokine production. Figure 6A
shows overproduction of the TH2 cytokine IL-4 in
spleens of germ-free mice compared with levels in con-
ventional mice. This result is consistent with previous
reports of the appreciably TH2-skewed profile of mice
devoid of bacterial contamination and reflects the hu-
Figure 6. Colonization of GF Mice with PSA-Producing B. fragilis Corrects TH1/TH2 Imbalances Associated with Cytokine-Mediated Pathol-
(A) IL-4 production from splenic CD4+T cells stimulated in vitro with α-CD3/α-CD28 (2 ?g/ml each) reveals that PSA is required to correct
the TH2 skew in GF mice. Compared with CNV mice, GF mice overproduce IL-4 (first and second bars). Intestinal colonization with B. fragilis
(third bar) reduces the expression of IL-4 from levels in GF mice. B. fragilis ?PSA colonization fails to correct the TH2 skew (fourth bar). Results
are representative of two experiments from pools of four mice. Error bars represent the ± SD from triplicate samples of a single experiment.
(B) IFNγ expression by splenic CD4+T cells during colonization indicates increased TH1 cytokine production in CNV compared to in GF mice
(first and second bars). PSA production by intestinal B. fragilis is required for the increase in IFNγ expression and immune homeostasis (third
bar); homeostasis is not seen in the absence of PSA (fourth bar). Error bars represent the ± SD from triplicate samples of a single experiment.
(C) Intracellular cytokine staining and FC of in vitro-stimulated (500 ng/ml PMA, 5 ?g/ml ionomycin) cultures of splenic CD4+T cells for 4 hr
in the presence of brefeldin A show that IFNγ (horizontal axis) is produced specifically by CD4+T cells (vertical axis) during bacterial coloniza-
tion. PSA production by B. fragilis is required for the specific increase in IFNγ expression to levels similar to those for CNV mice. GF and
B. fragilis ?PSA-colonized mice express low levels of TH1 cytokines. Results are representative of three experiments with pools of 3 to 5 mice.
(D) Thymic histology of germ-free mice (H&E) colonized with wild-type or B. fragilis ?PSA for over 1 year reveals follicles (arrows) within the
inner medullary compartment in the absence of PSA. None of five B. fragilis-colonized compared to 3 of 5 B. fragilis ?PSA-colonized mice
(E) FC of thymic tissues recovered from groups of differentially colonized GF mice (ten per group) reveals the anomalous presence of CD19+
B cells in B. fragilis ?PSA-colonized mice, a condition likely resulting from increased TH2 cytokine production in the absence of PSA.
production by the B. fragilis mutant during colonization
of germ-free mice results in low-level production of TH1
cytokines (Figure 6B). These results were corroborated
by intracellular cytokine staining of splenic lympho-
cytes from each group, which confirms that IFNγ pro-
duction is attributable to CD4+T cells (Figure 6C). The
production of IL-2, another TH1 cytokine, by CD4+T
cells in gnotobiotic mice also requires PSA produc-
tion (data not shown). Together, these results demon-
strate that intestinal colonization of germ-free mice by
Gut Bacterial Molecule Directs Immune Development
B. fragilis alone is sufficient to establish a proper sys-
temic TH1/TH2 balance within the host—a fundamental
aspect of the mammalian immune response.
The Timeless Struggle between Good
and Bad (Bacteria)
The origins of microbiology lie in the study of bacterial
pathogens. Louis Pasteur and Robert Koch were driven
to identify the microbial etiologies underlying infections
(Falkow, 2004). Investigation into pathogens has re-
vealed microbial molecules that mediate aspects of
the infectious process, many of which have been ex-
perimentally identified as virulence factors in animal
models of disease (Monack et al., 2004). Countless mo-
lecular and cellular inflammatory responses to these
molecules are well documented, and the results of
some interactions lead to development of immunologic
memory and resistance to reinfection. This acquired im-
munity reflects the immune system’s ability to recog-
nize and adapt to specific foreign molecules. However,
bacterial infections by pathogens are rare and opportu-
nistic. The vast majority of human encounters with bac-
teria involve harmless organisms found in our environ-
ment or as commensals in our bodies (Hooper and
Gordon, 2001). These are the species with which we
have coevolved. Unlike pathogens, certain commensals
may be beneficial to human development and physiology.
Absence of PSA Production by B. fragilis during
Colonization Is Associated with TH2-Mediated
Pathologies of the Thymus
Throughout our studies, specimens obtained at nec-
ropsy were subjected to histological examination. We
noticed a rare pathology of the thymus exclusively in
mice colonized with B. fragilis ?PSA. Thymic tissues
from germ-free mice colonized with wild-type B. fragilis
appear normal, with a darker-staining outer corona and
a uniform and homogeneous inner medullary compart-
ment. Surprisingly, at >1 year of age, the majority of
mice colonized with B. fragilis ?PSA display the out-
growth of B cell-like follicles in the thymic medulla (Fig-
ure 6D). Flow cytometry shows that these tissues con-
tain CD19+B cells (Figure 6E) not found in the normal
thymus. This rare condition appears to be similar to hu-
man thymic hyperplasia, in which B cells are found in
follicles of the medulla (Kasper et al., 2005). The latter
condition is associated with numerous autoimmune
disorders, most notably myasthenia gravis, a B cell-
mediated pathology (Malhotra et al., 1992; Infante and
Kraig, 1999). These disorders, as well as B cell out-
growths, are mediated by overproduction of TH2 cyto-
kines by CD4+T cells (Zhang et al., 1997; Janeway et
al., 2001). It is compelling to speculate that the inability
to restore proper TH1/TH2 balance in germ-free mice
through appropriate commensal colonization results in
an aberrant TH2 response, which may lead to immune-
The Gut Reaction to Symbiosis
How is the immunomodulatory signal provided by PSA
in the intestine ultimately transduced to the rest of the
immune system? DCs have received much attention as
mediators of nonclassical T cell responses (Morelli and
Thomson, 2003). Whereas other APCs provide proin-
flammatory instructions during infection, DCs uniquely
signal protective and tolerance-inducing reactions by T
cells (Colonna et al., 2004). This function may underlie
DC positioning at mucosal and epidermal surfaces con-
stantly encountering antigens that may not necessarily
induce inflammation (e.g., antigens associated with
commensal bacteria or food; inhaled antigens). DCs
have evolved various mechanisms for sampling anti-
gens from the gastrointestinal luminal compartment
(Rescigno et al., 2001; Mowat, 2003). Concordantly,
DCs are the APCs that mediate the generation and ac-
tivity of regulatory T cells (Tregs), a CD4+T cell subset
involved in suppression of immune reactions and direc-
tion of TH1/TH2 responses (Groux et al., 2004; Mills,
2004). We show here that PSA produced by an intesti-
nal symbiont is presented to T cells by DCs, with con-
sequent signaling, immune activation, and cytokine
production. Neither B. fragilis nor PSA is detectable in
spleens of colonized germ-free mice (data not shown),
although we find PSA in MLNs. DCs internalize com-
mensal bacteria in the gut and migrate only as far as
MLNs to mediate immune responses (Macpherson and
Uhr, 2004). As peripheral Tregs are known to traffic to
sites of inflammation where they control T cell reactions
(Kohm et al., 2002; Zou et al., 2004), T cell migration
from lymph nodes to spleen may mediate the cellu-
lar signal-transduction mechanisms required for PSA-
mediated immune development at extraintestinal sites.
PSA of Intestinal Bacteria as a Symbiosis Factor
The vertebrate brain and immune system are the only
organs that require environmental interactions for de-
velopment. We report here that maturation of the mam-
malian immune system requires the specific direction of
an immunomodulatory molecule provided by symbiotic
bacteria. B. fragilis, a ubiquitous constituent of the mam-
malian lower gastrointestinal microflora, elaborates a
ZPS (PSA) that directs the development of CD4+T cells;
the eventual result is the correction of immunologic de-
fects found in the absence of bacterial colonization. Im-
paired systemic CD4+T cell maturation, aberrant TH1/
TH2 lineage differentiation, and defective lymphoid-organ
development are all corrected by PSA production dur-
ing B. fragilis colonization. DCs in GALT apparently
sample B. fragilis and/or PSA from the intestine and,
after activation, migrate to lymphoid organs and signal
TH1 lineage differentiation through the IL-12/Stat4 path-
way. As TH1 cytokine production opposes the TH2 de-
fault phenotype (Kidd, 2003), this process may contrib-
ute to protection from disease by creating appropriate
cytokine balances in the immune system. The estab-
lishment of proper host immunologic function through
responses to PSA identifies this molecule as a factor
required for symbiosis.
Are Bacterial Symbionts the Victims
of Good Hygiene?
In 1989, it was postulated that reduced exposure to in-
fectious bacterial agents early in life due to improved
formed according to guidelines of the HMS Office for Research
B. fragilis strain NCTC 9343 and its isogenic PSA deletion mutant
have been described (Coyne et al., 2001). For mouse colonization,
w1 × 108cfu of bacteria grown in BHI medium were spread on
food and bedding. Mice were colonized for at least 60 days before
examination. Results are for mice up to 8 months postcolonization
(except thymic pathology, which requires aging over 1 year).
sanitation and antibiotic use explained increases in
allergy among residents of industrialized countries, a
concept termed the hygiene hypothesis (Strachan, 1989).
In a proposed modified counterregulation model, im-
proved hygiene limits exposure to immunomodulatory
molecules of beneficial commensal bacteria that pro-
vide protection from unrelated immune diseases (Wills-
Karp et al., 2001). Substantial evidence from the past
40 years indicates that incidences of atopy (allergy pre-
disposition) and asthma have increased dramatically in
Western nations but not in developing countries. Sev-
eral epidemiologic studies indicate that atopic and
nonatopic people differ in gut flora composition (Bjork-
sten et al., 1999; Kirjavainen et al., 2001; Kalliomaki et
al., 2001). Thus, hygiene-associated deviations in gut
microflora composition may be the environmental factor
underlying development of atopy and asthma in geneti-
cally predisposed individuals.
“Does the microbiota regulate immune responses
outside the gut?” (Noverr and Huffnagle, 2004). Asthma
and allergies are nonintestinal immunologic disorders
mediated by overproduction of TH2 cytokines and of
IgE—also a component of the TH2 response (Umetsu
et al., 2002). Here we show that colonization with the
ubiquitous commensal B. fragilis corrects the systemic
TH2 bias found in the absence of bacterial colonization
through specific production of TH1 cytokines. Systemic
expression of IFNγ, the characteristic TH1 marker, inhib-
its TH2 cytokine reactions and reduces IgE production
(Kidd, 2003), providing a mechanism for establishment
of immunologic cytokine balance. Germ-free mice and
atopic patients with altered gut microflora have a highly
TH2-skewed cytokine profile (Shi and Walker, 2004).
Colonization of germ-free mice with PSA-producing
B. fragilis restores normal cytokine production. Probi-
otic studies with Lactococcus and Bifidobacterium
species suggest that other prominent gut bacteria may
also produce immunomodulatory molecules (Guarner
and Malagelada, 2003; Rastall, 2004). Perhaps perva-
sive antibiotic use among children in industrialized
countries leadstoclearanceofsymbionts suchas B.frag-
ilis at an essential point in immune development and
results in the absence of molecules with PSA-like activ-
ities. Aberrant immune development without specific
direction by this class of immunomodulatory molecules
may lead to a default pathway of TH2 cytokine overpro-
duction and the onset of atopic and asthmatic disor-
ders. Future investigations must determine the function
of immunomodulatory molecules of symbiotic bacteria
in the critical balance between mammalian health and
Lymphocyte Isolation from Tissues
Lymphocytes were isolated from tissues (Tzianabos et al., 2000). In
brief, spleens, thymus, and MLNs were disrupted into single-cell
preparations and enriched for lymphocytes over a Histopaque 1083
gradient (Sigma, St. Louis). Cells were washed with PBS and used
directly or fixed with 0.5% PFA for 1 hr at 4°C.
Flow Cytometry, Fluorescence-Activated Cell Sorting,
and Intracellular Cytokine Staining
Directly fluorochrome-conjugated monoclonal antibodies were
used (BD Pharmingen, San Diego, California). For surface staining,
lymphocyte preparations were washed twice in FC buffer (PBS with
2% FBS) and resuspended in 100 ?l. 1 × 106cells were incubated
with antibodies at 2 ?g/ml for 30 min at 4°C. For intracellular cyto-
kine staining, cells were resuspended in 100 ?l of Cytofix/Cytoperm
buffer for 30 min at 4°C, washed with Perm/Wash buffer, and incu-
bated with fluorochrome-conjugated anti-cytokine antibodies for
30 min at 4°C. Cells were washed and analyzed on a model FC500
Cytometer (Beckman Coulter, Fullerton, California), and all data
were analyzed with RXP Analysis Software (Beckman Coulter).
FACS was performed on a BD FACSAria, and cell purity was al-
PSA Purification and Animal Treatment
PSA from B. fragilis NCTC 9343 was prepared (Tzianabos et al.,
1992). For some studies, PSA was treated with acetic anhydride to
neutralize the positively charged amino group (Tzianabos et al.,
1994). Mice received 50 ?g of PSA in 1.5% sodium bicarbonate/
PBS intraperitoneally or intragastrically three times a week for 2
weeks. Rag2−/−mice received 30 ?g of PSA (PBS controls) in al-
ternating intragastric and subcutaneous treatments three times a
In Vitro Cytokine Stimulation and Proliferation Assays
For analysis of splenic cytokines, lymphocytes were isolated as
above. CD4+T cells were purified with the MACS CD4 Sorting Kit
(Miltenyi Biotec, Auburn, California). Cell purity was >97% CD4+.
The remaining lymphocytes were used as APCs after γ irradiation.
3 × 105cells of each type were mixed in a 48-well plate, and
α-CD3/α-CD28 (2 ?g/ml) was added. Supernatants recovered after
stimulation for 48 and 72 hr were analyzed by ELISA.
For cocultures, CD4+T cells were purified from splenic lympho-
cytes with a CD4+T Cell Subset Kit (R&D Systems). Cell purity was
>95%. BMDCs from femurs of mice were purified. Cells were cul-
tured for 8 days in C-RPMI-10 with GM-CSF (20 ng/ml; Biosource,
Camarillo, California). Cells were >90% CD11c+. 1 × 106purified
CD4+T cells were mixed with 1 × 106purified CD11c+BMDCs and
incubated at 37°C/5% CO2.
ELISA plates were from precoated kits (Biosource). T cell prolifer-
ation assays were done with 1 × 105cells of each type (APCs irradi-
ated) after incubation for 96 hr. For the last 8 hr before harvest,
wells were pulsed with [3H]thymidine (1 ?Ci/well). Cells were washed,
harvested, and counted by liquid scintillation (Wallac; now Perkin-
Elmer, Boston). Data were expressed as mean cpm ± SD for tripli-
Mice and Bacterial Strains
Conventional SPF mice of strains C57BL/6NTac, BALB/cAnNTac,
B6.SJL-Ptprca/BoAiTac-H2-Ab1tm1GLMN13 (MHC II−/−), and B6.SJL-
Ptprca/BoAiTac (MHC II−/−control) were purchased from Taconic
Farms (Germantown, New York). C.129S2-Stat4tm1Gru(Stat4−/−) mice
were purchased from Jackson Laboratory (Bar Harbor, Maine). These
mice were housed in conventional cages at Harvard Medical
School. Germ-free Swiss-Webster mice were from Taconic Farms
and were housed in sterile isolators (Class Biologically Clean, Mad-
ison, Wisconsin). Mice were screened weekly for bacterial, viral,
and fungal contamination. All procedures with mice were per-
Histological Tissue Analysis
Paraffin-embedded mouse tissues were stained with H&E. Sections
were evaluated in blinded fashion by a single pathologist (R.T.
Gut Bacterial Molecule Directs Immune Development
bial symbiosis in the mammalian intestine: exploring an internal
ecosystem. Bioessays 20, 336–343.
Hooper, L.V., Falk, P.G., and Gordon, J.I. (2000). Analyzing the mo-
lecular foundations of commensalism in the mouse intestine. Curr.
Opin. Microbiol. 3, 79–85.
Hooper, L.V., Midtvedt, T., and Gordon, J.I. (2002). How host-micro-
bial interactions shape the nutrient environment of the mammalian
intestine. Annu. Rev. Nutr. 22, 283–307.
Infante, A.J., and Kraig, E. (1999). Myasthenia gravis and its animal
model: T cell receptor expression in an antibody mediated autoim-
mune disease. Int. Rev. Immunol. 18, 83–109.
Janeway, C.A., Travers, P., Walport, M., and Shlomchik, M. (2001).
Immunobiology (New York: Garland Publishing).
Kalka-Moll, W.M., Tzianabos, A.O., Bryant, P.W., Niemeyer, M.,
Ploegh, H.L., and Kasper, D.L. (2002). Zwitterionic polysaccharides
stimulate T cells by MHC class II-dependent interactions. J. Immu-
nol. 169, 6149–6153.
Kalliomaki, M., Kirjavainen, P., Eerola, E., Kero, P., Salminen, S., and
Isolauri, E. (2001). Distinct patterns of neonatal gut microflora in
infants in whom atopy was and was not developing. J. Allergy Clin.
Immunol. 107, 129–134.
Kapsenberg, M.L. (2003). Dendritic-cell control of pathogen-driven
T-cell polarization. Nat. Rev. Immunol. 3, 984–993.
Kasper, D.L., Braunwald, E., Fauci, A.S., Hauser, S.L., Longo, D.L.,
and Jameson, J.L. (2005). Harrison’s Principles of Internal Medicine
(New York: McGraw-Hill Publishing).
Kidd, P. (2003). Th1/Th2 balance: the hypothesis, its limitations, and
implications for health and disease. Altern. Med. Rev. 8, 223–246.
Kirjavainen, P.V., and Gibson, G.R. (1999). Healthy gut microflora
and allergy: factors influencing development of the microbiota.
Ann. Med. 31, 288–292.
Kirjavainen, P.V., Apostolou, E., Arvola, T., Salminen, S.J., Gibson,
G.R., and Isolauri, E. (2001). Characterizing the composition of in-
testinal microflora as a prospective treatment target in infant aller-
gic disease. FEMS Immunol. Med. Microbiol. 32, 1–7.
Kohm, A.P., Carpentier, P.A., Anger, H.A., and Miller, S.D. (2002).
Cutting edge: CD4+CD25+ regulatory T cells suppress antigen-
specific autoreactive immune responses and central nervous sys-
tem inflammation during active experimental autoimmune enceph-
alomyelitis. J. Immunol. 169, 4712–4716.
Kononen, E., Jousimies-Somer, H., and Asikainen, S. (1992). Rela-
tionship between oral gram-negative anaerobic bacteria in saliva
of the mother and the colonization of her edentulous infant. Oral
Microbiol. Immunol. 7, 273–276.
Krinos, C.M., Coyne, M.J., Weinacht, K.G., Tzianabos, A.O., Kasper,
D.L., and Comstock, L.E. (2001). Extensive surface diversity of a
commensal microorganism by multiple DNA inversions. Nature 414,
Macatonia, S.E., Hosken, N.A., Litton, M., Vieira, P., Hsieh, C.S.,
Culpepper, J.A., Wysocka, M., Trinchieri, G., Murphy, K.M., and
O’Garra, A. (1995). Dendritic cells produce IL-12 and direct the de-
velopment of Th1 cells from naive CD4+ T cells. J. Immunol. 154,
Macpherson, A.J., and Harris, N.L. (2004). Interactions between
commensal intestinal bacteria and the immune system. Nat. Rev.
Immunol. 4, 478–485.
Macpherson, A.J., and Uhr, T. (2004). Induction of protective IgA by
intestinal dendritic cells carrying commensal bacteria. Science 303,
Malhotra, V., Tatke, M., Khanna, S.K., and Gondal, R. (1992). Thymic
histology in myasthenia gravis. Indian J. Chest Dis. Allied Sci. 34,
Mills, K.H. (2004). Regulatory T cells: friend or foe in immunity to
infection? Nat. Rev. Immunol. 4, 841–855.
Monack, D.M., Mueller, A., and Falkow, S. (2004). Persistent bacte-
rial infections: the interface of the pathogen and the host immune
system. Nat. Rev. Microbiol. 2, 747–765.
Morelli, A.E., and Thomson, A.W. (2003). Dendritic cells: regulators
We thank Drs. A. Onderdonk, R. Cisneros, and R.T. Bronson for
expert technical and intellectual contributions; Drs. M. Wessels, C.
Nagler-Anderson, L. Comstock, J. McCoy, and H. Ton-That for criti-
cal reviews of the manuscript; and members of the Kasper and
Tzianabos labs for assistance throughout the work. This study was
supported by a WHO Fellowship to C.H.L. (WPRO 0003/03) and a
Postdoctoral Fellowship from the Helen Hay Whitney Foundation
to S.K.M. This work was supported by funding from the NIH (NIH/
NIAID R01AI039576) to D.L.K.
Received: January 12, 2005
Revised: February 28, 2005
Accepted: May 4, 2005
Published: July 14, 2005
Adkins, B. (2000). Development of neonatal Th1/Th2 function. Int.
Rev. Immunol. 19, 157–171.
Amsen, D., Blander, J.M., Lee, G.R., Tanigaki, K., Honjo, T., and
Flavell, R.A. (2004). Instruction of distinct CD4 T helper cell fates
by different notch ligands on antigen-presenting cells. Cell 117,
Banchereau, J., and Steinman, R.M. (1998). Dendritic cells and the
control of immunity. Nature 392, 245–252.
Bjorksten, B., Naaber, P., Sepp, E., and Mikelsaar, M. (1999). The
intestinal microflora in allergic Estonian and Swedish 2-year-old
children. Clin. Exp. Allergy 29, 342–346.
Bowman, L.M., and Holt, P.G. (2001). Selective enhancement of
systemic Th1 immunity in immunologically immature rats with an
orally administered bacterial extract. Infect. Immun. 69, 3719–3727.
Brubaker, J.O., Li, Q., Tzianabos, A.O., Kasper, D.L., and Finberg,
R.W. (1999). Mitogenic activity of purified capsular polysaccharide
A from Bacteroides fragilis: differential stimulatory effect on mouse
and rat lymphocytes in vitro. J. Immunol. 162, 2235–2242.
Cash, H.L., and Hooper, L.V. (2005). Commensal bacteria shape in-
testinal immune system development. ASM News 71, 77–83.
Cobb, B.A., Wang, Q., Tzianabos, A.O., and Kasper, D.L. (2004).
Polysaccharide processing and presentation by the MHCII path-
way. Cell 117, 677–687.
Colonna, M., Trinchieri, G., and Liu, Y.J. (2004). Plasmacytoid den-
dritic cells in immunity. Nat. Immunol. 5, 1219–1226.
Coyne, M.J., Tzianabos, A.O., Mallory, B.C., Carey, V.J., Kasper,
D.L., and Comstock, L.E. (2001). Polysaccharide biosynthesis locus
required for virulence of Bacteroides fragilis. Infect. Immun. 69,
Dobber, R., Hertogh-Huijbregts, A., Rozing, J., Bottomly, K., and
Negelkerken, L. (1992). The involvement of intestinal microflora in
the expansion of CD4+ T cells with a naïve phenotype in the periph-
ery. Dev. Immunol. 2, 141–150.
Falkow, S. (2004). Molecular Koch’s postulates applied to bacterial
pathogenicity—a personal recollection 15 years later. Nat. Rev.
Microbiol. 2, 67–72.
Groux, H., Fournier, N., and Cottrez, F. (2004). Role of dendritic cells
in the generation of regulatory T cells. Semin. Immunol. 16, 99–106.
Guarner, F., and Malagelada, J.R. (2003). Gut flora in health and
disease. Lancet 361, 512–519.
Heufler, C., Koch, F., Stanzl, U., Topar, G., Wysocka, M., Trinchieri,
G., Enk, A., Steinman, R.M., Romani, N., and Schuler, G. (1996).
Interleukin-12 is produced by dendritic cells and mediates T helper
1 development as well as interferon-gamma production by T helper
1 cells. Eur. J. Immunol. 26, 659–668.
Hooper, L.V. (2004). Bacterial contributions to mammalian gut de-
velopment. Trends Microbiol. 12, 129–134.
Hooper, L.V., and Gordon, J.I. (2001). Commensal host-bacterial re-
lationships in the gut. Science 292, 1115–1118.
Hooper, L.V., Bry, L., Falk, P.G., and Gordon, J.I. (1998). Host-micro-
of alloimmunity and opportunities for tolerance induction. Immunol.
Rev. 196, 125–146.
Mowat, A.M. (2003). Anatomical basis of tolerance and immunity to
intestinal antigens. Nat. Rev. Immunol. 3, 331–341.
Neurath, M.F., Finotto, S., and Glimcher, L.H. (2002). The role of
Th1/Th2 polarization in mucosal immunity. Nat. Med. 8, 567–573.
Noverr, M.C., and Huffnagle, G.B. (2004). Does the microbiota regu-
late immune responses outside the gut? Trends Microbiol. 12,
Pereira, P., Forni, L., Larsson, E., Cooper, M., Heusser, C., and Cou-
tinho, A. (1986). Autonomous activation of B and T cells in antigen-
free mice. Eur. J. Immunol. 16, 685–688.
Prescott, S.L., Macaubas, C., Holt, B.J., Smallacombe, T.B., Loh,
R., Sly, P.D., and Holt, P.G. (1998). Transplacental priming of the
human immune system to environmental allergens: universal skew-
ing of initial T cell responses toward the Th2 cytokine profile. J.
Immunol. 160, 4730–4737.
Rakoff-Nahoum, S., Paglino, J., Eslami-Varzaneh, F., Edberg, S.,
and Medzhitov, R. (2004). Recognition of commensal microflora by
toll-like receptors is required for intestinal homeostasis. Cell 118,
Rastall, R.A. (2004). Bacteria in the gut: friends and foes and how
to alter the balance. J. Nutr. 134, 2022S–2026S.
Rescigno, M., Rotta, G., Valzasina, B., and Ricciardi-Castagnoli, P.
(2001). Dendritic cells shuttle microbes across gut epithelial mono-
layers. Immunobiology 204, 572–581.
Rook, G.A., and Brunet, L.R. (2002). Give us this day our daily
germs. Biologist (London) 49, 145–149.
Sheikh, A., and Strachan, D.P. (2004). The hygiene theory: fact or
fiction? Curr. Opin. Otolaryngol. Head Neck Surg. 12, 232–236.
Shi, H.N., and Walker, A. (2004). Bacterial colonization and the de-
velopment of intestinal defences. Can. J. Gastroenterol. 18, 493–
Strachan, D.P. (1989). Hay fever, hygiene, and household size. BMJ
Thery, C., and Amigorena, S. (2001). The cell biology of antigen
presentation in dendritic cells. Curr. Opin. Immunol. 13, 45–51.
Trinchieri, G. (2003). Interleukin-12 and the regulation of innate re-
sistance and adaptive immunity. Nat. Rev. Immunol. 3, 133–146.
Tzianabos, A.O., and Kasper, D.L. (2002). Role of T cells in abscess
formation. Curr. Opin. Microbiol. 5, 92–96.
Tzianabos, A.O., Pantosti, A., Baumann, H., Brisson, J.R., Jennings,
H.J., and Kasper, D.L. (1992). The capsular polysaccharide of Bac-
teroides fragilis comprises two ionically linked polysaccharides. J.
Biol. Chem. 267, 18230–18235.
Tzianabos, A.O., Onderdonk, A.B., Rosner, B., Cisneros, R.L., and
Kasper, D.L. (1993). Structural features of polysaccharides that in-
duce intra-abdominal abscesses. Science 262, 416–419.
Tzianabos, A.O., Onderdonk, A.B., Zaleznik, D.F., Smith, R.S., and
Kasper, D.L. (1994). Structural characteristics of polysaccharides
that induce protection against intra-abdominal abscess formation.
Infect. Immun. 62, 4881–4886.
Tzianabos, A.O., Russell, P.R., Onderdonk, A.B., Gibson, F.C., 3rd,
Cywes, C., Chan, M., Finberg, R.W., and Kasper, D.L. (1999). IL-2
mediates protection against abscess formation in an experimental
model of sepsis. J. Immunol. 163, 893–897.
Tzianabos, A.O., Finberg, R.W., Wang, Y., Chan, M., Onderdonk,
A.B., Jennings, H.J., and Kasper, D.L. (2000). T cells activated by
zwitterionic molecules prevent abscesses induced by pathogenic
bacteria. J. Biol. Chem. 275, 6733–6740.
Umetsu, D.T., McIntire, J.J., Akbari, O., Macaubas, C., and De-
Kruyff, R.H. (2002). Asthma: an epidemic of dysregulated immunity.
Nat. Immunol. 3, 715–720.
Von Hertzen, L.C., and Haahtela, T. (2004). Asthma and atopy—the
price of affluence? Allergy 59, 124–137.
Wills-Karp, M., Santeliz, J., and Karp, C.L. (2001). The germless
theory of allergic disease: revisiting the hygiene hypothesis. Nat.
Rev. Immunol. 1, 69–75.
Yamanaka, T., Helgeland, L., Farstad, I.N., Fukushima, H., Midtvedt,
T., and Brandtzaeg, P. (2003). Microbial colonization drives lympho-
cyte accumulation and differentiation in the follicle-associated epi-
thelium of Peyer’s patches. J. Immunol. 170, 816–822.
Zhang, G.X., Navikas, V., and Link, H. (1997). Cytokines and the
pathogenesis of myasthenia gravis. Muscle Nerve 20, 543–551.
Zou, L., Barnett, B., Safah, H., Larussa, V.F., Evdemon-Hogan, M.,
Mottram, P., Wei, S., David, O., Curiel, T.J., and Zou, W. (2004).
Bone marrow is a reservoir for CD4+CD25+ regulatory T cells that
traffic through CXCL12/CXCR4 signals. Cancer Res. 64, 8451–