478 | JUNE 2004 | VOLUME 4 www.nature.com/reviews/immunol
evolution on both sides. The bacteria benefit
from the stable habitat, which is rich in energy
sources from the food we ingest; and we sal-
vage heat energy from compounds such as cel-
lulose, which would otherwise be indigestible
because we lack the necessary enzymes. Also,
some bacterial compounds, including short-
chain fatty acids and phylloquinone (vitamin
K1), are used in host anabolic pathways.
Moreover, the commensal microflora compete
with incoming foreign microorganisms for
space and resources, thereby making it more
difficult for true pathogens to become estab-
lished. However, this seemingly ideal balance is
sometimes disturbed: the development of
INFLAMMATORY BOWEL DISEASE
dence of which is 1 in 500 in Western popula-
tions) and possibly some autoimmune diseases
are associated with immune responses to envi-
ronmental microorganisms or mild pathogens.
Here,we discuss the pervasive way in
which the environmental flora shapes both
the mucosal and systemic immune systems.
It is clear that immune composition and
lymphoid structures differ when this flora is
present compared with when it is absent;
however, we are only beginning to under-
stand the mechanisms and functions of the
adaptive changes that occur when environ-
mental bacteria colonize the host. One of the
consequences is to physically limit the live
bacteria to the intestinal lumen. Yet, the
functional consequences of the systemic
that the combined number of genes in these
bacteria exceeds the total number of eukaryotic
genes by a factor of at least 100.
Usually, we ‘peacefully’ coexist with our
commensal microflora, in a good example of
MUTUALISM,which has been established through
Although we might shudder at the thought
of billions of bacteria living in our lower
intestine, we are colonized by these
passengers shortly after birth. However, the
relationship is mostly of mutual benefit, and
they shape our immune system throughout
life. Here, we describe our developing
understanding of the far-reaching effects
that the commensal flora have on mucosal
and systemic immunity and their relevance
to the effects of hygiene on human disease.
Most studies of the immune system aim to
understand the way in which it responds to
infectious pathogens. By analysing immune
mechanisms in animal models of infectious
disease, we learn how the immune system
responds to biologically important challenges.
Ye t clinically apparent microbial infections
are the exceptions in our harmonious coexis-
tence with vast numbers of non-pathogenic
microorganisms; these microorganisms enter
our body from the environment shortly after
birth and colonize the mucous membranes
and skin epithelia. In the lower intestine
alone, the density of commensal bacteria in
the lumen reaches 10
organisms per gram of
intestinal contents, with approximately 1,000
(BOX 1).So when the physiol-
ogist J.B.S. Haldane once commented that
even the Archbishop of Canterbury is 65%
, he omitted to mention that the head
of the Church of England also consists of
more bacterial cells than eukaryotic cells and
Interactions between commensal
intestinal bacteria and the immune
Andrew J. Macpherson and Nicola L. Harris
Box 1 | Intestinal microorganisms
Bacteria are the main type of microorganism present in the mammalian intestine, although
other types are also found, including protozoa and fungi. The stomach and small intestine
have relatively low bacterial densities (10
organisms per gram or ml of luminal contents
in mice, consisting mainly of acid-tolerant lactobacilli and streptococci). The distal portion
of the small intestine, the ileum, is a transition zone with higher bacterial densities (10
gram) and species diversity, but the most dense colonization is in the colon (10
gram), which hosts more than 400 bacterial species. In the lower intestine, anaerobes
predominate, particularly the Bacteroides,bifidobacteria, fusobacteria and peptostreptococci
(each group present at approximately 10
per gram); by contrast, aerobes and facultative
aerobes, including enterobacteria and lactobacilli, are present at only moderate densities
There are two main difficulties in understanding and measuring these complex flora. First,
a comparison of two techniques used to assess faecal bacterial numbers — counting colonies
of culturable bacteria and estimating numbers using smears — shows that less than 50% of
intestinal bacteria can be cultured. This is because of the precise oxygen requirements of
some species and their fastidious (and largely unknown) nutrient requirements. Second,
although most measurements have been made using faecal bacteria, the intestine is not a
homogeneous environment — groups of bacteria can also exist on the surface of the mucus
layer or deep within it.
Fortunately, there are ways of overcoming the difficulties in culturing intestinal bacteria.
The 1.5-kb gene encoding 16S ribosomal RNA is present in multiple copies in bacterial
chromosomes, and it is highly polymorphic. Therefore, the nucleotide sequence of this
gene (obtained after amplification by PCR) can be used to determine the species of each
,and the gene can serve as a target (in species-specific in situ hybridization)
for studying the spatial arrangement of each bacterial group.
122. Cvitkovitch, D. G., Li, Y. H. & Ellen, R. P. Quorum sensing
and biofilm formation in Streptococcal infections. J. Clin.
Invest. 112, 1626–1632 (2003).
123. Chill, J. H., Quadt, S. R., Levy, R., Schreiber, G. &
Anglister, J. The human type I interferon receptor:
NMR structure reveals the molecular basis of ligand
binding. Structure (Camb) 11, 791–802 (2003).
124. Zhao, X., Singh, M., Malashkevich, V. N. & Kim, P. S.
Structural characterization of the human respiratory
syncytial virus fusion protein core. Proc. Natl Acad.
Sci. USA 97, 14172–14177 (2000).
125. Sauer, F. G. et al. Structural basis of chaperone
function and pilus biogenesis. Science 285,
126. Braig, K., Adams, P. D. & Brunger, A. T. Conformational
variability in the refined structure of the chaperonin GroEL at
2.8 A resolution. Nature Struct. Biol. 2, 1083–1094 (1995).
127. Schwarz, U. S. Phase behavior of amphiphilic systems.
Acta Physica Polonica B 29, 1815–1826 (1998).
Competing interests statement
The authors declare that they have no competing financial interests.
The following terms in this article are linked online to:
Entrez Gene: http://www.ncbi.nlm.nih.gov/Entrez/
CD14 | CD36 | IFN-α | TLR2 | TLR3 | TLR4 | TLR5 | TLR7
See online article: S1 (table) | S2 (table) | S3 (table) | S4 (table)
Access to this interactive links box is free online.
NATURE REVIEWS | IMMUNOLOGY VOLUME 4 | JUNE 2004 | 479
show that although commensal bacteria are
killed within hours by macrophages, they can
survive for several days inside dendritic cells
(DCs). However, DCs that have been loaded
with commensal bacteria in the Peyer’s
patches do not penetrate farther than the
mesenteric lymph nodes, so DCs that are
primed by live bacteria are usually restricted to
the mucosal immune system. It is important
to note that this in vivo priming effect is lost if
the bacteria are killed by heat treatment, so it is
not mediated by lipopolysaccharide and/or
other bacteria-derived immunostimulatory
molecules. (Although by internalizing live bac-
teria, DCs might concentrate these bacterial
compounds.) So, IgA production, and proba-
bly intestinal T-cell responses, can be selec-
tively induced by DCs loaded with commensal
bacteria, and this increased local secretion of
IgA limits the penetration of commensal bac-
.It has long been established that both B
and T cells are activated in the Peyer’s patches,
after which they recirculate through the
intestinal lymphatics, enter the bloodstream at
the level of the thoracic duct and home back
to the intestinal lamina propria. Because DCs
containing live commensal bacteria are con-
fined to the mucosal immune system, the
induction of T- and B-cell responses is focused
at the mucosa where it is required.
Although there is a strong locally induced
immune response, because commensal bacte-
ria do not usually penetrate beyond the
mesenteric lymph nodes, we believe that the
systemic immune system is largely ignorant of
these organisms. An example of this is that
responses that prime the production of IgG
specific for Enterobacter cloacae (the main aer-
obe in our SPF mouse colony) are not seen in
unmanipulated mice colonized with these
.This occurs as a result of IGNORANCE,
rather than tolerance, because intravenous
injection of a small dose of E. cloacae causes a
highly reproducible priming response
priming of systemic immunity to commensal
bacteria is largely unnecessary, because the
innate immune system can destroy the few
organisms that do penetrate the intestinal
barrier — mice only become unable to coexist
with their intestinal flora when there are seri-
ous deficiencies in the phagocytic biocidal
mechanisms that generate reactive oxygen
and nitric oxide radicals
So, on the whole, we believe that mice
are systemically ignorant of particulate
(live) commensal intestinal bacteria; how-
ever, soluble bacterial degradation products
that reach the systemic circulation are
probably responsible for the differences in
the organization of secondary-lymphoid
and the different concentrations
immune alterations are also important,
because there is evidence that the hygiene
status of humans influences predisposition
to allergy and/or autoimmunity. We believe
that there are important differences between
the observed improvements in human
hygiene and the current animal models that
are used to study the immunological conse-
quences of these improvements, and we con-
sider that these differences are crucial for
understanding and modelling the effects of
hygiene on systemic immunopathology.
Mucosal immune adaptation
There is no question that the host is highly
adapted to the presence of commensal
intestinal bacteria. The evidence for this comes
from comparing germ-free mice
(FIG. 1), which
have no commensal microflora, with spe-
cific-pathogen-free (SPF) animals of the
same strain, which contain a simple flora
(BOX 2).The mucosal immune system is
undeveloped in germ-free animals: they have
PEYER’S PATCHES that contain few
germinal centres, as well as greatly reduced
numbers of IgA-producing plasma cells and
LAMINA PROPRIA CD4
immunological abnormalities in germ-free
animals are not confined to the mucosal
immune system: the spleen and lymph nodes
are relatively structureless, with poorly formed
B- and T-cell zones
(FIG. 2) and abnormal high
endothelial venule morphology
also have hypogammaglobulinaemic serum,
mainly because of reduced levels of IgG
Furthermore, the gene-expression profile of
the intestinal epithelial-cell layer is altered in
the absence of commensal bacteria
these abnormalities can be reversed within
several weeks of colonizing germ-free animals
with commensal bacteria, which can be
achieved by placing an SPF mouse in a cage
that contains germ-free animals
experimental colonization of germ-free ani-
mals might seem artificial, but similar colo-
nization occurs in every neonate within days
of its birth
Systemic immune ignorance
The physical barrier that separates the large
numbers of commensal intestinal bacteria
from the underlying tissues is only a simple
(single cell) epithelial layer; although it is
reinforced by a layer of mucus, the secretion of
IgA and the production of antibacterial mole-
cules (such as
DEFENSINS) by the epithelium
(FIG. 3).Considering this, it is not surprising
that in both humans and experimental ani-
mals some bacteria can penetrate this barrier
to reach deeper tissues
a b c
Figure 1 | Keeping germ-free mice in an isolator. It has been possible to keep experimental
animals in entirely germ-free conditions for the past 50 years or more, by initially delivering the pups
by sterile Caesarean section and hand rearing them aseptically. a | After this, it is less labour intensive
to keep the mice free from colonization by environmental organisms by breeding them in an isolator,
which is ventilated with sterile filtered air under positive pressure. Gnotobiotic (germ-free) animal
husbandry uses sterile food, water and bedding. It is important to note that germ-free animals have
no bacteria in the intestine or on other body surfaces, whereas specific-pathogen-free (SPF) mice
are only devoid of known mouse pathogens and do contain intestinal bacteria. b | A germ-free isolator
can be loaded with sterile supplies using a transport ring, which is small enough to fit in an animal-unit
autoclave. c | The ring is connected to a port on the isolator using a flexible plastic sleeve that
contains a glove. (The inside of the sleeve is sterilized by spraying it with peracetic acid immediately
before it is connected to the isolator and the transport ring.) The internal door on the isolator and the
external seal on the transport ring are then opened from the outside using the gloves, and sterile
materials are brought in. Samples are taken routinely from the husbandry materials and the mice,
to ensure that the colony remains germ free. Mice can either be experimentally manipulated in the
isolator, or they can be transferred into a sterile lamina-flow hood, by connecting the isolator to
a port on the side of the hood. Germ-free mice can easily be recolonized with bacteria by placing
a single mouse that has normal flora into the cage.
480 | JUNE 2004 | VOLUME 4 www.nature.com/reviews/immunol
referred to as antigen free (because they lack
full-length proteins), the animals still con-
sume milk (during lactation in the neonatal
period) and groom themselves. We argue
that the important difference between the
background immune stimulation of germ-
free animals fed a sterile (autoclaved) natural-
ingredient diet and those fed an elemental
diet probably results from the microbial
material contaminating the autoclaved food
(rather than the presence of full-length pro-
teins), because soluble proteins introduced
into the intestine are generally thought to be
tolerogenic rather than immunogenic
relative impact on immune reactivity of
reducing the levels of contaminating micro-
bial molecules or full-length proteins could
easily be tested by selective experimental
Compared with germ-free mice that are
fed autoclaved food, germ-free mice fed an
elemental diet have highly reduced levels
of serum IgG and IgA, reduced numbers of
splenic IgG-producing cells and fewer circu-
.Despite this, serum
IgM concentrations and the diversity of the
REPERTOIRE are preserved
mice were challenged with a model antigen
together with adjuvant, they mounted a
strong specific IgG response, but the poly-
clonal response that leads to increased levels
of non-antigen-specific IgG was consider-
ably diminished. This reduction in the poly-
clonal response was also seen in germ-free
animals that were fed sterile food (in com-
parison with SPF mice), but it was more
marked in germ-free mice that were fed the
.In addition, the usual
increase in the frequency of somatically
hypermutated immunoglobulins (of the IgG
isotype) that occurs during ageing also
depends on exposure to environmental anti-
gens (from commensal gut microorganisms),
because this has been shown to occur to a
lesser extent in germ-free mice
than in SPF
mice. So, bacteria-driven alterations that
result from ‘bathing’ the immune system in
immunostimulatory bacterial molecules
cause a baseline level of immune activation
this increases the degree of polyclonal stimu-
lation to protein antigens administered with
adjuvants and might also improve the
immune response against invasive pathogens.
It is well known that patients with agamma-
globulinaemia are protected against infec-
tions with encapsulated bacteria by treatment
with pooled gammaglobulin preparations
from healthy donors. Furthermore, experi-
ments on rats have shown that crossreactive
antibodies can sufficiently bind the repeti-
tive capsular polysaccharides of bacteria to
are fed an elemental diet (containing hydro-
lysed amino acids and purified lipids and
carbohydrates) rather than autoclaved food,
which contains (dead) bacterial material.
Although these elemental diets are often
and diversity of serum immunoglobulins in
germ-free and SPF animals
dence is provided by studies
that the hypogammaglobulinaemia of germ-
free animals is more pronounced when they
with intestinal bacteria
c Intestinal IgA
(inset: Peyer's-patch IgA)
b Intestinal CD4
(inset: intestinal CD8)
a Splenic CD4
(inset: splenic CD8)
Figure 2 | The presence of intestinal bacteria has a large impact on lymphoid structures of
both the intestine and systemic tissues. Histological sections of the spleen and intestine are shown
for a germ-free wild-type C57BL/6 mouse and for a C57BL/6 mouse colonized with intestinal bacteria.
In the absence of intestinal bacteria, the spleens have relatively few germinal centres and poorly formed
T-cell (pink) and B-cell zones (a). The intestines of germ-free mice have low numbers of lamina propria
cells (brown) (b), greatly reduced numbers of IgA-producing cells (brown) (c) and hypoplastic
Peyer’s patches. The histological abnormalities of germ-free mice reverse within weeks of colonization
with intestinal bacteria, which can be achieved, for example, by placing a normal colonized mouse in
the same cage. These images are reproduced with permission from
REF. 3 (2001) Elsevier.
Box 2 | The ‘Schaedler flora’ in experimental animals
As shown in
FIG. 1, experimental animals can be bred and maintained in a sterile environment.
These germ-free or gnotobiotic animals have no microorganisms in the gut or on other body
surfaces. Starting with germ-free mice, Russell Schaedler and colleagues
at the Rockefeller
Institute for Medical Research, New York, USA, carried out experiments in the 1960s in which
they reintroduced simple mixtures of defined intestinal bacteria. They showed that inoculation
with coliform bacteria alone led to high stable levels of these bacteria, which are only a minor
component of the gut flora in a normal mouse. To attempt to introduce a simple, balanced flora
that would colonize the gut without compromising the health of the mice, a cocktail of eight
species eventually became popular: Escherichia coli var. mutabilis,Streptococcus faecalis,
Lactobacillus acidophilus, Lactobacillus salivarius,group N Streptococcus, Bacteroides distasonis,
a Clostridium species and a species of extremely oxygen-sensitive (EOS) spiral-shaped (fusiform)
bacteria. In 1987, the National Cancer Institute of the United States revised this ‘Schaedler flora’
to include a Flexistipes fusiform bacterium and three further EOS fusiform species
Specific-pathogen-free (SPF) mice are defined on the basis of a negative screen for specific known
pathogens. They are usually derived from a commercial breeding colony, which might or might not
have been rederived (by Caesarean section or embryo transfer) in a germ-free environment. The
most rigorous derivation of SPF mice is to be derived germ free and then colonized with modified
It is clear that the flora of experimental animals generated in this way is far simpler than
in humans (approximately 1,000 species) or in conventional experimental animals. Although
this defined gut population should remove much of the background variability observed in
experiments, it is often assumed that the flora remains stable while subsequent generations
of animals are bred (in SPF facilities). However, the composition of the flora (and particularly
the anaerobes) is not often checked, and it is relatively easy for new species to be introduced,
either from human handlers or from animals sourced from other SPF suppliers. Scientists are only
aware of the mouse pathogens that are routinely tested for, and the immunological community is
generally ignorant of the environmental flora of the experimental animals that they use.
NATURE REVIEWS | IMMUNOLOGY VOLUME 4 | JUNE 2004 | 481
to assessing whether experiments with ‘ultra-
clean’ mice (that is, germ-free mice or SPF
mice with a restricted flora) can help us to
understand the patterns of human disease.
Hygiene is even more complex to investi-
gate in the human population, because poor
hygiene usually occurs together with several
confounding factors: poor nutrition, lifestyles
with lower technology, and less access to
medical care (for registering the incidence of
disease). The importance of these variables
needs to be considered in epidemiological
Autoimmunity and allergy, which result
from inappropriate and overactive immune
responses, are disadvantages arising from
our ability to combat infectious disease. The
‘hygiene hypothesis’ states that as we
improve hygiene, there are fewer infectious
challenges, and the subsequent response of
the immune system leads to allergy
induce the fixation of complement and the
opsonization of the bacteria
. One disadvan-
tage is that the peripheral self-tolerance that
is usually generated by the presentation of
proteins on immature DCs
might be com-
promised by bystander activation of the DCs
in the presence of bacterial degradation
The presence of commensal intestinal
bacteria, therefore, has clear structural and
functional consequences for the systemic
immune system. Because even high doses of
commensal bacteria do not penetrate to sys-
temic secondary-lymphoid structures
immune changes in germ-free animals are
accentuated when they are fed a purified diet
of hydrolysed amino acids, we propose that
the baseline setting of the immune system is
a response to contamination of the immune
environment with soluble breakdown
products from microorganisms.
Interactions between the systemic and
mucosal immune systems have traditionally
been studied by examining oral tolerance.
Such studies involve the administration of a
T-CELL-DEPENDENT ANTIGEN,either by ingestion or
mucosal application, then the measurement of
suppression of the response to later systemic
immunization. Although oral tolerance can be
induced in germ-free animals
,the kinetics of
the systemic response to immunization are
,probably because of the immaturity
of systemic secondary-lymphoid organization
under germ-free conditions. This makes the
interpretation of oral-tolerance results in
germ-free animals too complicated to allow
direct comparison with SPF animals.
Improving human hygiene
We now turn from examining the effects of
experimentally manipulating the density and
diversity of environmental antigens on mice
DC uptake of
B cell T cell
Induction of IgA-producing
Figure 3 | Immune defences against commensal intestinal bacteria. Commensal bacteria are present at a high density in the intestinal lumen (up to 10
per gram of luminal contents). Most commensal bacteria reside outside the layer of mucus that covers the intestinal epithelial cells. Some bacteria can be killed by
antibacterial molecules, such as defensins, which are produced by the epithelial cells. Bacteria that penetrate the enterocyte epithelial layer are rapidly killed by the
macrophages in the lamina propria. Commensal bacteria can also penetrate the specialized follicle-associated epithelium, containing M cells, which lies over the
Peyer’s patches. These bacteria are also rapidly killed by macrophages, but small numbers can survive for several days in dendritic cells (DCs). This enables the
interaction of DCs with T and B cells in the Peyer’s patches and/or the migration of DCs to the draining mesenteric lymph nodes. (DCs that contain live bacteria induce
IgA-producing plasma cells more effectively than heat-killed bacteria.) Although DCs loaded with commensal bacteria can traffic to the mesenteric lymph nodes, the
lymph nodes function as a barrier, and the loaded DCs cannot penetrate farther to reach the systemic secondary-lymphoid tissues. The result is that the induction of
immune responses by live bacteria is confined to the mucosa itself. Following activation, B- and T-cell blasts can leave the mesenteric lymph nodes through the
efferent lymph, enter the bloodstream at the thoracic duct and home back to the intestinal mucosa.
482 | JUNE 2004 | VOLUME 4 www.nature.com/reviews/immunol
during thymic selection can result in
.Patients with mutations in
the forkhead box P3 (FOXP3)gene (required
for the development of CD4
tory T cells
) also show systemic autoimmu-
nity, eczema and increased serum IgE levels
so a global lack of regulatory T cells does
affect immune-system homeostasis.Yet, in a
lymphocyte-replete human or experimental
animal, the ability of infections to alter regu-
latory T-cell populations is probably weak,
considering the poor antigen-specific prolif-
erative expansion of regulatory T cells
Therefore, we think that it is improbable that
functional alterations in regulatory T cells
explain the relationship between improved
hygiene and increasing incidence of allergy
and autoimmunity in the population.
Hygiene in experimental animals
Because we can manipulate the population of
environmentally derived microflora by con-
trolling the husbandry of rodents, we should
also be able to investigate the underlying
basis of the hygiene effect. It is crucial to
appreciate that whereas germ-free mice have
no microorganisms colonizing their body
surfaces, neither germ-free nor SPF animals
have been exposed to any known pathogen.
Hygienic humans have a diverse intestinal and
body-surface flora, but they take measures to
reduce their exposure to pathogens. The envi-
ronmental conditions that we manipulate for
experimental animals (when comparing
mice with a restricted commensal flora and
germ-free mice, neither of which are
exposed to pathogens) are therefore quite dif-
ferent from improvements in human hygiene
(in which early exposure to pathogens is
reduced). Humans have a much more
diverse (approximately 1,000 species) and
dense (approximately 10
bacteria per gram
of luminal contents) commensal bacterial
flora than SPF mice
(BOX 2).In fact, in con-
trast to humans, most rodent models of
autoimmunity actually show reduced inci-
dence in conditions of improved hygiene
Similarly, spontaneous and induced models
of inflammatory bowel disease are abrogated
in SPF or germ-free conditions
rare exceptions to this: notably, autoimmune
gastritis following neonatal thymectomy, the
incidence of which is unchanged in germ-
; and diabetes in non-obese
diabetic mice, in which the incidence is
There are potentially three ways in
which microorganisms interact with the
immune system: an infection in which live
microorganisms can proliferate systemi-
cally; the penetration through the mucosa
immune deviation, thereby skewing immune
responses away from the neonatal T
it has been difficult to show this phenomenon
directly, and it does not explain three other epi-
demiological observations: first, there is a simi-
lar negative association between atopy and
infection with helminths (which are known to
with an immunodeficiency resulting from
genetic lesions that affect T
pathways do not have an increased inci-
dence of allergic disease
; and third, the
increase in the incidence of allergy in recent
decades has been accompanied by similar
increases in autoimmune diabetes
— conditions that are usu-
ally considered to be T
Because the demographics of autoimmunity
are similar to those of allergy, they need to
be considered together when modelling the
effects of hygiene.
Another possible explanation for the
effects of altered hygiene on allergy and/or
autoimmunity is that there are differences in
the induction of regulatory T cells
Certain infectious agents, such as Bordetella
pertussis,have been reported to induce the
development of T regulatory 1 (TR1) cells
as a method of evading host responses
similar induction of regulatory T cells has
also been described following infection with
hepatitis C virus
, Helicobacter hepatis
.In addition, a strong bias in
There is reasonable clinical epidemiological
support for this theory. Children from fami-
lies of lower socio-economic status or with
more siblings have decreased incidence of
atopy, presumably because of exposure to
more infectious agents
.There is also an
inverse correlation between previous infec-
tion with mycobacteria or viruses, including
hepatitis A virus (HAV), and the subsequent
development of asthma
, and other
show that children brought up on
farms are protected from the development of
asthma. However, it is usually unclear which
of these infections can actually induce a pro-
tective effect and which are surrogate markers
of poor hygiene in a complex environment.
For those exposed to HAV, protection
against asthma development is more pro-
nounced in individuals carrying a six-amino-
acid insertion at position 157 of the TIM1
(T-cell immunoglobulin domain and mucin
domain 1) gene, which encodes the cell-surface
receptor through which HAV infects human
.This receptor is also expressed by acti-
T helper 2 (T
unclear whether the polymorphism in TIM1
alters HAV infectivity or alters the immune
response against the virus. However, the results
indicate that, in this case, HAV itself is respon-
sible for protection against allergy, rather than
just being a surrogate for another protective
One suggested explanation for the observed
protection against allergy is that interaction
with ‘unhygienic’ microorganisms causes
Box 3 | The ‘hygiene hypothesis’
In 1989, Strachan
coined the term ‘hygiene hypothesis’. This hypothesis states that a leading
cause of the increased incidence of allergy in today’s population is the decrease in exposure
to common infections during early life, which occurs as a result of smaller family size and
improved hygienic conditions. Considerable epidemiological and experimental evidence
supports the hypothesis, including studies examining airborne viruses, mycobacteria, orofaecal
microorganisms and helminths
.Two popular theories that offer explanations for the
hygiene hypothesis are immune deviation and counter regulation.
It is well known that the T helper 1 (T
1)-cell cytokine interferon-γ can suppress the
differentiation of T
and the production of IgE
,which are associated with atopy.
Because the neonatal immune system has been described as showing a T
proponents of immune deviation argue that exposure to microorganisms that induce T
responses is required to prevent the development of atopic (T
The counter-regulation model states that the production of immunoregulatory factors
after exposure to microorganisms limits the development of unrelated immune-mediated
.Indeed, experimental infection of mice with Mycobacterium vaccae can elicit a
population of CD4
regulatory T cells that attenuate ovalbumin-induced airway
inflammation through the production of interleukin-10 (IL-10) and transforming growth
.IL-10-producing cells that are induced by infection with enteric helminths have
also been shown to protect mice from immunopathology associated with the subsequent
ingestion of a food allergen
We argue here that neither explanation is likely to account entirely for the long-term
consequences of altered hygiene conditions and that alterations in T- and B-cell repertoires
after pathogenic infections probably contribute more to the differences in incidence of
NATURE REVIEWS | IMMUNOLOGY VOLUME 4 | JUNE 2004 | 483
lymph nodes were smaller and nephritis
was reduced compared with animals on the
natural-ingredient diet, thereby demon-
strating the possible impact of microbial
products on the resulting immunopathology.
Modelling human hygiene effects
Instead of the current method of comparing
germ-free and colonized animals, in which
large differences in lymphoid structure are
apparent, we think that deliberate, defined
experimental infections will be required to
model the way in which improvements in
human hygiene during early life lead to a
greater incidence of allergy and autoimmunity.
Indeed, it is possible to show short-term
alterations in susceptibility to induced aller-
gic responses following defined experimental
infections. For example, previous infection
with Mycoplasma or Mycobacterium bovis
Bacillus Calmette-Guerin (BCG) can attenu-
ate airway inflammation that is induced
experimentally using ovalbumin, at least
when mice are challenged within 1–2 weeks
of clearing the infection
monary infection with influenza virus can
also provide protection against bronchial
hyperresponsiveness in mice
.This effect is
dependent on interferon-γ production by
lung-resident memory CD8
T cells, which
can be re-activated by nonspecific stimuli
encountered during allergen challenge
These studies provide evidence for the ability
of pathogenic infections to alter unrelated
immune responses through immune deviation
mechanisms, but they do not model the long-
term effects of early exposure to pathogenic
agents and how this prevents the development
of allergy and/or autoimmunity.
Infections result in an immune response
that is partly specific (T-cell clones specific for
peptides derived from pathogens and high-
affinity neutralizing antibodies specific for
and partly nonspecific
(class-switch recombination of natural anti-
body specificities, resulting mainly from
bystander help provided by specific T-cell
.We propose that infection causes
an alteration of the T-cell repertoire that
could also account for the hygiene effect,
without necessarily involving T
immune deviation or T-cell-mediated regula-
tion. The presence of a large quantity of anti-
gen can eliminate or inactivate T-cell clones;
EXHAUSTION has well-recognized
effects that abrogate antiviral
immune responses. Exhaustion of
virus-specific T cells can also clearly occur in
wild-type immunocompetent mice that are
infected with lymphocytic choriomeningitis
.Because T cells that are specific for
of small numbers of commensal environ-
mental organisms that cannot proliferate
efficiently; and systemic penetration of solu-
ble microbial molecules that activate
on immune cells. The biggest differ-
ence between humans living under hygienic
conditions and those under ‘primitive’ (non-
hygienic) conditions is the decreased expo-
sure of the former to systemic infections,
whereas SPF mice are not exposed to
pathogens and even their intestinal bacteria
do not usually reach the systemic immune
system. We think that reduced exposure to
clinically obvious infections is therefore
probably crucial for the ‘hygiene effect’ in
humans, whereas the decreased nonspecific
immune activation in germ-free animals
(because of the absence of soluble microbial
products) is the key to the beneficial effect in
many autoimmune models
Evidence of the potential importance of
microbial products (and the possibility of
confusion with specific infections) is demon-
strated by the induction of autoimmune
haemolytic anaemia in mice transgenic for
pathogenic autoantibodies in this strain are
produced by B1 cells derived from the pleuro-
peritoneal lineage. In germ-free conditions,
there are few B1 cells, and no autoimmune dis-
ease develops. In SPF conditions, the animals
have B1 cells but still show no haemolytic
anaemia. The immunopathology is only seen
when the strain is maintained in conven-
tional conditions; however, defined infec-
tions of SPF animals have not yet revealed
the causative pathogen. In this case, it is clear
that a pathogen can trigger autoimmunity;
yet, because injection of lipopolysaccharide
has the same effect, the consequences of
pathogenic infection probably result from
increased exposure to microbial products.
A second example is provided by study-
ing the role of environmental antigens in the
spontaneous development of autoimmunity
in MRL/lpr mice, which have a mutation in
the Fas (CD95) gene and therefore have
defective lymphocyte apoptosis
conventional conditions, these animals
spontaneously develop an autoimmune
syndrome that is characterized by lympho-
proliferation (mainly of a CD4
subset), high serum immunoglobulin levels
(including multiple autoantibodies), vasculitis
and nephritis. When the MRL/lpr strain was
bred in a germ-free environment but fed an
autoclaved natural-ingredient diet, there was
no difference in lymphoproliferation or
autoimmune pathology compared with ani-
mals bred in conventional conditions.
However, when the germ-free MRL/lpr ani-
mals were fed a sterile elemental diet, the
A family of proteins exhibiting bactericidal properties.
They are secreted by immune cells (particularly neu-
trophils), intestinal Paneth cells and epithelial cells.
Non-responsiveness of the immune system resulting
from the deletion of specific thymocytes (central toler-
ance) and the deletion or functional inactivation of
specific T cells in the periphery (peripheral tolerance)
in the presence of large quantities of antigen.
Non-responsiveness of the immune system in the pres-
ence of a given antigen, despite the existence of specific
T and B cells capable of mounting a functional response.
INFLAMMATORY BOWEL DISEASE
Immune-mediated inflammation of the bowel. There are
two main forms: Crohn’s disease, which is a granuloma-
tous segmental inflammation affecting any part of the
intestine, and ulcerative colitis, which is a mucosal
inflammation involving the rectum and extending for a
variable distance along the colon. In developed countries,
the incidence of inflammatory bowel disease is approxi-
mately 1 in 50,000. It usually starts in early adult life and
continues afterwards with a relapsing, remitting course.
The layer of the intestine between the epithelial cells and
the most superficial smooth-muscle layer.
Resemblance between epitopes contained within
microbial and host proteins, leading to crossreactivity
of T cells in the host.
The relationship between two different species that live
in close proximity and benefit from one another.
Collections of lymphoid tissue located in the mucosa of
the small intestine, with an outer epithelial layer contain-
ing specialized epithelial cells, called M cells.
The spectrum of B or T cells. Defined according to the
specificities of the B-cell- or T-cell-receptors that are pre-
sent immediately before onset of a clinically important
To generate an antibody response to a T-cell-dependent
protein antigen requires recognition of the antigen (in
the context of MHC molecules) by helper T cells and
cooperation between those antigen-specific T cells and
B cells that recognize the same antigen.
Cell-associated pattern-recognition receptors that
recognize molecules unique to microorganisms,
resulting in immune-cell activation and production
of pro-inflammatory molecules.
484 | JUNE 2004 | VOLUME 4 www.nature.com/reviews/immunol
16. Shiloh, M. U. et al. Phenotype of mice and macrophages
deficient in both phagocyte oxidase and inducible nitric
oxide synthase. Immunity 10, 29–38 (1999).
17. Crabbe, P. A., Nash, D. R., Bazin, H., Eyssen, H. &
Heremans, J. F. Immunohistochemical observations on
lymphoid tissues from conventional and germ-free mice.
Lab. Invest. 22, 448–457 (1970).
18. Benveniste, J., Lespinats, G. & Salomon, J. Serum and
secretory IgA in axenic and holoxenic mice. J. Immunol.
107, 1656–1662 (1971).
19. Hooijkaas, H., Benner, R., Pleasants, J. R. &
Wostmann, B. S. Isotypes and specificities of
immunoglobulins produced by germ-free mice
fed chemically defined ultrafiltered ‘antigen-free’
diet. Eur. J. Immunol. 14, 1127–1130 (1984).
20. Wostmann, B. S., Pleasants, J. R. & Bealmear, P.
Dietary stimulation of immune mechanisms. Fed.
Proc. 30, 1779–1784 (1971).
21. Wostmann, B. S. & Pleasants, J. R. The germ-free
animal fed chemically defined diet: a unique tool.
Proc. Soc. Exp. Biol. Med. 198, 539–546 (1991).
22. Garside, P. & Mowat, A. M. Oral tolerance. Semin.
Immunol. 13, 177–185 (2001).
23. Bos, N. A. & Ploplis, V. A. Humoral immune response
to 2,4-dinitrophenyl–keyhole limpet hemocyanin in
antigen-free, germ-free and conventional BALB/c
mice. Eur. J. Immunol. 24, 59–65 (1994).
24. Williams, G. T., Jolly, C. J., Kohler, J. & Neuberger, M. S.
The contribution of somatic hypermutation to the diversity
of serum immunoglobulin: dramatic increase with age.
Immunity 13, 409–417 (2000).
25. Bakker, R., Lasonder, E. & Bos, N. A. Measurement of
affinity in serum samples of antigen-free, germ-free and
conventional mice after hyperimmunization with 2,4-
dinitrophenyl–keyhole limpet hemocyanin, using surface
plasmon resonance. Eur. J. Immunol. 25, 1680–1686
26. Moxon, E. R. & Anderson, P. Meningitis caused by
Haemophilus influenzae in infant rats: protective immunity
and antibody priming by gastrointestinal colonization with
Escherichia coli. J. Infect. Dis. 140, 471–478 (1979).
27. Probst, H. C., Lagnel, J., Kollias, G. & van den Broek, M.
Inducible transgenic mice reveal resting dendritic cells as
potent inducers of CD8
T cell tolerance. Immunity 18,
28. Sudo, N. et al. The requirement of intestinal bacterial
flora for the development of an IgE production system
fully susceptible to oral tolerance induction. J. Immunol.
159, 1739–1745 (1997).
29. Strachan, D. P. Hay fever, hygiene, and household size.
BMJ 299, 1259–1260 (1989).
30. Matricardi, P. M. et al. Cross sectional retrospective
study of prevalence of atopy among Italian military
students with antibodies against hepatitis A virus.
BMJ 314, 999–1003 (1997).
31. Matricardi, P. M. et al.Exposure to foodborne and
orofecal microbes versus airborne viruses in relation
to atopy and allergic asthma: epidemiological study.
BMJ 320, 412–417 (2000).
32. Shirakawa, T., Enomoto, T., Shimazu, S. & Hopkin, J. M.
The inverse association between tuberculin responses
and atopic disorder. Science 275, 77–79 (1997).
33. Kilpelainen, M., Terho, E. O., Helenius, H. &
Koskenvuo, M. Farm environment in childhood
prevents the development of allergies. Clin. Exp.
Allergy 30, 201–108 (2000).
34. Riedler, J., Eder, W., Oberfeld, G. & Schreuer, M. Austrian
children living on a farm have less hay fever, asthma and
allergic sensitization. Clin. Exp. Allergy 30, 194–200 (2000).
35. Riedler, J. et al.Exposure to farming in early life and
development of asthma and allergy: a cross-sectional
survey. Lancet 358, 1129–1133 (2001).
36. Von Ehrenstein, O. S. et al. Reduced risk of hay fever
and asthma among children of farmers. Clin. Exp.
Allergy 30, 187–193 (2000).
37. McIntire, J. J. et al.Hepatitis A virus link to atopic
disease. Nature 425, 576 (2003).
38. McIntire, J. J. et al.Identification of Tapr (an airway
hyperreactivity regulatory locus) and the linked Tim
gene family. Nature Immunol. 2, 1109–1116 (2001).
39. Busse, W. W. & Lemanske, R. F. Asthma. N. Engl. J. Med.
344, 350–362 (2001).
40. Scrivener, S. et al.Independent effects of intestinal
parasite infection and domestic allergen exposure on
risk of wheeze in Ethiopia: a nested case-control study.
Lancet 358, 1493–1499 (2002).
41. Lammas, D. A., Casanova, J. L. & Kumararatne, D. S.
Clinical consequences of defects in the IL-12-dependent
interferon-γ (IFN-γ) pathway. Clin. Exp. Immunol. 121,
probably reflects poor secondary-lymphoid
tissue organization and low baseline activa-
tion, resulting from reduced penetration of
soluble microbial products throughout the
The epidemiological evidence for the
hygiene hypothesis is persuasive, but to dissect
the mechanisms using in vivo models requires
studying long-term effects of experimental
pathogenic infection on the later induction of
allergic and/or autoimmune disease. Pathogens
can certainly leave their imprint on the com-
position and structure of the immune sys-
tem. The converse of the hygiene hypothesis,
MIMICRY — which is the triggering of immuno-
pathology by a previous infection — also
provides a clinically persuasive explanation
for the development of rheumatic fever and
Guillain–Barré syndrome, but it is not easy to
model satisfactorily. Understanding the long-
term pathogenic and environmental influences
on immune-system function is yet another
frontier for immunologists to cross.
Andrew J. Macpherson and Nicola L. Harris are at
the Institute of Experimental Immunology,
Universitätsspital, Schmelzbergstrasse 12, CH8091
Correspondence to A.J.M.
1. Clark, R. W. The Life and Work of J. B. S. Haldane
(Oxford Univ. Press, 1984).
2. Sartor, R. B. Pathogenesis and immune mechanisms
of chronic inflammatory bowel diseases. Am. J.
Gastroenterol. 92, 5S–11S (1997).
3. Macpherson, A. J., Hunziker, L., McCoy, K. & Lamarre, A.
IgA responses in the intestinal mucosa against pathogenic
and non-pathogenic microorganisms. Microbes Infect.
3, 1021–1035 (2001).
4. Macpherson, A. J., Martinic, M. M. & Harris, N. The
functions of mucosal T cells in containing the indigenous
flora of the intestine. Cell. Mol. Life Sci. 59, 2088–2096
5. Bauer, H., Horowitz, R. E., Levenson, S. M. & Popper, H.
The response of lymphatic tissue to the microbial flora.
Studies on germfree mice. Am. J. Pathol. 42, 471–479
6. Manolios, N., Geczy, C. L. & Schrieber, L. High
endothelial venule morphology and function are
inducible in germ-free mice: a possible role for
interferon-γ. Cell. Immunol. 117, 136–151 (1988).
7. Benveniste, J., Lespinats, G., Adam, C. & Salomon, J. C.
Immunoglobulins in intact, immunized, and contaminated
axenic mice: study of serum IgA. J. Immunol. 107,
8. Hooper, L. V. et al.Molecular analysis of commensal
host–microbial relationships in the intestine. Science
291, 881–884 (2001).
9. Mackie, R., Sghir, A. & Gaskins, H. R. Developmental
microbial ecology of the neonatal gastrointestinal tract.
Am. J. Clin. Nutr. 69, 1035S–1045S (1999).
10. Ganz, T. Defensins: antimicrobial peptides of innate
immunity. Nature Rev. Immunol. 3, 710–720 (2003).
11. Berg, R. D. The indigenous gastrointestinal microflora.
Trends Microbiol. 4, 430–435 (1996).
12. Wells, C. L., Maddaus, M. A. & Simmons, R. L.
Proposed mechanisms for the translocation of
intestinal bacteria. Rev. Infect. Dis. 10, 958–979 (1988).
13. O’Boyle, C. J. et al. Microbiology of bacterial
translocation in humans. Gut 42, 29–35 (1998).
14. Macpherson, A. J. & Uhr, T. Induction of protective IgA
by intestinal dendritic cells carrying commensal bacteria.
Science 303, 1662–1665 (2004).
15. Macpherson, A. J. et al.A primitive T cell-independent
mechanism of intestinal mucosal IgA responses to
commensal bacteria. Science 288, 2222–2226 (2000).
particular microbial antigens are selectively
deleted or inactivated during exhaustion, the
consequence is unresponsiveness to particular
peptide antigens, rather than generalized
immunosuppression. Moreover, because
there is a hierarchy of the dominance of T-cell
epitopes in complex pathogens
toire attrition would probably result in the
dominance of alternative epitopes, rather
than increasing susceptibility to the organism;
however, the immune response to more simple
crossreactive proteins (such as allergens or
autoantigens) would be lost. Therefore, specific
infections can shape the T-cell repertoire and
possibly limit the ability of B cells to generate
high-affinity antibodies that are capable of
causing immunopathology. An alternative
model is that an increased density of intestinal
,resulting in higher pene-
tration of soluble microbial molecules, might
modify the T-cell repertoire by facilitating tol-
erance through activation and clonal deletion
of specific T cells that would be repeatedly
exposed to non-replicating antigen. Moreover,
it has been shown that functional exhaustion
can explain the tolerization against experimen-
tal allergic encephalomyelitis or renal tubular
interstitial nephritis that can be induced fol-
lowing administration of autoantigen in
incomplete Freund’s adjuvant
specific exhaustion or tolerance of cross-
reactive T cells that is required to model this
repertoire-dependent version of the hygiene
hypothesis has not yet been studied in detail,
and this will require sophisticated sequential
studies of the CD4
T-cell repertoire in vari-
In this perspective, we discuss the fact that
live commensal bacteria are confined to the
mucosal immune compartment because
they are readily killed by macrophages and
can only survive in dendritic cells; therefore,
the systemic immune system is essentially
ignorant of these organisms. In humans,
epidemiological associations have been
made between improved conditions of
hygiene and increased incidence of allergy
and autoimmunity. We argue that although
most animal models of autoimmunity are
ameliorated under SPF or germ-free condi-
tions, it is not reasonable to compare this
with improved conditions of human
hygiene, because the experimental condi-
tions only affect organisms that are not usu-
ally seen in large numbers by the systemic
immune system. The reduced systemic
immune responsiveness observed when the
number of colonizing environmental micro-
organisms is low (in ultraclean animals)
NATURE REVIEWS | IMMUNOLOGY VOLUME 4 | JUNE 2004 | 485
77. Moskophidis, D., Lechner, F., Pircher, H. & Zinkernagel, R.
Virus persistence in acutely infected immunocompetent
mice by exhaustion of antiviral cytotoxic effector T cells.
Science 362, 759–761 (1993).
78. Probst, H. C., Dumrese, T. & van den Broek, M.
Competition for APC by CTLs of different specificities is
not functionally important during induction of antiviral
responses. J. Immunol. 168, 5387–5391 (2002).
79. Kalliomäki, M. et al.Distinct patterns of neonatal gut
microflora in infants in whom atopy was and was not
developing. J. Allergy Clin. Immunol. 107, 129–134 (2001).
80. Heeger, P. S. et al. Revisiting tolerance induced by
autoantigen in incomplete Freund’s adjuvant. J. Immunol.
164, 5771–5781 (2000).
81. Weisburg, W. G., Barns, S. M., Pelletier, D. A. & Lane, D. J.
16S ribosomal DNA amplification for phylogenetic study.
J. Bacteriol. 173, 697–703 (1991).
82. Schaedler, R. W., Dubos, R. & Costello, R. Association
of germfree mice with bacteria isolated from normal
mice. J. Exp. Med. 122, 77–83 (1965).
83. Orcutt, R. P., Gianni, F. J. & Judge, R. J. Development of
an ‘Altered Schaedler flora’ for NCI gnotobiotic rodents.
Microecol. Ther. 17, 59 (1987).
84. Wills-Karp, M., Santeliz, J. & Karp, C. L. The germless
theory of allergic disease: revisiting the hygiene
hypothesis. Nature Rev. Immunol. 1, 69–75 (2001).
85. Matricardi, P. M. & Ronchetti, R. Are infections protecting
from atopy? Curr. Opin. Allergy Clin. Immunol. 1, 413–419
86. Maggi, E. et al. Reciprocal regulator effects of IFN-γ
and IL-4 on the in vitro development of human T
2 clones. J. Immunol. 148, 2142–2147 (1992).
87. Snapper, C. M. & Paul, W. E. Interferon-γ and B cell
stimulatory factor-1 reciprocally regulate Ig isotype
production. Science 236, 944–947 (1987).
88. Prescott, S. L. et al. Transplacental priming of the human
immune system to environmental allergens: universal
skewing of initial T cell responses toward the T
cytokine profile. J. Immunol. 160, 4730–4737 (1998).
89. Holt, P. G., Rowe, J., Loh, R. & Sly, P. D.
Developmental factors associated with risk for atopic
disease: implications for vaccine strategies in early
childhood. Vaccine 21, 3432–3445 (2003).
90. Zuany-Amorim, Z. et al. Suppression of airway
eosinophilia by killed Mycobacterium vaccae-induced
allergen-specific regulatory T cells. Nature Med.
8, 625–629 (2002).
91. Bashir, M. E. H., Anderson, P., Fuss, I. J., Shi, H. N. &
Nagler-Anderson, C. An enteric helminth infection
protects against an allergic response to dietary antigen.
J. Immunol. 169, 3284–3292 (2002).
Competing interests statement
The authors declare that they have no competing financial interests.
The following terms in this article are linked online to:
CD4 | CD8 | CD25 | CD45 | Fas | FOXP3 | IL-10 | interferon-γ |
TIM1 | transforming growth factor-β1
Access to this interactive links box is free online.
42. Stene, l. C. & Nafstad, P. Relation between occurrence
of type 1 diabetes and asthma. Lancet 357, 607–608
43. Kero, J., Gissler, M., Hemminki, E. & Isolauri, E. Could
1 and T
2 diseases coexist? Evaluation of asthma
incidence in children with coeliac disease, type 1
diabetes, or rheumatoid arthritis: a register study.
J. Allergy Clin. Immunol. 108, 781–783 (2001).
44. McGuirk, P., McCann, C. & Mills, K. H. G. Pathogen-
specific T regulatory 1 cells induced in the respiratory
tract by a bacterial molecule that stimulates interleukin 10
production by dendritic cells: a novel strategy for evasion
of protective T helper type 1 responses by Bordetella
pertussis. J. Exp. Med. 195, 221–231 (2002).
45. MacDonald, A. J. et al. CD4 T helper type 1 and
regulatory T cells induced against the same epitopes
on the core protein in hepatitis C virus-infected
persons. J. Infect. Dis. 185, 720–727 (2002).
46. Kullberg, M. C. et al. Bacteria-triggered CD4
T regulatory cells suppress Helicobacter hepaticus-
induced colitis. J. Exp. Med. 196, 505–515 (2002).
47. Doetze, A. et al.Antigen-specific cellular
hyporesponsiveness in a chronic human helminth
infection is mediated by T
3/TR1-type cytokines IL-10
and transforming growth factor-β but not by a T
2 shift. Int. Immunol. 12, 623–630 (2000).
48. Hori, S., Takahashi, T. & Sakaguchi, S. Control of
autoimmunity by naturally arising regulatory CD4
T cells. Adv. Immunol. 81, 331–371 (2003).
49. Shih, F. F., Mandik-Nayak, L., Wipke, B. T. & Allen, P. M.
Massive thymic deletion results in systemic autoimmunity
through elimination of CD4
T regulatory cells.
J. Exp. Med. 199, 323–335 (2004).
50. Hori, S., Nomura, M. & Sakaguchi, S. Control of
regulatory T cell development by the transcription
factor Foxp3. Science 299, 1057–1061 (2003).
51. Gambineri, E., Torgerson, T. R. & Ochs, H. D. Immune
dysregulation, polyendocrinopathy, enteropathy, and
X-linked inheritance (IPEX), a syndrome of systemic
autoimmunity caused by mutations of FOXP3, a
critical regulator of T-cell homeostasis. Curr. Opin.
Rheumatol. 15, 430–435 (2003).
52. Jordan, M. S. et al. Thymic selection of CD4
regulatory T cells induced by an agonist self-peptide.
Nature Immunol. 2, 301–306 (2001).
53. Dewhirst, F. E. et al. Phylogeny of the defined murine
microbiota: altered Schaedler flora. Appl. Environ.
Microbiol. 65, 3287–3292 (1999).
54. East, J., Prosser, P. R., Holborow, E. J. & Jaquet, H.
Autoimmune reactions and virus-like particles in germ-
free NZB mice. Lancet 1, 755–757 (1967).
55. Unni, K. K., Holley, K. E., McDuffie, F. C. & Titus, J. L.
Comparative study of NZB mice under germ-free and
conventional conditions. J. Rheumatol. 2, 36–44 (1975).
56. Goverman, J. et al.Transgenic mice that express a
myelin basic protein-specific T cell receptor develop
spontaneous autoimmunity. Cell 72, 551–560 (1993).
57. Murakami, M. et al.Effects of breeding environments
on generation and activation of autoreactive B1 cells
in anti-red blood cell autoantibody transgenic mice.
J. Exp. Med. 185, 791–794 (1997).
58. Penhale, W. J. & Young, P. R. The influence of the normal
microbial flora on the susceptibility of rats to experimental
autoimmune thyroiditis. Clin. Exp. Immunol. 72, 288–292.
59. Maldonado, M. A. et al. The role of environmental
antigens in the spontaneous development of
autoimmunity in MRL-lpr mice. J. Immunol. 162,
60. Kuhn, R., Lohler, J., Rennick, D., Rajewsky, K. & Muller, W.
Interleukin-10-deficient mice develop chronic enterocolitis.
Cell 75, 263–274 (1993).
61. Sadlack, B. et al. Ulcerative colitis-like disease in mice
with a disrupted interleukin-2 gene. Cell 75, 253–261
62. Mombaerts, P. et al. Spontaneous development of
inflammatory bowel disease in T cell receptor mutant
mice. Cell 75, 274–282 (1993).
63. Simpson, S. J., de Jong, Y. P., Comiskey, M.
& Terhorst, C. T cells in mouse models of gut
inflammation. Chem. Immunol. 71, 118–138 (1998).
64. Murakami, K. et al. Germ-free condition and the
susceptibility of BALB/c mice to post-thymectomy
autoimmune gastritis. Autoimmunity 12, 69–70 (1992).
65. Suzuki, T. et al.in Immune Deficient Animals in
Biomedical Research (eds Rygaard, J., Graem, N. &
Sprang-Thomsen, M.) 112–116 (Karger, Basel, 1987).
66. Leiter, E. in The Role of Microorganisms in Non-
infectious Disease (eds de Vries, R. R. P., Cohen, I. R.
& van Rood, J. J.) 39–55 (Springer, Berlin, 1990).
67. Hunziker, L. et al. Hypergammaglobulinemia and
autoantibody induction mechanisms in viral infections.
Nature Immunol. 4, 343–349 (2003).
68. Chu, H. W., Honour, J. M., Rawlinson, C. A., Harbeck, R. J.
& Martin, R. J. Effects of respiratory Mycoplasma
pnemoniae infection on allergen-induced bronchial
hyperresponsiveness and lung inflammation in mice.
Infect. Immun. 71, 1520–1526 (2003).
69. Erb, K. J., Holloway, J. W., Sobeck, A., Moll, H. &
Le Gros, G. L. Infection of mice with Mycobacterium
bovis-Bacillus Calmette-Guerin (BCG) suppresses
allergen-induced airway eosinophilia. J. Exp. Med.
187, 561–569 (1998).
70. Marsland, B. J. et al.Bystander suppression of allergic
airway inflammation by lung resident memory CD8
T cells. Proc. Natl Acad. Sci. USA 101, 6116–6121
71. Bachmann, M. F. et al. The role of antibody concentration
and avidity in antiviral protection. Science 276, 2024–2027
72. Freer, G. et al.Vesicular stomatitis virus Indiana
glycoprotein as a T cell-dependent and -independent
antigen. J. Virol. 68, 3650–3655 (1994).
73. Sangster, M. Y. et al.An early CD4
immunoglobulin A response to influenza infection in the
absence of key cognate T–B interactions. J. Exp. Med.
198, 1011–1021 (2003).
74. Oxenius, A., Zinkernagel, R. M. & Hengartner, H.
Comparison of activation versus induction of
unresponsiveness of virus-specific CD4
T cells upon acute versus persistent viral infection.
Immunity 9, 449–457 (1998).
75. Hunziker, L., Klenerman, P., Zinkernagel, R. M. & Ehl, S.
Exhaustion of cytotoxic T cells during adoptive
immunotherapy of virus carrier mice can be prevented
by B cells or CD4
T cells. Eur. J. Immunol. 32, 374–382
76. Klein, L. et al.Visualizing the course of antigen-specific
CD8 and CD4 T cell responses to a growing tumor.
Eur. J. Immunol. 33, 806–814 (2003).