The major HISTOCOMPATIBILITY complex (MHC) (BOX 1)
was originally identified as a region that encodes
a family of molecules that are responsible for the
rejection of transplanted organs.However,these so-
called ‘transplantation antigens’ were soon found to
perform a much broader recognition function
in the context of the ADAPTIVE IMMUNE SYSTEM. MHC
class I molecules bind peptides derived from proteo-
lysis of intracellular proteins. They present these
peptides on the cell surface,where they are monitored
by CYTOTOXIC T LYMPHOCYTES(CTLs) and are identified as
self or non-self.This discrimination is at the core of
our ability to rid ourselves of bacterial and viral
infections, and to curb the development of some
There is increasing evidence that MHC class I
molecules have additional functions in the developing
and adult CNS. These findings have revived a
long-standing debate about the ability of neurons
to express MHC class I proteins, and have broad
implications for both normal and abnormal brain
development and function.In this review,we discuss
evidence from various systems regarding MHC class I
expression, signalling and function in neurons, as
well as the potential roles of this class of molecules in
Neuronal expression of MHC class I
Members ofthe MHC class I family are expressed on
the surface of most nucleated cells.For many years,
however,CNS immune privilege (the relative slowness
or failure of the immune response in the CNS) was
considered strong evidence that normal neurons do not
express MHC class I genes1,2.In addition,numerous
studies failed to detect MHC class I protein in normal
untreated brain slices3,cultured neurons4or neuron-
like immortalized cell lines2,5–8(reviewed in REF. 9).
By contrast,neurons do express high levels of MHC
class I protein after various treatments,including axo-
tomy10,11,ventral-horn root avulsion12,viral or parasitic
infection13–15,exposure to cytokines4,8,16–18and pharma-
cological manipulation ofelectrical activity4,19,20.
Recent results,however,indicate that even normal,
uninjured neurons express both classical and non-
classical MHC class I genes in vivo(FIG.1).MHC class I
mRNA and/or protein has been detected in diverse
neuronal populations,including motor nuclei,substantia
nigra pars compacta12,21,dorsal root ganglia neurons22,
dopaminergic nigral cells23, developing and adult
hippocampal pyramidal cells17,20,sensory neurons of
the vomeronasal organ (VNO)24,25,brainstem12,23and
spinal12,19motor neurons,and cortical pyramidal cells20,21
(FIG.1).Some ofthese studies also confirm that MHC
IMMUNE SIGNALLING IN NEURAL
PLASTICITY AND DISEASE
Lisa M.Boulanger* and Carla J.Shatz‡
Research has long supported the view that the brain is immunologically privileged, in part because
normal, uninfected neurons were not thought to express major histocompatibility complex (MHC)
class I molecules. Recently, however , it has been shown that neurons normally express MHC class I
molecules in vivo. Furthermore, accumulating evidence indicates that neuronal MHC class I does
not simply function in an immune capacity, but is also crucial for normal brain development,
neuronal differentiation, synaptic plasticity and even behaviour. These findings point to new
directions for research, and imply that immune proteins could be involved in the origin and
expression of neurological disorders.
The ability oftissues to be
successfully grafted.Also refers
to the genetic systems that
determine tissue rejection
through immune responses of
ADAPTIVE IMMUNE SYSTEM
The system that coordinates the
T cells to an antigen.The process
is mediated by clonal selection
NATURE REVIEWS |NEUROSCIENCE
VOLUME 5 |JULY 2004 |5 2 1
Pacific Hall 1212A,
9500 Gilman Drive,
Correspondence to L.M.B.
CYTOTOXIC T LYMPHOCYTE
(CTL).An effector cell ofthe
adaptive immune system that
binds MHC class I and induces
cytolysis ofcells bearing non-self
peptides derived from cytosolic
pathogens.Most CTL express
the co-receptor CD8.
A subset of lymphocytes that
are defined by their
development in the thymus
and by the expression of
receptors associated with CD3
proteins.T cells mediate
cellular adaptive immunity,
whereas B lymphocytes
(B cells) mediate humoral
A method that allows the
separation ofcells that express a
specific protein by tagging them
with a fluorescent antibody
against the molecule ofinterest.
A laser beam excites the
fluorescent tag,and the emission
oflight triggers the cell sorting.
A site on an antigen that is
recognized by an antibody or
5 2 2 |JULY 2004 |VOLUME 5
R E V I E W S
are dissociated and grown in vitro1.So,both the entry and
activity ofimmune effectors seem to be tightly regulated
in the CNS.Our current understanding ofCNS immune
responses indicates that CTL-mediated lysis,which is
commonly used to detect MHC class I protein function,
would yield false negative results in neurons.However,it
does not explain why so many studies have failed to detect
MHC class I expression in neurons using more direct
There are several possible explanations.First,many
ofthe studies relied on immunostaining ofaldehyde-
fixed brain,using antibodies that were developed for
FLUORESCENCE-ACTIVATED CELL SORTINGoflive immune cells.
The EPITOPES that are detected by these antibodies are
often exquisitely sensitive to fixation1.Indeed,even in
mice that overexpress functionally measurable MHC
class I protein under a neuron-specific promoter,
MHC protein was not detected by immunocytochemical
methods in fixed tissue slices,reflecting the challenges of
the technique and the lack of suitable antibodies1.In
addition,each antibody might detect only a fraction of
the hundreds ofMHC class I gene and allelic variants
(BOX 1).So-called non-classical MHC class I proteins were
most strongly detected in one study ofthe rat brain12,
but most ofthe antibodies that were used in previous
CNS studies do not crossreact with this population.
Furthermore, outside the nervous system, antibody
binding to MHC class I molecules is known to be influ-
enced by the peptide being presented by the MHC class I
gene33,a feature that might be different for neurons.
class I expression in neurons can be further increased by
treatments such as axotomy,exposure to cytokines19and
changes in electrical activity20,21.
It is becoming clear that brain immune privilege is
not absolute — most viral and bacterial infections of
neurons are cleared through an active immune response
(albeit with different kinetics than in other tissues26–28),
and foreign brain tissue can induce MHC class I-
mediated transplantation immunity29–31.Also,several
lines ofevidence indicate that the partial CNS immune
privilege that is observed under some conditions cannot
be attributed simply to a lack ofMHC molecules on the
neuronal cell surface.Even neurons that overexpress
MHC class I genes are resistant to CTL-mediated lysis
following viral infection,despite increased CTL infiltra-
tion of the brain and enhanced viral clearance from
MHC-overexpressing infected neurons1.
So,immune privilege probably reflects the regulation
ofthe immune response at other levels.For example,the
entry ofimmune effector cells (including T CELLS) into
the brain is regulated by the blood–brain barrier (BBB).
Although this barrier is permeable to activated T cells
(FIG.2),and overall permeability increases with systemic
infection,the BBB excludes most T cells from the brain,
under normal conditions32.Additionally,factors released
from neurons or glia might interfere with CTL–neuronal
interactions,neuronal lysis or T-cell viability,contributing
to an immunosuppressive microenvironment in the
brain.For example,MHC class I-overexpressing neurons
that resist T-cell lysis in vivocan be efficiently lysed ifthey
Figure 1 |Expression of mRNA for three different major
histocompatibility complex (MHC) class I molecules in a
coronal section of adult mouse brain. Blue, H-2D; red, T22;
green, Qa-1. Image courtesy of G. S. Huh and C.J .S.
Box 1 | The MHC gene family: members, nomenclature and functions
The major histocompatibility complex (MHC) is a tightly linked cluster ofgenes that
has been found in every vertebrate genome to date120.The mouse MHC,also called the
H-2 (for histocompatibility-2) region,is analogous to RT in the rat and HLA (human
leukocyte antigen) in humans,but there are interspecies differences in the number and
identity ofindividual MHC genes.The mouse MHC,on chromosome 17,contains over
200 genes,most ofwhich are divided into three broad categories:class I (equivalent to
HLA A,B and C in humans);class II (HLA DP,DQ,and DR in humans);and class III,
which includes components ofthe complement system.The class I genes are further
distinguished as encoding classical (Ia) or non-classical (Ib) MHC class I genes121.
MHC class I genes encode heavy chains (~45 kDa),most ofwhich non-covalently
bind the 12 kDa light chain β2-microglobulin (β2m),which is encoded on
chromosome 2.There are also an increasing number ofMHC-like molecules,many of
which are encoded outside the MHC locus.These molecules can share striking
sequence,structural and functional features with MHC molecules,including the
ability to present molecular cargo at the cell surface,and to bind β2m and
MHC genes are highly POLYMORPHIC,and they display some ofthe highest allelic
diversity in the genome — more than 50 alleles ofthe MHC class Ia gene H-2D alone
have been characterized.MHC class Ia molecules are key players in the adaptive
immune system,and their extraordinary polymorphism enables them to present a
diverse array ofantigenic peptides.MHC alleles are co-dominantly expressed,
increasing the diversity ofMHC proteins that can be expressed by a given individual.
MHC class Ib products are homologous to the class Ia molecules,but are usually less
polymorphic.In many cases,their roles and expression patterns are unknown122,123.
Individuals who express certain MHC class I haplotypes might have an increased
probability ofdeveloping certain neurological diseases,including autism93and
narcolepsy124.The source ofgenetic linkage between the MHC region and these and
other neurological disorders remains a mystery (see main text).
NATURE REVIEWS |NEUROSCIENCE
VOLUME 5 |JULY 2004 |5 2 3
R E V I E W S
(β2m),bind and present peptides (TABLE 1),and have
high sequence and structural homology with members
of the classical MHC class I proteins, although they
display more restricted expression patterns and little or
no polymorphism.Interestingly,recent studies located
one family of these ‘orphan’ molecules exclusively
within the VNO,a small pit in the anterior nasal cavity
of some mammals that is specialized to detect
pheromones (see below)24,25.
In situhybridization with specific probes for individ-
ual classical and non-classical MHC class I genes reveals
a complex pattern ofMHC class I mRNA expression in
the healthy adult brain20,21,24,25(FIG.1).MHC class I genes
display overlapping but distinct neuronal expression
patterns,and these patterns are particularly dynamic
during normal development20,21.Along with the fact
that MHC class I expression can be regulated by natu-
rally occurring electrical activity,these results indicate
that the precise timing and level ofMHC class I expres-
sion might be crucial for its function in the brain.An
important step towards understanding the role ofMHC
class I molecules in the brain is to determine which of
the many MHC class I proteins are expressed in neu-
rons,and to characterize the specific expression profile
of each MHC class I product in the developing and
Physiological functions of neuronal MHC class I
Investigations into the function ofMHC class I genes in
the brain were prompted by the identification ofMHC
class I family members in genomic screens ofspecific
neuronal populations.The first hint ofa non-immune
function for MHC class I molecules in neurons came
from an unbiased functional screen for genes that
are involved in activity-dependent plasticity in
the developing visual system20. MHC class I gene
expression was found to decrease after activity blockade
Biological factors probably also contributed to the
divergent findings.MHC class I expression in neurons
and some other non-neuronal CNS cell types is low
compared with the levels that are seen in tissues such
as the spleen20and the endothelial cells that line
CNS blood vessels.More sensitive techniques,such as
in situhybridization and RNASE PROTECTIONassays,might
improve the detection of MHC class I molecules in
neurons, as PCR-based analysis consistently reveals
neuronal MHC class I mRNA expression,both in vitro
and in vivo14,21,24,25,34.
The age and identity of the neurons also affect the
ability to detect MHC class I protein.MHC class I protein
is expressed by only a subset ofneurons at any given time,
and it is developmentally regulated in the brain,the high-
est levels being seen in the perinatal period20,21.Another
probable source ofdiscrepancy in expression data is the
wide variety of systems used.Many studies tested for
MHC class I expression on acutely dissociated neurons
in vitro,whereas immortalized neuron-like cell lines,such
as OBL-21,CHP-126,RN33B and HEK293 cells,were
used in others2,5–8.So,MHC class I protein diversity and
expression patterns,as well as problems with detection
methods and systems,probably all contribute to difficul-
ties in detecting MHC class I molecules in neurons.
The role ofneural activity in regulating MHC class I
expression is also a contentious issue. In one set of
experiments,blockade ofsodium-based action potentials
with tetrodotoxin (TTX) led to upregulation ofMHC
class I gene expression17,whereas in another set,TTX
treatment led to an equally dramatic downregulation20.
However,these studies differ fundamentally.The first
study examined TTX-treated (action potential blocked)
hippocampal pyramidal neurons that had been dissoci-
ated from embryonic day 18 (E18) rats in vitro17,and
electrical silencing itselfdid not upregulate MHC class I
expression unless it was paired with interferon (IFN)
treatment. The second study showed that blocking
spontaneously generated,endogenous electrical activity
in the fetal cat brain decreased MHC class I expression
in the dorsal lateral geniculate nucleus (dLGN) in vivo20.
Subsequent invivoexperiments showed that MHC class I
expression is increased in the rat hippocampus after
seizures20.Therefore,in both in vivoexperiments,activ-
ity is associated with elevated levels ofneuronal MHC
class I protein.It might therefore be expected that the
relatively low connectivity (and subsequent low levels
ofactivity) in vitromight result in lower basal levels of
MHC class I expression in various culture systems.
Regulation ofMHC class I expression is also known to
change with development:adult dorsal root ganglion
(DRG) sensory neurons are refractory to IFNγ35,
whereas the same neurons from rats at E15 respond
with robust upregulation ofMHC class I expression18.
Although MHC class I genes are best known for their
‘classical’ products,which are crucial for the adaptive
immune response mediated by T cells,most class I genes
actually code for ‘non-classical’ MHC class I products
(FIG.3),many ofwhich have no known function in the
immune system.Some non-classical class I proteins asso-
ciate with the MHC class I light chain β2 microglobulin
Having multiple alleles at a
A technique that is used to
measure the quantity ofmRNA
that corresponds to a given gene
in an RNA sample.A labelled
RNA probe that is
complementary to the relevant
sequence is hybridized with the
RNA sample;any RNA that does
not hybridize with the probe is
then digested away using
mRNA can then be quantified
on an electrophoresis gel.
Figure 2 |T cells can enter the CNS. Pseudocoloured
transmission electron micrograph shows adoptively transferred
myelin-basic-protein-restricted CD4+T cells (green) entering
the CNS through the tight endothelial layer of the blood–brain
barrier (BBB) during induction of experimental autoimmune
encephalomyelitis. Image courtesy of Hartmut Wekerle,
Max-Planck Institute for Neurobiology.
5 2 4 |JULY 2004 |VOLUME 5
R E V I E W S
result of general immune abnormalities, but rather
reflect a novel non-immune function for MHC class I
molecules in the CNS21,36.In addition,they are not likely
to be the result ofiron overload,a known phenotype in
β2m-deficient mice39,for the following reasons.First,
CD3ζ–/–mice,which share the specific neuroanatomical
and synaptic defects ofβ2m–/–;TAP1–/–mice,do not share
their defects in iron trafficking (see below). Second,
β2m-deficient mice do not exhibit a significant increase
in cerebral iron levels, despite striking increases in
hepatic iron,perhaps owing to the ability of the BBB
to restrict plasma iron entry40.Third,iron overload in
these mice is cumulative,with pathology emerging as
the mice age.These experiments were conducted on
very young mice (1 week–2 months ofage),before iron
accumulations are detectable41.
A second screen identified several members ofthe
non-classical MHC class I family of genes that are
expressed specifically in the mammalian VNO25.This
cluster of MHC class I genes (encoded in the M10
region,FIG.3) is selectively expressed in subpopulations
of sensory neurons in the VNO,such that each VNO
neuron expresses only one or a few ofthese MHC class I
molecules24,25,and specific MHC class I molecules are
co-expressed with specific receptors ofthe pheromone
system,the V2Rs.In heterologous cells,co-expression of
the pheromone receptor EC1-V2R, the MHC class I
light chain β2m and the MHC class I heavy chain M10.5
led to surface expression of the pheromone receptor
in vitro,whereas in the absence ofβ2m,surface expres-
sion ofthe VN4/V2R pheromone receptor in the lumen
ofthe VNO was compromised in vivo24.β2m-deficient
mice do not display normal male–male aggression24,
indicating that MHC class I is important in social inter-
actions (recently reviewed in REF.42).
These studies are important in that they provide
unexpected insights into the role of MHC class I in
normal neuronal function, both centrally and in the
periphery.They also provide a valuable starting point for
further studies.One caveat ofboth studies is that they
make use of mice that are deficient in β2m, which
reduces cell-surface expression of most MHC class I
genes.In addition,these mutations are systemic and pre-
sent from birth.β2m is expressed throughout the body
with TTX,specifically during the period when sponta-
neous retinal activity is needed for synaptic refinement of
overlapping eye-specific inputs to LGN neurons to form a
mature,segregated pattern ofconnections20.Subsequent
examination revealed that MHC class I expression closely
parallels the spatiotemporal pattern ofactivity-dependent
plasticity in the developing and adult mammalian brain,
including the early postnatal retina and LGN,and the
adult cerebellum and hippocampus20,21,36. Together,
these observations indicated that MHC class I molecules
might be involved in activity-dependent structural and
These tantalizing correlations were tested directly by
examining two forms ofactivity-dependent plasticity in
mice that were genetically deficient for MHC class I pro-
teins21.Rather than knocking out the class I genes them-
selves,the authors used mice that lacked two crucial
players in the MHC class I protein expression pathway
— β2m,the obligatory light chain ofmost MHC class I
(REF.37),and TAP1,a transporter that is required to load
peptides onto MHC class I molecules for delivery to the
cell surface38.In the absence ofthese two proteins,there
is little stable expression ofMHC class I molecules at the
In these MHC class I-mutant mice,retinal afferents
fail to segregate into eye-specific layers (FIG.4) despite the
presence of normal retinal activity21, indicating that
MHC class I might be involved in translating neural
activity into developmental changes in connectivity21,36.
These and other activity-dependent changes in the pattern
ofconnections are thought to involve functional weaken-
ing and strengthening ofsynapses — phenomena that
have been best-studied in the mammalian hippocampus.
In MHC-deficient mice,adult hippocampal long-term
potentiation is enhanced,whereas long-term depression
is absent21(FIG.4).These results indicate a crucial role for
MHC class I in functional weakening and structural
retraction ofsynaptic connections21,36.Remarkably,the
same specific defects were found in mice that were
mutant for CD3ζ,a component ofmany receptors for
MHC class I proteins in the immune system.However,
more severely immunocompromisedRAG1-deficient
mice did not share these defects.Therefore,brain pheno-
types in MHC-deficient mice are not the nonspecific
T16 T17T18 T19T22 T23
Q1Q2 Q4Q5 Q6 Q7Q8Q9
M7 M8 M1
T24T1 T2T5 T6 T7T8 T9T10 T11
M5M4 M6M3 M2
0 1,5001,6001,7001,800 1,9002,000
50100 1502002500 1020 30 4050 607080
Figure 3 |The mouse major histocompatibility complex (MHC) class I region. Mouse chromosome 17 showing classical
(MHC class Ia; white) and non-classical (MHC class Ib; blue shades) genes.
NATURE REVIEWS |NEUROSCIENCE
VOLUME 5 |JULY 2004 |5 2 5
R E V I E W S
Ly49 proteins43–45.These receptors include players in
both the INNATEand adaptive immune systems.
A crucial aspect ofstudies into the function ofMHC
class I molecules in neurons is the identification of
MHC class I receptors that are expressed endogenously
in the CNS,either on neurons or on non-neural cells.
One clue to the possible identity ofa neuronal MHC
class I receptor came from the finding that CD3ζ— a
required signalling component ofTCRs and also some
natural killer cell receptors46,47— is expressed in a
restricted set ofneurons and is regulated during devel-
opment21.Outside the brain,cells do not express CD3ζ
stably on their cell surface unless they also express
TCR48,indicating that TCR or a related receptor might
be present in the developing and adult CNS21.CD3ζ-
deficient mice exhibit defects in developmental and
adult activity-dependent plasticity that are indistin-
guishable from those that are seen in MHC class I-
deficient mice, indicating that at least some of the
neuronal functions of MHC class I molecules might
involve a CD3ζ-containing receptor21. In addition,
much ofthe signalling machinery that lies downstream
ofMHC class I receptors in the immune system is also
present in neurons,and some components are already
known to have a role in activity-dependent plasticity
(reviewed in REF.36).
Despite evidence for a neuronal role for CD3ζ,so far
no complete MHC class I-binding receptors have been
identified in neurons. However, recent studies have
detected mRNA that encodes a component ofthe TCR
— TCRβ — in adult and developing brain tissue49,50.
TCRβmRNA is dynamically regulated during mouse
brain development:in neonates,expression is highest in
thalamic nuclei,including the LGN,at postnatal days 4
and 7,which corresponds to the period when visual
inputs are undergoing activity-dependent refinement49.
However,mice that are genetically deficient for TCRβ
undergo normal visual refinement,unlike those lacking
CD3ζ, indicating that TCRβ-containing receptors
are not required for this process49. In adults, TCRβ
and at many stages ofdevelopment,so interpretation of
the origins offunctional and behavioural phenotypes in
full knockouts must be made with caution.Nevertheless,
this approach is useful as a first-pass to demonstrate
functional requirements for MHC class I molecules in
the nervous system.
It will be important to clarify the location,identity and
timing ofMHC class I protein expression that is relevant
for its functional and structural effects in neurons,
perhaps using conditional knockouts or reagents that
acutely and/or locally manipulate MHC class I expres-
sion.Possible interactions between MHC class I and
other proteins,including pheromone receptors,should be
verified using methods that permit discrimination
ofdirect interactions from participation in large,multi-
protein complexes.It will also be important to attempt
acute rescue ofMHC class I-deficient activity-dependent
plasticity and behavioural phenotypes to determine
whether they are the result of ongoing MHC class I
signalling or the product ofMHC class I function in a
MHC class I binding partners
How do MHC class I proteins generate and transduce sig-
nals in neurons at the molecular level? Almost all MHC
class I mRNAs encode transmembrane proteins (the
exception being a handful ofsecreted MHC class I pro-
teins), but the cytoplasmic domains of MHC class I
molecules are small,indicating that a binding partner
might be required for intracellular as well as intercellular
MHC class I signalling.Several candidate partners have
been identified on the basis ofthe known functions of
MHC class I molecules outside the CNS (FIG.5).
Immunoreceptors.In the immune system,MHC class I
signalling largely involves interactions with other trans-
membrane immune proteins.Known binding partners
for MHC class I proteins in the immune system include
T-cell antigen receptors (TCRs),NKG2/CD94 receptors,
human KIR and LIR proteins,CD8 dimers and mouse
INNATE IMMUNE SYSTEM
The system that mediates the
early phases ofthe host response
to a group ofrelated pathogens.
Innate immune responses,
unlike adaptive immune
responses,do not increase with
repeated exposure to a given
Table 1 | MHC class 1 proteins expressed by neurons in the mouse CNS
Classical class I (class Ia)
D and K
Antigenic peptides YesAntigen presentation,
activation/inhibition of CD8+T cells,
NK cells;dysfunction implicated
in autoimmune disease, cancer
Non-classical class I (class 1b)*
M1, M10? YesPheromone receptor targeting,
Recognized by T cells,used for
control of NK (and presumably
Qa-1 Signal peptide of
class Ia molecules,
T22Yes Cell activation marker;
γδ-T cells? 21
* see REF. 12. MHC, major histocompatibility complex; NKT, natural killer T cell.
5 2 6 |JULY 2004 |VOLUME 5
R E V I E W S
non-rearranged genomic loci49,50.One of these brain
TCRβtranscripts encodes a hypothetical 23 kDa protein,
but immunoprecipitation and Western blotting ofbrain
tissue with antibodies specific for the TCRβ constant
region failed to detect proteins ofany size49,50.Further-
more,another obligatory component ofthe functional
immune TCR,TCRα,was not detected in neurons49.It is
possible that TCR subunit proteins are expressed only
transiently or at low levels in neurons,or they might not
be recognized by reagents that were developed for use in
the immune system.It is also possible that other CD3ζ-
containing receptors are involved in the actions ofMHC
class I molecules in the brain.In support ofthis sugges-
tion,DIgR1,a putative activating immunoreceptor that
might bind CD3ζ-like proteins,and DIgR2,its inhibitory
partner,are expressed in neurons in several regions ofthe
brain,including the hippocampus,striatum,cortex and
cerebellum (J.Syken and C.J.S.,unpublished data).
It is possible that CD3ζ might also influence
neuronal development and plasticity through a distinct,
convergent mechanism that is independent of MHC
class I.Alternatively, CD3ζ-containing receptors on
neurons or other cells might compete for MHC class I
interactions with other functionally relevant cell surface
Non-immunoreceptor transmembrane proteins.MHC
class I-like molecules are encoded outside the MHC
class I region,and they share sequence,structural and
functional features with the MHC class I products
(BOX 1).There is evidence that several ofthese molecules
function outside the immune system by binding non-
immunoreceptor transmembrane proteins51(FIG. 5).
Indeed,disruption ofone such interaction is the proba-
ble cause of a common heritable disease, hereditary
haemochromatosis (HH) (FIG.6).HH is a disorder of
dietary iron homeostasis,and it is usually caused by a
point mutation in a gene that encodes an MHC class I-
like molecule,HFE52.HFE forms a complex with the
transferrin receptor (TFR)53,54,which binds iron-loaded
transferrin.The HFE–TFR interaction regulates iron
homeostasis,and the point mutation that is found in
~83% ofHH patients52probably interferes with HFE
surface expression55,leading to toxic accumulation of
iron in many tissues54.
Another MHC class I-like molecule,the neonatal Fc
receptor (FcRn),is involved in transport ofmaternal
immunoglobulin across the fetal intestinal epithelium.
FcRn shares the distinctive MHC class I fold,a structure
on the extracellular domain that is specialized to bind
peptide antigens in MHC class Ia and some Ib proteins.
In class Ia molecules, the fold is open and forms a
groove that holds the peptide,whereas in FcRn,the fold
is closed and cannot bind peptides56.Instead,the mole-
cule uses parts ofthis domain to bind the Fc portion of
These results demonstrate that some MHC class I-
like proteins can bind non-immunoreceptor macro-
molecules58and affect their transport between different
cellular compartments.Consistent with this role,recent
results indicate that the M10 non-classical MHC class I
transcripts are strongly expressed in layers 5 and 6 of
the neocortex, whereas expression in the thalamus
In the immune system,functional TCRβ transcripts
are encoded by loci that have undergone SOMATIC
RECOMBINATION, whereas all TCRβ transcripts that are
detected in the brain are the product ofdirect splicing of
Gene segment rearrangements
during lymphocyte development
that lead to the production ofa
wide variety ofcomplete,
variable regions for T-cell
antigen receptors and
fEPSP slope (% of baseline)
0.11 10 100
Figure 4 |Major histocompatibility complex (MHC)-deficient mice have specific defects
in activity-dependent plasticity. a | Mice that are genetically deficient for β2m and TAP1 lack
stable cell surface expression of most MHC class I, owing to lack of the required light chain and
peptide36. b | Endogenous activity arising in the eyes drives refinement of an initially overlapping
projection from the retina to the lateral geniculate nucleus (top) into the mature, laminar,
segregated pattern (bottom)36. c,d | Mice that are deficient for MHC class I genes or CD3ζ fail to
undergo normal activity-dependent refinement of the developing visual projection (c) and display
systematic shifts in hippocampal synaptic plasticity (d). fEPSP, field excitatory postsynaptic
potential; WT, wild type. Parts c and d modified, with permission, from REF.21 (2000) American
Association for the Advancement of Science.
NATURE REVIEWS |NEUROSCIENCE
VOLUME 5 |JULY 2004 |5 2 7
R E V I E W S
In some cases,the immune surveillance that holds
cancers at bay can go awry,leading to neuron-directed
autoimmunity.Paraneoplastic neurological degenera-
tions (PNDs) are a group ofneurodegenerative disorders
that develop in some patients with non-neuronal
cancers67. PNDs are triggered by a T-cell-mediated
immune response against cancer antigens that are also
expressed in neurons (onconeural antigens),leading to
autoimmunitywith neurological symptoms.In addition,
humoral immunity (in the form ofantibodies against
specific neuronal antigens) has an important role.Several
PNDs have been identified,each with distinct neurologi-
cal symptoms and each linked to characteristic tumour
types68.It is unknown why different neurons are targeted
in each PND,as the identified target antigens are often
much more widely expressed.One possibility is that the
affected neurons must express both the target antigen
and the appropriate MHC class I molecule to present it.
In support ofthis idea,PNDs primarily affect the limbic
areas and cerebellum,which are sites ofhigh MHC class I
protein expression in the adult brain20,21.
class I protein might also be neuroprotective.Although
MHC class I-restricted T cells are thought to contribute
to clinical demyelination in multiple sclerosis and
other disorders,new evidence indicates that innate and
adaptive immune responses can facilitate CNS repair
at a later stage by restricting a prominent secondary
wave of damage (reviewed in REF. 69). For example,
anti-myelin basic protein T cells arrest the progression
ofthis secondary degeneration through an unknown
Developmental and behavioural disorders
Because MHC class I is involved in normal brain devel-
opment and plasticity21,36,it is conceivable that altered
MHC class I function could contribute to the disrup-
tion ofthese processes.This possibility is consistent with
reports that several neurological disorders also have
associated immune symptoms.This complex ofsymp-
toms might reflect crosstalk between the immune and
nervous systems,or alternatively,might result from a
disruption ofshared molecular machinery.In addition
to MHC class I, key players in the CELLULAR IMMUNE
RESPONSEinclude adhesion molecules,pro-inflammatory
cytokines,chemokines and proteases,many ofwhich
also have important roles in neuronal development and
function.So,an immune response might have direct
neuronal consequences,because both systems use the
same molecular machinery,albeit with different cellular
In this section,we will discuss the evidence that MHC
class I is involved in three common neurological disor-
ders:schizophrenia,autism and dyslexia (a full discussion
of the symptoms and aetiology of these disorders is
beyond the scope ofthis review).
Schizophrenia.Schizophrenia affects about 1% ofthe
population worldwide,and it is thought to be a neuro-
developmental disorder71.Schizophrenia has a strong
proteins might be required for proper cell-surface
expression ofmembers ofthe V2R family ofpheromone
receptors in the VNO24,raising the possibility that there
are other such examples.
Implications for neuronal disorders
Products ofthe MHC class I region have been linked to
a wide variety ofdisorders with neurological symptoms,
including spinocerebellar ataxia,Huntington’s disease,
Parkinson’s disease,multiple sclerosis,amyotrophic lat-
autism (reviewed in REFS 59–62).This diversity ofassocia-
tions is a product ofthe role ofMHC class I in adaptive
immunity,as well as its extraordinary genetic diversity.
In addition, disruptions of MHC class I function in
either direction — too weak (permitting rampant infec-
tions and tumour expansion) or too strong (causing
transplant rejection and AUTOIMMUNITY) — can lead to
clinical disturbances.It is becoming clear,however,that
this long list ofdisorders implicating MHC class I mole-
cules also reflects the crucial functions ofMHC class I
proteins beyond the immune system.
Targeted immune attack ofneurons.As some neurons
express MHC class I,they could be susceptible to targeted
attack by the body’s own immune system63,64.Such an
attack could be triggered by many stimuli, including
neurotropic viral or bacterial infection,or cancer.For
example,neurons and other cell types upregulate the
expression of MHC class I proteins after spinal cord
infection13.It is unclear whether increases in MHC pro-
tein in these and other cases are a symptom or a cause of
neuronal damage.Interestingly,in an animal model of
virus-induced demyelination (Theiler’s murine
encephalomyelitis),MHC class I-deficient mice are pro-
tected against degeneration following demyelination65,
indicating a causal role for MHC class I in some aspects of
neuronal damage.Cytotoxic T cells can enter the CNS
when they are activated,as well as during periods ofsys-
temic stress,so they could potentially participate in MHC
class I-mediated neuronal damage32,66.
Immune responses directed at
CELLULAR IMMUNE RESPONSE
An adaptive immune response
that is dominated by antigen-
specific T cells,as opposed to
humoral immunity,which is
primarily mediated by
Figure 5 |Modes of major histocompatibility complex
(MHC) class I protein–protein interactions. In the immune
system, MHC class I proteins can interact with other immune
proteins (yellow) on the same cell or on other cells, including
T cells and natural killer cells. Outside the immune system,
MHC class I-like proteins are known to interact with non-
immune proteins (white) on the same cell, and in some cases
might modulate their surface expression.
5 2 8 |JULY 2004 |VOLUME 5
R E V I E W S
In neurons,cytokines markedly upregulate expression
ofMHC class I proteins in vitro4,8,16–19.Therefore,it is pos-
sible that cytokine-induced changes in neuronal MHC
class I expression,at a time when MHC class I is involved
in sculpting developing neuronal circuits,might cause
neurodevelopmental abnormalities that lead to schizo-
phrenia.In addition,cytokine-induced changes in MHC
class I expression in the adult might acutely affect synaptic
function during psychotic episodes.It will be important
to determine whether cytokines induce similar changes in
MHC class I expression in vivo.
Schizophrenia is typified by a baffling array ofneuro-
logical symptoms,including enlarged lateral ventricles89
— a phenotype that is occasionally seen in MHC class I-
deficient mice21.Furthermore,anatomical studies of
post-mortem brain tissue from people with schizophre-
nia show an apparent defect in normal developmental
pruning ofsynaptic connections90,also one ofthe main
phenotypes ofMHC class I-deficient mice21.Although
schizophrenia is thought to be a neurodevelopmental
disorder,symptoms are not typically observed until late
adolescence or early adulthood, so the bulk of the
reported syndrome ofsymptoms could be the result,
not the cause,ofaberrant synaptic organization or loss
Autism.Several studies have indicated a genetic link
between autism and genes within the MHC class I
region91–93(but see also REF.94).In particular,an unusual
number ofchildren with autism share all or part ofthe
extended MHC HAPLOTYPEB44-SC30-DR4 (REFS 91,92,95).
Although this linkage is not statistically significant,
recent studies on hundreds ofmultiplex families have
consistently failed to identify any genes that are signifi-
cantly linked,despite a clear genetic component to the
disorder on the basis of inheritance patterns96. This
implies that autism is a POLYGENIC disorder and/or that
interactions with environmental factors are required.
Consistent with a role for environmental factors,
identical twin concordance rates are less than 100%
(REF.96).As with schizophrenia,there is evidence that
maternal viral infection can increase the risk ofthe child
There are also reports of elevated incidence of
immune disorders in patients with autism and their
first-order relatives. These include abnormal T-cell
populations and cell-mediated immunity100–102,reduced
natural killer cell activity103, and abnormal humoral
and autoantibody responses104–109.These abnormalities
are not sufficient to cause autism,however,as first-order
relatives that do not have autism often share immune
system dysfunction110.Rather,an environmental factor
(such as an immune challenge) might interact with these
and other — possibly genetic — factors to cause autism.
Although autism is usually diagnosed in children
around age three or four,pathology probably precedes
diagnosis.For example,recent studies reveal abnormal
brain development by the early postnatal period,which
is manifested in increased head circumference111.Larger
brain volumes could be the result ofa failure to remove
inappropriate connections during the course ofnormal
genetic component72,and over 60 studies to date have
noted a genetic correlation between schizophrenia and
MHC class I,although these results remain controversial
(REFS 73,74;reviewed in REF.75).Environmental factors
probably also have a crucial role,because identical twin
concordance rates are only ~50% (REFS 76,77).
One possible environmental risk factor for schizo-
phrenia might be an infectious insult. Reports have
correlated maternal viral infection with an increased
chance ofschizophrenia in the child78.Furthermore,an
animal model ofrespiratory infection ofpregnant mice
induces a syndrome with many parallels to schizophrenia
in the offspring,including abnormal social interactions,
increased anxiety in novel or stressful situations,defects
in prepulse inhibition,thinning ofthe neocortex and
hippocampus,and pyramidal cell atrophy79.Therefore,
changes in maternal or fetal immune signalling might
influence the development ofaberrant neuronal con-
nectivity and function in schizophrenia. Genetic
immune abnormalities could contribute to the relative
vulnerability to infection,as well as the ability to mount
an appropriate immune response.
There is widespread but contentious evidence of
immune abnormalities in people with schizophrenia,
including increased serum autoantibodies80,interleukin-6
(IL-6)81and soluble IL-2 receptor82, but decreased
expression of IL-2 and IFNγ83.These features,along
with adolescent onset,stress triggers and variability of
course,are shared with known autoimmune disorders84.
Patients who have a first degree relative with schizophre-
nia are significantly more likely to also have a parent or
sibling with an autoimmune disease85.Conversely,there
is a strong negative correlation between schizophrenia
and two specific autoimmune disorders — insulin-
dependent diabetes mellitus86and rheumatoid
arthritis87.Psychotic episodes have been shown to be
preceded by raised levels ofimmune cytokines in the
cerebrospinal fluid88, and treatment with cytokines
can provoke psychiatric symptoms83. CYTOKINES are
known to affect brain development and adult synaptic
transmission and plasticity,although the mechanisms
Proteins that affect the
behaviour ofother cells through
specific cytokine receptors.
Cytokines that are made by
lymphocytes are often called
lymphokines or interleukins.
The combination ofalleles that
is expressed by a given
individual.The MHC genes are
usually inherited as a haplotype
from each parent.
A term that refers to several loci
that encodes proteins ofsimilar
Figure 6 |HFE point mutation causes hereditary hemochromatosis. HFE interacts with
the transferrin receptor, which is involved in internalization and unloading of iron-loaded
transferrin. A point mutation in the HFE gene prevents interactions between HFE and its
obligatory light chain β2m, preventing stable cell surface expression of HFE, and leading to
enhanced dietary iron absorption.
NATURE REVIEWS |NEUROSCIENCE
VOLUME 5 |JULY 2004 |5 2 9
R E V I E W S
disease in a predisposed individual,and the identity,
time, duration and intensity of infection could all
potentially affect the range, timing and severity of
symptoms on a continuum.
The potential involvement ofMHC class I genes in
these and other neurological disorders might also pro-
vide a clue to the puzzle ofthe specificity ofneuronal
destruction.Neurodevelopmental and neurodegenera-
tive disorders usually produce distinctive patterns of
neuronal loss,disruption or damage.Why are some
neurons targeted and others spared? One possibility is
that expression of specific MHC class I molecules
might render subsets of neurons temporarily and
selectively vulnerable to autoimmune attack or MHC
class I-mediated disruption of development or func-
tion.Additional specificity could be conferred by the
timing of MHC class I expression relative to distinct
developmental events,as well as the co-expression of
crucial cofactors or target proteins.A key first test
of this hypothesis will be to compare the spatial and
temporal expression patterns ofindividual MHC class
I proteins with the populations ofneurons involved in
these disorders.Intriguingly,rat MHC class I and β2m
proteins are expressed most strongly by the dopamin-
ergic and motor neurons that are most susceptible to
neurodegeneration in Parkinson’s disease (FIG.1) and
amyotrophic lateral sclerosis in humans21,23.
Owing to the probable multiplicity of disease
aetiologies,it will be important to develop objective —
perhaps molecular or biochemical — diagnostic criteria
that discriminate subsets of patients with different
pathophysiology and disease course. This must be
accompanied by the development ofappropriate ani-
mal models,in which MHC class I protein expression
and function are monitored along with behavioural and
pharmacological abnormalities. It would also be of
interest to determine MHC class I protein expression,
function and dysfunction in patients with these disor-
ders.Subsequent studies could examine the value of
immunodiagnostics and immunotherapy specifically in
the pool ofpatients that is most likely to be responsive;
that is,those with a probable immune aetiology.
Determining the extent to which neuronal and immune
system MHC class I functions share similar mechanisms
is of vital importance.Many MHC class I receptors49
anddownstream signal transduction components36
are expressed in neurons.Which of these,if any,are
required for the neuronal functions ofMHC class I? Are
other aspects ofMHC class I function in the immune
system — for instance,peptide loading — instructive or
permissive in the neuronal capacity of MHC class I?
What is the role of the genetic diversity in neuronal
functions ofMHC class I?
Considerations raised in this review indicate new
approaches to the study of brain development and
plasticity.Understanding the role ofMHC class I in the
brain might also provide unexpected new avenues for
the diagnosis,treatment and prevention ofneurological
development,a phenotype that has been identified in
specific regions ofthe MHC-deficient mouse brain21.
It is also ofinterest that populations ofneurons that are
known to be specifically affected in autism,including
cerebellar Purkinje cells,normally express high levels of
MHC class I proteins20,21.
Dyslexia.Developmental dyslexia is the most common
childhood learning disorder.Children with dyslexia
have difficulties in reading,despite sufficient intelli-
gence, education and social environment. There is
clearly a strong genetic component to dyslexia,
although MONOZYGOTIC twin concordance is less than
100%,indicating a role for environmental contribu-
tions112.Linkage and association studies have identified
at least six loci that correlate with dyslexia.The second
locus to be identified — DXY2 — maps to the MHC
complex113, a finding that has been confirmed by
several independent studies (reviewed in REF. 112).
Linkage ofreading-related phenotypes to this region is
currently one of the most consistent findings in the
genetics ofhuman cognition112.
There are also controversial reports ofco-morbidity in
patients with dyslexia and their immediate relatives for
immune disorders,including Hashimoto’s thyroiditis,
ulcerative colitis,rheumatoid arthritis,Crohn’s disease
and systematic lupus erythematosus114,115.In addition,
unusual autoantibodies have been detected in the serum
ofmothers whose children have dyslexia116,117.
Notably,people with dyslexia exhibit specific diffi-
culties in visual and auditory discrimination118— tasks
that involve pathways that express high levels ofMHC
class I during mouse development21,59,119.Furthermore,
development ofprimary visual projections is abnormal
in mice that are deficient for cell surface MHC
class I proteins21,indicating that changes in MHC class I
expression are sufficient to disrupt the mature form
ofimportant visual pathways.Therefore,it would be of
great interest to determine whether MHC class I is also
expressed in primate and human visual structures,
including the LGN, and whether this expression
is disrupted in post-mortem brains from people
Critique.Clearly,further experiments are necessary to
test the possible significance ofMHC class I in each of
the above neurological disorders,as the genetic links to
MHC class I (or any other gene) remain contentious.
Many tantalizing human studies rely on small pools of
patient data,which are in some cases poorly controlled.
In addition,methodological and aetiological hetero-
geneity are likely to contribute to the conflicting results.
All three disorders discussed here are complex and
probably involve multiple predisposing genes,as well as
environmental influences.An infectious event is a tan-
talizing candidate environmental factor, because it
could account for two puzzling aspects ofthese disor-
ders:the fact that immune abnormalities are merely
predisposing, and the maddening heterogeneity of
these syndromes in terms ofsymptoms.Infection could
provide the mandatory ‘second hit’ that produces
A term that refers to identical
twins,which develop from a
5 3 0 |JULY 2004 |VOLUME 5
R E V I E W S
1. Rall, G. F., Mucke, L. & Oldstone, M. B. Consequences of
cytotoxic T lymphocyte interaction with major
histocompatibility complex class I-expressing neurons in
vivo. J . Exp. Med. 182, 1201–1212 (1995).
J oly, E., Mucke, L. & Oldstone, M. B. Viral persistence in
neurons explained by lack of major histocompatibility class I
expression. Science253, 1283–1285 (1991).
Lampson, L. A., Whelan, J . P. & Siegel, G. Functional
implications of class I MHC modulation in neural tissue. Ann.
NY Acad. Sci. 540, 479–482 (1988).
Fujimaki, H., Hikawa, N., Nagoya, M., Nagata, T. & Minami, M.
IFN-γ induces expression of MHC class I molecules in adult
mouse dorsal root ganglion neurones. Neuroreport7,
J oly, E. & Oldstone, M. B. Neuronal cells are deficient in
loading peptides onto MHC class I molecules. Neuron8,
Drew, P. D. et al. Regulation of MHC class I and β2-
microglobulin gene expression in human neuronal cells.
Factor binding to conserved cis-acting regulatory
sequences correlates with expression of the genes.
J . Immunol. 150, 3300–3310 (1993).
White, L. A., Keane, R. W. & Whittemore, S. R.
Differentiation of an immortalized CNS neuronal cell line
decreases their susceptibility to cytotoxic T cell lysis invitro.
J . Neuroimmunol. 49, 135–143 (1994).
Lampson, L. A. & Fisher, C. A. Weak HLA and β2-
microglobulin expression of neuronal cell lines can be
modulated by interferon. Proc. Natl Acad. Sci. USA 81,
Lampson, L. A. Interpreting MHC class I expression and
class I/class II reciprocity in the CNS: reconciling divergent
findings. Microsc. Res. Tech. 32, 267–285 (1995).
10. Olsson, T. et al. γ-Interferon-like immunoreactivity in
axotomized rat motor neurons. J . Neurosci. 9, 3870–3875
11. Maehlen, J ., Schroder, H. D., Klareskog, L., Olsson, T. &
Kristensson, K. Axotomy induces MHC class I antigen
expression on rat nerve cells. Neurosci. Lett. 92, 8–13
12. Lidman, O., Olsson, T. & Piehl, F. Expression of nonclassical
MHC class I (RT1-U) in certain neuronal populations of the
central nervous system. Eur. J . Neurosci. 11, 4468–4472
13. Redwine, J . M., Buchmeier, M. J . & Evans, C. F. In vivo
expression of major histocompatibility complex molecules
on oligodendrocytes and neurons during viral infection. Am.
J . Pathol. 159, 1219–1224 (2001).
14. Pereira, R. A., Tscharke, D. C. & Simmons, A. Upregulation
of class I major histocompatibility complex gene expression
in primary sensory neurons, satellite cells, and Schwann
cells of mice in response to acute but not latent herpes
simplex virus infection in vivo. J . Exp. Med. 180, 841–850
15. Foster, J . A., Quan, N., Stern, E. L., Kristensson, K. &
Herkenham, M. Induced neuronal expression of class I
major histocompatibility complex mRNA in acute and
chronic inflammation models. J . Neuroimmunol. 131, 83–91
16. Wong, G. H., Bartlett, P. F., Clark-Lewis, I., Battye, F. &
Schrader, J . W. Inducible expression of H-2 and Ia antigens
on brain cells. Nature310, 688–691 (1984).
17. Neumann, H., Cavalie, A., J enne, D. E. & Wekerle, H.
Induction of MHC class I genes in neurons. Science269,
18. Neumann, H., Schmidt, H., Cavalie, A., J enne, D. &
Wekerle, H. Major histocompatibility complex (MHC) class I
gene expression in single neurons of the central nervous
system: differential regulation by interferon (IFN)-γ and tumor
necrosis factor (TNF)-α. J . Exp. Med. 185, 305–316 (1997).
19. Linda, H. et al. Expression of MHC class I and β2-
microglobulin in rat spinal motoneurons: regulatory
influences by IFN-γ and axotomy. Exp. Neurol. 150,
20. Corriveau, R. A., Huh, G. S. & Shatz, C. J . Regulation of
class I MHC gene expression in the developing and mature
CNS by neural activity. Neuron21, 505–520 (1998).
This study first identified MHC class I in an unbiased
screen for genes involved in activity-dependent
plasticity. MHC was found to be expressed by
neurons and regulated by activity in the adult and
developing mammalian brain.
21. Huh, G. S. et al. Functional requirement for class I MHC in
CNS development and plasticity. Science290, 2155–2159
Using mice deficient for most MHC class I genes, this
study determined that MHC class I is required for
normal activity-dependent refinement of developing
visual projections as well as normal long-term
potentiation and long-term depression in the adult
22. Neumann, H., Schmidt, H., Wilharm, E., Behrens, L. &
Wekerle, H. Interferon-γ gene expression in sensory
neurons: evidence for autocrine gene regulation. J . Exp.
Med. 186, 2023–2031 (1997).
23. Linda, H., Hammarberg, H., Piehl, F., Khademi, M. & Olsson, T.
Expression of MHC class I heavy chain and β2-
microglobulin in rat brainstem motoneurons and nigral
dopaminergic neurons. J . Neuroimmunol. 101, 76–86
24. Loconto, J . et al. Functional expression of murine V2R
pheromone receptors involves selective association with the
M10 and M1 families of MHC class Ib molecules. Cell112,
MHC class Ib (non-classical) genes were identified in
an unbiased screen for genes that were coexpressed
with a subset of pheromone receptors in the VNO,
where they might be involved in pheromone
receptor delivery to the cell surface, as well as gender
25. Ishii, T., Hirota, J . & Mombaerts, P. Combinatorial
coexpression of neural and immune multigene families in
mouse vomeronasal sensory neurons. Curr. Biol. 13,
In parallel with the above study, members of the H2-M
non-classical MHC class I family were identified in
vomeronasal sensory neurons and were found to be
expressed in complex and nonrandom cellular
association with specific pheromone receptors.
26. Binder, G. K. & Griffin, D. E. Interferon-γ-mediated site-
specific clearance of alphavirus from CNS neurons. Science
293, 303–306 (2001).
27. Dalziel, R. G., Lampert, P. W., Talbot, P. J . & Buchmeier, M. J .
Site-specific alteration of murine hepatitis virus type 4
peplomer glycoprotein E2 results in reduced neurovirulence.
J . Virol. 59, 463–471 (1986).
28. Fazakerley, J . K., Parker, S. E., Bloom, F. & Buchmeier, M. J .
The V5A13.1 envelope glycoprotein deletion mutant of
mouse hepatitis virus type-4 is neuroattenuated by its
reduced rate of spread in the central nervous system.
Virology187, 178–188 (1992).
29. Duan, W. M., Westerman, M., Flores, T. & Low, W. C.
Survival of intrastriatal xenografts of ventral mesencephalic
dopamine neurons from MHC-deficient mice to adult rats.
Exp. Neurol. 167, 108–117 (2001).
30. Veng, L. M. et al. Xenografts of MHC-deficient mouse
embryonic mesencephalon improve behavioral recovery in
hemiparkinsonian rats. Cell Transplant. 11, 5–16 (2002).
31. Pakzaban, P., Deacon, T. W., Burns, L. H., Dinsmore, J . &
Isacson, O. A novel mode of immunoprotection of neural
xenotransplants: masking of donor major histocompatibility
complex class I enhances transplant survival in the central
nervous system. Neuroscience65, 983–996 (1995).
32. Ransohoff, R. M., Kivisakk, P. & Kidd, G. Three or more
routes for leukocyte migration into the central nervous
system. Nature Rev. Immunol. 3, 569–581 (2003).
33. Catipovic, B. et al. Analysis of the structure of empty and
peptide-loaded major histocompatibility complex molecules
at the cell surface. J . Exp. Med. 180, 1753–1761 (1994).
34. Pereira, R. A. & Simmons, A. Cell surface expression of H2
antigens on primary sensory neurons in response to acute
but not latent herpes simplex virus infection in vivo. J . Virol.
73, 6484–6489 (1999).
35. Turnley, A. M., Starr, R. & Bartlett, P. F. Failure of sensory
neurons to express class I MHC is due to differential SOCS1
expression. J . Neuroimmunol. 123, 35–40 (2002).
36. Boulanger, L. M., Huh, G. S. & Shatz, C. J . Neuronal
plasticity and cellular immunity: shared molecular
mechanisms. Curr. Opin. Neurobiol. 11, 568–578 (2001).
37. Zijlstra, M. et al. β2-microglobulin deficient mice lack CD4–8+
cytolytic T cells. Nature344, 742–746 (1990).
38. Van Kaer, L., Ashton-Rickardt, P. G., Ploegh, H. L. &
Tonegawa, S. TAP1 mutant mice are deficient in antigen
presentation, surface class I molecules, and CD4–8+T cells.
Cell71, 1205–1214 (1992).
References 37 and 38 detail development and
characterization of two lines of mutant mice deficient
in MHC class I cell surface expression — valuable
tools for studying MHC class I functions in the
immune system and beyond.
39. de Sousa, M. et al. Iron overload in β2-microglobulin-
deficient mice. Immunol. Lett. 39, 105–111 (1994).
40. Moos, T., Trinder, D. & Morgan, E. H. Cellular distribution of
ferric iron, ferritin, transferrin and divalent metal transporter 1
(DMT1) in substantia nigra and basal ganglia of normal and
β2-microglobulin deficient mouse brain. Cell Mol. Biol.
(Noisy-le-grand) 46, 549–561 (2000).
41. Rothenberg, B. E. & Voland, J . R. β2-knockout mice
develop parenchymal iron overload: a putative role for class I
genes of the major histocompatibility complex in iron
metabolism. Proc. Natl Acad. Sci. USA 93, 1529–1534
42. Dulac, C. & Torello, A. T. Molecular detection of pheromone
signals in mammals: from genes to behaviour. Nature Rev.
Neurosci. 4, 551–562 (2003).
43. Moretta, A. et al. Major histocompatibility complex class I-
specific receptors on human natural killer and T
lymphocytes. Immunol. Rev. 155, 105–117 (1997).
44. Ugolini, S. & Vivier, E. Regulation of T cell function by NK cell
receptors for classical MHC class I molecules. Curr. Opin.
Immunol. 12, 295–300 (2000).
45. Trowsdale, J . et al. The genomic context of natural killer
receptor extended gene families. Immunol. Rev. 181, 20–38
46. Bakker, A. B., Wu, J ., Phillips, J . H. & Lanier, L. L. NK cell
activation: distinct stimulatory pathways counterbalancing
inhibitory signals. Hum. Immunol. 61, 18–27 (2000).
47. Kane, K. P., Silver, E. T. & Hazes, B. Specificity and function
of activating Ly-49 receptors. Immunol. Rev. 181, 104–114
48. Klausner, R. D., Weissman, A. M., Baniyash, M., Bonifacino,
J . S. & Samelson, L. E. The role of the ζ-chain in the
expression, structure and function of the T cell receptor. Adv.
Exp. Med. Biol. 254, 21–24 (1989).
49. Syken, J . & Shatz, C. J . Expression of T cell receptor
β-locus in central nervous system neurons. Proc. Natl Acad.
Sci. USA 100, 13048–13053 (2003).
This study found striking and dynamic expression of
mRNA encoding an unrecombined β β-subunit of the
TCR in the developing and adult mouse CNS,
including in the thalamic nuclei and deep layers of
50. Nishiyori, A., Hanno, Y., Saito, M. & Yoshihara, Y. Aberrant
transcription of unrearranged T-cell receptor β-gene in
mouse brain. J . Comp. Neurol. 469, 214–226 (2004).
51. Wilson, I. A. & Bjorkman, P. J . Unusual MHC-like molecules:
CD1, Fc receptor, the hemochromatosis gene product, and
viral homologs. Curr. Opin. Immunol. 10, 67–73 (1998).
52. Feder, J . N. et al. A novel MHC class I-like gene is mutated
in patients with hereditary haemochromatosis. Nature
Genet. 13, 399–408 (1996).
53. Parkkila, S. et al. Association of the transferrin receptor in
human placenta with HFE, the protein defective in hereditary
hemochromatosis. Proc. Natl Acad. Sci. USA 94,
54. Feder, J . N. et al. The hemochromatosis gene product
complexes with the transferrin receptor and lowers its affinity
for ligand binding. Proc. Natl Acad. Sci. USA 95,
55. Feder, J . N. et al. The hemochromatosis founder mutation in
HLA-H disrupts β2-microglobulin interaction and cell surface
expression. J . Biol. Chem. 272, 14025–14028 (1997).
56. Burmeister, W. P., Gastinel, L. N., Simister, N. E., Blum, M. L.
& Bjorkman, P. J . Crystal structure at 2.2 Å resolution of the
MHC-related neonatal Fc receptor. Nature372, 336–343
57. Burmeister, W. P., Huber, A. H. & Bjorkman, P. J . Crystal
structure of the complex of rat neonatal Fc receptor with Fc.
Nature372, 379–383 (1994).
58. Ojcius, D. M., Delarbre, C., Kourilsky, P. & Gachelin, G. MHC
and MHC-related proteins as pleiotropic signal molecules.
FASEB J . 16, 202–206 (2002).
59. Stein, J . The magnocellular theory of developmental
dyslexia. Dyslexia 7, 12–36 (2001).
60. Rubinstein, G. Schizophrenia, rheumatoid arthritis and
natural resistance genes. Schizophr. Res. 25, 177–181
61. Torres, A. R., Maciulis, A. & Odell, D. The association of
MHC genes with autism. Front. Biosci. 6, D936–943
62. Tiwari, J . L. & Terasaki, P. I. HLA and Disease Associations
(Springer-Verlag, New York, 1985).
63. Howard, J . & Thompson, I. First class way to develop a
brain. Nature396, 219–221 (1998).
64. Darnell, R. B. Immunologic complexity in neurons. Neuron
21, 947–950 (1998).
65. Rivera-Quinones, C. et al. Absence of neurological deficits
following extensive demyelination in a class I-deficient
murine model of multiple sclerosis. NatureMed. 4, 187–193
66. Neumann, H., Medana, I. M., Bauer, J . & Lassmann, H.
Cytotoxic T lymphocytes in autoimmune and degenerative
CNS diseases. Trends Neurosci. 25, 313–319 (2002).
67. Darnell, R. B. Onconeural antigens and the paraneoplastic
neurologic disorders: at the intersection of cancer, immunity,
and the brain. Proc. Natl Acad. Sci. USA 93, 4529–4536
68. Albert, M. L. & Darnell, R. B. Paraneoplastic neurological
degenerations: keys to tumour immunity. Nature Rev.
Cancer4, 36–44 (2004).
69. Schwartz, M., Moalem, G., Leibowitz-Amit, R. & Cohen, I. R.
Innate and adaptive immune responses can be beneficial for
CNS repair. Trends Neurosci. 22, 295–299 (1999).
NATURE REVIEWS |NEUROSCIENCE Download full-text
VOLUME 5 |JULY 2004 |5 3 1
R E V I E W S
70. Yoles, E. et al. Protective autoimmunity is a physiological
response to CNS trauma. J . Neurosci. 21, 3740–3748
71. Rothermundt, M., Arolt, V. & Bayer, T. A. Review of
immunological and immunopathological findings in
schizophrenia. Brain Behav. Immun. 15, 319–339 (2001).
72. Karayiorgou, M. & Gogos, J . A. A turning point in
schizophrenia genetics. Neuron19, 967–979 (1997).
73. Schwab, S. G. et al. A genome-wide autosomal screen for
schizophrenia susceptibility loci in 71 families with affected
siblings: support for loci on chromosome 10p and 6. Mol.
Psychiatry5, 638–649 (2000).
74. Smeraldi, E., Bellodi, L. & Cazzullo, C. L. Further studies on
the major histocompatibility complex as a genetic marker for
schizophrenia. Biol. Psychiatry11, 655–661 (1976).
75. Wright, P., Nimgaonkar , V. L., Donaldson, P. T. & Murray, R. M.
Schizophrenia and HLA: a review. Schizophr. Res. 47, 1–12
76. Cardno, A. G. et al. Heritability estimates for psychotic
disorders: the Maudsley twin psychosis series. Arch. Gen.
Psychiatry56, 162–168 (1999).
77. Torrey, E. F. Are we overestimating the genetic contribution
to schizophrenia? Schizophr. Bull. 18, 159–170 (1992).
78. Brown, A. S. et al. Maternal exposure to respiratory
infections and adult schizophrenia spectrum disorders: a
prospective birth cohort study. Schizophr. Bull. 26, 287–295
79. Fatemi, S. H. et al. Prenatal viral infection leads to pyramidal
cell atrophy and macrocephaly in adulthood: implications for
genesis of autism and schizophrenia. Cell. Mol. Neurobiol.
22, 25–33 (2002).
This and several other elegant studies have shown
that prenatal viral infection is a significant risk factor
for neurodevelopmental abnormalities, perhaps
through changes in cytokine expression in the fetal
80. Shima, S., Yano, K., Sugiura, M. & Tokunaga, Y. Anticerebral
antibodies in functional psychoses. Biol. Psychiatry29,
81. Wojtanowska, M. & Rybakowski, J . Changes of humoral
and cellular immunity in schizophrenia. Psychiatr. Pol. 30,
82. Arolt, V., Rothermundt, M., Wandinger, K. P. & Kirchner, H.
Decreased in vitroproduction of interferon-γ and interleukin-2
in whole blood of patients with schizophrenia during
treatment. Mol. Psychiatry5, 150–158 (2000).
83. Nawa, H., Takahashi, M. & Patterson, P. H. Cytokine and
growth factor involvement in schizophrenia — support for
the developmental model. Mol. Psychiatry5, 594–603
84. Ganguli, R. et al. Autoimmunity in schizophrenia: a review of
recent findings. Ann. Med. 25, 489–496 (1993).
85. Wright, P. et al. Autoimmune diseases in the pedigrees of
schizophrenic and control subjects. Schizophr. Res. 20,
86. Finney, G. O. J uvenile onset diabetes and schizophrenia?
Lancet2, 1214–1215 (1989).
87. Eaton, W. W., Hayward, C. & Ram, R. Schizophrenia and
rheumatoid arthritis: a review. Schizoph. Res. 6, 181–192
88. McAllister , C. G. et al. Increases in CSF levels of interleukin-2
in schizophrenia: effects of recurrence of psychosis and
medication status. Am. J . Psychiatry152, 1291–1297 (1995).
89. Degreef, G. et al. Volumes of ventricular system subdivisions
measured from magnetic resonance images in first-episode
schizophrenic patients. Arch. Gen. Psychiatry49, 531–537
90. Keshavan, M. S., Anderson, S. & Pettegrew, J . W. Is
schizophrenia due to excessive synaptic pruning in the
prefrontal cortex? The Feinberg hypothesis revisited.
J . Psychiatr. Res. 28, 239–265 (1994).
91. Warren, R. P. et al. Possible association of the extended
MHC haplotype B44-SC30-DR4 with autism.
Immunogenetics 36, 203–207 (1992).
92. Daniels, W. W. et al. Increased frequency of the extended or
ancestral haplotype B44-SC30-DR4 in autism.
Neuropsychobiology32, 120–123 (1995).
93. Warren, R. P. et al. Strong association of the third
hypervariable region of HLA-DR β1 with autism.
J . Neuroimmunol. 67, 97–102 (1996).
94. Yonan, A. L. et al. A genomewide screen of 345 families for
autism-susceptibility loci. Am. J . Hum. Genet. 73, 886–897
95. Torres, A. R., Maciulis, A., Stubbs, E. G., Cutler, A. & Odell, D.
The transmission disequilibrium test suggests that HLA-DR4
and DR13 are linked to autism spectrum disorder. Hum.
Immunol. 63, 311–316 (2002).
96. Lauritsen, M. & Ewald, H. The genetics of autism. Acta
Psychiatr. Scand. 103, 411–427 (2001).
97. Chess, S. Follow-up report on autism in congenital rubella.
J . Autism Child. Schizophr. 7, 69–81 (1977).
98. Desmond, M. M. et al. Congenital rubella encephalitis.
Course and early sequelae. J . Pediatr. 71, 311–331 (1967).
99. Stubbs, E. G., Ash, E. & Williams, C. P. Autism and
congenital cytomegalovirus. J . Autism Dev. Disord. 14,
100. Stubbs, E. G. & Crawford, M. L. Depressed lymphocyte
responsiveness in autistic children. J . Autism Child.
Schizophr. 7, 49–55 (1977).
101. Warren, R. P. et al. Deficiency of suppressor-inducer
(CD4+CD45RA+) T cells in autism. Immunol. Invest. 19,
102. Denney, D. R., Frei, B. W. & Gaffney, G. R. Lymphocyte
subsets and interleukin-2 receptors in autistic children.
J . Autism Dev. Disord. 26, 87–97 (1996).
103. Warren, R. P., Foster, A. & Margaretten, N. C. Reduced
natural killer cell activity in autism. J . Am. Acad. Child
Adolesc. Psychiatry26, 333–335 (1987).
104. Warren, R. P., Burger, R. A., Odell, D., Torres, A. R. &
Warren, W. L. Decreased plasma concentrations of the C4B
complement protein in autism. Arch. Pediatr. Adolesc. Med.
148, 180–183 (1994).
105. Dalton, P. et al. Maternal neuronal antibodies associated
with autism and a language disorder. Ann. Neurol. 53,
106. Warren, R. P. et al. Increased frequency of the null allele at
the complement C4b locus in autism. Clin. Exp. Immunol.
83, 438–440 (1991).
107. Warren, R. P. et al. Detection of maternal antibodies in
infantile autism. J . Am. Acad. Child Adolesc. Psychiatry29,
108. Singh, V. K., Warren, R., Averett, R. & Ghaziuddin, M.
Circulating autoantibodies to neuronal and glial filament
proteins in autism. Pediatr. Neurol. 17, 88–90 (1997).
109. Connolly, A. M. et al. Serum autoantibodies to brain in
Landau-Kleffner variant, autism, and other neurologic
disorders. J . Pediatr. 134, 607–613 (1999).
110. Krause, I., He, X. S., Gershwin, M. E. & Shoenfeld, Y. Brief
report: immune factors in autism: a critical review. J . Autism
Dev. Disord. 32, 337–345 (2002).
111. Courchesne, E., Carper, R. & Akshoomoff, N. Evidence of
brain overgrowth in the first year of life in autism. J AMA 290,
112. Fisher, S. E. & DeFries, J . C. Developmental dyslexia:
genetic dissection of a complex cognitive trait. Nature Rev.
Neurosci. 3, 767–780 (2002).
113. Cardon, L. R. et al. Quantitative trait locus for reading
disability on chromosome 6. Science266, 276–279 (1994).
This study was the first to show a strong genetic link
between the MHC class I region and developmental
114. Pennington, B. F., Smith, S. D., Kimberling, W. J ., Green, P. A.
& Haith, M. M. Left-handedness and immune disorders in
familial dyslexics. Arch. Neurol. 44, 634–639 (1987).
115. Hugdahl, K., Synnevag, B. & Satz, P. Immune and
autoimmune diseases in dyslexic children.
Neuropsychologia 28, 673–679 (1990).
116. Behan, W. M., Behan, P. O. & Geschwind, N. Anti-Ro
antibody in mothers of dyslexic children. Dev. Med. Child
Neurol. 27, 538–540 (1985).
117. Vincent, A. et al. Maternal antibody-mediated dyslexia?
Evidence for a pathogenic serum factor in a mother of two
dyslexic children shown by transfer to mice using
behavioural studies and magnetic resonance spectroscopy.
J . Neuroimmunol. 130, 243–247 (2002).
118. Witton, C. et al. Sensitivity to dynamic auditory and visual
stimuli predicts nonword reading ability in both dyslexic and
normal readers. Curr. Biol. 8, 791–797 (1998).
119. Stein, J . The neurobiology of reading difficulties.
Prostaglandins Leukot. Essent. Fatty Acids 63, 109–116
120. Kulski, J . K., Shiina, T., Anzai, T., Kohara, S. & Inoko, H.
Comparative genomic analysis of the MHC: the evolution of
class I duplication blocks, diversity and complexity from
shark to man. Immunol Rev190, 95–122 (2002).
121. Gunther, E. & Walter, L. Comparative genomic aspects of
rat, mouse and human MHC class I gene regions.
Cytogenet. Cell Genet. 91, 107–112 (2000).
122. Heinrichs, H. & Orr, H. T. HLA non-A,B,C class I genes: their
structure and expression. Immunol. Res. 9, 265–274 (1990).
123. Hedrick, S. M. Dawn of the hunt for nonclassical MHC
function. Cell70, 177–180 (1992).
124. Mignot, E. Genetic and familial aspects of narcolepsy.
Neurology50, S16–22 (1998).
We are very grateful to N. Colaco and M. Majdan for critical reading
of the manuscript. L.M.B. is supported by a J unior Fellowship from
the Harvard Society of Fellows.
Competing interests statement
The authors declare that they have no competing financial interests.
The following terms in this article are linked online to:
β2m | CD3ζ | DIgR1 | DIgR2 | HFE | IFNγ | IL-2 | IL-6 | MHC class I
| TAP1 | TCRα | TCRβ | RAG1
amyotrophic lateral sclerosis | hereditary haemochromatosis |
Huntington’s disease | multiple sclerosis | narcolepsy | Parkinson’s
disease | schizophrenia
Encyclopedia of Life Sciences: http://www.els.net/
major histocompatibility complex | major histocompatibility
complex: disease associations | major histocompatibility complex:
interaction with peptides
The Boulanger Lab: http://www-biology.ucsd.edu/faculty/
The Shatz Lab:
Access to this interactive links box is free online.