Non-Classical MHC Class I
Molecules (MHC-Ib)
Rachel L Allen, Centre for Infection and Immunity, St George’s, University of London,
London, UK
L Hogan, Centre for Infection and Immunity, St George’s, University of London, London, UK
Major histocompatibility complex class I (MHC-I) mole-
cules are a family of structurally related proteins that
were first characterised through their central role in
adaptive immunity. Classical MHC class I (MHC-Ia) mole-
cules comprise a heavy chain complexed with b2m to
display short peptide fragments on the surface of all
nucleated cells. As antigen-presenting proteins, the clas-
sical role of MHC-Ia is to present these peptides for
recognition by T-cell receptors expressed by cytotoxic
T-cells. MHC-Ia are also recognised by innate immune
receptors on other leukocyte subsets. A diverse group of
structurally related proteins, collectively termed non-
classical MHC class I molecules or MHC-Ib, exists to per-
form a range of alternative immune roles. Here, the
authors describe the biology of human MHC-Ib molecules
and their recognition as antigen-presenting and antigen-
independent ligands for various immune receptors.
Introduction
Classical major histocompatibility complex class I proteins
(MHC-Ia) are encoded within the MHC, on human chro-
mosome 6. Being also known as human leukocyte antigens
(HLA) in humans, these proteins are found on the surface
of all nucleated cells. The best-characterised role for MHC-
Ia is to obtain short peptide samples derived from the pool
of proteins expressed by a cell, and then carry them to the
eLS subject area: Immunology
How to cite:
Allen, Rachel L; and Hogan, L (November 2013) Non-Classical MHC
Class I Molecules (MHC-Ib). In: eLS. John Wiley & Sons, Ltd:
Chichester.
DOI: 10.1002/9780470015902.a0024246
eLS & 2013, John Wiley & Sons, Ltd. www.els.net
Advanced article
. Introduction
Article Contents
. Alternative Forms of MHC-Ia Proteins
. Nonclassical MHC Class I Proteins (MHC-Ib)
. TCR Recognition of MHC-Ia and MHC-Ib; Classical and
Alternative T-Cell Subsets
. Other Immune Receptors for MHC-Ia and MHC-Ib
. MHC-Ib Proteins encoded within the Major
Histocompatibility Complex
. Outside the MHC: Other MHC-Ib and MHC-I-Like
Proteins
. Nonclassical MHC-I in Disease
. Conclusions
Online posting date: 15th November 2013
cell membrane where they are presented for surveillance by
cytotoxic T-cells (CTLs). MHC-Ia are members of the
Immunoglobulin (Ig) superfamily and contain three Ig
domains (a1, a2 and a3), a transmembrane section and a
short cytoplasmic tail. The peptide presentation function
of MHC-Ia is dependent on the a1 and a2 domains, which
fold together to form an antigen-binding groove. This
antigen-binding groove is lined with pockets within which
the antigenic peptide can be anchored. The a3 domain of
the protein forms a noncovalent association with
b2-microglobulin (b2m) to provide structural support for
the antigen-binding groove (as shown in Figure 1). See also:
Histocompatibility Antigen Complex of Man; Major
Histocompatibility Complex (MHC)
MHC-Ia proteins are highly polymorphic, with hun-
dreds of variants identified to date. The majority of these
polymorphisms are located in or around the antigen-
binding groove. Variations in the amino acid residues that
line the pockets of the groove have a reciprocal effect on
bound peptide, by determining which amino acid ‘anchor
residues’ the pockets can accommodate. Allelic variations
thus determine the repertoire of antigenic peptides that an
MHC-Ia can bind and present for T-cell recognition. Each
TCR on CTLs recognises a unique combination of an
antigenic peptide and the MHC-I protein that presents
it (Garcia et al., 1996). TCR recognition of peptides
presented by b2m-associated MHC-Ia is a focal point
of the adaptive immune response. See also: Major
Histocompatibility Complex: Interaction with Peptides;
T Lymphocytes: Cytotoxic
Alternative Forms of MHC-Ia Proteins
Classical MHC-Ia proteins can exist in alternatively fol-
ded, nonclassical forms. In addition to the heterodimer
complex that presents peptides for recognition by the TCR,
MHC-Ia can fold into alternative structures lacking b2m.
1
 Non-Classical MHC Class I Molecules (MHC-Ib)

Receptor(s)  TCR, KIR, LILR

Contents of antigen Short peptide fragment

MHC-la binding groove

Heavy chain 1 2 3

2m associated? Usually 2m-associated,

1 2 2m-independent forms
are found on activated
leukocytes
2m 3
Cell attachment Transmembrane

Expression Ubiquitous

Figure 1 MHC-Ia. Within the MHC-Ia structure, the antigen-binding groove is formed by the a1 and a2 domains of the MHC-I heavy chain. Individual
pockets within the groove are defined by biochemical characteristics, which, in turn, determine the nature of the peptide ligands that can be anchored
within them. The classical ab T-cell Receptor (TCR) binds diagonally across the antigen-binding groove. This orientation allows the receptor to recognise
both the MHC-I protein itself and the antigenic peptide presented within the groove. The a3 domain and b2m provide structural support for the antigen-
binding groove in addition to forming binding sites for other receptors such as CD8 and LILR.

‘Heavy chain’ forms of MHC-Ia lacking b2m are found on
the surface of activated leukocytes to the extent that they
are considered a marker of lymphocyte activation (Arosa
et al., 2007). These heavy chain structures are not recog-
nised by the classical TCR but instead engage with alter-
native receptors in cis and/or act as ligands for innate
immune receptors in trans. HLA-C alleles and the disease-
associated HLA B27 allele show a particular tendency to
fold into heavy chain structures (Arosa et al., 2007).
their classical antigen-presenting role and into innate
immunity. These observations have led to the suggestion
that it may be more accurate to define MHC-I proteins on
the basis of their gene location. Thus, genes encoded at the
HLA-A, -B and -C loci are defined as classical MHC-Ia,
whereas other MHC-I-like molecules, whether encoded
within or outside the MHC, are termed nonclassical or
MHC-Ib proteins (Kumanovcis et al., 2003).

TCR Recognition of MHC-Ia and

Nonclassical MHC Class I Proteins
(MHC-Ib)
MHC-Ia are characterised by a number of factors, namely
their polymorphism, ability to present peptide antigen for
recognition by TCR and ubiquitous expression on nucle-
ated cells. Closely related proteins known as ‘nonclassical
MHC-I’, or MHC-Ib, have a similar structure to their
classical counterparts, but in general they are less poly-
morphic, present different types of antigen, are recognised
by alternative immune receptors or may only be expressed
at the cell surface under certain conditions. MHC-Ib were
originally defined by such differences, but as these proteins
have become better characterised, an increasing degree of
overlap of features with MHC-Ia has been observed; there
are examples of MHC-Ib which are polymorphic, MHC
encoded or recognised by TCRs. Further to this, members
of the MHC-Ib family may be ubiquitously expressed or
present peptide antigens. Indeed, genetic analysis has
shown that the MHC-Ia genes of some species are more
similar to human MHC-Ib than to our classical MHC-Ia
alleles. Meanwhile, in recent years, the biology of classical
MHC-Ia themselves has been shown to extend beyond
MHC-Ib; Classical and Alternative
T-Cell Subsets
All MHC-Ia and several MHC-Ib proteins are recognised
by the TCR for antigen. The antigen recognition domain of
the TCR is composed of a heterodimer of two polypeptide
chains, either the TCR-a and -b subunits or the TCR-g
and -d subunits. The classical TCR required for adaptive
immune recognition of MHC-Ia contains an ab hetero-
dimer expressed by CTLs, which use CD8 as a coreceptor.
On antigen encounter, the orientation of an ab TCR across
its ligand allows the receptor to recognise both the pre-
sented peptide and the MHC-Ia allele that presents it.
To do so, the TCR binds diagonally across the groove,
forming direct contacts with both the peptide and MHC-Ia
a1 and a2 domains on either side of it (Garcia et al., 1996).
A large repertoire of ab TCRs is required to recognise the
full range of peptide antigens presented in complex with an
individual’s multiple MHC-Ia alleles. TCR diversity is
achieved by combinatorial gene rearrangement of variable
(V), diverse (D) and joining (J) gene segments, a process
that takes place in the thymus to generate the a and
b transcripts for each unique TCR. Va chains are formed

2 eLS & 2013, John Wiley & Sons, Ltd. www.els.net


Table 1 Major histocompatibility complex (MHC)-Ia- and MHC-Ib-specific populations of T-cells and their T-cell receptors
(TCRs)
T-cell population TCRs
Classical T-cells Polyclonal ab TCR
Mucosal associated invariant T-cells Invariant ab TCR Va7.2/Ja33 with Vb2 or Vb13
Invariant Natural Killer T-cells Invariant ab TCR Va24/Vb11
Type II Natural Killer T-cells Oligoclonal ab TCR
Intraepithelial lymphocytes Invariant gd TCR Vd1
from V and J segments, whereas Vb chains are generated
from combinations of V, D and J segments. This gene
rearrangement can generate a circulating repertoire of
T-cells that has been estimated to include 106 different
b chains alone (Bonarius et al., 2006). In the absence of
immune challenge, the circulating repertoire of multiple
TCR is maintained. During an infection, naive pre-
cursor(s) bearing a TCR that is capable of recognising a
specific peptide/MHC-Ia combination must then pro-
liferate to provide an expanded clonal population of cells
present in sufficient numbers to mount an effective immune
response. This requirement for clonal expansion results in a
delay between an initial immune challenge and a primary
adaptive immune response. See also: T-cell Receptors
For the MHC-Ib proteins that are subject to abTCR
recognition, a more restricted usage of TCR gene segments
is usually observed, as indicated in Table 1. For example, the
mucosal associated invariant T-cell subset (MAIT), which
forms a large proportion of circulating T cells, expresses a
TCR composed of a Va7.2/Ja33 a-chain paired with a TCR
b-chain containing either Vb2 or Vb13 gene segments. This
invariant TCR recognises an MHC-Ib protein known as
MR1. Unlike the binding footprint of classical TCRs, only
one chain of the MAIT TCR makes contact with its MHC-
Ib ligand (Reantragoon et al., 2012). One advantage of
restricted TCR usage is that a population of preexisting
MAIT cells is already present in sufficient numbers to
provide an immediate response on encountering their
ligand, with no need for clonal expansion. Hence, MAIT
can be regarded as agents of innate immunity, rather than
part of the adaptive immune response.
The natural killer T-cell (NKT) subset displays proper-
ties of both innate and adaptive immunity. NKT cells can
be subdivided into type I (invariant NKT (iNKT)) cells,
which bear a semi-invariant TCR (Va24/Vb11 in humans)
and type II NKT cells, which express an oligoclonal TCR
repertoire. In a similar scenario to that observed for MAIT,
the semi-invariant TCR on iNKT cells does not follow the
diagonal binding profile used by classical TCR but instead
adopts a tilted orientation and binds parallel to the MHC
with only one chain of the receptor contacting its antigen-
presenting ligand CD1d. Again, due to their restricted
TCR usage, iNKT cells are present in sufficient numbers
and provide a rapid cytokine response on recognition of
their ligand without the time delay required for clonal
expansion. In the case of type II NKT cells, both a- and b-
chains of the TCR contact CD1d, as the TCR forms a
eLS & 2013, John Wiley & Sons, Ltd. www.els.net
Non-Classical MHC Class I Molecules (MHC-Ib)
MHC ligand(s)
MHC-Ia
MR1
CD1d
CD1d
MICA/MICB
diagonal footprint across the antigen-presenting groove.
However, only one chain of the type II NKT TCR contacts
the antigen presented by CD1d. Type II NKT cell recep-
tors, therefore, combine features of iNKT and classical
TCR recognition (Girardi et al., 2012). See also: Natural
Killer T Cells
T-cells bearing a gd TCR are less common and less well
characterised than their ab TCR counterparts. Thismay be
due in part to their recognition of alternative, non-MHC-
Ia ligands, which are expressed only in particular tissues
and/or at times of cellular stress. Unlike ab TCR, gd
receptors do not necessarily recognise antigen presented by
an MHC protein but may recognise the MHC-like protein
itself (Chien and Konigshofer, 2007). Human gd T-cells can
be allocated into subsets on the basis of their Vd gene usage,
which appears to correspond to the locations in which they
are found. For example, intraepithelial lymphocytes bear
receptors using the Vd1 gene segment and are the major gd
T-cell subset within the epithelium. Vd2 T-cells circulate in
peripheral blood. See also: Lymphocytes: Gamma Delta;
Lymphocytes: Intraepithelial
Other Immune Receptors for MHC-Ia
and MHC-Ib
A number of innate immune receptors can recognise
MHC-Ia and/or MHC-Ib. These are often referred to as
natural killer (NK) receptors (although they are not
necessarily expressed on NK cells) and fall into the cate-
gories of lectin-like receptors or members of the Ig super-
family. Like the MHC molecules themselves, genes for NK
receptors are colocated within immune gene complexes.
Members of the lectin-like receptor family are found within
the natural killer complex (NKC) on human chromosome
12p13.1, whereas MHC-specific members of the Ig-
like receptor family are located within the leukocyte
receptor complex (LRC) on human chromosome 19q13.4
(Trowsdale, 2001).
Two closely related MHC-I-specific receptor families are
encoded within the LRC. Members of the killer Ig-like
receptor (KIR) family are found on NK cells and T-cells,
whereas members of the leukocyte ig-like receptors (LILR)
are predominantly expressed on cells of the myeloid lineage.
The KIR and LILR families include receptors that are
designated as activating or inhibitory on the basis of the
3
 Non-Classical MHC Class I Molecules (MHC-Ib)
intracellular motifs through which they signal. Inhibitory
KIR and LILR carry long cytoplasmic tails bearing
immunoreceptor tyrosine-based inhibitory motifs (ITIMs).
In contrast, activating receptors have a truncated cyto-
plasmic tail that lacks any signalling domains of its own.
Instead, these receptors signal by association with adaptor
proteins such as DAP12, FcRg or DAP10 using immunor-
eceptor tyrosine-based activating motifs (ITAMs) in their
cytoplasmic tails. This dependence on adaptor proteins is
indicated by the location of the genes for DAP12 and
DAP10 within the LRC, where the KIR and LILR are
themselves encoded (Trowsdale, 2001).
Recognition of MHC-Ia and -Ib by members of the KIR
and LILR families can elicit a wide range of functional
effects, including the regulation of NK cell activity and
modulation of antigen-presenting cell phenotype (Trows-
dale, 2001; Anderson and Allen, 2009; Pilsbury et al.,
2010). Although they lack the fine specificity for individual
peptide/MHC complexes shown by their classical TCR
counterparts, KIR and LILR do exhibit some degree of
ligand specificity. Individual KIR recognise subsets of
MHC-Ia alleles. Inhibitory LILR show a broad specificity
for MHC-Ia and -Ib; however, they bind some alleles more
strongly than others. Activating members of the LILR
family show a preference for b2m-independent forms of
HLA-C alleles (Jones et al., 2012).
Several lectin-like receptors specific for MHC-Ib are
encoded within the NKC on human chromosome 12. These
include NKG2-A, -B, -C, -E and -H and the CD94 protein
with which they form heterodimer complexes. As with the
LRC-encoded receptors, receptors in the NKG2 family can
be designated as activating or inhibitory according to the
presence or absence of ITIM motifs in their cytoplasmic
domains. NKG2A and B are thus designated as inhibitory,
whereas activating members of the family, such as CD94/
NKG2C, associate with the adapter protein DAP12 to
signal through its ITAM domains.
HLA-E
1 2
2m 3
Figure 2 HLA-E. The overall structure of HLA-E is very similar to that for classical MHC-Ia. A defining feature of this MHC-Ib protein is that strict structural
constraints within the antigen-binding groove restrict the repertoire of peptides that can be bound.
4 eLS & 2013, John Wiley & Sons, Ltd. www.els.net
MHC-Ib Proteins encoded within the
Major Histocompatibility Complex
MHC-I proteins were first named after the gene complex in
which the classical members of the family are encoded – the
MHC. Various MHC-I proteins are found within this
region of chromosome 6 and include the classical HLA-A,
-B and -C in addition to the nonclassical MHC-Ib proteins
HLA-E, -F and -G, Hfe, MHC class I polypeptide-related
sequence A (MICA) and MHC class I polypeptide-related
sequence B (MICB).
HLA-E
Of the MHC-Ib proteins encoded on chromosome 6,
HLA-E shows the greatest similarity to classical alleles in
terms of its ligand-binding and immune-recognition
properties; HLA-E associates with b2m to bind short
peptides within its antigen-binding groove (Figure 2) and
can be recognised by innate immune receptors with a
degree of peptide specificity (Hoare et al., 2008) as well as
by classical TCR.
The crystal structure of HLA-E revealed a peptide-
binding groove containing five pockets that accommodate
‘anchor’ residues of the bound peptide, as compared with
the two or three that are usually present in MHC-Ia
(O’Callaghan et al., 1998). A high number of anchor resi-
dues impose strict constraints on the sequence of peptides
that can be bound by HLA-E. The cognate peptide, which
can achieve these structural constraints, is derived from the
leader sequence of MHC-Ia proteins. A leader sequence is
required to direct newly synthesised MHC-Ia into the
endoplasmic reticulum during protein translation. Having
served this purpose, the leader sequence is cleaved off and
then subsequently processed for binding to HLA-E. Once
an MHC-Ia leader peptide has been complexed within its
Receptor(s) CD94/NKG2A-C and TCR
Contents of antigen Short peptide fragment
binding groove derived from leader
sequence of MHC-la
Heavy chain 1 2 3
2m associated? Yes
Cell attachment Transmembrane
Expression Ubiquitous
 Non-Classical MHC Class I Molecules (MHC-Ib)

antigen-binding groove, HLA-E progresses to the cell
surface for immune surveillance. As its cell surface
expression is dependent on peptides derived from MHC-Ia,
HLA-E acts as a ubiquitous marker of healthy nucleated
cells alongside HLA-Ia. Any loss of MHC-Ia expression
(as frequently observed in viral infection or cancer) will
result in a lack of leader peptide available to bind HLA-E,
which will then be retained inside the cell rather than
progressing to the cell surface.
Reduced levels in surface expression of HLA-E will have
immune consequences for a cell. This MHC-Ib protein has
a predominantly inhibitory effect on NK cell functions due
to its recognition by members of the CD94/NKG2 receptor
family (Lee et al., 1998; Braud et al., 1998). Thus, loss of
HLA-E on a target cell can result in NK cell-mediated
destruction. CD94 is the major determinant of HLA-E
binding, with the associated NKG2 subunit responsible for
signal transduction. As described above, NKG2 receptors
can be activating or inhibitory in nature, so control systems
must be in place to ensure an appropriate outcome fol-
lowing HLA-E recognition. The various NKG2 receptors
are expressed in a sequential order with inhibitory mem-
bers of the family appearing before activating receptors
(Iwaszko and Bogunia-Kubik, 2011). A further level of
control is imposed by the strength of receptor/ligand
interaction; activating CD94/NKG2 receptors appear to
have a much lower affinity for HLA-E than their inhibitory
counterparts (Sullivan et al., 2008). Together these control
mechanisms ensure that the main function of HLA-E is to
act as a marker of healthy MHC-Ia production by a cell
and protect it from NK cell lysis.
Under certain conditions such as stress or infection, an
alternative repertoire of peptides may bind to HLA-E.
Cellular stress induces the upregulation of heat shock
proteins, and peptide fragments of these can bind and be
presented by HLA-E. This reduces the strength of binding
to CD94/NKG2 inhibitory receptors and results in target
cell lysis. It can also lead to HLA-E-restricted T-cell
recognition. Microbial-derived peptides presented by
HLA-E have been identified for pathogens including
Mycobacterium tuberculosis to influence pathogen-specific
T-cell responses (Van Hall et al., 2010).
HLA-F
The biology of HLA-F is less well studied than that of its
MHC-encoded class Ia and Ib neighbours, so relatively
little is known about the recognition or functions of this
protein. HLA-F readily associates with b2m but does not
bind peptide or present any known antigen (Figure 3; Lepin
et al., 2000; Goodridge et al., 2010). In the majority of cells,
HLA-F remains intracellular, but cell surface HLA-F is
seen on leukocytes following cell activation (Lee et al.,
2010), where it associates with b2m-free heavy chain forms
of MHC-Ia (Goodridge et al., 2010). As with the b2m-free
MHC-Ia heavy chains themselves, cell surface HLA-F
may, therefore, be regarded as a marker of immune cell
activation. It is interesting to note that the only lympho-
cytes that do not exhibit cell surface expression of HLA-F
on activation are regulatory T-cells (Lee et al., 2010).
Together these observations may indicate a functional role
for HLA-F. HLA-F is known to act as a ligand for the
inhibitory innate immune receptors LILRB1 and LILRB2
but is yet to be studied in the context of activating LILR.
However, it is known that the b2m-free heavy chains, with
which HLA-F associate, are preferential ligands for acti-
vating members of the LILR family (Jones et al., 2012).
Interaction of b2m-free MHC-Ia heavy chains with HLA-F
may thus regulate the balance of LILR activity.
HLA-G
HLA-G is a powerful immune inhibitory protein. Its
expression is usually limited to placental trophoblast cells,
where it can be produced in multiple isoforms and is
thought to protect fetal tissue from attack by the maternal
immune system.
A crystal structure of HLA-G defined an antigen-
binding groove that is capable of binding peptide but
which, like HLA-E, structurally limits the repertoire of

Receptor(s) LILRB1 and LILRB2

HLA-F

Contents of antigen Empty

binding groove
1 2
Heavy chain 1 2 3
2m associated? Yes
2m 3
Cell attachment Transmembrane

Expression Activated lymphocytes

Figure 3 HLA-F. Relatively little is known about this MHC-encoded protein. Although it associates with b2m, this MHC-Ib protein does not appear to play
any role in antigen presentation. To date, the only known receptors for HLA-F are members of the LILR family.

eLS & 2013, John Wiley & Sons, Ltd. www.els.net 5

 Non-Classical MHC Class I Molecules (MHC-Ib)

Receptor(s) KIR2DS4, LILRB1 and LILRB2

HLA-G
Contents of antigen Short peptide fragment
binding groove

1 2 Heavy chain 1 2 3

2m associated? Can be 2m associated or

2m-idependent
2m 3
Cell attachment Transmembrane

Expression Trophoblast

Figure 4 HLA-G. HLA-G exists in multiple isoforms and shows a restricted tissue distribution. The standard form of HLA-G is similar to that of MHC-Ia
alleles, although structural constraints limit the repertoire of peptides that can bind within the antigen-binding groove.

peptides it can bind (Figure 4) (Shiroishi et al., 2006). HLA- Hfe

G exists in multiple isoforms. These HLA-G isoforms vary
in their b2m association and their existence as soluble or cell
surface forms. Of the seven known HLA-G isoforms, four
are found at the cell surface (HLA-G1 to HLA-G4) and
three are produced as soluble proteins (HLA-G5 to HLA-
G7) (Carosella et al., 2011a). Some HLA-G isoforms exist
as dimers, due to the unusual ability of this MHC-Ib to
form disulphide-bonded homodimers through otherwise
unpaired cysteine residues in its extracellular domains
(Shiroishi et al., 2006).
HLA-G acts as a ligand for several innate immune
receptors in the LRC-encoded KIR and LILR families.
The first receptor interaction documented for HLA-G was
with an Ig-like receptor, KIR2DL4. KIR2DL4 is an aty-
pical member of the KIR family; it is one of three ‘frame-
work’ receptors that are found in all haplotypes of the
otherwise highly variable KIR gene complex and is highly
conserved. In contrast to other members of the KIR family
found at the cell surface, KIR2DL4 appears to be located
in endosomes, where it interacts with soluble HLA-G iso-
forms that have been endocytosed (Rajagopalan and
Long, 2012). The functional consequences of this intra-
cellular recognition continue to be studied. HLA-G is also
the highest-affinity self-ligand for the inhibitory receptors
LILRB1 and LILRB2. The strength of this LILR/HLA-G
interaction is further enhanced for disulphide-bonded
dimers of HLA-G (Shiroishi et al., 2006). Through its
engagement of LILRB1 and LILRB2, HLA-G has the
potential to inhibit the immune functions of T-, B-, NK
and antigen-presenting cells. It is known to favor the dif-
ferentiation of tolerogenic antigen-presenting cells and
regulatory T-cells (Anderson and Allen, 2009).
In addition to direct inhibition of immune functions,
HLA-G can exert indirect immunoregulation via HLA-E.
When the leader peptide of HLA-G is bound within its
antigen-presenting groove, HLA-E shows a stronger affi-
nity for CD94/NKG2 than for MHC-Ia-derived leader
peptides in addition to increased inhibitory signalling
(Iwaszko and Bogunia-Kubik, 2011).
Originally known as HLA-H (a nomenclature now used to
refer to an unrelated pseudogene), mutations in this
MHC-Ib protein were found to be responsible for hae-
mochromatosis, an iron overload disease. Following this
discovery, the gene was renamed Hfe (for High Iron (Fe)).
Hfe associates with b2m but does not present any antigen
(Figure 5). Instead, it associates in cis with the transferrin
receptor (Bennet et al., 2000) to regulate iron absorption.
Mutations that prevent Hfe from associating with b2m will
result in haemochromatosis (Jacobs et al., 2007). Hence, it
appears that Hfe has no direct role in the immune system.
However, it is worth noting that iron homoeostasis can
influence disease outcome following infection, with several
pathogens known to target iron uptake (Drakesmith and
Prentice, 2008).
MICA and MICB
Of the MHC-Ib proteins encoded within the MHC, MICA
and MICB may be considered as bearing the least similarity
to the classical MHC-Ia. These proteins do not associate
with b2m and do not appear to have a functional antigen-
presenting groove (Figure 6). They do, however, exhibit
a high degree of polymorphism; more than 70 MICA
and 30 MICB alleles have been identified to date
(Fernandez-Messina et al., 2012). Owing to their extensive
polymorphisms, MIC proteins play a role as minor histo-
compatibility antigens in transplantation.
The appearance of MICA or MICB on the surface of
epithelial cells may act as a danger signal in response to
stress, infection or transformation. MICA and MICB
trigger signalling through NKG2D, an NKC-encoded
member of the C-type lectin family, which is expressed on
NK cells, T-cells and macrophages. Unlike other members
of the NKG2 family, functional NKG2D exists as a
homodimer and transmits activating signals via associa-
tion with the adaptor protein DAP10. MICA is somewhat
promiscuous in its receptor-binding property, as it may

6 eLS & 2013, John Wiley & Sons, Ltd. www.els.net

 Non-Classical MHC Class I Molecules (MHC-Ib)

Receptor Transferrin receptor

Hfe
(TfR)

Contents of antigen Empty

binding groove
1 2
Heavy chain 1 2 3

3 2m associated? Yes

2m
Cell attachment Transmembrane

Expression Ubiqitous

Figure 5 Hfe. This MHC-Ib protein associates with b2m but does not appear to play any role in antigen presentation. Rather than immune function, the
principal activity of Hfe is in Iron regulation.

Receptor(s) NKG2D and  TCR

MICA and MICB

Contents of antigen Empty

binding groove
1 2
Heavy chain 1 2 3

3 2m associated? No

Cell attachment Transmembrane

Expression Stress-induced

Figure 6 MICA and MICB. MIC proteins do not associate with b2m and do not appear to play a role in antigen presentation. These transmembrane
proteins are upregulated in response to stress stimuli.

also be recognised by gd TCR bearing the Vd1 gene seg-
ment (Groh et al., 1998; Groh et al., 1996; Wu et al., 2002).
NKG2D and Vd1 TCR may compete for binding to
MICA. It appears that the interaction with the gd TCR
is of low affinity but long lasting, whereas binding to
NKG2D is almost 1000-fold stronger but of shorter
duration (Xu et al., 2011).
are upregulated in response to cellular stress. They are also
polymorphic but less than MIC genes. It is not clear why
there are multiple ligands for NKG2D or why the receptor
itself is expressed on multiple cell types (NK cells, CD8+
ab T-cells and gd T-cells from the intestinal epithelium).
Recognition of certain NKG2D ligands by gd TCR adds a
further layer of complexity to this system.

Outside the MHC: Other MHC-Ib and
MHC-I-Like Proteins
ULBPs
A further set of MHC-Ib proteins encoded within another
region of chromosome 6 acts as ligands for NKG2D
(reviewed in Fernandez-Messina et al., 2012). UL16-
binding proteins (ULBPs) contain domains that resemble
the a1 and a2 domains of MHC-Ia and usually exist as
structures that are attached to the membrane by a glyco-
phosphatidylinositol (GPI) tail (Figure 7). The relevance of
GPI linkage is not clear but may locate the protein to
specific regions of the cell membrane within lipid rafts. Like
their fellow NKG2D ligands MICA and MICB, ULBPs
CD1
CD1 proteins are predominantly expressed by professional
antigen-presenting cells (Pei et al., 2012). These b2m-asso-
ciated MHC-Ib proteins have a specialised antigen-binding
groove which enables them to present glycolipid or phos-
pholipid antigens for immune recognition (Figure 8). Struc-
tural analysis has revealed in fine detail how the hydrophobic
region of a lipid becomes engaged within the pockets of
CD1d, leaving the polar part of the molecule available for
immune receptor recognition (Zajonc and Kronenberg,
2007). See also: Glycolipid Presentation by CD1
In humans, the CD1 family can be subdivided into three
subsets: group 1 (CD1a, CD1b and CD1c), group 2 (CD1d)
and group 3 (CD1e), but only one of these, CD1d, is found
in mice. CD1d is the best-characterised member of the CD1
family. This is due in part not only to its presence in the

eLS & 2013, John Wiley & Sons, Ltd. www.els.net 7

 Non-Classical MHC Class I Molecules (MHC-Ib)

Receptor NKG2D and  TCR

Contents of antigen Empty

binding groove

Heavy chain 1 2

ULBPs 2m associated? No

Cell attachment GPI-linked (ULBP1-3 and ULBP5)

or
1 2 Transmembrane (ULBP2 and
ULBP4-5)

Expression Stress-induced

Figure 7 ULBPs. Like their MIC counterparts, ULBPs do not play a role in antigen presentation. ULBPs do not associate with b2m and consist of a1 and a2
domains linked to the cell surface by a GPI tail.

Receptor  TCR and  TCR

CD1
Contents of antigen glycolipid and
binding groove phospholipid antigens

Heavy chain 1 2 3
1 2
2m associated? Yes

2m 3 Cell attachment? Transmembrane

Expression Professional antigen

presenting cells

Figure 8 CD1. CD1 molecules are antigen-presenting MHC-Ib proteins, which bind glycolipid or phospholipid antigens within their binding groove and
present these for recognition by receptors on NKT cells. They are expressed by professional antigen-presenting cells.

murine system (which enables knockout studies to be
performed) but also to the identification of a-galacto-
sylceramide (aGalCer), a high-affinity ligand for the
antigen-binding groove of CD1d (Kawano et al., 1997).
aGalCer is derived from marine sponges. Although it is a
nonphysiological ligand, complexes of aGalCer/CD1d can
be used to identify the iNKT cells that recognise CD1d.
CD1 presentation allows NKT cells to respond to glyco-
lipids from a variety of microorganisms (Zeisseg and
Blumberg, 2011). It also appears that the repertoire of
lipids presented naturally by CD1d may alter following
immune stimulation (reviewed in Hofstetter et al., 2011).
Different lipid antigens may be recognised by different
subsets of NKT cells. For example, sulfatide is a self-
antigen that is presented by CD1d for recognition by type
II NKT cells (reviewed in Zajonc and Kronenberg, 2007).
NKT cells exert their effects via cytokine secretion, with the
nature of the cytokines generated varying between different
CD1 presented ligands and their cell surface localisation
(Im et al., 2009). See also: Natural Killer T Cells
Although less well characterised than CD1d, some
ligands and functions are known for other members of the
CD1 family. The various CD1 proteins differ in their
trafficking patterns, a feature that may bring individual
members of the family into contact with different sources
and types of antigens (Adams, 2013). CD1a is pre-
dominantly found on Langerhans cells in the skin, where it
is recognised by Th22 cells (De Jong et al., 2010). CD1e
appears to act as a chaperone rather than an antigen-pre-
senting molecule. It is located intracellularly in lysosomes,
where it enables loading of glycolipid antigens into the
groove of other CD1 proteins (De la Salle et al., 2005).
MR1
MR1 is a further example of an MHC-Ib protein that
specialises in the presentation of nonpeptide antigens
(Figure 9). It is highly conserved and is ubiquitously
expressed, albeit at low levels. The MHC class II invariant
chain traffics newly synthesised MR1 to endosomes, which

8 eLS & 2013, John Wiley & Sons, Ltd. www.els.net

 Non-Classical MHC Class I Molecules (MHC-Ib)

Receptor  TCR
MR1
Contents of antigen Vitamin B metabolites
binding groove

1 2 Heavy chain 1 2 3

2m associated? Yes

2m 3
Cell attachment? Transmembrane

Expression Ubiquitous

Figure 9 MR1. MR1 is an antigen-presenting MHC-Ib protein, which presents nonpeptide antigens. These ligands are recognised by receptors on MAIT.

Receptor  TCR

Contents of antigen Glycolipid antigen

Endothelial protien C receptor binding groove
(EPCR)
Heavy chain 12

2m associated? No
1 2
Cell attachment Transmembrane

Figure 10 EPCR. This MHC-Ib protein resembles CD1 and presents phospholipids for recognition by gd T-cells. It does not associate with b2m and consists
of a1 and a2 domains linked to the cell surface by a transmembrane domain.

may allow it to bind ligands from exogenous sources such
as endocytosed microbes. MR1-specific MAIT have been
shown to be triggered by MR1+ targets only in the pre-
sence of bacteria or yeast, although not all strains of bac-
teria appear to be able to elicit MAIT reactivity. It has
recently been shown that bacterial strains that activate
MAIT produce a vitamin B metabolite that is presented to
them by MR1 (Kjer-Nielson et al., 2012).
MR1 engages receptors on MAIT, as described above
(reviewed in Hofstetter et al., 2011). These innate lym-
phocytes take their name from their predominant location
in the intestinal lamina propria but can also be found in the
circulation. MAIT express chemokine receptors that
enable trafficking to the intestine and liver but do not
express CCR7, a chemokine receptor required for traf-
ficking to lymph nodes (Le Bourhis et al., 2011). It is
thought that postnatal expansion of the MAIT population
is triggered by commensal gut flora colonisation (Le
Bourhis et al., 2011).
Other MHC-Ib
MHC-I-like proteins that are subject to immune recogni-
tion continue to be identified and may play a variety of
immune and nonimmune roles. For example, endothelial
protein C receptor (EPCR) resembles CD1 and is a trans-
membrane protein bearing an a1-a2 domain platform
that can bind phospholipids (Figure 10) (Willcox et al.,
2012). Its primary role is to regulate coagulation, but it
can be upregulated on endothelial cells in response to
stressors such as cytomegalovirus (CMV) infection or
transformation where it acts as a ligand for gd T-cells
(Willcox et al., 2012). In this case, both EPCR and costi-
mulatory proteins are required to elicit a functional
response from gd T-cells.
Nonclassical MHC-I in Disease
MHC-Ib proteins enable the immune system to provide
a rapid response to microbial antigens following in-
fection and/or to self-derived danger signals elicited during
pathologies such as cancer. As some members of this
protein family are upregulated by cellular stress, inap-
propriate expression and recognition of MHC-Ib
could also contribute to autoimmune or -inflammatory
pathologies.

eLS & 2013, John Wiley & Sons, Ltd. www.els.net 9

 Non-Classical MHC Class I Molecules (MHC-Ib)
Infection
MHC-Ib proteins capable of antigen presentation are
particularly relevant during infection. Animal models
deficient in MR1 show increased bacterial load following
infection and it has been shown that MAIT may migrate to
the site of mycobacterial infection (Le Bourhis et al., 2011;
Gold et al., 2010).
CD1-presented glycolipid antigens derived from several
Gram-negative and -positive bacteria, including Sphino-
gomonas, Borrelia and Streptococcus sp., have been iden-
tified and NKT cell responses appear to be important in the
control of these infections (Pei et al., 2012). Knockout mice
studies have shown that absence of CD1d also increases
susceptibility to certain viral infections (Zeisseg and
Blumberg, 2011), an effect that is reflected in human
patients deficient in a CD1 antigen-loading protein
(Zeisseg and Blumberg, 2011). In the context of viral
infection, CD1 is subject to specific immune evasion
mechanisms (Hofstetter et al., 2011); Herpes simplex and
Epstein–Barr viruses are known to interfere with CD1d,
suggesting that NKT cells play an important role in con-
trolling these infections (Zeisseg and Blumberg, 2011).
The relevance of MHC-Ib in infection is highlighted by
the existence of viral immune evasion mechanisms. The
human CMV (HCMV) genome encodes a variety of pro-
teins that target the activity of MHC-Ia and -Ib. Loss of
MHC-Ia from the cell surface, as seen during HCMV
infection, would confer an advantage for the virus by
allowing it to evade CTL recognition but could be pre-
dicted to leave the infected cell susceptible to NK cell
killing through loss of MHC-Ia-mediated inhibition.
HCMV overcomes the deleterious effects of MHC-Ia
downregulation by encoding a viral peptide that can bind
within the HLA-E antigen-binding groove to maintain the
functional effects of this protein (Van Hall et al., 2010).
HCMV also encodes its own MHC-I homologue, a protein
known as UL18, which acts as a high-affinity ligand for the
inhibitory receptor LILRB1 (Cosman et al., 1997), whereas
other viral proteins modulate the expression of NKG2D
ligands (Fernandez-Messina et al., 2012). Other viruses
known to target NKG2D ligands include human herpes-
virus-7 and Kaposi sarcoma-associated herpesvirus.
Antitumour responses
MIC gene expression is triggered by a number of stress
stimuli, as indicated by the presence of heat shock elements
in their promoter region (Trowsdale, 2001). Like their
fellow NKG2D ligands, the ULBPs, these proteins have
been shown to be upregulated in numerous types of cancer
(Fernandez-Messina et al., 2012). Stress-induced MHC-Ib
proteins act as a self-induced danger signal and would,
therefore, be expected to play a role in antitumour
responses. Accordingly, gd T-cells, iNKT cells and Type II
NKT cells appear to provide protection against cancer
progression (Groh et al., 1999; Zajonc and Kronenberg,
10 eLS & 2013, John Wiley & Sons, Ltd. www.els.net
2007; Pei et al., 2012). Expression of stress-induced MHC-
Ib may not always be beneficial; soluble forms of these
proteins can be shed by tumours, with their presence
leading to loss of the activating immune receptor
(Fernandez-Messina et al., 2012).
Ectopic expression of an MHC-Ib with powerful
immune inhibitory effects may lead to a loss of antitumour
responses. Inappropriate expression of HLA-G has been
implicated in various pathologies and is frequently
observed in tumour cells (Bukur et al., 2012). The tolero-
genic effects of HLA-G are further enhanced by the ability
of this protein to upregulate the immune inhibitory
receptors that recognise it (Carosella et al., 2011).
Conclusions
The biology of MHC-Ib proteins and their recognition is a
complex field. Future studies will continue to identify novel
ligands and roles for these proteins. Although the roles of
MHC-Ib in health and disease are not yet well char-
acterised, the accumulating body of knowledge on these
proteins already provides some insight into their functions
and highlights the interrelated functions of innate and
adaptive immunity in the response to immune challenge.
References
Adams EJ (2013) Diverse antigen presentation by the Group 1
CD1 molecule CD1d. Molecular Immunology 55: 182–185.
Anderson KJ and Allen RL (2009) Regulation of T-cell immunity
by leucocyte immunoglobulin-like receptors: innate immune
receptors for self on antigen presenting cells. Immunology 127:
8–17.
Arosa FA, Santos SG and Powis SJ (2007) Open conformers: the
hidden face of MHC-I molecules. Trends in Immunology 28:
115–123.
Bennet MJ, Lebron JA and Bjorkman PJ (2000) Crystal structure
of the hereditary haemochromatosis protein HFE complexed
with transferrin receptor. Nature 403: 46–53.
Bonarius HP, Baas F, Remmerswaal EB et al. (2006) Monitoring
the T-cell receptor repertoire at single-clone resolution. PLoS
ONE 1(1): e55. doi:10.1371/journal.pone.0000055.
Braud VM, Allan DS, O’Callaghan CA et al. (1998) HLA-E binds
to natural killer receptors CD94/NKG2A B and C. Nature 391:
795–799.
Bukur J, Jasinski S and Seliger B (2012) The role of classical and
non-classical HLA antigens in human tumours. Seminars in
Cancer Biology 22: 350–358.
Carosella ED, Gregori S and LeMaoult J (2011) The tolerogenic
interplay(s) among HLA-G, myeloid APCs, and regulatory
cells. Blood 118: 6499–6505.
Carosella ED, Gregori S, Rouas-Freiss N et al. (2011a) The role of
HLA-G in immunity and hematopoiesis. Cellular and Mole-
cular Life Sciences 68: 353–368.
Chien YH and Konigshofer Y (2007) Antigen recognition by
gammadelta T cells. Immunology Reviews 215: 46–58.

Cosman D, Fanger N, Borges L et al. (1997) A novel immu-
noglobulin superfamily receptor for cellular and viral MHC
class I molecules. Immunity 7: 273–282.
De Jong A, Pena-Cruz V, Cheng TY et al. (2010) CD1a-auto-
reactive T cells are a normal component of the human ab T cell
repertoire. Nature Immunology 11: 1102–1109.
De la Salle H, Mariotti SM, Angenieux CM et al. (2005) Assis-
tance of microbial glycolipid antigen processing by CD1e.
Science 310: 1321–1324.
Drakesmith H and Prentice A (2008) Viral infection and iron
metabolism. Nature Reviews Microbiology 6: 541–542.
Fernandez-Messina L, Reyburn HT and Vales-Gomez M (2012)
Human NKG2D ligands: cell biology strategies to ensure
immune regulation. Frontiers in Immunology 3: 299.
Garcia KC, Degano M, Stanfield RL et al. (1996) An alphabeta T
cell receptor structure at 2.5 A and its orientation in the TCR-
MHC complex. Science 274: 209–219.
Girardi E, Maricic I, Wang J et al. (2012) Type II natural killer T cells
use features of both innate-like and conventional T cells to
recognise sulfatide self-antigens. Nature Immunology 13: 851–856.
Gold MC, Cerri S, Smyk-Pearson S et al. (2010) Human mucosal-
associated invariant T cells detect bacterially infected cells.
PLoS Biology 8: e1000407.
Goodridge JP, Burian A, Lee N and Geraghty DE (2010)
HLA-F complex without peptide binds to MHC class I
protein in the open conformer form. Journal of Immunology
184: 6199–6208.
Groh V, Bahram S, Bauer S et al. (1996) Cell stress-regulated
human major histocompatibility complex class I gene expressed
in gastrointestinal epithelium. Proceedings of the National
Academy of Sciences of the USA 93: 12445–12450.
Groh V, Steinle A, Bauer S and Spies T (1998) Recognition of
stress-induced MHC molecules by intestinal epithelial gam-
madelta T cells. Science 279: 1737–1740.
Groh V, Rhinehart R, Secrist H et al. (1999) Broad tumor-asso-
ciated expression and recognition by tumor-derived gamma
delta T cells of MICA and MICB. Proceedings of the National
Academy of Sciences of the USA 96: 6879–6884.
Hoare HL, Sullivan LC, Clements CS et al. (2008) Subtle changes
in peptide conformation profoundly affect recognition of
the non-classical MHC class I molecule HLA-E by the
CD94-NKG2 natural killer cell receptors. Journal of Molecular
Biology 377: 1297–1303.
Hofstetter AR, Sullivan LC, Lukacher AE and Brooks AG (2011)
Diverse roles of non-diverse molecules: MHC class Ib molecules
in host defense and control of autoimmunity. Current Opinion
in Immunology 23: 104–110.
Im JS, Arora P, Bricard G et al. (2009) Kinetics and cellular site of
glycolipid loading control the outcome of natural killer T cell
activation. Immunity 30: 888–898.
Iwaszko M and Bogunia-Kubik K (2011) Clinical significance of
the HLA-E and CD94/NKG2 interaction. Archivum Immuno-
logiae et Therapiae Experimentalis 59: 353–367.
Jacobs EM, Verbeek AL, Kreeftenberg HG et al. (2007) Changing
aspects of HFe-related hereditary haemochromatosis and
endeavours to early diagnosis. Netherlands Journal of Medicine
65: 419–424.
Jones DC, Kosmoliaptsis V, Apps R et al. (2012) HLA class I
allelic sequence and conformation regulate leukocyte Ig-like
receptor binding. Journal of Immunology 186: 2990s–2997s.
eLS & 2013, John Wiley & Sons, Ltd. www.els.net
Non-Classical MHC Class I Molecules (MHC-Ib)
Kawano T, Cui J, Koezuka Y et al. (1997) CD1d-restricted and
TCR-mediated activation of valpha14 NKT cells by glyco-
sylceramides Science 278: 1626–1629.
Kjer-Nielson L, Patel O, Corbett AJ et al. (2012) MR1 presents
microbial vitamin B metabolites to MAIT T cells. Nature
10.1038/nature11605
Kumanovcis A, Takada T and Fischer-Lindahl K (2003) Geno-
mic organisation of the mammalian MHC. Annual Review of
Immunology 21: 629–657.
Le Bourhis L, Guerri L, Dusseaux M et al. (2011) Mucosal-
associated invariant T cells: unconventional development and
function. Trends in Immunology 32: 212–218.
Lee N, Ishitani A and Geraghty DE (2010) HLA-F is a surface
marker on activated lymphocytes. European Journal of Immu-
nology 40: 2308–2318.
Lee N, Llano M, Carretero M et al. (1998) HLA-E is a major
ligand for the natural killer inhibitory receptor CD94/NKG2A.
Proceedings of the National Academy of Sciences of the USA 95:
5199–5204.
Lepin EJ, Bastin JM, Allan DS et al. (2000) Functional char-
acterization of HLA-F and binding of HLA-F tetramers to
ILT2 and ILT4 receptors. European Journal of Immunology 30:
3552–3561.
O’Callaghan CA, Tormo J, WIllcox BE et al. (1998) Structural
features impose tight peptide binding specificity in the non-
classical MHC molecule HLA-E. Molecular Cell 1: 531–541.
Pei B, Vela JL, Zajonc D and Kronenberg M (2012) Interplay
between carbohydrate and lipid in recognition of glycolipid
antigens by natural killer T cells. Annals of the New York
Academy of Sciences 1253: 68–79.
Pilsbury LE, Allen RL and Vordermeier M (2010) Modulation of
Toll like receptor activity by Leukocyte Ig-like Receptors and
their effects during bacterial infection. Mediators of Inflamma-
tion 2010: 536478.
Rajagopalan S and Long EO (2012) KIR2DL4 (CD158d): An
activation receptor for HLA-G. Frontiers in Immunology 3: 258.
Reantragoon R, Kjer-Neilsen L, Patel O et al. (2012) Structural
insight into MR1-mediated recognition of the mucosal asso-
ciated invariant T cell receptor. Journal of Experimental Med-
icine 209: 761–774.
Shiroishi M, Kuroki K, Ose T et al. (2006) . Efficient leukocyte Ig-
like receptor signalling and crystal structure of disulphide-
linked HLA-G dimer. Journal of Chemical Biology 281: 10439–
10447.
Sullivan LC, Clements CS, Rossjohn J and Brooks AG (2008) The
major histocompatibility complex class Ib molecule HLA-E at
the interface between innate and adaptive immunity. Tissue
Antigens 72: 415–424.
Trowsdale J (2001) Genetic and functional relationships between
MHC and NK receptor genes. Immunity 15: 363–374.
Van Hall T, Oliveira CC, Joosten SA and Ottenhoff TH
(2010) The other Janus face of Qa-1 and HLA-E: diverse pep-
tide repertoires in times of stress. Microbes and Infection 12:
910–918.
Willcox CR, Pitard V, Netzer S et al. (2012) Cytomegalovirus and
tumour stress surveillance by binding of a human gd T cell
antigen receptor to endothelial protein C Receptor. Nature
Immunology 13: 872–879.
Wu J, Groh V and Spies T (2002) T cell antigen receptor
engagement and specificity in the recognition of stress-inducible
11
 Non-Classical MHC Class I Molecules (MHC-Ib)
MHC class I-related chains by human epithelial gamma delta T
cells. Journal of Immunology 169: 1236–1240.
Xu B, Pizzaro JC, Holmes MA et al. (2011) Crystal structure
of a gd T-cell receptor specific for the human MHC class I
homolog MICA. Proceedings of the National Academy of
Sciences of the USA 108: 2414–2419.
Zajonc DM and Kronenberg M (2007) CD1 mediated cell
recognition of glycolipids. Current Opinion in Structural
Biology 17: 521–529.
Zeisseg S and Blumberg RS (2011) Primary immunodeficiency
associated with defects in CD1 and CD1-restricted T cells.
Annals of the New York Academy of Sciences 1250: 14–24.
12 eLS & 2013, John Wiley & Sons, Ltd. www.els.net
Further Reading
Apps R, Gardner L and Moffett A (2008) A critical look at
HLA-G. Trends in Immunology 29: 313–321.
Hayday A (2000) [gamma][delta] cells: a right time and a right
place for a conserved third way of protection. Annual Review of
Immunology 18: 975–1026.
Kasahara M and Yoshida S (2012) Immunogenetics of the
NKG2D ligand family. Immunogenetics 64: 855–867.
Willcox BE, Willcox CR, Dover LG and Besra G (2007)
Structures and functions of microbial lipid antigens presented by
CD1. Current Topics in Microbiology and Immunology 314: 73–110.