Cutting Edge: A Naturally Occurring Mutation in CD1e
Impairs Lipid Antigen Presentation1
Sylvie Tourne,2*‡§Blandine Maitre,*†‡§Anthony Collmann,¶Emilie Layre,?
Sabrina Mariotti,¶Franc ¸ois Signorino-Gelo,*‡§Caroline Loch,#Jean Salamero,**††
Martine Gilleron,?Catherine Ange ´nieux,*‡§Jean-Pierre Cazenave,†‡§Lucia Mori,¶
Daniel Hanau,*‡§Germain Puzo,?Gennaro De Libero,¶and Henri de la Salle2*‡§
The human CD1a–d proteins are plasma membrane
molecules involved in the presentation of lipid Ags to T
cells. In contrast, CD1e is an intracellular protein present
in a soluble form in late endosomes or lysosomes and is
essential for the processing of complex glycolipid Ags such
as hexamannosylated phosphatidyl-myo-inositol, PIM6.
CD1e is formed by the association of ?2-microglobulin
with an ?-chain encoded by a polymorphic gene. We re-
port here that one variant of CD1e with a proline at po-
sition 194, encoded by allele 4, does not assist PIM6
presentation to CD1b-restricted specific T cells. The
immunological incompetence of this CD1e variant is
mainly due to inefficient assembly and poor transport
of this molecule to late endosomal compartments. Al-
though the allele 4 of CD1E is not frequent in the pop-
ulation, our findings suggest that homozygous individ-
uals might display an altered immune response to
complex glycolipid Ags. The Journal of Immunology,
2008, 180: 3642–3646.
from the cell surface (2). CD1e also participates in the presen-
CD1e never transits through the plasma membrane but is di-
rectly targeted from Golgi compartments to early endosomes
before reaching late endosomes and lysosomes (3). In these lat-
n humans, CD1a–d present lipids to T cells (1). They ac-
quire self-lipid ligands in the endoplasmic reticulum
(ER),3where they are assembled, and lipid Ags in the en-
ter compartments CD1e facilitates the processing of complex
glycolipids, which is required for their presentation by CD1b
molecules. In particular, CD1e is essential in the processing of
sosomal ?-mannosidase (4).
CD1, like MHC class I molecules, is composed of a trans-
membrane ?-chain that noncovalently associates with the ?2-
microglobulin (?2m). The ?-chain folds in three structural ?
domains (?1–3), with the ?1 and ?2 domains delimiting a hy-
drophobic pocket-containing groove in which lipid ligands
bind. For CD1e, the ?-chain is cleaved between the ?3 and the
transmembrane domains in late endosomal compartments,
generating by this way soluble CD1e, which represents the
CD1e active form (4, 5). The human CD1 genes are poorly
polymorphic. Only two alleles have been described for CD1A,
B, C, and D (6, 7), the polymorphism of CD1B and C being
silent (6, 7). CD1E is the most polymorphic CD1 gene, six al-
leles having been reported (6, 8, 9). Among individuals from
diverse ethnic backgrounds, alleles 1 and 2 display a frequency
of 49 and 51%, respectively (6), whereas the four other alleles
have been described once (8, 9). The polymorphic nucleotides
are located in exons 2 or 3 of the CD1E gene, encoding the ?1
and ?2 domains, respectively (see Fig. 1) (6, 8, 9).
The impact of the polymorphism of the CD1 gene on the
structure and function of the encoded protein has been poorly
on the cell surface and display similar structural characteristics
(10), whereas no significant correlation between the CD1 ge-
notype and susceptibility to Mycobacterium malmoense pulmo-
nary disease (11) or chronic dysimmune neuropathies (12)
could be inferred.
*Unite ´ 725 “Biology of Human Dendritic Cells” and†Unite ´ 311, Institut National de la
Sante ´ et de la Recherche Me ´dicale (INSERM), Strasbourg, France;‡Etablissement Fran-
c ¸ais du Sang-Alsace, Strasbourg, France;§Universite ´ Louis-Pasteur, Strasbourg, France;
¶Experimental Immunology, Department of Research, Basel University Hospital, Basel,
Switzerland;?Institut de Pharmacologie et de Biologie Structurale, Centre National de la
Recherche Scientifique (CNRS) Unite ´ Mixte de Recherche (UMR) 5089, Department of
Molecular Mechanisms of Mycobacterial Infections, Toulouse, France;#Institut de Ge ´ne ´-
tique et de Biologie Mole ´culaire et Cellulaire, CNRS/INSERM, Universite ´ Louis-Pasteur,
Illkirch-Graffenstaden, France; and **Imaging Center and††“Molecular Mechanisms of
Intracellular Transport,” UMR144 CNRS, Institut Curie, Paris, France
Received for publication November 5, 2007. Accepted for publication January 25, 2008.
This article must therefore be hereby marked advertisement in accordance with 18 U.S.C.
Section 1734 solely to indicate this fact.
1This work was supported by Institut National de la Sante ´ et de la Recherche Me ´dicale,
Etablissement Franc ¸ais du Sang-Alsace and Agence Nationale de Recherche Microbiolo-
gie-Maladie Emergentes (ANR-05-MIME-006), the European Union funded TB-VAC
tuberculosis vaccine project (LSHP-CT-2003-503367), Swiss National Fund Grant
3100A0-109918, and the Basel Cancer League (Krebsliga Beider Basel). B.M. was the re-
cipient of a grant from Association de Recherche et de De ´veloppement en Me ´decine et en
Sante ´ Publique (ARMESA).
2Address correspondence and reprint requests to Dr. Sylvie Tourne or Dr. Henri de la
Salle, Institut National de la Sante ´ et de la Recherche Me ´dicale, Unite ´ 725, Etablissement
Franc ¸ais du Sang-Alsace, 10 Rue Spielmann, 67065 Strasbourg Cedex, France. E-mail
addresses: firstname.lastname@example.org and email@example.com
3Abbreviations used in this paper: ER, endoplasmic reticulum; ?2m, ?2-microglobulin;
Endo-H, endoglycosidase H; GM1, monosialoganglioside GM1; MFI, median fluores-
cence intensity; PIM, phosphatidyl-myo-inositol-mannoside; PIM6, hexamannosylated
phosphatidyl-myo-inositol; PNGase F, peptide N-glycosidase F; rs, recombinant soluble
(prefix); WB, Western blotting.
Copyright © 2008 by The American Association of Immunologists, Inc. 0022-1767/08/$2.00
CD1E gene on the function of the encoded proteins.
Materials and Methods
DNA constructs and cell transfection
expression vector (Clontech) and transfected, as previously described (5), into
the melanoma M10 cells already transfected, or not, with CD1b cDNA. The
cDNA encoding recombinant soluble (rs) CD1e-4, from residue Asp21to res-
idue Ser305, was cloned in a pMTV5His plasmid (Invitrogen Life Technolo-
gies) and coexpressed with human ?2m in Drosophila S2 cells as described (4).
The anti-CD1e mAb VIIC7, 1.22 and 20.6, have been described (5). Poly-
clonal IgGs specific for the denatured CD1e H chain were obtained by immu-
nizing a rabbit with a recombinant protein consisting of the CD1e ?1, ?2, and
?3 domains fused to GST. An anti-CD1b mAb (clone 4A7.6) was purchased
from Beckman Coulter and an anti-CD63 mAb (clone H5C6) was conjugated
IgG was obtained from respectively Jackson ImmunoResearch or DakoCyto-
mation. HRP-conjugated polyclonal Abs specific for rabbit and mouse Ig were
IgG1 (clone 679.1 Mc7) and IgG2a (clone U7.27) mAbs from Beckman
Coulter were used as isotypic controls and normal mouse serum was obtained
Immunostaining and analyses
Cells, fixed and permeabilized for intracellular labeling, were stained as de-
under a Leica SP5 AOBS confocal microscope (Leica Microsystems). The spe-
cific median fluorescence intensities (MFIs) were obtained by subtracting the
stained with the Ab of interest. Colocalization was quantified as described (3).
Cells, labeled or not with [35S]methionine and cysteine as previously described
(5), were resuspended in lysis buffer containing 1% Triton X-100, Nonidet
of protease inhibitors (Complete Mini; Roche Applied Science) for 30 min on
ice and then cleared by a 10-min centrifugation at 13,000 ? g.
Immunoprecipitations and Western blotting (WB)
Immunoprecipitations and treatment with glycosidases were performed as de-
scribed (5) before separation on SDS-PAGE. When necessary, gels were fixed,
dried, and exposed to autoradiography or to PhosphorImager screens. The ra-
dioactive signals were quantified by Quantity One software after scanning on a
Typhoon Trio PhosphorImager (GE-Healthcare). For WB, proteins were
transferred to nitrocellulose membranes (Bio-Rad) and CD1e molecules were
detected according to the usual protocols using West Pico chemiluminescent
substrate (Perbio Science).
Ag presentation assays
APCs (3 ? 104cells) were preincubated for 2 h at 37°C with monosialogan-
glioside (GM1; Matreya) or Mycobacterium tuberculosis PIM6, extracted, and
in triplicate) GG33A recognizing the CD1e-independent Ag GM1 (14) and
DL15A30 recognizing PIMs in a CD1b-restricted manner, obtained as de-
scribed in Ref. 4, were then added and, after 36 h, GM-CSF release was mea-
sured by ELISA (R&D Systems), as well as IFN-? release (data not shown).
In vitro mannosidase digestion assays
In vitro digestion assays of PIM6by mannosidase and mass spectrum analyses
were performed as previously described (4).
Results and Discussion
CD1e with P194 is not able to assist CD1b presentation of PIM6Ag
Human CD1e is essential for the lysosomal processing of PIM6
into antigenic molecule(s) presented by CD1b (4). This func-
most common allele (6). In this study we investigated whether
the products of the five other alleles of CD1E gene display sim-
We constructed cDNAs encoding CD1e molecules with res-
idues specifically found in the proteins encoded by the alleles 1,
2, 3, 4, 5, and 6 of CD1E, referred to in this paper as CD1e-1,
CD1e-2, CD1e-3, CD1e-4, CD1e-5, and CD1e-6, respec-
tively (Fig. 1B). M10 cells expressing CD1b (M10-CD1b),
which have already been used as APCs to study CD1e function
(4), were stably transfected with each of these cDNAs. Clones
expressing the CD1e variants were selected by immunofluores-
cence staining of fixed, permeabilized cells with the conforma-
tion-dependent mAbs 1.22 and 20.6, previously raised and se-
all of the variants were recognized by 1.22 and all but CD1e-3
were recognized by 20.6 (data not shown), which means that
arginine 164, substituted by a tryptophan in CD1e-3, contrib-
utes to the epitope recognized by the mAb 20.6. These analyses
also showed that all of the clones displayed a similar MFI of
ilar amounts of CD1e (data not shown).
These doubly transfected cells were used as APCs for the
CD1b-restricted presentation of PIM6to the specific T cell
clone DL15A30. The GM-CSF release of T cells in response to
different concentrations of Ag revealed that cells coexpressing
CD1b and CD1e-1, CD1e-2, CD1e-3, CD1e-5, or CD1e-6
were able to present PIM6to T cells in a dose-dependent man-
ner (Fig. 2A). In contrast, PIM6presentation by M10-CD1b-
CD1e-4 cells was significantly reduced and comparable to that
data were obtained by measuring IFN-? release and by using
another CD1b-restricted and PIM6-specific T cell clone (data
The poor capacity of CD1e-4-expressing cells to assist PIM6
presentation was not due to a lower expression of CD1b, be-
cause FACS analyses showed that the specific MFIs of M10-
CD1b-CD1e-4 and -CD1e-2 cells stained with an anti-CD1b
mAb were comparable (346 and 225, respectively, with M10
CD1e-2 and the locations of the residues affected by the polymorphism of
CD1E. B, Nature of the amino acids located at positions 102, 106, 149, 164,
and 194 in the sequences of the different CD1e variants.
CD1e residues affected by the polymorphism of the CD1E
3643 The Journal of Immunology
cells having a specific MFI of 0.4) (data not shown). This re-
duced T cell response was also not caused by a general defect in
Ag presentation of CD1e-4 expressing cells, as these cells very
efficiently presented the CD1e-independent Ag GM1 to the
specific CD1b-restricted T cell clone GG33A (Fig. 2A).
presentation was due to incapacity of the protein to participate
in the processing of PIM6into PIM2by ?-mannosidase, we
produced rsCD1e-4 molecules in insect cells. The activity of
rsCD1e-4 in the in vitro enzymatic digestion of PIM6by
?-mannosidase is reported in Fig. 2B. MALDI-TOF analyses
for rsCD1e-2 (4), ?-mannosidase was able to generate PIM2
molecules. Thus, the rsCD1e-4 produced in insect cells was
biochemically functional in vitro.
CD1e-4 exhibits a singular intracellular fate
Because CD1e-4 appeared to be active in biochemical but not
in Ag presentation assays, we examined whether CD1e-4 un-
CD1e variants. We first established that the other CD1e vari-
ants displayed subcellular localization (in late endosomal com-
Endo-H) similar to those of CD1e-2 (data not shown), and we
then chose to compare CD1e-1 and CD1e-4, which only differ
at residue 194 (Fig. 1B).
We first showed that CD1e-4 remained strictly intracellular,
as does CD1e-1 (data not shown). To examine the intracellular
localization of the protein, fixed and permeabilized transfected
cells were costained with anti-CD1e and anti-CD63 mAbs and
analyzed by confocal microscopy. CD1e-4 colocalized with
partments, like CD1e-1 (Fig. 3A). However, quantitative mea-
surements revealed that CD1e-4 colocalized with CD63 about
four times less than CD1e-1 did (Fig. 3B).
During its intracellular transport, CD1e becomes Endo-H
resistant in the Golgi apparatus and soluble in the late endoso-
mal compartments. To investigate the biochemical maturation
of CD1e-4, transfected cells were metabolically labeled for 45
min, chased for 0, 4, or 8 h, and lysed in buffer containing Tri-
ton X-100, after which CD1e molecules were immunoprecipi-
tated with the mAb 20.6, treated or not with PNGase F or
which was progressively converted into a cleaved form during
chase, cleaved CD1e-4 remained undetectable after 4 h (Fig.
4A) and up to 8 h of chase (data not shown). Moreover, when
CD1e-4 was immunoprecipitated with the mAb 20.6 from de-
tergent-solubilized extracts of transfected cells, no cleaved pro-
revealed the uncleaved form, as did the mAb VIIC7 (Fig. 4B).
In contrast, cleaved CD1e molecules were detected in control
experiments using extracts from cells expressing CD1e-2 or
CD1e-1 (Fig. 4B). Pulse-chase experiments also revealed that
unlike CD1e-1, which became Endo-H resistant during chase,
CD1e-4 remained Endo-H sensitive (Fig. 4A). Nonetheless,
WB analyses of CD1e, immunoprecipitated from detergent-
solubilized extracts of transfected cells, demonstrated that at
least a small fraction of CD1e-4 molecules became Endo-H re-
sistant (Fig. 4C), consistent with its detection in CD63?com-
partments (Fig. 3A). Similar results were obtained when pulse-
chase experiments were performed using the mAb 1.22 instead
of 20.6 or CHAPS instead of Triton X-100 (data not shown).
M10 (not transfected), M10-CD1b, M10-CD1e-2, M10-CD1b-CD1e-1,
-CD1e-2, -CD1e-3, -CD1e-4, -CD1e-5, or -CD1e-6 cells were preincubated
with different doses of PIM6or GM1 before addition of DL15A30 or GG33A
T cells, respectively. GM-CSF release was used to quantify the T cells. The
experiments illustrated have been performed twice with the same results. B, As-
sistance of rsCD1e-4 in the in vitro digestion of PIM6by ?-mannosidase (?-
by MALDI-TOF mass spectrometry. Relative abundance of the different PIM
glycoforms is presented.
A, Assistance of CD1e-4 in PIM6presentation to T cells.
M10-CD1b-CD1e-1 or -CD1e-4 cells were stained with the mAb 20.6 and
revealed with cyanine 3-conjugated polyclonal Abs and an Alexa Fluor 488-
focal view of a part of the cell. B, The colocalization of CD1e and CD63 was
quantified by counting CD63 and CD1e single and double positive structures
total numbers of double positive organelles in CD1e-1- or CD1e-4-expressing
ratios of the numbers of double positive (DP) to single positive vesicles
(DP?CD1e?and DP?CD63?, respectively). The SD for each condition is
Subcellular distribution of CD1e-4. A, Fixed, permeabilized
Altogether, these observations suggested that either only
small amounts of CD1e-4 reach late endosomal compartments
or that CD1e-4 is degraded in these compartments.
CD1e assembly and traffic are affected by the L194P substitution
To study the stability of CD1e-4 in late endosomal compart-
ments, M10-CD1b-CD1e-1 or -CD1e-4 cells were metaboli-
cally labeled during 5 h in the absence or presence of bafilomy-
cin, which blocks acidification of the late endosomal
compartments. Immunoprecipitation of CD1e with the mAb
20.6 followed or not by a treatment with glycosidases showed
that uncleaved Endo-H resistant forms of both CD1e-1 and
CD1e-4 accumulated in cells treated with bafilomycin (Fig.
amounts of CD1e-4 reach endosomes (Fig. 5A).
This defect might result from an altered association of the
CD1e ?-chain with ?2m in the ER. Indeed, the exit of CD1a,
CD1b, and CD1c from the ER has been reported to be strictly
dependent on the association of CD1 ?-chains with ?2m (15),
the plasma membrane not associated with ?2m (16). In agree-
noprecipitated with the CD1e-4 ?-chain as compared with the
CD1e-1 ?-chain using the mAb VIIC7 (Fig. 5B) or 20.6 (data
not shown). This difference was also apparent when Nonidet
vides no argument in favor of a direct participation of leucine
194 in the binding of CD1e to ?2m (17). Nevertheless, proline
teins by introducing an elbow into the structure. Owing to the
proximity of residue 194 to the ?3 domain, which interacts
with ?2m, the association of ?2m with the CD1e-4 ?-chain
might be affected. Next to residue 194, at position 193, is a
conserved aromatic amino acid, a phenylalanine that contrib-
utes to the A? pocket of the CD1 binding groove. As CD1e
binds lipids (4), the L194P substitution could affect the stereo
arrangement of this aromatic residue and, consequently, the
capture of an endogenous lipid required for the appropriate as-
sembly and stability of CD1e molecules in the ER.
the first example of a functional defect in the family of CD1
molecules. It alters the assembly and, consequently, the intra-
cellular transport and function of the encoded molecule. Be-
cause the functional impairment resulting from this polymor-
phism relates to the capacity of the CD1e molecule to
participate in the immune response to complex glycolipids, it is
tempting to speculate that individuals homozygous for the al-
tuberculosis lipid Ags requiring endosomal processing. It will be
important to determine whether the polymorphism of the
CD1E gene represents a factor of susceptibility to human dis-
eases involving the immune response to complex glycolipid
We are especially grateful to J. Mulvihill for excellent editorial assistance. We
also thank the members of the RIO-Cell and Tissue Imaging Facility of UMR
144, CNRS-Institut Curie for their help with imaging approaches.
The authors have no financial conflict of interest.
cells were metabolically labeled for 45 min and chased for 0 or 4 h. CD1e mol-
Endo-H (H) or not (?) and separated by SDS-PAGE. The positions of the
uncleaved and cleaved CD1e forms are indicated by an arrowhead and an ar-
row, respectively. The positions of Endo-H-sensitive and Endo-H-resistant
species are indicated by a larger arrowhead followed by s (sensitive) and r (re-
sistant), respectively (at 4-h chase time point). B, Cell extracts were prepared
from M10-CD1b (?) or M10-CD1b-CD1e-1 (1), CD1e-2 (2), or CD1e-4
(4), and CD1e molecules were immunoadsorbed using the mAb 20.6. Un-
cleaved forms were revealed by WB with the mAb VIIC7, while uncleaved and
cleaved forms were detected with polyclonal rabbit IgGs specific for the dena-
tured luminal part of CD1e. Symbols are defined in A. C, Cell extracts were
prepared from M10-CD1b-CD1e-1 or -CD1e-4 cells and uncleaved CD1e
molecules were immunoadsorbed with the mAb VIIC7, digested with PNGase
F (F) or Endo-H (H) or not (?) and analyzed by WB using the mAb VIIC7.
Symbols are defined in A.
Maturation of CD1e-4. A, M10-CD1b-CD1e-1 or -CD1e-4
ments. A, M10-CD1b-CD1e-1 or -CD1e-4 cells were metabolically labeled in
the presence or absence of bafilomycin (0.1 ?M). After 5 h, CD1e molecules
were immunoadsorbed with the mAb 20.6, digested with PNGase F (F) or
Endo-H (H) or not (?) and separated by SDS-PAGE. Symbols are defined in
Fig. 4A. B, M10-CD1b (?) cells or M10-CD1b-CD1e-1 (4) or M10-CD1b-
CD1e-4 (4) cellswere metabolically labeled for 1 h and cell extracts were pre-
pared in Triton X-100, Nonidet P-40, digitonin, or CHAPS. CD1e molecules
were immunoadsorbed with the mAb VIIC7, separated by SDS-PAGE, and
exposed to PhosphorImager screens.
Assembly and export of CD1e-4 to late endosomal compart-
3645The Journal of Immunology
References Download full-text
1. De Libero, G., and L. Mori. 2005. Recognition of lipid antigens by T cells. Nat. Rev.
Immunol. 5: 485–496.
2. Sugita, M., D. C. Barral, and M. B. Brenner. 2007. Pathways of CD1 and lipid an-
Immunol. 314: 143–164.
3. Angenieux, C., V. Fraisier, B. Maitre, V. Racine, N. van der Wel, D. Fricker,
F. Proamer, M. Sachse, J. P. Cazenave, P. Peters, et al. 2005. The cellular pathway of
CD1e in immature and maturing dendritic cells. Traffic 6: 286–302.
4. de la Salle, H., S. Mariotti, C. Angenieux, M. Gilleron, L. F. Garcia-Alles, D. Malm,
T. Berg, S. Paoletti, B. Maitre, L. Mourey, et al. 2005. Assistance of microbial glyco-
lipid antigen processing by CD1e. Science 310: 1321–1324.
5. Angenieux, C., J. Salamero, D. Fricker, J. P. Cazenave, B. Goud, D. Hanau, and
H. de La Salle. 2000. Characterization of CD1e, a third type of CD1 molecule ex-
pressed in dendritic cells. J. Biol. Chem. 275: 37757–37764.
6. Han, M., L. I. Hannick, M. DiBrino, and M. A. Robinson. 1999. Polymorphism of
human CD1 genes. Tissue Antigens 54: 122–127.
7. Oteo, M., J. F. Parra, I. Mirones, L. I. Gimenez, F. Setien, and E. Martinez-Naves.
1999. Single strand conformational polymorphism analysis of human CD1 genes in
different ethnic groups. Tissue Antigens 53: 545–550.
8. Mirones, I., M. Oteo, J. F. Parra-Cuadrado, and E. Martinez-Naves. 2000. Identifi-
cation of two novel human CD1E alleles. Tissue Antigens 56: 159–161.
9. Tamouza, R., R. Sghiri, R. Ramasawmy, M. G. Neonato, L. E. Mombo, J. C. Poirier,
V. Schaeffer, C. Fortier, D. Labie, R. Girot, et al. 2002. Two novel CD1 E alleles
identified in black African individuals. Tissue Antigens 59: 417–420.
10. Oteo, M., P. Arribas, F. Setien, J. F. Parra, I. Mirones, M. Gomez del Moral, and
E. Martinez-Naves. 2001. Structural characterization of two CD1A allelic variants.
Hum. Immunol. 62: 1137–1141.
11. Jones, D. C., C. M. Gelder, T. Ahmad, I. A. Campbell, M. C. Barnardo, K. I. Welsh,
malmoense pulmonary disease. Tissue Antigens 58: 19–23.
12. De Angelis, M. V., F. Notturno, C. M. Caporale, M. Pace, and A. Uncini. 2007.
13. Gilleron, M., C. Ronet, M. Mempel, B. Monsarrat, G. Gachelin, and G. Puzo. 2001.
Acylation state of the phosphatidylinositol mannosides from Mycobacterium bovis ba-
cillus Calmette Guerin and ability to induce granuloma and recruit natural killer T
cells. J. Biol. Chem. 276: 34896–34904.
glycolipids as T-cell autoantigens. Eur. J. Immunol. 29: 1667–1675.
15. Bauer, A., R. Huttinger, G. Staffler, C. Hansmann, W. Schmidt, O. Majdic,
W. Knapp, and H. Stockinger. 1997. Analysis of the requirement for ?2-microglobu-
lin for expression and formation of human CD1 antigens. Eur. J. Immunol. 27:
16. Balk, S. P., S. Burke, J. E. Polischuk, M. E. Frantz, L. Yang, S. Porcelli, S. P. Colgan,
and R. S. Blumberg. 1994. ?2-Microglobulin-independent MHC class Ib molecule
expressed by human intestinal epithelium. Science 265: 259–262.
complexes. Nat. Rev. Immunol. 5: 387–399.