M 3-[N-morpholino]propanesulfonic acid [ionic
strength = 0.10 (NaCI) (pH 7.0)] and 5% dimethyl
sulfoxide containing 20 FM antibody. The con-
centration of esters 5a and 5b was 1 mM, amine 6
was 2 mM, and, because of limited solubility, 5c
was 100[LM.After 20 s and 6 min, an aliquot was
withdrawn and quenched with perchloric acid to a
final pH of approximately 2.5. The quenched
sample was injected into an HPLC column (Wa-
ters 600E) equipped with an analytical C18 re-
versed-phase column (VYDAC, Hesperia, CA)
and eluted with an acetonitrile-water (0.1% tri-
fluoroacetic acid) gradient with the detector set at
280 nM. We identified the product by comparing
the retention time of the product formed during
the reaction with that of authentic samples. The
relative rates were determined by dividing the
amount of 7 formed by the antibody with that
formed in the control reaction.
12. The control experiments were carried out under
the same conditions as described (11), except
antibody was not added to the reaction mixture.
13. A. R. Fersht and W. P. Jencks, J. Am. Chem. Soc.
92, 5442 (1970).
14. This is the lower limit with an estimated spontane-
ous rate of 0.06 FM min-1.
15. R. Hirschmann et al., unpublished results.
16. A. D. Napper, S. J. Benkovic, A. Tramontano, R.
A. Lerner, Science 237, 1041 (1987).
17. A. G. Cochran, T. Pham, R. Sugasawara, P. G.
Schultz, J. Am. Chem. Soc. 113, 6670 (1991).
18. J. R. Jacobsen, J. R. Prudent, L. Kochersperger,
S. Yonkovich, P. G. Schultz, Science 256, 365
(1992). In this case, the dissociation constant of
18 nM for the product relative to a value of 0.77
mM for the alcohol substrate effectively inhibited
the reaction after one turnover.
19. A. L. Heard and G. T. Young, J. Chem. Soc. 1963,
5807 (1963); N. L. Benoiton, in The Peptides, S.
Udenfriend and J. Meienhofer, Eds. (Academic
Press, San Diego, 1983), vol. 5, chap. 4.
20. R. Stanfield, M. Takimoto-Kamimura, J. Rini, A. T.
Profy, I. Wilson, Structure 1, 83 (1993); J. Moult et
al., J. Mol. Biol. 74, 137 (1976).
21. J. H. Arevalo, M. J. Tausig, I. A. Wilson, Nature
365, 859 (1993).
22. B. S. Green, in Catalytic Antibodies (CIBA Foun-
dation Symposium 159, Wiley, West Sussex, Unit-
ed Kingdom, 1991), p. 139.
23. J. D. Stewart and S. J. Benkovic, Chem. Soc. Rev.
22, 213 (1993).
24. R.H. acknowledges J. R. Huff, E. Thornton, and T.
Widlanski for substantive discussions. We thank D.
Schloeder for assistance in the preparation of the
monoclonal antibodies and B. Jiang for assisting
S.D.T. and P.A.B. Supported by NIH (Institute of
General Medical Sciences) through grant GM-
(R.H. and A.B.S.), Bachem (R.H.), and
Merck Research Laboratories (R.H.). C.M.T. is a
recipient of a William Georgetti Scholarship. S.D.T.
is a recipient of a Natural Sciences and Engineer-
ing Research Council of Canada postdoctoral fel-
lowship. K.M.Y. is a recipient of an NIH (National
Cancer Institute) postdoctoral fellowship.
9 February 1994; accepted 26 May 1994
Homozygous Human TAP Peptide Transporter
Mutation in HLA Class
Henri de la Salle,* Daniel Hanau, Dominique Fricker,
Arlette Urlacher, Adrian Kelly, Jean Salamero, Stephen H. Powis,
Lionel Donato, Huguette Bausinger, Michel Laforet,
Matjaz Jeras, Daniele Spehner, Thomas Bieber, Annie Falkenrodt,
Jean-Pierre Cazenave, John Trowsdale, Marie-Marthe Tongio
Human lymphocyte antigen (HLA) class
are largely dependent for expression on small peptides supplied to them by transporter
associated with antigen processing (TAP) protein. An inherited human deficiency in theTAP
transporter was identified in two siblings suffering from recurrent respiratory bacterial in-
fections. The expression on the cell surface of class
ofCD1awas normal, andthe cytotoxicity of natural killer cellswas affected. In addition, CD8+
ac Tcellswere present in lowbutsignificantnumbersandwerecytotoxic inthe mostseverely
affected sibling, who also showed an increase in CD4+CD8+ T cells and yb T cells.
I proteins of the major histocompatibility complex
I proteins was very low, whereas that
Class I molecules of the major histocompat-
ibility complex (MHC) present peptides to
CD8' T cells. These peptides derive from
proteolytic degradation in the cytosol. They
are subsequently imported by a peptide trans-
porter into the lumen of the endoplasmic
reticulum where they associate with class I
H. de Ia Salle, D. Hanau, D. Fricker, A. Urlacher, H.
Bausinger, M. Laforet, M. Jeras, T. Bieber, M.-M.
Tongio, Laboratoire d'Histocompatibilit6, Centre R6-
gional de Transfusion Sanguine, 67085 Strasbourg,
A. Kelly, S. H. Powis, J. Trowsdale, Imperial Cancer
Research Fund Laboratories, Lincoln's Inn Fields,
London WC2A 3PX, UK.
J. Salamero, INSERM U 255, Institut Curie, 75231
L. Donato, Service de P6diatrie, H6pitaux Universi-
taires de Strasbourg, H6pital de Hautepierre, 67098
D. Spehner, INSERM U 74 et Laboratoire commun
Universit6 Louis Pasteur/Synth6labo, FacultM de M6-
decine, 67085 Strasbourg, France.
A. Falkenrodt, Laboratoire d'Hematologie Normale et
Pathologique, Centre R6gional de Transfusion San-
guine, 67085 Strasbourg, France.
J.-P. Cazenave, INSERM U 311, Centre R6gional de
Transfusion Sanguine, 67085 Strasbourg, France.
*To whom correspondence should be addressed.
molecules (1). The peptide transporter is a
heterodimeric protein formed of two homolo-
gous polypeptides encoded by the TAPI and
TAP2 genes located in the MHC class II
region (2-4). In mutant cell lines that do not
express this transporter, most02-microglobu-
lin (12M)-class I heavy chain complexes do
not acquire peptides. Such "empty" complex-
es are unstable at physiological temperature
and are inefficiently transported through the
Golgi compartment. Consequently, most of
the class I heavy chains remain unsialylated.
Relatively few peptide-free class I molecules
reach the cell surface where they can be
stabilized by exogenous class I-specific pep-
tides (5). Mice in which the TAPI gene has
been disrupted by homologous recombination
have almost no detectable CD8+ T cells and
no alloreactive cells when bred in germ-free
conditions (6). In this report we describe a
human peptide transporter (TAP) genetic
A TAP deficiency was observed in family
E which is from Morocco and includes the
parents, who are first cousins, and five chil-
8 JULY 1994
dren. The first case, EFA, was a 15-year-old
female who had chronic bacterial sinobron-
chial infections but no history of viral infec-
tions. Human lymphocyte antigen (HLA)
serotyping did not detect HLA class I mole-
cules on her peripheral blood mononuclear
cells (PBMCs). A deficiency in
ruled out. A complete HLA typing of the
family (Fig. 1A) showed that two children in
the family, EFA and a 6-year-old brother,
EMO, are HLA-homozygous and express class
II but not class I molecules. The other mem-
bers of the family, who are heterozygous or
HLA different (EAH) express both class I and
class II antigens. These results suggest that the
defect is genetically linked to the MHC.
Because "absence" of class I molecules as
determined by serological typing methods
does not mean complete absence of these
molecules from the cell surface, we labeled
PBMCs with the W6/32 class I monomorphic
monoclonal antibody (mAb) and analyzed
them by flow cytometry (Fig. 1B). The fluo-
rescence intensity was 1% (T cells and mono-
cytes) to 3% (B cells) that ofnormal PBMCs.
Epstein-Barr virus (EBV)-transformed B cell
lines (named ST-XXX) were generated from
B cells from EMO and EAH. Flow cytometry
showed a 99% reduction of class I molecules
on the ST-EMO cells compared with the
ST-EAH cell line. Because both cell lines
expressed equivalent amounts of class
mRNA, detected by Northern (RNA) blot
analysis, we looked for defects in processing of
class I molecules. The time course ofpolysac-
charide side chain maturation in ST-EMO
(class I-) and SCHIOW9013, a control
HLA-A3-homozygous cell line, showed that
HLA-A3 molecules became resistant to en-
doglycosidase H (endo H) (Fig. 1C); thus,
they were transported to the cis-Golgi com-
partment. However, whereas endo H-resis-
tant molecules remained stable after a 2-hour
chase in SCH1OW9013, those of ST-EMO
decreased with time. In the extracts of ST-
EMO, 32M was poorly detected, showingthat
it was loosely associated with the class I heavy
on March 4, 2014
on March 4, 2014
on March 4, 2014
on March 4, 2014
on March 4, 2014
chains and probably dissociated during the
immunoprecipitation. Isoelectric focusing of
class I molecules immunoprecipitated from
ST-EMO with the W6/32 and HLA-A3 mAb
GAP A3 (7) showed that most class I heavy
chains remained unsialylated (Fig. 1D), sug-
gesting that the class I molecules were unsta-
ble or poorly transported toward the trans-
Golgi compartment. These observations are
reminiscent of those reported with the TAP-
deficient BM36.1 cell line (3).
These results are consistent with a defi-
ciency in the peptide transporter. Therefore,
protein immunoblot analysis of protein ex-
tracts from ST-EMO (class I-) was done (Fig.
2A) with polyclonal antibodies to TAP1 and
TAP2, generated from immunization with
synthetic peptides of the COOH-terminal
domain of the two subunits. The TAP1 but
not the TAP2 protein was detectable in the
cell extracts from ST-EMO, whereas both
were present in extracts of the normal B cell
line LCL721. Antibodies to LMP2 and LMP7
revealed normal expression of the two genet-
ically linked proteasome subunits. Thus, the
TAP2 protein is either missing or truncated,
and the deficiency of class I molecules
probably due to a mutation affecting only the
To localize the mutation, we reverse tran-
scribed RNA samples from ST-EMO (class
I-) and ST-EAH (class I'), and TAP2 com-
plementaryDNA (cDNA) was amplified with
amounts of cDNA were amplified from each
RNA with each set ofoligonucleotides, show-
ing that the defect is not due to the absence or
instability of TAP2 mRNA. The complete
sequence ofthe TAP2 cDNA in the ST-EMO
cell line was determined. A single mutation
was identified, a C to T substitution changing
the CGA Arg codon at amino acid 253 into a
TGA stop codon. This was confirmed by
direct sequencing of polymerase chain reac-
tion (PCR) amplification products from ge-
nomic DNA obtained from different members
of the family (8). The two affected children
displayed the C to T transition. The TAP2
sequence ofEAH was normal and the hetero-
zygous genotypes ofEHA and EMA included
sets of oligonucleotides. Equivalent
both sequences. Thus, the genetic defect re-
sulted from a premature stop mutation in the
To determine whetherP2M-classI heavy
chain complexes could be stabilized on the
surface of the ST-EMO (TAP2-) cells by
appropriate synthetic peptides, as observed on
TAP-deficient cells (5), we incubated ST-
EMO cells that express the HLA*0301 allele
in serum-free medium supplemented with hu-
cytometry with mAb GAP A3 showed that
addition of pn2a, but not control peptide,
stabilized the HLA-A3 molecules (Fig. 2B).
Thus, the absence of class I antigens on the
cell surface appears to be the sole consequence
of a failure to load j32M-class I heavy chain
complexes with peptides.
We assessed the effect of the TAP defi-
ciency on the expression ofclass Ib molecules
by investigating the presence of CD1a on
epidermal Langerhans cells of a patient. Re-
sults of immunochemical staining techniques
on a skin biopsy indicated that CDla mole-
or HLA-A2 (9). Flow
1 02 10o 10
102-o101 102102 1041l02101
102 103 104
a b c d e
Fig. 1. Class
tion of HLAhaplotypes. Class
11 histocompatibility antigens expressed
on PBMCs and B lymphocytes, respec-
tively, were typed by serology and DNA
methods (27). (-), Antigen-negative by
serological typing; C*, not identifiable by
serology typing (C blank). The common
DOA1*0301, DQB1*0302, DPB1*0301. (B) Expression of class
cules on PBMCs. The PBMCs from EHA (TAP2+'-) and from EMO
(TAP2-'-) were double labeled with mAb W6/32 and mAbs to CD14
(monocyte specific), CD19 (B cell-specific), or CD3 (T cell-specific) and
analyzed by flow cytometry. Similar results as for EMO were obtained for
EFA. The y-axis data are for the antibody named in the upper left, and the
x-axis data are for W6/32. (C) Endo H resistance of HLA-A3 heavy chains
and stability of HLA-A3 molecules. Homozygous HLA-A3 SCHUl OW9013
EBV cells (lanes a to e) and ST-EMO (TAP2-) cells (lanes f to j) were
labeled 30 min with 35S-methionine and chased with cold methionine for
0 (a and f), 1 (b and g), 2 (c and h), 3 (d and i), and 4 hours (e and j). Cell
extracts were immunoprecipitated with HLA-A3 mAb (GAP A3), treated
(+) or not treated (-) with endo H, and analyzed by SDS-PAGE. Similar
deficiency. (A) Segrega-
patterns of endo H resistance were obtained after immunoprecipitation
with W6/32. Arrow indicates 32M. (D) Defect in sialylation of class heavy
chains. ST-EMO cells (lanes a, d, and i), HLA-A3+ (SCHU1OW9013 cell
line) (lanes b, e, g, and h), and B63+ (peripheral blood lymphocytes from
a normal donor) (lanes c and f) control cells were labeled for 4 hours with
35S-methionine, and class molecules wereimmunoprecipitatedwith mAb
W6/32 (lanes a to f) or GAP A3 (lanes g, h, and i), treated (+) or not
treated (-) with neuraminidase, and analyzed by gel isofocusing electro-
phoresis (28). Sialylated heavy chains are indicated by dots, unsialylated
heavy chains by arrows. (E) Induction of the expression of CD1a on
monocyte-derived dendritic cells. Monocytes from EMO (histogram 1)
and a normal donor (histogram 2) were isolated by plastic adherence and
incubated 7 days
expression of CD1 a was determined by cell cytometry.
in the presence of GM-CSF and IL-4 (29), and
8 JULY 1994
8qreb of HLA alWi~s
A30 B18 CwS
DR7 DR53 DO2
DR4 DR53 DO3
DR4 DR53 D3
DR4 DRS3 D3
DR4 DR53 D3
cules were normally expressed, whereas no
class I molecules could be detected on the
whole skin section (10). Moreover, expres-
sion of CD1a molecules could be induced at
normal levels on dendritic cells obtained by
differentiation of monocytes with granulo-
and interleukin4 (IL-4) (Fig. 1E). Thus, in
contrast to HLA class I molecules, the class Ib
CD1a molecule does not require TAP for
Blood cell populations of EFA and EMO
were analyzed by flow cytometry and cytotox-
icity tests. Flow cytometry showed that the
numberofCD3-CD16+CD56' natural killer
(NK) cells was normal in both children as was
the number of CD8' NK cells (Table 1).
However, no cytotoxic activity against the
class I-negative K562 cell line was observed
(Fig. 3A). Thus, this class I deficiency does
not affect the formation of CD8' NK cells
but, like the murineP2Mdeficiency (11),
affectsNK activity. T cell population numbers
were compared with the numbers observed in
normal subjects (12). Normal numbers of
CD4+CD8- T cells were observed but, con-
Flg. 2. (A) Protein immu-
noblot analysis of MHC-
from the ST-EMO
line (lanes 2, 4, 6, and 8)
and from the LCL721 normal B cell line (lanes 1, 3,5, and
7) wereanalyzed (30). Carboxyl-terminal peptide antise-
ra for TAP1 (3) (lanes 1 and 2), TAP2A (lanes 3 and 4),
TAP2B (4) (lanes 5 and 6), and LMP2 and LMP7 antisera
(lanes 7 and 8) were used. LCL721 expresses both
TAP2A and TAP2B alleles (4). (B) Stabilization by HLA-
A3 peptides. ST-EMO (TAP2-) cells were incubated for
18 hours at 370C in serum-free medium supplemented
with humanP2M (10pug/ml)and pn2a (KLYEKYIYK) or
an HLA-A2-specific negative control peptide (GLF-
GGGGGV) (124gg/ml) (9). The HLA-A3 homozygous SCHU1OW9013 cell line was similarly incubated
without peptide. Expression of HLA-A3 was followed by flow cytometry with the mAb GAP A3. Shaded
curve:isotypic control. The single letter abbreviations for amino acid residues are as follows: E, Asp; F,
Phe; G, Gly; K, Lys; L, Leu; T, Thr; V, Val; and Y, Tyr.
1 23 4 5 67 8
Allogenic cytotoxity ot
Fig. 3. (A) Cytotoxic activity of NK
cells and (B) CD8+ T cells. (A) Cyto-
toxic activity of PBMCs from two nor-
mal donors (open symbols) or from
the two siblings were tested against
the class I-negative K562 cell line
with different effector to target (E:T)
ratios. (B) The upper section shows
the cytotoxicity of PBMCs from EFA,
EMO, and two healthy controls, FRO
and KOU, when stimulated with allo-
geneic irradiated PBMCs from two
unrelated donors, FLA and MAR, and
tested against the same targets be-
fore or after depletion of CD8+ T cells
(31). The lower section shows the
allogeneic cytotoxicity from PBMCs from EFA against MAR before (black squares) or after (open
squares) depletion of an or yb T cells. Percentage of specific lysis obtained for effector to target ratios
(E:T) (30:1, 15:1, and 5:1) is given. Autologous specific lysis was <10% (32).
Depletion ofy6 T cells
trary to the TAP1- or the I2M-deficient
11), CD8+CD4- T cells were
found in PBMCs from EMO (7% ofT cells,
normal range 30%) and EFA (20% of T
cells). The peripheral blood of EFA con-
tained high numbers ofCD4+CD8' double-
positive T cells (10% of T cells, normal
range <2%), which have been observed in
only a few individuals (13). The presence of
these CD4+CD8' T cells may be the con-
sequence of a thymic failure, but their ab-
sence in the PBMCs of the younger brother
healthy subjects (14), a high proportion of
y&-positive T cells was observed in EFA
(33% T cells, normal range 5 to 10%),
among which 33% were CD8' and 1%
CD4+. Lower numbers of yi T cells were
observed in EMO, but one-third were also
CD8+. Thus, the higher number ofCD8+ T
cells in EFA results from an expansion ofy&
T cells, the numbers of cad CD8+ T cells
being low in the two siblings. Because the
percentage of CD8' y& T cells was normal
(14), their development seems not to be as
strictly dependent on the expression of class
I antigens, unlike the positive selection of
gens, such as CD1a, that are expressed
normally in these TAP-deficient individuals,
may be involved in their selection. The
expansion of y5 T cells in EFA may be a
consequence of her lung disease or may
compensate for the low numbers of CD8+
aj3 T cells.
Functions ofhelperT cells and cytotoxicT
cells were tested in mixed lymphocyte culture
(MLC) and cell-mediated lympholysis (CML)
assays, respectively. In MLC tests, EFA and
tI3 T cells. Nonclassical class I anti-
Table 1. Lymphocyte subpopulations from pe-
ripheral blood. The PBMCs were isolated on
Ficoll Hypaque, labeled with mAb, and ana-
lyzed on FACSort (Becton-Dickinson). Results
are expressed as percent of lymphocytes. The
number of lymphocytes in blood were 2.1 x 106
per milliliter (EMO) and 3.9 x 106 per milliliter
(EFA) (standard values are 1.9 to 4.8 x 106 per
apT cell receptor
y5 T cellreceptor
8 JULY 1994
EMO lymphocytes had normal proliferation
indexes when cultured with unrelated stimu-
latory cells carrying different class II types,
suggesting that their CD4' T cells were nor-
mal (15). In contrast, EFA and EMO T cells
differed in CML tests, and unlike TAP1
knock-out mice, an allogeneic cytotoxicity
activity from CD8' cells was observed in
PBMCs from EFA (Fig. 3B). Depletion ofoad
or yb T cells in PBMCs from EFA demon-
strated that the allogeneic response was me-
diated by CD8' otl T cells but not by -y T
cells. Local development of cytotoxic T cells
was observed in 12M-deficient mice after in-
traperitoneal injection of allogeneic MHC
class I-positive cells (16), and CD8' T cells
appeared in the spleens after viral infections
(17) or rejection of skin grafts (18). The
presence of these CD8+ T cells may be
explained by low cell surface expression of
class I molecules (11), which may allow the
positive selection ofT cells expressing T cell
receptors with a high affinity for the class I
molecules (19). Our observations show that,
similar to the class I 12M-deficient mice,
CD8+ T cells can develop in a peptide trans-
porter-deficient background where cell sur-
face expression ofclass I molecules is low. The
differences between the TAP- humans and
TAP- mice in this regard could be that the
two children had more extensive exposure to
pathogens than the laboratory animals did. In
agreement with this hypothesis, EFA, the
eldest child, had developed four times more
an CD8+ T cells than her brother.
With a low number ofCD8+ oa3 T cells,
the immune response is likely to be less
efficient. Nevertheless, antibody titers in se-
rum show that the patients were infected by
several viruses. Indeed, although they have
never been vaccinated against herpes, mea-
sles, varicella, mumps, or cytomegalovirus,
EFA had high titers of antibodies against all
of these viruses and EMO had antibodies
against the first three of them (20). These
observations suggest that antibody-depen-
dent responses may be essential in the anti-
viral defense of these patients. NK cells,
although they are unable to lyse the K562
cells, may also be involved because complete
absence of NK cells may lead to severe
infections by herpes, cytomegalovirus, or
varicella (21), a pathology not observed in
the two siblings. Finally, because antigen
presentation in peptide transporter-deficient
cell lines occurs in some cases (22), the
significant numbers of CD8+ T cells that
persist may still play a role.
HLA class I deficiency, the so-called type I
bare lymphocytes syndrome (BLS), is a rare
disease that, in most of the cases described so
far, has not been linked to the MHC (23,
24). BLS may be lethal in early childhood,
but in some instances, such as these two
individuals with a TAP transporter deficien-
cy, the disease may manifest itself later in life.
EFA suffered from chronic colonization of
lung by bacteria. EMO was initially healthy
for 6 years until he presented with a pulmo-
nary impairment similar to his sister. A similar
pathology has been observed in cases with an
HLA class I deficiency, where the expression
of class II molecules was not clearly deter-
mined (24, 25). This pathology may be an
indication of the physiological significance of
the presentation ofbacterial antigens by class
I molecules that occurs in macrophages (26).
Why the pathology is restricted to lungs is not
understood: it may reflect an important role of
macrophages in this tissue. Moreover, previ-
ous viral infections may have injured lung
tissue and favored susceptibility to bacterial
pathogens. This last hypothesis is compatible
with the observation that the TAP deficiency
may go unnoticed in early childhood and
develop later, depending on the incidence of
viral infections. These observations may be
useful in identifying other cases of class I
deficiency associated with absence of TAPs
and in characterizing the immune responses
that are invoked to compensate for defects in
REFERENCES AND NOTES
1. J. J. Neefjes, F. Momburg, G. J. Hammerling,
Science 261, 769 (1993); J. C. Shepherd et al.,
Cell 74, 577 (1993).
2. S. Bahram, D. Arnold, M. Bresnahan, J. L. Stromin-
ger, T. Spies, Proc. NatI. Acad. Sci. U.S.A. 88,10094
(1991); S. Beck etal., J. Mol. Biol. 228, 433 (1992);
J. J. Monaco, Immunol. Today 13, 173 (1992).
3. A. Kelly et al., Nature 355, 641 (1992).
4. S. Powis et al., Proc. Nati. Acad. Sci. U.S.A. 89,
5. A. Townsend et al., Nature 340, 443 (1989); H.-G.
Ljunggren etal., ibid. 346, 476 (1990); K. L. Rock, C.
Gramm, B. Benacerraf, Proc. NatI. Acad. Sci. U.S.A.
88, 4200 (1991); G. R. Otten etal., J. Immunol. 148,
6. L. Van Kaer, P. G. Ashton-Rickardt, H. L. Ploegh,
S. Tonegawa, Cell 71, 1205 (1992).
7. A. E. Berger, J. E. Davis, P. Creswell, Hybridoma
1, 87 (1982).
8. Genomic DNA from different members of the
family were amplified with the use ofAGCTGAAC-
TCACGGCTGAGCT and CACCTTCTCCGCTGC-
TATTGT oligonucleotides and sequenced.
9. M. DiBrino et al., Proc. Nati. Acad. Sci. U.S.A. 90,
10. D. Hanau and T. Bieber, unpublished observa-
tions. Skin sections were fixed and stained as
described [G. F. Murphy, A. K. Khan, T. J. Harrist,
M. C. Mihm, Br. J. Dermatol. 108, 423 (1983); T.
Bieber et al., J. Invest. Dermatol. 93, 215 (1989)].
11. Reviewed in D. H. Raulet, Adv. Immunol. 55, 381
M. DeBruyere, Immunol. Today 13, 215 (1992).
13. S. Furukawa et al., Blood 80, 3253 (1992); P.
Sala, E. Tonutti, C. Feruglio, F. Florian, A. Colom-
batti, ibid. 82, 1546 (1993).
14. V. Groh etal., J. Exp. Med. 169, 1277 (1989).
15. M.-M. Tongio, unpublished observations.
16. S. Apasov and M. Sitkovsky, Proc. Nati. Acad. Sci.
U.S.A. 90, 2837 (1993).
17. P. C. Doherty, Curr. Opin. Immunol. 5, 479 (1993).
18. M. Zijistra et al., J. Exp. Med. 175, 885 (1992); E.
Lamous6-Smith, V. K. Clements, S. Ostrand-Ro-
senberg, J. Immunol. 151, 6283 (1993).
19. R. Glas, C. Ohlen, P. Hoglund, K. Karre, J. Exp.
Med. 179, 661 (1994).
20. H. de la Salle, unpublished observations.
21. C. A. Biron, K. S. Byron, J. L. Sullivan, Nat. Immun.
I. Hannet, F. Erkeller-Yuksel, P. Lydyard, V. Deneys,
8 JULY 1994
Cell Growth Regul. 7, 47 (1988); C. A. Biron, K. S.
Byron, J. L. Sullivan, N. Engi. J. Med. 320, 1731
22. Reviewed in M. T. Heemels and H. Ploegh, Curr.
Biol. 3, 380 (1993); S. A. Hammond, R. C. Bollinger,
T. W. Tobery, R. F. Siciliano, Nature364, 158 (1993).
23. J. L. Touraine, H. Betuel, G. Souillet, M. Jeune, J.
Pediatr. 93, 47 (1978); R. K. B. Schuurman etal.,
Clin. Immunol. Immunopathol. 14, 418 (1979); R.
Payne, F. M. Brodsky, B. M. Peterlin, L. M. Young,
Hum. Immunol. 6, 219 (1983); A. Marcadet et al.,
N. Engl. J. Med. 312,1287 (1985).
24. H. Maeda et al., Immunogenetics 21, 549 (1985).
25. Y. Sugiyama, H. Maeda, K. Okumara, F. Takaku,
Chest 89, 398 (1986); Y. Sugiyama, S. Kudo, S.
Kitamura, Nippon Kyobu Shikkan Gakkai Zasshi
27, 980 (1989).
26. J. D. Pfeifer etal., Nature 361, 359 (1993); M. G.
Rittig, T. Haupl, G. R. Burmester, Int. Arch. Allergy
Immunol. 103, 4 (1994).
27. Serological typing was performed as described [P.
I. Terasaki and J. D. MacClelland, Nature 204, 998
(1964); F. Vardal et al., Tissue Antigens 28, 301
(1986)]. Molecular biology typing was determined
as described [J.-M. Tiercy, C. Goumaz, B. Mach, M.
Jeannet, Transplantation51, 1110 (1991); A. Kimura
and T. Sasazuki, in HLA 1991, K. Tsuji, M. Aizawa, T.
Sasazuki, Eds. (Oxford Univ. Press, Oxford, 1992),
vol. 1, pp 397-419; H. Zetterquist and 0. Olerup,
Hum. Immunol. 34,64 (1992); 0. Olerup, A. Aldener,
A. Fogdell, Tissue Antigens 41, 119 (1993); M. J.
Browning et al., Proc. Natl. Acad. Sci. U.S.A. 90,
2842 (1993)]. The A*3002 subtype was determined
by sequencing PCR fragments with the use of 5'-
CGGAATGTGAAGGCCCAC-3' and 5'-TCTCAACT-
GCTCCGCCCG-3' oligonucleotides. The cDNA of
HLA-C was amplified with the use of 5'-CATOT-
CAGGGTGAGGGGCTC-3' and 5'-CCCACTCCAT-
GAGGTATTTC-3 oligonucleotides and sequenced.
This typing showed that both parents expressed two
complete sets of class
bility molecules and shared a common HLA haplo-
type. EFA and her 6-year-old brother, EMO, were
homozygous for this haplotype and expressed only
haplotype (EMA and EKA), or carrying the other
parental haplotypes (EAH), expressed both class
and class 11 molecules. The HLA-C antigen of the
common haplotype was not identifiable by serology
(HLA-C blank) but was determined by DNA se-
number of the sequence in National Center for
Biotechnology Information, U07230. The subtype is
identical to HLA-Cw14*1401 except for two amino
acid substitutions (Thr to Ser and Val to Met at
position 178 and 248). DNA methods were used to
confirm serological typing and demonstrated that
class 11 genes and HLA-A alleles common to the two
siblings were inherited without any recombination.
28. J. J. Neefjes, B. S. Breuer-Vriesendorp, G. A. Van
Seveter, P. Ivanyi, H. Ploegh, Hum. Immunol. 16,
29. S. Porcelli, C. T. Morita, M. B. Brenner, Nature
360, 593 (1992).
30. Total NP-40 cell lysateswere preparedfrom 5 x 1 4
cells and separated by SDS-polyacrylamide gel
electrophoresis (SDS-PAGE) in reducing buffer. Pro-
teins were transferred to Hybond C Extra nitrocellu-
lose and incubated with approximately 1:5000 dilu-
tions of the relevant primary antisera and a 1 :1000
swine antibody to rabbit immunoglobulin G as the
second antibody. Enhanced chemiluminescence
(Amersham) was used for detection. Antibodies
against LMP2 and LMP7 are described in A. Kellyet
al. [Nature 353, 667 (1991)] and R. Glynne et al.
[Eur. J. Immunol. 23, 860 (1993)].
31. Natural killer and allogeneic cytotoxicitieswere as-
sayed as described in W. E. Seaman et al., J. Clin.
Invest. 67, 1324 (1981) and in the Report from the
European CML Study Groupon the ThirdEuropean
CML Workshop, Tissue Antigens 16, 335 (1980),
respectively. Depletionsof CD8+ T cells were done
withimmunomagneticbeads coated with antibod-
ies to CD8. To depletea(3 and yb T lympho-
cytes,we incubated cells with relevant mAbs and
and class 11 histocompati-
11 antigens. Children heterozygous for this
is available, accession
then with antibodies to murine IgG-coated immuno-
magnetic beads. Depletions were confirmed by flow
32. H. de la Salle et al., data not shown.
33. We thank P. Creswell for mAb GAP A3, R. Dujol for
photography, and P. Binnert, N. Froelich, B. Pfeif-
fer, A. Rutz, and A. Schell for technical assist-
ance. Supported by INSERM (CRE 930606) and
the Centre Regional de Transfusion Sanguine de
13 January 1994; accepted 16 May 1994
Gem: An Induced, Immediate Early Protein
Belonging to the Ras Family
Jean Maguire,* Thomas Santoro, Peter Jensen, Ulrich Siebenlist,
John Yewdell, Kathleen Kellyt
A gene encoding a 35-kilodalton guanosine triphosphate (GTP)-binding protein, Gem, was
cloned from mitogen-induced human peripheral blood T cells. Gem and Rad, the product of
a gene overexpressed in skeletal muscle in individuals with Type 11 diabetes, constitute a new
familyofRas-related GTP-binding proteins. The distinct structuralfeaturesofthisfamily include
the G3 GTP-binding motif, extensive amino- and carboxyl-terminal extensions beyond the
Ras-related domain, and a motifthatdetermines membrane association. Gemwastransiently
expressed in human peripheral blood T cells in response to mitogenic stimulation; the protein
was phosphorylated on tyrosine residues and localized to the cytosolic face of the plasma
membrane. Deregulated Gem expression prevented proliferation of normal and transformed
3T3 cells. These results suggest that Gem is a regulatory protein, possibly participating in
receptor-mediated signal transduction at the plasma membrane.
Genes that are transcribed early after mito-
genic activation of resting cells are thought
to be crucial for subsequent cell proliferation
and expression of differentiated
functions. We have cloned mitogen-induced
genes from human peripheral blood T cells
on the basis of differential expression of
mRNA between resting and stimulated cells
(1). One such clone, pAT 270, is now
shown to encode a protein we have termed
Gem because it binds GTP and is induced by
mitogens. Human Gem is encoded by a
single copy gene (2), and its 2127-base pair
(bp) complementary DNA (cDNA) was
similar in size to the corresponding mRNA
(2200 bp) and predicted an open reading
frame of 296 amino acids (Fig. 1). The Gem
protein contains a core sequence (amino
acids 75 to 240) that is highly related to
members of the Ras superfamily of small
GTP-binding proteins; the flanking NH2-
and COOH-terminal sequences are unrelat-
ed to Ras. Gem is most closely related to Rad
(-60% identity) (Fig. 1), a protein encoded
by a gene that is overexpressed in skeletal
muscle from individuals with Type II diabe-
tes relative to skeletal muscle from normal or
J. Maguire, T. Santoro, P. Jensen, K. Kelly, Laboratory
of Pathology, National Cancer Institute, National Insti-
tutes of Health, Bethesda, MD 20892, USA.
U. Siebenlist, Laboratory of Immunoregulation, Nation-
al Institute of Allergy and Infectious Diseases, National
Institutes of Health, Bethesda, MD 20892, USA.
J. Yewdell, Laboratory of Viral Diseases, National
Institute of Allergy and Infectious Diseases, National
Institutes of Health, Bethesda, MD 20892, USA.
*Present address: Department of Pathology, Universi-
ty of Chicago, Chicago, IL 60637, USA.
tTo whomcorrespondenceshould be addressed.
Type I diabetic individuals (3). The greatest
similarity between Gem and Rad exists in
regions that correspond to the guanine nu-
cleotide-binding domains of Ras. Gem and
Rad diverge in the putative effector, or G2,
region, suggesting that they interact with
Gem initiates from the first ATG codon
(nucleotide 175) whereas Rad has been pre-
dicted to initiate from an intemal ATG. The
predicted start site for Gem was confirmed by
immunoprecipitation with antibodies to the
predicted NH2-terminus and by in vitro tran-
scription and translation analyses (2).
Mutational analyses of GTP-binding pro-
teins and crystallographic studies of the H-ras
oncogene product have defined regions of
sequence consensus that interact with various
positions of the guanine nucleotide (4-6).
The guanine specificity consensus sequences
NKXD and EXSA (X represents any amino
acid) are perfectly conserved in Gem, with
appropriate spacing, as NKSD (residues 191
to 194) and ETSA (residues 219 to 222),
GXXXXGK, which participates in interac-
tions with the a andPphosphates of the
guanine nucleotide, is also conserved in Gem
as the sequence GEQGVGK (residues 82 to
Gem contains an unusual motif in the
G3 (DXXG) region, which putatively par-
ticipates in binding and hydrolysis of the
GTP y phosphate
ENKG (residues 134 to 137), contains the
invariant glycine residue but has a conser-
(5). The sequence,
Fig. 1. Predicted amino
acid sequences encod-
ed by cDNAs for human
and murine Gem (H- and
and comparison to hu-
man Rad (3) and human
open reading frame was
determined for two inde-
pendent human and a
cDNA clone. Amino ac-
ids conserved in at least
three proteins are in bold
teins are bold and un-
gaps inserted to allow
for optimal alignment of
bers on the right indicate
GTP-binding regions are
indicated in italics (4, 5).
amino acid residues are
as follows: A, Ala; C,
Cys; D, Asp; E, Glu; F,
Phe; G, Gly; H, His; I lle;
K, Lys; L, Leu; M, Met; N,
Asn; P, Pro; 0, Gin; R, Arg; S. Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. GenBank accession numbers
for the human and murine Gem cDNA sequences are 010550 and U10551, respectively.
M-Gem EEHCRRTWSSDSTDSVIS ..SESGNTYXRVVLIaEOC
..............................EXKLVVV9 Gf*~M TIQL
M-Gem HDSMDSDCEVLGEDTXERTLVVEMSATIILWJIWZNK0ENZ. .WLHDHC
D X, T
all four pro- G4
Rad MAMOpAYVIXSVTDKGS ASZZLVQ9LUAR4QDjVZIIaaU.SDLV
c-H-Rasl .AIT2ESRQAQDL&RSYGIPY KTRQGGEDAZYTL3WEIZHKLRE
Rad LA ARQATRRT8SLGKA
8 JULY 1994