(Received for publication, October 8, 1996, and in revised form, December 2, 1996)
From the Centre d'Immunologie INSERM/CNRS de
Marseille-Luminy, Case 906, 13288 Marseille Cedex 09, France,
** Laboratoire d'Immunologie Cellulaire et Clinique, INSERM U255,
Institut Curie, Paris, 75 231 France, § Laboratoire
d'Immunologie Cellulaire, CNRS URA625, CERVI, Groupe hospitalier
Pitié-Salpêtrière, 75013 Paris, France,
¶ Istituto Nazionale per la Ricerca sul Cancro, Via R. Benzi, 10, 16132 Genova, Italy, and
Istituto di Istologia ed Embriologia
Generale, University of Genova, Genova, 16 132 Italy
Natural killer and T cells express at their
surface, members of a multigenic family of killer cell inhibitory
receptors (KIR) for major histocompatibility complex Class I molecules.
KIR engagement leads to the inhibition of natural killer and T cell
activation programs. We investigated here the functional reconstitution
of KIR in a non-lymphoid cell type. Using stable transfection in the
RBL-2H3 mast cell line, we demonstrated that (i) KIR can inhibit signals induced by FcRI
or CD3
polypeptides that bear
immunoreceptor tyrosine-based activation motifs; (ii) two distinct
immunoreceptor tyrosine-based inhibition motifs-bearing receptors,
i.e. KIR and Fc
RIIB, use distinct inhibitory pathways
since KIR engagement inhibits the intracellular Ca2+
release from endoplasmic reticulum stores, in contrast to Fc
RIIB, which only inhibits extracellular Ca2+ entry; (iii) KIR
require co-ligation with an immunoreceptor tyrosine-based activation
motif-dependent receptor to mediate their inhibitory function. This latter finding is central to the mechanism by which KIR
selectively inhibit only the activatory receptors in close vicinity.
Taken together our observations also contribute to define and extend
the family of immunoreceptor tyrosine-based inhibition motif-bearing
receptors involved in the negative control of cell activation.
NK1 cells are cytotoxic lymphocytes that can induce the lysis of target cells by two mechanisms (1-3). Antibody-dependent cell cytotoxicity leads to the lysis of antibody-coated target cells, whereas natural cytotoxicity leads to the antibody-independent lysis of a variety of cell targets, including primarily virus-infected cells and tumor cells. Induction of both natural cytotoxicity and antibody-dependent cell cytotoxicity programs is controlled by the engagement of killer cell inhibitory receptors (KIR) for MHC Class I molecules; recognition of MHC Class I molecules expressed on target cell surface by KIR induces a general inhibition of both NK cell cytotoxic programs (4-8). When expressed on T cells, KIR also inhibits CD3/TCR-dependent activation programs (9, 10).
Similarly, a class of low affinity Fc receptors, FcRIIB, has been
shown to inhibit B cells and T cells as well as mast cell activation
(11-15). The molecular dissection of Fc
RIIB intracytoplasmic domain
revealed the inhibitory function of a 13 amino acid ITIM (11-15).
Within this sequence, we and others have reported the conservation of a
(I/V)XYXX(L/V) stretch in human, as well as mouse, KIR and Fc
RIIB (16, 17). Upon tyrosine phosphorylation of the
common (I/V)XYXX(L/V) motif, both KIR and
Fc
RIIB bind the SH2-tandem protein-tyrosine phosphatases SHP-1 and
SHP-2, which are reported to be involved in the inhibitory pathways
used by KIR and Fc
RIIB (15-19). These data prompted us to further
investigate the mechanisms by which KIR exert their inhibitory function
and may thus be defined as members of the family of ITIM-bearing
receptors involved in the attenuation of hematopoietic cell
activation.
In this attempt, we reconstituted an HLA-Cw3-specific KIR (p58.2) in a
transfected mast cell/basophil-like RBL-2H3 cell line (RTIIB). RTIIB
cells express two distinct ITAM-dependent receptors: the
endogenous FcRI antigen receptor and a transfected CD25/CD3
chimeric molecule, as well as the murine Fc
RIIb2 ITIM-bearing receptor (12, 13). We also reconstituted a naturally occurring activatory form of p58.2, the p50.2 KAR (killer cell activatory receptor), in RTIIB cells. The p50.2 KAR expresses a shorter
intracytoplasmic domain, which does not contain any
(I/V)XYXX(L/V) stretch. This activatory receptor
has been reported to trigger T and NK cell activation programs (20,
21).
The model of KIR and KAR reconstitution in RTIIB cells presents several
useful features. First, RTIIB cell activation induced by FcRI as
well as by CD25/CD3
receptors is inhibited by Fc
RIIB (12, 13).
Second, RTIIB, NK, and T cells express multimeric receptor complexes,
that are built according to a similar architecture. Indeed, Fc
RI,
the antibody-dependent cell cytotoxicity receptor complex
(Fc
RIIIA), and the CD3·TCR complexes include associated polypeptides that are necessary and sufficient for their cell surface
expression as well as their transducing properties. Human Fc
RIIIA
includes CD16 in non-covalent association with CD3
and FcR
(Fc
RI
) homo- and heterodimers (22-25). On T cells, the vast majority of CD3·TCR complexes include CD3
homodimers, but
TCR
+ T cells, CD3/TCR+ large granular
lymphocytes, and subsets of CD8
intraepithelial lymphocytes
express FcR
in homodimers or in heterodimers with CD3
(26-28).
Therefore, the effect of KIR on CD25/CD3
- and Fc
RI-signaling pathways in RTIIB cells can be directly compared with the effect of KIR
on CD3/TCR- as well as antibody-dependent cell cytotoxicity receptor complex-mediated lymphocyte activation. Third, RTIIB, NK
cells, and cytotoxic T lymphocyte express intracytoplasmic granules
that undergo regulated exocytosis upon ITAM-dependent receptor aggregation. Intracytoplasmic granules expressed in NK and T
lymphocytes contain an array of molecules involved in
lymphocyte-mediated cell cytotoxicity, such as perforin and granzymes
(29). Although lymphocyte-mediated cell cytotoxicity has not been
formally proven to depend upon regulated exocytosis (30), the drastic
reduction of both NK and T cell-mediated cytotoxicity in perforin
/
mice supports the fact that regulated exocytosis is involved in
lymphocyte-mediated cell cytotoxicity (31). In this regard, the action
of KIR on serotonin release in RTIIB cells might be relevant to the
inhibition of lymphocyte-mediated cell cytotoxicity by KIR.
Our results show that (i) human KIR reconstituted in RTIIB cells
inhibit CD3- as well as Fc
RI-induced activation; (ii)
transcription-factor independent effector cell responses,
i.e. intracytoplasmic calcium ([Ca2+]i) increase and serotonin release, can be
inhibited by KIR in RTIIB cells, whereas (iii) KAR are not functional
in the same cells, and finally (iv) KIR inhibitory function requires KIR and ITAM-containing receptor cross-linking. These results also
provide new insights on the importance of the elements of the signaling
cascade for the functions of a novel receptor family, characterized
either by intact intracytoplasmic ITIM (Fc
RIIB and KIR) or by their
mutated version (KAR).
The following mAb and antisera were obtained
from Immunotech, Marseille, France: mouse anti-human CD25 mAb (B1.49.9,
IgG2a), rat anti-human CD25 mAb (33B3.1, IgG2a), mouse anti-human p58.2 mAb (GL183, IgG1), mouse anti-rat Ig (H + L) F(ab)2 (MAR),
goat anti-mouse Ig (H + L) F(ab)
2 (GAM), donkey anti-mouse
Ig (H + L) F(ab)
2 (DAM), and donkey anti-rat Ig (H + L)
F(ab)
2 (DAR). DAM and DAR are mouse Ig- and rat Ig-
specific, respectively. Rat IgE mAb (LO-DNP-30), mouse IgE mAb
(2682-I), rat anti-Fc
RII/III mAb (2.4G2) and mouse anti-CD25 mAb
(7G7, IgG1) have been described earlier (13). The mouse anti-rat
Fc
RI
(BC4, IgG1) was a generous gift from Drs. R. Siraganian
(National Institutes of Health, Bethesda, MD) and M. Benhamou (CERVI,
Paris). Mouse IgE mAb was used as a dilution of hybridoma supernatants.
All other mAbs were used as protein A/G-purified mAb. GL183, 2.4G2, and
7G7 mAb were used as F(ab)
2; LO-DNP-30, 2682-I, B1.49.9,
and 33.B3.1 were used as intact mAbs. LO-DNP-30 and 2682-I are directed
against DNP and TNP. MAR F(ab)
2 was trinitrophenylated
using trinitro-benzene sulfonic acid (Eastman Kodak Co.), and 1 mol of
MAR F(ab)
2 was substituted with an average number of 10 TNP mol. This TNP-F(ab)
2 MAR was used to cross-link mouse
anti-TNP IgE and rat anti-Fc
RII/III 2.4G2.
All cells were cultured in Dulbecco's modified
Eagle's medium supplemented with 10% FCS and penicillin (100 IU/ml)
and streptomycin (100 µg/ml). RBL-2H3 cell transfectants expressing
murine FcRIIb2 and CD25/CD3
chimeric molecules (RTIIB) have been
previously described. The CD25/CD3
chimeric molecule includes the
complete human CD25 ecto- and transmembrane domains linked to the
complete mouse CD3
intracytoplasmic domain (32). RTIIB cells were
transfected by electroporation using either the 183.6 cDNA-encoding
p58.2 (RTIIB.p58) or the 183.Act1 cDNA-encoding p50.2 (RTIIB.p50),
in the RSV-5.gpt expression vector (33). Stable transfectants were established by culture in the presence of xanthine (250 µg/ml), hypoxanthine (13.6 µg/ml), and mycophenolic acid (2 µg/ml). Two representative clones of each transfection series (RTIIB.p58A, RTIIB.p58B and RTIIB.p50A, RTIIB.p50B) were examined in parallel and
gave similar results. Unless indicated, results from one clone of each
transfection series are shown.
The primary mAb was incubated with cells on ice for 20-30 min, followed by three washes with phosphate-buffered saline supplemented with 0.2% bovine serum albumin. The secondary staining was performed using fluorescein-conjugated rabbit anti-mouse IgG (Immunotech), followed by three phosphate-buffered saline/0.2% bovine serum albumin washes and resuspension in PBS containing 1% formaldehyde. Cells were analyzed on a FACS-Scan apparatus (Becton Dickinson, Mountain View, CA).
Single-cell Ca2+ Video ImagingCells (2 × 105/sample) were allowed to adhere overnight onto glass
coverslips in culture medium. Adherent cells were washed and then
incubated at 37 °C for 40 min in RPMI 1640 medium supplemented with
10% FCS with a 103 dilution of mouse IgE (2682-I) in the
presence or absence of either 1 µg/ml GL183 or 1 µg/ml 2.4G2. 1 µM Fura-2/AM (Molecular Probes, Inc., Eugene, OR) in
dimethyl sulfoxide premixed with 0.2 mg/ml pluronic acid (Molecular
Probes) was then added to the medium for 20 min. Cells were washed, and
then measurements of intracellular Ca2+
([Ca2+]i) mobilization were performed at 37 °C
in MS buffer (140 mM NaCl, 5 mM KCl, 10 mM HEPES, pH 7.4, 1 mM CaCl2, 1 mM MgCl2) with a Nikon Diaphot 300 microscope
and an IMSTAR imaging system as described previously (34). Briefly,
each [Ca2+]i image (taken every 6 s) was
calculated from the ratio of the average of four fluorescence images
after 340 nm excitation and four fluorescence images after 380 nm
excitation. Calibrations were done as previously reported (35).
[Ca2+]i values were calculated according to
Grynkiewicz et al. (36). The stimulation was done by adding
GAM F(ab)
2 or TNP-F(ab)
2 MAR to the MS buffer
for a final concentration of 50 µg/ml and 10 µg/ml, respectively.
The measurements of intracellular calcium store release were done by
replacing the MS buffer with a stimulation buffer (150 mM
NaCl, 5 mM KCl, 1 mM MgCl2, 0.5 mM EGTA, 20 mM HEPES, pH 7.3, 50 µg/ml F(ab)
2
GAM) at the time of stimulation. For each experiment, results were
shown as the average variation of [Ca2+]i
(nM) as a function of time for a population of 20-30 cells.
RTIIB cell transfectants harvested using
trypsin EDTA were examined for serotonin release as described (12).
Briefly, cells were resuspended in RPMI 1640 medium supplemented with
10% FCS at 1 × 106 cells/ml, and incubated at
37 °C for 1 h with 2 µCi/ml [3H] serotonin
(Amersham Corp., Les Ulis, France), washed, resuspended in RPMI 1640 medium, 10% FCS, incubated for another hour at 37 °C, washed again,
resuspended in the same medium and distributed in 96-well microculture
plates at 2 × 105 cells/well. Cells were then
incubated for 1 h with mouse or rat IgE, with mouse or rat
anti-CD25 mAb in the absence or in the presence of GL183
F(ab)2, in a final volume of 50 µl. Cells were washed
three times in 200 µl of phosphate-buffered saline and once in 200 µl of RPMI 1640 medium, 10% FCS, and then 25 µl of culture medium
were added to each well and cells were warmed for 15 min at 37 °C
before challenge. Cells were challenged for 30 min at 37 °C with 25 µl of prewarmed GAM, DAM, or DAR F(ab)
2 as indicated.
Reactions were stopped by adding 50 µl of ice-cold medium and by
placing plates on ice. 50 µl of supernatants were mixed with 1 ml of
emulsifier safe scintillation liquid (Packard Instruments, Groningen,
The Netherlands) and counted in a LS6000 Beckman counter. The
percentage of serotonin release was calculated using as 100% the cpm
contained in 50 µl harvested from wells containing the same number of
cells and lysed in 100 µl of 0.5% SDS, 0.5% Nonidet P-40.
Stable
RTIIB cell transfectants expressing murine FcRIIB as well as a
CD25/CD3
chimeric molecule at their surface, were further
transfected with two distinct NK cell MHC Class I receptor p58.2 and
p50.2 cDNA constructions. In NK cells, p58.2 (KIR) exerts an
inhibitory effect whereas p50.2 (KAR) is an activating molecule (20,
21). Despite this striking difference, both p58.2 and p50.2 are
HLA-Cw3-specific receptors and are recognized by the GL183 mAb.
Representative transfected clones used thereafter were selected for
their matched cell-surface expression of p58.2 (RTIIB.p58) or p50.2
(RTIIB.p50) HLA-Cw3 receptors (Fig. 1).
Inhibition of ITAM-dependent Serotonin Release by KIR Reconstituted in RTIIB Cells
RTIIB cells can be induced to
release serotonin by one of two ways: aggregation of the endogenous rat
FcRI receptor complex or aggregation of the CD25/CD3
chimeric
molecule (12, 37). Mouse IgE binding to Fc
RI is not sufficient to
induce RTIIB serotonin release, and aggregation of Fc
RI receptors
was obtained using GAM F(ab)
2. The integrity of ITAM
expressed by both receptors is required for RTIIB serotonin release
(38) indicating that both Fc
RI- and CD25/CD3
-induced serotonin
release utilizes ITAM-dependent signaling mechanisms. Using
RTIIB cells expressing p58.2 KIR or p50.2 KAR in addition to Fc
RI
and CD25/CD3
, we investigated whether the expression or the
aggregation of reconstituted HLA-Cw3 receptors modulate
ITAM-dependent serotonin release.
As shown in Fig. 2, A and B,
aggregation of FcRI or CD25/CD3
receptors induces a
dose-dependent serotonin release of RTIIB cells expressing
either p58.2 KIR (RTIIB.p58) or p50.2 KAR (RTIIB.p50). The larger
serotonin release elicited by anti-CD25 in RTIIB.p58 compared with
RTIIB.p50 cells most likely reflects the difference in surface
expression of CD25/CD3
in the two cell types. When p58.2 KIR were
aggregated using anti-p58.2 mAb (GL183), no RTIIB serotonin release was
observed. This result is consistent with the lack of detectable changes
in NK and T cell function upon anti-p58.2 mAb stimulation. However, no
serotonin release was detected in response to GL183 mAb in RTIIB.p50
cells, in contrast with the stimulatory effect of p50.2 KAR reported in
both NK and T cells (21, 39).
In a second set of experiments, p58.2 KIR versus p50.2 KAR
and FcRI receptors were co-aggregated using mouse IgE, mouse GL183 and GAM. Using saturating concentrations of GL183, the serotonin release induced by Fc
RI was impaired in RTIIB.p58 cells (Fig. 3A). This inhibition was detectable at suboptimal
(<10
3 dilution), as well as optimal concentrations of
IgE (10
3 dilution). Using saturating concentrations of
both GL183 (1 µg/ml) and IgE (10
3 dilution), 72.8 ± 2.2% and 55.8 ± 9.9% (mean ± S.E., n = 3) inhibition of serotonin release was observed in two distinct
RTIIB.p58 clones. Similar results were obtained when serotonin release
was triggered via the CD25/CD3
chimeric molecule (Fig.
3C). Using saturating concentrations of both GL183 (1 µg/ml) and anti-hCD25 (3 µg/ml), serotonin release was inhibited by
39.3 ± 11.9% in the representative RTIIB.p58B clone.
In contrast, no significant inhibition or potentiation of
ITAM-dependent cell activation was detected when p50.2 KAR
was co-aggregated with either FcRI or CD25/CD3
surface receptors,
even at suboptimal concentrations of triggering IgE or anti-hCD25, in
the presence of a saturating concentration of GL183 (1 µg/ml) (Fig.
3, B and D). These results first demonstrate that
p58.2 KIR reconstituted in RTIIB cells are functional and inhibit
ITAM-dependent cell activation. Second, they show that the
integrity of the p58.2 intracytoplasmic sequence is required for
KIR-mediated inhibition of RTIIB serotonin release. Finally, these data
indicate that RTIIB cells provide an appropriate cellular environment
for a functional reconstitution of p58.2 KIR but not of p50.2 KAR.
In the next set of experiments,
RTIIB.p58 cells were stimulated via FcRI in the presence of
aggregated p58.2 in one of two ways: DAM was used to co-aggregate
Fc
RI and p58.2 KIR via mouse IgE and mouse GL183, respectively (Fig.
4C), or a combination of DAR and DAM was used
to independently aggregate Fc
RI and p58.2 KIR via rat IgE and mouse
GL183 respectively (Fig. 4D). As shown in Fig. 4A
(closed circles), Fc
RI-p58.2 KIR co-aggregation induced a
GL183 dose-dependent inhibition of Fc
RI-induced
serotonin release, consistent with our observation reported in Fig.
3A. By contrast, the independent aggregation of Fc
RI and
p58.2 KIR failed to inhibit Fc
RI-induced serotonin release at any
GL183 concentration (Fig. 4A, open circles).
Similar results were obtained when RTIIB.p58 serotonin release was
induced via the CD25/CD3
chimeric molecule (Fig. 4B).
These results demonstrate that KIR require a co-aggregation with
activatory receptors (Fc
RI or CD25/CD3
) to exert their inhibitory
functions. They also suggest that KIR-mediated inhibition takes place
in the immediate vicinity of KIR molecules. Consistent with this
conclusion, co-aggregation of Fc
RI and KIR only inhibits RTIIB cell
serotonin release induced by Fc
RI but not by the triggering of the
CD25/CD3
chimeric molecule (data not shown).
From these data, one can predict that an inhibitory effect of p58.2 KIR
should only be observed when the relative proportion of co-aggregated
KIR-activatory receptors is high enough. To test this point, we have
selected a clone expressing an unusually high level of CD25/CD3
chimeric molecules. In this particular clone, using saturating
concentrations of GL183 F(ab)
2 (1 µg/ml), only low
levels of inhibition (14.1 ± 5.0%) of CD25/CD3
-induced
serotonin release were observed. In addition, we have observed that in
all clones stimulated with supraoptimal IgE concentrations
(>10
3) KIR fail to inhibit Fc
RI-induced RTIIB
serotonin release. These data are compatible with an inhibition
requiring the co-aggregation of activatory and inhibitory
receptors.
Activation of
RTIIB cells via ITAM triggers [Ca2+]i
mobilization (40, 41). To test whether reconstituted human KIR inhibit
early ITAM-dependent activatory signals, measurements of
[Ca2+]i were performed using a single-cell
imaging system. RTIIB.p58 and RTIIB.p50 cells were stimulated via the
FcRI receptor complex in the absence or presence of GL183 using GAM
as a cross-linker. In both RTIIB.p58 and RTIIB.p50 cells, aggregation
of the Fc
RI receptor complex induced a large
[Ca2+]i response consisting of a peak followed by
a sustained plateau (Fig. 5A, dotted
line). When p58.2 KIR was co-aggregated with Fc
RI, the
IgE-induced [Ca2+]i peak was impaired (Fig.
5A, continuous line). Using saturating
concentrations of both GL183 F(ab)
2 (1 µg/ml) and IgE
(10
3 dilution), the [Ca2+]i reached
at the peak decreased from 1459 ± 92 nM
(n = 58 cells) to 756 ± 59 nM
(n = 60 cells) (mean ± S.E., p < 0.001) (Table I).
|
In contrast, when p50.2 KAR was co-aggregated with FcRI, no
significant alteration of IgE-induced [Ca2+]i
mobilization was observed. At the [Ca2+]i peak,
using saturating concentrations of both GL183 (1 µg/ml) and IgE
(10
3 dilution), [Ca2+]i was
829 ± 67 nM (n = 67 cells)
versus 873 ± 67 nM (n = 67 cells) (mean ± S.E.) for IgE versus IgE-GL183
co-aggregation, respectively (Table I).
To evaluate whether KIR inhibited ITAM-dependent release of
[Ca2+]i from intracellular stores and/or
[Ca2+]i influx, further experiments were
performed on RTIIB.p58 cells in the absence of extracellular
Ca2+. For these conditions, only a small peak corresponding
to the release of Ca2+ from intracellular stores was
observed upon IgE stimulation (Fig. 5C, dotted
line). Using saturating concentrations of both GL183 (1 µg/ml)
and IgE (103 dilution), [Ca2+]i
decreased from 625 ± 39 nM (n = 56 cells) to 368 ± 32 nM (n = 37 cells)
(mean ± S.E., p < 0.001) at the
[Ca2+]i peak (Table I, Fig. 5C).
Therefore, p58.2 KIR is able to inhibit the release of Ca2+
from intracellular stores upon co-aggregation with Fc
RI. In addition, no [Ca2+]i mobilization was detected
when p58.2 or p50.2 was aggregated on RTIIB.p58 and RTIIB.p50 cells in
the absence of IgE stimulation (data not shown).
These results show that p58.2 KIR impairs ITAM-induced Ca2+ mobilization in RTIIB cells. Furthermore, p50.2 KAR was not capable of mediating any detectable Ca2+ mobilization when expressed in RTIIB cells, in contrast to its stimulatory function reported in both T and NK cells.
FcRTIIB.p58 cells also
express FcRIIB ITIM-bearing receptors. Similar to KIR, Fc
RIIB was
shown to inhibit serotonin release in RTIIB cells (13). By contrast to
KIR, Fc
RIIB has been reported in B cells to only inhibit
Ca2+ entry (42). Therefore, we examined the effect of
Fc
RIIB-Fc
RI co-aggregation on Ca2+ mobilization, to
document whether the differential effect of KIR and Fc
RIIB on
Ca2+ mobilization is due to a difference between B cells
and RTIIB cells or is the consequence of distinct inhibitory strategies used by these two ITIM-bearing receptors. RTIIB.p58 cells pre-incubated with murine IgE (10
3 dilution) in the presence or absence
of 2.4G2 F(ab)
2 (1 µg/ml) were challenged with a
TNP-F(ab)
2 MAR (10 µg/ml). Aggregation of the Fc
RI
receptor complex induced a large [Ca2+]i response
consisting of a peak followed by a sustained plateau (Fig.
5B, dotted line). When Fc
RIIB was
co-aggregated with Fc
RI, the [Ca2+]i peak was
not impaired but the plateau was not sustained (Fig. 5B,
continuous line). Using saturating concentrations of both
2.4G2 F(ab)
2 (1 µg/ml) and IgE (10
3
dilution), the [Ca2+]i reached at the plateau
decreased from 515 ± 27 nM (n = 60 cells) to 253 ± 12 nM (n = 59 cells)
(mean ± S.E., p < 0.001) (Table I).
To dissect the effects of FcRIIB inhibition on Ca2+
mobilization, additional experiments were performed on RTIIB.p58 cells in the absence of extracellular Ca2+. In these conditions,
only a peak corresponding to the release of Ca2+ from
intracellular stores was observed upon IgE stimulation (Fig. 5D, dotted line). Using saturating concentrations
of both 2.4G2 (1 µg/ml) and IgE (10
3 dilution), no
inhibition of Ca2+ release was observed, but rather
[Ca2+]i increased from 712 ± 32 nM (n = 79 cells) to 901 ± 36 nM (n = 59 cells) (mean ± S.E.) at
the [Ca2+]i peak (Table I, Fig. 5D).
These results indicate that Fc
RIIB has no effect on the release of
Ca2+ from intracellular stores upon co-aggregation with
Fc
RI. In addition, no [Ca2+]i mobilization was
detected when Fc
RIIB was aggregated on RTIIB.p58 cells in the
absence of IgE stimulation (data not shown). Therefore p58.2 KIR and
Fc
RIIB ITIM-bearing receptors exert distinct regulatory function on
Fc
RI-dependent Ca2+ mobilization.
The molecular cloning of MHC Class I-specific KIR has been
reported both in human and in mouse (33, 43-45). Human KIR belong to a
multigenic family of Ig-like members and are characterized, in their
extracellular portions by two or three Ig-like domains, corresponding
to HLA-C-specific p58, HLA-Bw4-specific p70/NKB1, and HLA-A3-specific
NKAT4 receptors. By contrast, mouse KIR belong to a family of type II
integral dimeric lectin proteins (7). Despite these striking
differences of genetic origin, mouse and human KIR express one or two
conserved intracytoplasmic (I/L)XYXX(Y/V) ITIM
motifs, respectively (16-19). We and others have described that these
amino acid stretches recruit upon phosphorylation of the tyrosine
residue, the SH2-tandem protein-tyrosine phosphatases, SHP-1 and SHP-2
(16, 17). The implication of SHP-1 in the inhibition of
ITAM-dependent activatory signals also has been suggested
for B lymphocyte surface molecules expressing an intracytoplasmic (I/V)XYXX(L/V) amino acid stretch, such as
FcRIIB and CD22 (15, 46, 47). In the present paper, we have
investigated the influence of the cellular environment on the function
of KIR, and the results are compared with those obtained with Fc
RIIB
(11-15, 42, 48).
We report here that HLA-Cw3-specific human KIR expressed in RTIIB cells
can inhibit [Ca2+]i mobilization and serotonin
release induced by the FcRI receptor as well as by a CD25/CD3
chimeric molecule (Figs. 3 and 5). Therefore, KIR and Fc
RIIB share
several features. First, KIR and Fc
RIIB inhibit early steps of the
signaling cascade, which are transcription-independent and are likely
to reflect NK cell killing mechanisms, such as regulated exocytosis
(49). Second, KIR and Fc
RIIB control the signals induced via
polypeptides including only one ITAM (FcR
), as well as receptors
including three sequential ITAM (CD3
). However KIR and Fc
RIIB
appear to use distinct inhibitory strategies. Indeed KIR inhibits
Ca2+ release from ER stores, whereas Fc
RIIB only
inhibits influx from the extracellular compartment. In addition, upon
phosphorylation of the ITIM, Fc
RIIB recruits preferentially the
phosphatidylinositol 5-phosphatase SHIP
(48),2 whereas KIR recruits the SHP-1
tyrosine phosphatase (16-19).2 It remains to be elucidated
whether these two findings are related.
We also show that in the absence of intracytoplasmic (I/V)XYXX(L/V), the HLA Class I receptor, p50.2 KAR, fails to transmit an inhibitory signal. These data strongly support the involvement of (I/V)XYXX(L/V) in KIR inhibitory function. In addition, and in contrast to T and NK cells, RTIIB cells expressing KAR are not activated upon KAR aggregation (Figs. 2 and 3). KAR molecules present a charged amino acid residue in their transmembrane domains. In T and NK cells, but not in RTIIB cells, KAR are part of a multimeric complex including three KAR-associated polypeptides (KARAP).3 It is thus tempting to speculate that in NK and T cells, KAR do not function in isolation but rather associate with other proteins, including KARAP, to mediate their activatory function.
Finally, we show that the mere cell-surface expression of KIR does not
modulate RTIIB cell activation (Fig. 2), just as described for
endogenous KIR expressed in NK and T cells and in contrast to other
regulators of lymphocyte activation, such as CD45 (50). On the
contrary, co-aggregation between an activatory receptor (the FcRI
receptor complex or the CD25/CD3
chimeric molecule) and p58.2 KIR is
required for the inhibition of serotonin release by RTIIB cells (Fig.
4). Although our data are obtained in a heterologous cell system, they
imply that the KIR-inhibitory effect occurs in the immediate vicinity
of the molecule, a feature shared by Fc
RIIB (12, 13). The
requirement for a co-engagement of KIR with an activatory receptor may
reflect the necessity for KIR to be phosphorylated by a non-diffusible
protein-tyrosine kinase associated with the activatory receptor.
Indeed, the tyrosine phosphorylation of KIR is mandatory to the
recruitment of SHP-1. Thus, the same protein-tyrosine kinase might
induce the tyrosine phosphorylation of both ITAM and ITIM, which is
consistent with data showing that ITIM YXX(L/V) sequence is
a potential substrate for the Src family member protein-tyrosine kinase
used by ITAM-containing receptors, such as Lyn, Lck, and Fyn (51). An
additional basis for the obligation of proximity between activatory and
inhibitory molecules may be that the tyrosine-phosphorylated
targets of SHP-1 belong to the ITAM-dependent activation
cascade, and thus must be brought in close proximity to SHP-1.
The identification of the major targets for SHP-1 involved in the KIR
inhibitory pathways is not yet achieved. Nevertheless, the recognition
of target cells protected from NK cell lysis by the surface expression
of HLA-B alleles leads to an inhibition of phosphatidylinositol
4,5-biphosphate hydrolysis, resulting in the prevention of
[Ca2+]i mobilization in NK cells (52). Similarly,
our results show that KIR inhibit ITAM-induced
[Ca2+]i mobilization in RTIIB cells. Therefore,
phospholipase C-, and/or its potential upstream signaling
effector/adaptors, such as p36-38 (53), as well as the ITAM-binding
SH2-tandem protein-tyrosine kinase (ZAP-70 and
p72syk) may be SHP-1 targets involved in KIR
signaling pathways. This model is supported by recent data in B cells,
where p72syk, phospholipase C-
1, and SHP-1
are recruited to CD22 upon BCR cross-linking (54), as well as in T
cells, where ZAP-70 is reported to be a direct substrate for SHP-1
(55).
In conclusion, the balance between receptor-mediated activation and
inactivation is central to cell homeostasis. We propose that KIR belong
to a novel family of ITIM-containing receptors possessing the
intracytoplasmic (I/V)XYXX(L/V) ITIM. Members of this family are broadly expressed within the hematopoietic lineage and
include receptors for Ig (FcRIIB) and MHC Class I molecules (KIR),
as well as B cell molecule CD22 and a KIR homologue gp49B expressed on
macrophages and mast cells (56, 57). All of these receptors act in a
spatio-temporal coordinated way with ITAM-containing receptors to
regulate the amplitude as well as the rates of signal initiation and
termination in hematopoietic cells.
We thank Odile Malbec (INSERM U255) for help
in serotonin release assay and TNP-F(ab)2 MAR production,
Corinne Béziers-Lafosse for graphic expertise, and Valéry
Renard, Frédéric Vély, Bernard Malissen, Michel
Fougereau, Pierre Golstein (CIML), Wolf H. Fridman (INSERM U255), and
Marc Benhamou (CERVI) for helpful discussions and continuous
encouragement.