(Received for publication, May 22, 1997, and in revised form, June 6, 1997)
From the Cancer Biology Program, Division of
Hematology-Oncology, Department of Medicine, Beth Israel Deaconess
Medical Center, Boston, Massachusetts 02215 and the ¶ Department
of Pathology, Tufts University School of Medicine,
Boston, Massachusetts 02111
The inhibitory Fc receptor, FcRIIB, provides a
signal that aborts B cell antigen receptor activation, blocking
extracellular calcium influx. Because the protein-tyrosine phosphatase
SHP-1 binds tyrosyl phosphorylated Fc
RIIB and Fc
RIIB-mediated
inhibition is defective in motheaten (me/me)
mice, which do not express SHP-1, it was proposed that SHP-1 mediates
Fc
RIIB signaling in B cells (D'Ambrosio, D., Hippen, K. L.,
Minskoff, S. A., Mellman, I., Pani, G., Siminovitch, K. A.,
and Cambier, J. C. (1995) Science 268, 293-297).
However, SHP-1 is dispensable for Fc
RIIB-mediated inhibition of
Fc
RI signaling in mast cells (Ono, M., Bolland, S., Tempst, P., and
Ravetch, J. V. (1996) Nature 383, 263-266), prompting
us to re-examine the role of SHP-1 in Fc
RIIB signaling in B cells.
We generated immortalized sIgM+, Fc
RIIB+ cell lines from
me/me mice and normal littermates. Co-ligation of Fc
RIIB and the sIgM antigen receptor inhibits calcium influx in both cell
lines. Inhibition is reversed by preincubation with anti-Fc
RIIB antibodies, indicating that it is mediated by Fc
RIIB. The inositol 5
phosphatase SHIP is recruited to tyrosyl-phosphorylated Fc
RIIB in
both cell lines. Fc
RIIB-mediated CD19 dephosphorylation also occurs
in the presence or the absence of SHP-1. Our results establish that
SHP-1 is dispensable for Fc
RIIB-mediated inhibition of sIgM antigen
receptor signaling.
The sIgM antigen receptor
(BCR)1 is a cell surface
heteroligomeric structure consisting of a ligand binding component,
surface immunoglobulin (sIg), and accessory signal transducing
components, the Ig and Ig
chains. Activation of the BCR leads to
B lymphocyte proliferation, differentiation, and antibody secretion.
Upon BCR cross-linking, several protein-tyrosine kinases including Lyn, Fyn, Blk, Syk, and Btk become activated and cause the recruitment, tyrosine phosphorylation, and activation of secondary signaling molecules, including PI-3K, Shc, Gap, Vav, and phospholipase C
. BCR
accessory components, such as CD19 and CD22, also become tyrosyl phosphorylated and recruit secondary signaling molecules, which either
enhance or dampen the BCR signal. These initial events lead to Ras
activation, PI3-K activation, phosphoinositide turnover, and an
increase in intracellular free calcium (reviewed in Ref. 1).
By co-ligating the BCR and the inhibitory Fc receptor, FcRIIB,
immune complexes activate a negative feedback mechanism that inhibits
antibody production (reviewed in Refs. 2 and 3). Fc
RIIB-mediated
inhibition in B lymphocytes can be studied experimentally by comparing
responses evoked by cross-linking the BCR with F(ab
)2 anti-Mu, which engages the BCR alone, with those initiated with intact
anti-Mu Ig, which co-ligates Fc
RIIB and the BCR. Co-ligation results
in inhibition of BCR-evoked inositol triphosphate production and
extracellular calcium influx (4, 5). Sustained calcium influx is
required for multiple cellular processes, including proliferation and
differentiation (reviewed in Refs. 6 and 7). Low concentrations of
calcium ionophores override Fc
RIIB-mediated inhibition of A20 B cell
activation, indicating that inhibition of calcium influx is a (the) key
mechanism by which antigen-mediated signals are aborted by Fc
RIIB
signals (4). An analogous mechanism operates in mast cells, where
co-engagement of Fc
RIIB and Fc
RI results in inhibition of
Fc
RI-generated responses (9-11).
A single tyrosyl residue, situated within the 13-amino acid ITIM
(immune receptor tyrosine-based inhibitory motif) in the cytoplasmic
domain of FcRIIB, is required for Fc
RIIB-mediated inhibition of
BCR signaling (12). SHP-1, an SH2-domain containing protein-tyrosine
phosphatase (PTP), binds to the tyrosyl-phosphorylated Fc
RIIB ITIM
motif both in vitro and in vivo (13). Because B cells from motheaten (me/me) and motheaten
viable (mev/mev)
mice, which have defective SHP-1 alleles, failed to demonstrate Fc
RIIB-mediated inhibition, it was proposed that the Fc
RIIB inhibitory signal is mediated by SHP-1 (13, 14). Consistent with this
model, a recent report demonstrated that CD19 is dephosphorylated upon
co-cross-linking of Fc
RIIB and the BCR, leading to decreased binding
of the p85 subunit of PI-3K to CD19 (15). This led to the proposal that
SHP-1 regulates intracellular calcium levels by controlling CD19
tyrosyl phosphorylation, which in turn controls associated PI-3K
activity (15). Interestingly, me/me mice exhibit a plethora
of immune system defects, including hypergammaglobulinemia and a high
level of circulating immune complexes (reviewed in Ref. 16).
Conceivably, one or more of these disorders might be explained by
defective Fc
RIIB signaling.
However, the role of SHP-1 in FcRIIB-mediated inhibitory signaling
has been a subject of controversy. Mice lacking Fc
RIIB do not
exhibit uncontrolled antibody production when challenged with antigen
(9). Moreover, as revealed by studies of mast cells from
me/me mice, SHP-1 clearly is dispensable for
Fc
RIIB-mediated inhibition of Fc
RI signaling (17). Ono et
al. have proposed that the inositol phosphatase SHIP
(SH2 domain containing 5
-inositol phosphatase), which also binds to the Fc
RIIB ITIM, is
the key mediator of Fc
RIIB signaling (17). The situation is
complicated further by a report that the other mammalian SH2-containing
PTP, SHP-2, also binds to this ITIM, at least in vitro (14).
Although it is possible that Fc
RIIB employs unique signaling
mechanisms in different cell types, these findings have raised
questions as to whether SHP-1 actually is required for
Fc
RIIB-mediated inhibition in B cells.
To develop a model system for studying the role of SHP-1 in multiple B
cell signaling pathways, we generated B cell lines from
me/me (ME) and normal (N) littermate mice by immortalization with the retrovirus J2 (18, 19). N and ME cell lines express similar
surface markers and respond to BCR stimulation and FcRIIB/BCR co-ligation. We find that both cell lines exhibit comparably decreased calcium influx when Fc
RIIB is co-cross-linked with the BCR.
Consistent with previous reports, CD19 tyrosyl phosphorylation
decreases upon co-cross-linking of Fc
RIIB and the BCR in N cells.
However, dephosphorylation also is observed in the ME line. Our results establish that SHP-1 is dispensable for both biochemical and biological responses to Fc
RIIB engagement in B cells.
J2 producer cells (20) were obtained from Dr. Christopher Myers of the University of Tennessee, Memphis, TN. Virus-containing supernatants were collected from J2 cells growing in log phase for 24 h in RPMI 1640 supplemented with 20% heat-inactivated fetal calf serum, 12 mM NaHC03, 100 IU/ml penicillin G, 100 µg/ml streptomycin, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 µM 2-mercaptoethanol, and 12.5 mM HEPES, pH 7.4. Spleens from 12-14-day-old me/me and normal littermate mice were removed and red blood cells were lysed with Tris-NH4Cl. Immortalization was carried out by incubating total splenocytes at a concentration of 7 × 106 cells/ml with medium containing J2 retrovirus. After 15-30 days, cells that had proliferated were expanded and analyzed for sIgM by flow cytometry. sIgM+ cells were subsequently cloned by limiting dilution, expanded, and maintained in RPMI 1640 supplemented as described above, except that the fetal calf serum concentration was decreased to 10%.
Flow CytometryCells were stained with the appropriate
antibodies in the presence of fully supplemented RPMI. Antibodies used
for staining included anti-IgM and anti-FcRIIB (CD32/CD16; 2.4G2),
obtained from Pharmingen. Flow cytometric data were obtained on a
FACScan using Cell Quest acquisition and analytical software (Becton
Dickinson, San Jose, CA).
N and ME cells were loaded with 2 µM (final concentration) Fura-2 (Molecular Probes,
Eugene, OR) in a buffer containing 135 mM NaCl, 5 mM KCl, 1 mM CaCl2, 1 mM MgCl2, 5.6 mM glucose, 10 mM HEPES, pH 7.4, and 0.1% bovine serum albumin. Cells
were washed once and resuspended at 2 × 106 cells/ml.
Assays were performed as described previously (21) using a Delta Scan
spectrofluorometer (Photon Technologies Inc., South Brunswick, NJ).
Cells were activated with 5 µg/ml F(ab)2 goat anti-mouse
Mu or with 10 µg/ml rabbit anti-mouse Mu (Jackson ImmunoResearch)
either with or without preincubation with anti-FcR (2.4G2) antibody. To
monitor calcium influx, cells were activated with anti-Mu antibodies in
the presence of 1.5 mM EGTA, followed by the addition of 5 mM CaCl2. The emission values at 510 nm
following excitation at 340 or 380 nm were collected and analyzed using FELIX software (Photon Technologies). Analyses of data and generation of graphs were performed using Quattropro.
N and ME cells
were activated by incubation with 5-10 µg/ml F(ab)2
anti-Mu or with 10 µg/ml rabbit anti-mouse Mu (Jackson ImmunoResearch) at 37 °C. Activations were stopped by addition of
lysis buffer containing 1% Nonidet P-40, 150 mM sodium
chloride, 5 mM sodium flouride, 1 mM EDTA, 2 mM sodium orthovanadate, 25 mM HEPES, pH 7.4, and the protease inhibitors leupeptin (10 µg/ml), aprotinin (10 µg/ml), pepstatin (1 µg/ml), antipain (1 µg/ml), and
phenylmethylsulfonyl fluoride (10 µg/ml). Typically, 5 × 106 cells were activated per time point, and one-tenth of
the resulting lysate was removed before immunoprecipitation and
analyzed by anti-phosphotyrosine immunoblotting. In some cases,
activations were stopped by the addition of ice-cold phosphate-buffered
saline, and cell pellets were frozen in liquid N2 and
stored at
80 °C prior to lysis. Lysates were immunoprecipitated
for 1-3 h with the indicated antibodies, including anti-CD19
(Pharmingen), anti-Fc
RIIB (CD32/CD16; Pharmingen), and anti-SHIP (a
gift from Dr. Mark Coggeshall, Ohio State University). Anti-Fc
RIIB
antibodies were covalently cross-linked to protein G-agarose beads
prior to use. Immunoprecipitates were collected onto either protein
A-Sepharose (Pharmacia Biotech Inc.) or protein G-agarose (Oncogene
Sciences) beads, washed three times with lysis buffer, separated by
SDS-polyacrylamide gel electrophoresis, and transferred onto
polyvinylidene difluoride membranes (Immobilon, Millipore).
For anti-phosphotyrosine immunoblotting, membranes were blocked with 5% bovine serum albumin (Boehringer Mannheim) in TBST buffer (10 mM Tris, pH 7.4, 150 mM sodium chloride, 0.1% Tween 20), followed by incubation with 1 µg/ml monoclonal anti-phosphotyrosine 4G10 antibodies (Upstate Biotechnology Inc.) and subsequent incubation with secondary antibodies (donkey anti-mouse) conjugated to horseradish peroxidase (Amersham Corp.). All other blots were blocked in 5% milk in TBST and were developed in a similar manner with the indicated antibodies. Antibodies used for immunoblotting included: anti-SHP-1 polyclonal antiserum (1:1000 dilution) (22, 23), polyclonal anti-PTP1C and anti-PTP1D (1 µg/ml; Signal Transduction Labs), polyclonal anti-CD19 (a gift from Dr. John Cambier, Denver Colorado), and anti-SHIP (1:1000). Immunoreactive proteins were detected by Enhanced Chemiluminescence (Amersham Corp.). For reprobing, polyvinylidene difluoride membranes were stripped of antibodies, according to the manufacturer's suggestions (Amersham Corp.).
SHP-1 has been implicated as a negative regulator of antigen
receptor (BCR) signaling in B cells and, in particular, in FcRIIB signaling (13, 24, 25). Comparison of BCR signaling events in primary B
cells from me/me mice is difficult because these cells are
developmentally distinct and have reduced surface levels of sIgM and
Fc
RIIB compared with age-matched B cells from normal littermates
(24).2 To circumvent such
problems, we generated a set of immortalized B cell lines derived from
me/me and normal littermate mice, using the murine
retrovirus J2, which expresses the viral oncogenes raf and
myc (19, 20) (Fig.
1A). Previous work indicated
that this virus can generate immortal, factor-independent macrophage and B cell lines in vivo and ex vivo (20, 26).
Unlike B cell lines produced by Abelson MuLV, which are arrested at the
pro- or early pre-B cell stage and thus cannot signal through the BCR, many J2-derived lines are sIgM+, sIgD+, and retain BCR signaling capacity (8, 20) (Figs. 2 and
3). We obtained several normal cell lines
(N) and one me/me-derived line (ME) (Fig. 1). Southern analysis indicated that these lines are clonal (data not shown), and
immunoblotting confirmed that the ME line lacks SHP-1 expression (Fig.
1B). Notably, however, the N and ME lines express equal levels of SHP-2 (Fig. 1B). The surface phenotypes of the N
and ME cell lines are similar; of particular importance for the results discussed below, they express comparable levels of sIgM and Fc
RIIB (Fig. 1, C-F). Both lines also display similar levels of
sIgD, CD45 (B220), CD19, and CD22 but do not express CD5, CD11b, B7.1 or B7.2 (data not shown). Therefore, their surface phenotype more closely resembles primary B cells, rather than B-1a, germinal center or
memory B cells. Other properties of these cell lines will be discussed
in more detail
elsewhere.3
We assessed FcRIIB-mediated inhibition in both cell lines by
comparing their calcium response following BCR ligation alone or
together with Fc
RIIB. The change in the ratio (chelated/free) of
Fura-2 fluorescence was used to measure calcium responses to both
stimuli. We first tested the N cell line to determine whether J2
immortalization interferes with BCR activation and/or
Fc
RIIB-mediated inhibition. N cells activated with
F(ab
)2 anti-Mu (engagement of BCR alone) exhibited a
robust increase in intracellular free calcium. This response was
markedly diminished in N cells stimulated with intact anti-Mu (Fig.
2A). Preincubation with anti-Fc
RIIB antibodies (2.4G2)
abrogated the ability of intact anti-Mu to inhibit the calcium response
in N cells (Fig. 2B), indicating that inhibition occurred
solely via engagement of Fc
RIIB. N cells activated in the presence
of EGTA showed little difference in peak calcium release from
intracellular stores (Fig. 2C). Readdition of
CaCl2 after the release phase to allow measurement of
calcium influx revealed strong inhibition by intact anti-Mu (Fig.
2C). These results, which are consistent with those reported
for primary B cells and several B cell lines (see Introduction),
demonstrate that when engaged, Fc
RIIB inhibits BCR signaling in the
N cell line by preventing calcium influx. Importantly, the
cross-linking antibodies, although derived from different species,
evoked identical peak levels of calcium release, indicating that they
provide comparable activation stimuli (Fig. 2C). Likewise,
the calcium responses to F(ab
)2 and intact anti-Mu in the
presence of anti-Fc
RIIB (2.4G2) antibodies were virtually identical
(Fig. 2, compare A with B). Recently, we have
confirmed these results using F(ab
)2 and intact anti-Mu
IgG from the same species (rabbit) as cross-linking reagents (data not
shown). Taken together, these data show that J2-immortalized B cells
provide a valid model to study BCR signal transduction and its
regulation by Fc
RIIB.
To determine whether SHP-1 is required for negative regulation of
calcium influx by FcRIIB, we compared calcium responses in ME cells
following ligation of the BCR with either F(ab
)2 or intact
anti-Mu. Surprisingly, ME cells activated with intact anti-Mu also
showed complete inhibition of the calcium response compared with those
stimulated with F(ab
)2 anti-Mu (Fig. 2D); again, inhibition was reversed by preincubation with anti-Fc
RIIB antibodies (Fig. 2E). As in N cells, there was little
difference in peak calcium release from intracellular stores following
ligation of ME cells with either antibody in the presence of EGTA, but extracellular calcium influx was markedly reduced with intact anti-Mu
compared with F(ab
)2 anti-Mu (Fig. 2F). As with
N cells, we have recently confirmed these results using matched rabbit F(ab
)2 and intact anti-Mu IgG (data not shown). Because ME
cells exhibit Fc
RIIB-mediated inhibition of the BCR-induced calcium response, we conclude that, as in mast cells (17), SHP-1 is not required for FcR-mediated inhibition of B cell
signaling.
Although SHP-1 is dispensable for FcRIIB-mediated inhibition of BCR
calcium influx, it remained possible that it influences Fc
RIIB-evoked tyrosyl phosphorylation events. To address this issue,
we first examined the tyrosyl phosphorylation of Fc
RIIB itself.
Previous studies established that tyrosyl phosphorylation of Fc
RIIB
occurs upon co-cross-linking to the BCR, but not when the BCR alone is
engaged (12). Anti-phosphotyrosine immunoblots of Fc
RIIB
immunoprecipitates from N and ME cells activated by either
F(ab
)2 or by intact anti-Mu ligation revealed Fc
RIIB tyrosyl phosphorylation only upon co-cross-linking (Fig. 3A,
upper panel). Fc
RIIB was not significantly tyrosyl
phosphorylated either basally or upon F(ab
)2 anti-Mu
stimulation in ME cells, indicating that loss of SHP-1 does not lead to
aberrant Fc
RIIB tyrosyl phosphorylation (Fig. 3A,
upper panel). To determine if absence of SHP-1 resulted in
global differences in total cellular tyrosyl phosphorylation upon
Fc
RIIB/BCR co-cross-linking, we compared anti-phosphotyrosine immunoblots of total cell lysates from cells that were activated with
either F(ab
)2 or intact anti-Mu. In N cells, each stimulus yielded a comparable overall total cellular tyrosyl phosphorylation response (Fig. 3C), consistent with previous studies that
reported no gross difference in total tyrosyl phosphorylation between
intact versus F(ab
)2 anti-BCR stimulation (12).
Each stimulus also evoked comparable increases in total cellular
phosphotyrosine in ME cells (Fig. 3C). A detailed comparison
of BCR signal transduction between N, ME, and ME cells reconstituted
with wild type and mutant SHP-1 will be presented
elsewhere.3
CD19 has been reported to display decreased tyrosyl phosphorylation
upon FcRIIB co-cross-linking with the BCR. It has been suggested
that SHP-1 mediates the differential tyrosyl phosphorylation of this
co-receptor (15). To address these issues, we examined CD19 tyrosyl
phosphorylation in both cell lines. In N cells, anti-phosphotyrosine immunoblotting of CD19 immunoprecipitates revealed that tyrosyl phosphorylation of CD19 is slightly decreased when the BCR is co-cross-linked with Fc
RIIB (Fig. 3D), consistent with
the previous study (15). However, decreased CD19 tyrosyl
phosphorylation also was observed when the BCR was co-cross-linked with
Fc
RIIB in ME cells (Fig. 3D). These data suggest that
SHP-1 is not solely responsible for CD19 dephosphorylation
in response to Fc
RIIB engagement.
Our results clearly establish that SHP-1 is not required for
FcRIIB-mediated inhibition of BCR-induced calcium influx. These findings contrast with an earlier report, which suggested that SHP-1 is
necessary for Fc
RIIB-mediated inhibition in B cells (13). The
earlier study compared signaling in primary lymphocytes from normal and
me/me or
mev/mev mice. Primary B
cells from me/me mice are developmentally distinct from
those of their normal counterparts and are similar in phenotype to
anergized B cells, exhibiting decreased levels of BCR and increased expression of activation markers (24). The failure of the earlier workers to observe effects of Fc
RIIB engagement on the proliferative responses of me/me B cells may reflect developmental
differences caused by the lack of SHP-1, rather than the direct effect
of absence of SHP-1 on signaling via Fc
RIIB. Alternatively, the high
levels of circulating immune complexes in me/me mice may lead to desensitization of Fc
RIIB signaling (although
me/me B cells do retain surface Fc
RIIB; data not
shown).
Our results also establish that SHP-1 is not required for the
FcRIIB-evoked decrease in CD19 tyrosyl phosphorylation. Presumably, because there has been no demonstrated change in any protein-tyrosine kinase activity upon Fc
RIIB engagement (12), a PTP other than SHP-1
promotes CD19 dephosphorylation. SHP-2 has been shown to bind the
Fc
RIIB ITIM in vitro (14), raising the possibility that
it may be responsible for CD19 dephosphorylation in vivo. In
any case, the functional significance of CD19 dephosphorylation in
response to Fc
RIIB engagement remains unclear.
As in mast cells (17) and other B cell lines and primary B cells, SHIP,
an SH2-domain containing inositol phosphatase, associates with tyrosyl
phosphorylated FcRIIB and becomes tyrosyl phosphorylated in both N
and ME cells (Figs. 3, A, lower panel, and
B), raising the possibility that it mediates inhibitory
signaling by this receptor. The mechanism by which an inositol
phosphatase could promote inhibition of calcium influx is unclear.
Hydrolysis of phosphatidylinositol 3,4,5-trisphosphate and/or
phosphatidylinositol 4,5-bisphosphate could directly or indirectly
modulate a calcium channel responsible for influx. Such a channel(s)
remains to be identified in B cells. Dephosphorylation of CD19, with
consequent diminished recruitment of PI-3K (15), might lead to similar effects on phosphatidylinositol-containing lipids, as recruitment of
SHIP to Fc
RIIB. Therefore, it remains possible that Fc
RIIB may
function by recruiting both SHIP and at least SHP-1 or SHP-2. However,
it is important to note that we have been unable to detect stable
association of SHP-2 with tyrosyl phosphorylated Fc
RIIB in B cells
(data not shown). Clarifying the roles of SHIP and SHP-2 in Fc
RIIB
signaling in B cells will most likely require generation of cell lines
or mice that lack expression of these proteins.
We thank Dr. Christopher Meyers for the J2 producer cell line and Dr. Paul McLean for assistance in the Southern analysis of the ME and N cell lines. We also thank Drs. Andrew Scharenberg and Jean-Pierre Kinet (Beth Israel Deaconess Medical Center, Boston) for helpful comments and for use of the Photon Technologies instrument.
Dr. J. V. Ravetch's group also has shown
that SHP-1 is dispensable for FcRIIB-mediated inhibition of B cell
antigen receptor signaling, using DT40 B cells lacking SHP-1 as a
consequence of homologous recombination (Ono, M., Okada, H., Bolland,
S., Yanagi, S., Kurosaki, T., and Ravetch, J.V. (1997) Cell,
in press). In addition, Ono et al. show that SHP-1 is
required for KIR-mediated inhibition.