(Received for publication, March 4, 1997, and in revised form, March 20, 1997)
From the Departments of Medicine and
§ Immunology, Duke University Medical Center,
Durham, North Carolina 27710
The selectin adhesion molecules and chemoattractant receptors synergistically regulate leukocyte migration into lymphoid tissues and sites of inflammation, but little is known about how these families of receptors modulate each other's function. In this study, L-selectin was found to be phosphorylated in lymphoblastoid cell lines, and phosphorylation was enhanced by phorbol ester (phorbol 12-myristate 13-acetate (PMA)) treatment. Interactions between L-selectin and chemoattractant receptors were therefore examined using transfected rat basophilic leukemia cell lines (RBL-2H3) that expressed human L-selectin along with human leukocyte chemoattractant receptors. L-selectin was rapidly phosphorylated in cells treated with chemoattractants, thrombin, IgE receptor agonists, or PMA. Pertussis toxin or the protein kinase C inhibitor, staurosporine, completely blocked chemoattractant receptor-induced phosphorylation of L-selectin. PMA-induced phosphorylation was on serine residues within the cytoplasmic tail of L-selectin that have been well conserved during recent evolution. Although L-selectin phosphorylation was not essential for basal levels of adhesion through L-selectin in transformed cell lines, the rapid increase in ligand binding activity of L-selectin that occurs following leukocyte activation was blocked by staurosporine. These results demonstrate that L-selectin can be phosphorylated following engagement of chemoattractant receptors and suggest that this may be a physiologically relevant mechanism for the synergistic regulation of these receptors during leukocyte migration.
The selectin and integrin families of adhesion molecules regulate
leukocyte migration into lymphoid tissues and sites of inflammation (1-4). L-, P-, and E-selectin mediate the initial interactions of
leukocytes with endothelium that result in leukocytes rolling along the
venular wall (2). During their initial interactions with endothelial
cells, leukocytes encounter chemoattractants that bind to cell surface
receptors (5-9). Signal transduction through chemoattractant receptors
results in increased binding activity for L-selectin, 2
integrins, and
1 integrins (10), which stabilizes
leukocyte interactions with endothelial cells (3, 4).
While much is known regarding the independent functions of adhesion molecules and chemoattractant receptors, little is known about how these receptors modulate the function of each other. In one example, the ligand binding activity of L-selectin can be rapidly up-regulated by exposing leukocytes to a variety of pro-inflammatory agents including chemoattractants (10). Therefore, potential mechanisms by which chemoattractant receptor signaling may modulate L-selectin function were examined using the rat basophilic leukemia cell line, RBL-2H3 (RBL1 cells), as an in vivo model (11-13). Phosphorylation of L-selectin is a potential site for receptor regulation since the cytoplasmic domain of L-selectin contains numerous basic residues surrounding 2 serine residues that have been highly conserved during recent mammalian evolution (2, 14, 15). RBL cells stably transfected to co-express functional human chemoattractant receptors and L-selectin provide direct evidence that activation of chemoattractant receptors induces immediate phosphorylation of L-selectin through a protein kinase C (PKC)-dependent pathway.
Antibodies used in these studies included: mouse LAM1-116 (IgG2a) and LAM1-110 (IgG1) mAbs that react with human, mouse, and rat L-selectin (34); anti-CD83 mAb (HB15A IgG2b); and anti-human CD3 mAb (RW2-8C8). Antibodies were purified from ascites fluid by sodium sulfate precipitation and DEAE-Sepharose anion exchange column chromatography (Pharmacia Biotech Inc.). The 12CA5 mAb reactive with a 9-amino acid epitope tag was from Boehringer Mannheim.
Immunofluorescence staining of cells and cell lines was as described previously (16) using mAbs optimally diluted for immunostaining: FITC-conjugated LAM1-116 mAb or unconjugated LAM1-116 mAb detected with FITC-conjugated goat anti-mouse IgG antibodies (Caltag, South San Francisco, CA). Single color immunofluorescence analysis of 10,000 cells was performed on a FACScan flow cytometer (Becton Dickinson) with fluorescence intensity analyzed on a 4-decade log scale. The lectin activity of L-selectin was assessed by incubating transfected RBL cells with biotinylated polyphosphomonoester core polysaccharide (5 mg/ml) from yeast and FITC-conjugated streptavidin using methods similar to those previously described (10).
Cells and Cell LinesRBL cells or RBL cells expressing
epitope-tagged chemoattractant receptors were cultured as described
(11). RBL or 300.19 cells were co-transfected with L-selectin or
LM-N cDNA by electroporation, and clones were isolated as
described (13, 15). 300.19 cells transfected with human cDNA for
either L-selectin or L
cyto cDNA were as described (16, 17).
Human blood lymphocytes were isolated from heparin-anticoagulated
venous blood from healthy adult volunteers by centrifugation over
Ficoll density gradient medium (Nycomed, Oslo, Norway).
The cytoplasmic domain of human L-selectin is composed
of 12 amino acid residues (G323KKSKRSMNDPY334)
(14). The L-selectin cDNA that encodes a protein with the 2 cytoplasmic serine residues replaced by alanine residues (L-SS/AA) and
the cDNA that deletes the GKKSKRS sequence (LG-S) of the cytoplasmic domain were generated by a two-step polymerase chain reaction and were verified by sequence analysis. The L
cyto and L
M-N cDNAs were as described (15, 17). Epitope-tagged
chemoattractant receptors were as described (11-13).
Cells (5-10 × 106) were surface labeled with 125I, lysed, immunoprecipitated using the indicated mAbs, resolved by SDS-PAGE, and visualized by autoradiography as described (11-13). Metabolic labeling of cells with [32P]orthophosphate was as described (11-13). Labeled cells were incubated with or without agonists (thrombin receptor peptide, SFLLRN, 100 µM, Peninsula Laboratories, Belmont, CA; thrombin, 1 unit/ml; fMLP, 1 µM; PAF, 100 nM; PMA, 0.5 µM, all from Calbiochem; IL-8, 100 nM, Genzyme, Cambridge, MA; C5a, 100 nM, Sigma) for 3 min at 37 °C. Cells were also activated with IgE (0.2 µg/ml) plus antigen (dinitrophenyl-conjugated bovine serum albumin, 0.1 µg/ml) as described (11). In some cases the cells were also pretreated for 5 min with 100 nM staurosporine (Calbiochem) or overnight with 100 ng/ml of pertussis toxin (List Biological Laboratories) before stimulation with agonists. For lymphoblastoid cell lines, cells were incubated in the presence of protease inhibitors as described (18, 19). The cells were lysed after 3 min. Phosphorylated L-selectin or chemoattractant receptors were immunoprecipitated with the LAM1-116 (15 µg) or 12CA5 (10 µg) mAbs, respectively, analyzed by SDS-PAGE, and visualized by autoradiography.
Cell Binding AssayBinding of transfected 300.19 cells expressing native or modified forms of L-selectin to high endothelial venules (HEVs) was assessed as described (16, 20). Human blood lymphocytes were stimulated through the CD3 receptor as described (10) in the presence of 100 nM staurosporine or Me2SO carrier.
Previous attempts to demonstrate
phosphorylation of L-selectin by us and others have been unsuccessful
probably because of the rapid endoproteolytic release of L-selectin
from the cell surface following cellular activation (21-24).
Therefore, metabolically labeled human lymphoblastoid cell lines that
express L-selectin were cultured for 3 min with either medium or PMA, a
known activator of PKC, in the presence of a protease inhibitor that
blocks the endoproteolytic release of L-selectin (18, 19). These
inhibitors do not affect leukocyte activation or adhesive interactions
between leukocytes and endothelial cells (18, 19, 25). Following lysis
of the cells and immunoprecipitation with an L-selectin-specific mAb,
an appropriately sized phosphoprotein band was isolated (Fig. 1). In each case, PMA treatment of the cell lines
resulted in increased phosphorylation of this protein. Therefore, RBL
cells transfected with L-selectin cDNA were used as a model system
to verify that L-selectin was the phosphorylated protein
immunoprecipitated from lymphoblastoid cell lines.
RBL cells transfected with human L-selectin cDNA (LAM1 cells)
expressed high levels of either unmodified L-selectin (14) or a
modified form of L-selectin (LM-N) (15) that is fully functional but
not endoproteolytically released from the cell surface (Fig.
2A). RBL cells did not express endogenous
L-selectin as determined using two mAbs reactive with rat L-selectin
(LAM1-116 and LAM1-110, Fig. 2A). Human L-selectin
expressed in RBL cells was functionally and structurally intact since
it retained the ability to specifically bind polyphosphomonoester
core polysaccharide, which mimics the natural ligand of L-selectin
(Fig. 2B) (26) and was bound by seven mAbs directed against
distinct epitopes of its extracellular domains (data not shown). In
addition, only LAM1 cells displayed an ~80-kDa surface protein that
was immunoprecipitated with mAbs specific for L-selectin (Fig.
2C). L-selectin immunoprecipitated from unactivated RBL
cells was only weakly phosphorylated or not phosphorylated at
detectable levels. However, phosphorylation of L-selectin was markedly
increased following PMA treatment of LAM1 cells for 3 min (Fig.
3A, lanes 1 and 2). In
contrast, PMA treatment of untransfected RBL cells did not induce
phosphorylation of proteins in this size range (Fig. 3A,
lane 4).
Chemoattractant Receptors Induce L-selectin Phosphorylation
RBL cells that expressed the LM-N form of
L-selectin, as well as native thrombin and IgE receptors, were also
transfected with cDNA encoding epitope-tagged human receptors for
formyl peptides (fMLPR), a peptide component of complement activation
(C5aR), interleukin-8 (IL-8R), or platelet-activating factor (PAFR).
RBL clones expressing these chemoattractant receptors respond to
chemoattractants by activating similar signal transduction pathways as
do leukocytes including actin polymerization, phosphoinositide
hydrolysis, calcium mobilization, phospholipase D activation, and
degranulation (11-13). L-selectin expressed together with human
chemoattractant receptors (Fig. 3, B and C) was
phosphorylated following PMA treatment (Fig. 3, lanes 2, 6, and 10). Activation for 3 min of chemoattractant receptors
for fMLP, IL-8, and PAF with their respective ligands also resulted in
phosphorylation of L-selectin in co-transfected cells (lanes 7, 11, and 13). fMLP induced strong phosphorylation of
L-selectin, while PAF activation resulted in weaker phosphorylation (lanes 7 and 13). IL-8, fMLP, and PAF stimulation
also resulted in the homologous phosphorylation of their receptors
(lanes 8, 12, and 14) as well as
cross-phosphorylation of the PAFR by IL-8R activation as described
previously (11-13, 27). Activation of RBL cells through endogenous
thrombin, IgE receptors, or ectopic C5a receptors also resulted in
phosphorylation of L-selectin (Fig. 4A).
Activation of RBL cells through these receptors also results in
phosphorylation of the C5a receptor (11), which was simultaneously immunoprecipitated with L-selectin to provide an internal control for
receptor signaling (Fig. 4A). In addition, dose-response
studies showed that concentrations (1-3 nM) of C5a
analogous to those found at sites of inflammation induced L-selectin
phosphorylation (data not shown). Therefore, L-selectin was rapidly
phosphorylated following chemoattractant receptor activation.
L-selectin Is Phosphorylated on Serine Residues following PKC Activation
L-selectin was phosphorylated specifically on serine
residues following PMA treatment. Phosphorylated L-selectin was
immunoprecipitated from RBL cells transfected with a cDNA encoding
the LM-N form of L-selectin, while phosphorylated L-selectin was not
immunoprecipitated from cells expressing the L
M-N form of L-selectin
with the 2 serine residues in the cytoplasmic tail replaced with
alanine residues (L
MN-SS/AA) (Fig. 5). Since PKC is
activated by all of the agonists used above, its role in L-selectin
phosphorylation was examined using staurosporine, a PKC inhibitor (28).
Both PMA- and C5a-induced phosphorylation of L-selectin was completely inhibited by treating LAM1 cells with staurosporine (Fig.
4B, lanes 3 and 5). Pertussis toxin,
an inhibitor of signaling through Gi proteins (29), also
blocked C5a-induced phosphorylation of L-selectin, indicating that the
production of second messengers through G-protein activation is
required for chemoattractant receptor-induced phosphorylation of
L-selectin (Fig. 4B, lane 6). L-selectin
phosphorylation was detected at the earliest measurable time point of
7 s following fMLPR-induced activation of RBL cells (Fig.
4C). These results demonstrate that L-selectin was
immediately phosphorylated on serine residues following cellular
activation by a wide range of pro-inflammatory mediators that activate
PKC.
Role of Phosphorylation in L-selectin Function
Native
L-selectin was endoproteolytically released from the cell surface
within minutes following activation of RBL cells with PMA, while the
LM-N form of L-selectin was retained (data not shown). To determine
whether phosphorylation of L-selectin regulates its endoproteolytic
release from the cell surface, RBL and 300.19 cells, a mouse pre-B cell
line, were transfected with L-selectin cDNAs that encoded native
receptors with the 2 serine residues in the cytoplasmic tail replaced
with alanine residues (L-SS/AA), the 7-amino acid region containing the
serine residues (L
G-S) deleted, or the entire cytoplasmic tail
(L
cyto) deleted (see "Experimental Procedures"). In all cases,
the spontaneous or PMA-induced endoproteolytic release of L-selectin
was not measurably affected by these modifications (data not shown).
Therefore, endoproteolytic release of L-selectin was not regulated by
the cytoplasmic domain.
The cytoplasmic domain of L-selectin is required for receptor-mediated
adhesion in vivo and in vitro (17). Furthermore, the binding activity of L-selectin for ligand increases rapidly following lymphocyte activation through the T cell receptor (CD3) complex or neutrophil activation with cytokines (10). Since cross-linking CD3 activates PKC (30) a role for PKC-mediated phosphorylation in L-selectin-dependent binding is
possible. Unfortunately, it is not feasible to transfect normal
leukocytes with L-selectin cDNAs lacking the cytoplasmic serine
residues to test this directly. Likewise, it has not been possible to
demonstrate up-regulated L-selectin binding activity in RBL cells or
lymphoblastoid cell lines.2 Therefore, the
effects of blocking receptor phosphorylation on up-regulated L-selectin
binding activity were examined indirectly by treating cells with a PKC
inhibitor. Treatment of lymphocytes with staurosporine completely
inhibited the CD3-mediated increase in L-selectin-dependent
binding to HEVs and reduced, but did not eliminate, basal L-selectin
binding (Fig. 6). Since leukocyte binding to HEVs, both
basal and up-regulated, is completely inhibited by blocking L-selectin
function (10) and chemoattractant receptor-induced adhesion through
integrins is not PKC-dependent (31), the current results
suggest that rapid PKC-mediated phosphorylation of L-selectin may
up-regulate its binding activity in vivo.
The effects of receptor phosphorylation on basal L-selectin
binding activity were examined directly using
cDNA-transfected 300.19 cells. The L-SS/AA-modified L-selectin did
not significantly affect 300.19 cell binding to HEVs of peripheral
lymph nodes (L-selectin, 7.8 ± 0.8 cells/HEV; L-SS/AA, 6.3 ± 1.2 cells/HEV; p < 0.1) in four experiments. By
contrast, the LG-S mutation (0.6 ± 0.3 cells/HEV) was similar
to that of L
cyto (1.2 ± 1.0 cells/HEV; p < 0.0001 versus L-selectin) and showed very weak if any
binding to HEVs. Therefore, although phosphorylation was not necessary
for basal adhesion through L-selectin, the region containing the serine residues within the cytoplasmic domain was required.
The finding that L-selectin was rapidly phosphorylated in lymphoblastoid cell lines (Fig. 1) and transfected RBL cells (Figs. 3 and 4) immediately following chemoattractant or PMA stimulation suggests a physiologically relevant role for L-selectin phosphorylation in the inflammatory response. Indeed, signaling of RBL cells through endogenous thrombin and IgE receptors or ectopic receptors for fMLP, C5a, IL-8, and PAF all induced L-selectin phosphorylation (Figs. 3 and 4). That all of these receptors activate PKC-dependent pathways, that L-selectin phosphorylation was completely blocked by a PKC inhibitor (Fig. 4B), and that L-selectin was phosphorylated on serine residues (Fig. 5) suggest that L-selectin may be phosphorylated directly by PKC. Although phosphorylation of L-selectin in human neutrophils freshly isolated from blood could not be demonstrated, this is likely to result from the fact that L-selectin from neutrophils is heavily glycosylated and runs as a diffuse and broad band when analyzed by SDS-PAGE (22, 32). Also, metabolic labeling of these terminally differentiated cells is inefficient, and endoproteolytic release of L-selectin from neutrophils is not completely inhibited in the presence of protease inhibitors (18). These factors in combination are likely to explain the previous inability to demonstrate L-selectin phosphorylation in native leukocytes by us and others (21-24). Nonetheless, a requirement for L-selectin phosphorylation may explain why the amino acid sequence of the cytoplasmic domain of L-selectin has been highly conserved during recent mammalian evolution (particularly the serine residues) (2).
Phosphorylation of L-selectin occurs within seconds following cell exposure to fMLP (Fig. 4C), which is consistent with an in vivo role for L-selectin in leukocyte-endothelium interactions. The ligand binding activity of L-selectin is also rapidly up-regulated within seconds following lymphocyte activation or leukocyte stimulation with pro-inflammatory agents that activate PKC (10). The finding that activation-induced up-regulation of L-selectin adhesive function is staurosporine-sensitive (Fig. 5) suggests that these two events are likely to be related. L-selectin phosphorylation was not required for basal adhesion through this receptor since staurosporine does not eliminate basal adhesion of lymphocytes to HEVs (Fig. 6), and replacement of the serine residues in L-selectin did not inhibit its ability to bind to HEVs. However, deletion of the cytoplasmic region containing the serine residues or the entire cytoplasmic domain eliminated all adhesion. This suggests that this region of L-selectin may mediate intermolecular associations critical for L-selectin function. The functional outcome of these intermolecular interactions could be regulated by phosphorylation of L-selectin in native leukocytes but not in transfected cells. Alternatively, PKC-dependent phosphorylation of other proteins could indirectly affect L-selectin-dependent adhesion and lead to its enhanced adhesive function following leukocyte activation.
Phosphorylation and up-regulated binding activity of L-selectin are both induced more rapidly than endoproteolytic release of L-selectin from the cell surface (10). Consistent with this, phosphorylation of L-selectin on serine residues is not required for endoproteolytic release of L-selectin since the elimination of serine residues within the cytoplasmic domain of L-selectin did not affect receptor cleavage. Whether serine phosphorylation of L-selectin is required for signal transduction through L-selectin following its ligation remains an open issue. Recent studies have demonstrated rapid tyrosine phosphorylation of L-selectin following cross-linking by antibodies (33). However, ligation of L-selectin through conserved ligand-binding regions within the lectin domain induces rapid and potent intercellular adhesion in human, mouse, and rat leukocytes that is also induced in cell lines expressing L-selectin lacking the cytoplasmic serine residues (L-SS/AA) or the cytoplasmic tyrosine residue (34). Although it is not currently feasible to express mutant L-selectin molecules in primary leukocytes to assess the biological significance of L-selectin phosphorylation in vivo, the current studies provide a rationale for further studies examining this issue.
The finding that L-selectin phosphorylation was inhibited by both a PKC inhibitor and pertussis toxin suggests that L-selectin phosphorylation may be one of the rapid G-protein-regulated activation events involved in leukocyte interactions with vascular endothelium under physiologic flow conditions (35, 36). In current models of leukocyte recruitment to inflammatory sites, initial interactions between selectins and their ligands result in rolling followed by chemoattractant-mediated integrin activation leading to firm adhesion (3, 4, 35). Our results suggest that activation of chemoattractant receptors induces L-selectin phosphorylation through PKC-dependent pathways. Phosphorylation of L-selectin may induce transient changes in L-selectin binding activity that may contribute directly to leukocyte interactions with endothelial cells and account in part for lineage-specific differences in leukocyte migration (10).
We thank Drs. S. Rosen, H. Metzger, and S. F. Schlossman for providing reagents and Xiu Qin Zhang for expert technical assistance.