(Received for publication, May 5, 1997, and in revised form, July 1, 1997)
From ICOS Corporation, Bothell, Washington 98021
Intercellular adhesion molecule-3 (ICAM-3), a
ligand for 2 integrins, elicits a variety of
activation responses in lymphocytes. We describe a functional mapping
study that focuses on the 37-residue cytoplasmic region of ICAM-3.
Carboxyl-terminal truncations delineated portions involved in T cell
antigen receptor costimulation, homotypic aggregation, and cellular
spreading. Truncation of the membrane distal 25 residues resulted in
loss of T cell antigen receptor costimulation as determined by
interleukin 2 secretion. Aggregation and cell spreading were sensitive
to truncation of the membrane distal and proximal thirds of the
cytoplasmic portion. Phosphoamino acid analysis revealed that ICAM-3
from activated cells contained phosphoserine and
phosphopeptide mapping identified Ser489 as a site of
phosphorylation in vivo. Mutation of Ser489 or
Ser515 to alanine blocked interleukin 2 secretion,
aggregation and cell spreading, while mutation of other serine residues
affected only a subset of functions. Ser489 was a
phosphorylation site in vitro for recombinant protein
kinase C
. Finally, treatment of Jurkat cells with chelerythrine
chloride, a protein kinase C inhibitor, prevented ICAM-3-triggered
spreading. This study delineates separable regions and amino acid
residues within the cytoplasmic portion of ICAM-3 that are important
for T cell function.
ICAM-31 (CD50) is a
member of the Ig superfamily sharing sequence and functional attributes
with ICAM-1, -2, -4 (Landsteiner-Weiner blood group glycoprotein), and
-5 (telencephalin). ICAM-3 binds to LFA-1 (CD11a/CD18) and the newly
described integrin d/CD18 (1). It is constitutively
expressed at high levels by most hematopoietic cells, leading to the
suggestion that it is involved in early activation steps of an
inflammatory response (2-4).
ICAM-3 has been functionally characterized with respect to the five Ig-like extracellular domains. The first amino-terminal domain binds LFA-1 via conserved residues also found in ICAM-1 (5). These conserved sequences of ICAM-3 and -1 may bind to distinct sites of the I domain of LFA-1, suggesting that non-conserved domain 1 residues might contribute to integrin binding (6, 7).
Numerous intracellular signaling events have also been observed to be
affected by ICAM-3 engagement. Specifically, activation of
intracellular calcium flux and stimulation of tyrosine kinase activity
possibly via non-receptor tyrosine kinases p56lck and
p59fyn were seen (8, 9). ICAM-3 engagement has also been
observed to up-regulate 1 and
2 integrin
function, and to trigger phosphorylation of the
cyclin-dependent kinase cdc2 (10-12). Little information, however, is available regarding the molecular mechanisms of these phenomena.
Here we report that ICAM-3 engagement initiates several distinct aspects of lymphocyte function, which involve the 37-amino acid cytoplasmic portion. For these analyses, we developed and characterized an ICAM-3-deficient human T-leukemic Jurkat cell line. Using these cells and gene transfer techniques, a functional map of the cytoplasmic region of ICAM-3 with respect to TCR accessory molecule function, homotypic aggregation, and cell spreading was generated. These data pinpoint serine residues, particularly serine 489, as critical for ICAM-3 function.
Jurkat 77 (J77, a gift from Dr. S. Burakoff, Dana Farber Cancer Research Institute, Boston, MA) and the ICAM-3-deficient J77.50.3 cells were maintained in RPMI complete medium (RPMI supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 µg/ml streptomycin, 2 mM L-glutamine, and 1 mM sodium pyruvate) in humidified 5% CO2 at 37 °C.
Monoclonal AntibodiesMurine mAb used in this study were as
follows. Anti-ICAM-3 (CD50) mAb ICR1.1 (IgG2a), ICR2.1 (IgG1), ICR9.2
(IgG2a), and anti-ICAM-1 (CD54) mAb 18E3D (IgG1) were generated by the
ICOS hybridoma facility. Hybridoma lines secreting anti-TCR (CD3e) OKT3
(IgG2a), anti-CD11a TS1/22 (IgG1), anti-CD18 TS1.18 (IgG1), anti-CD45
4B2 (IgG2a), and anti-MHC class I W6/32 (IgG2a) were obtained from
American Type Culture Collection, Rockville, MD. These mAb were
purified by protein A column chromatography of mouse ascites fluid.
Purified mAb were dialyzed against and stored in PBS. Isotype-matched
control mAb used were UPC10 (IgG1) and MOPC 21 (IgG2a) (Sigma). Anti-HA mAb 12CA5 (IgG2a) was from Boehringer Mannheim. Fluorescein
isothiocyanate-conjugated sheep anti-mouse IgG F(ab)2 was
purchased from Sigma.
A variant of Jurkat 77 cells deficient in the expression of ICAM-3 (J77.50.3) was generated by two rounds of indirect staining and cell sorting using a mixture of ICR1.1 and 9.2 mAb (4). J77.50.3 cells were compared with the parental line for surface expression of numerous membrane proteins by indirect cytofluorometry (FACSCAN, Becton-Dickinson, Mountain View, CA) and found to exhibit similar levels for all except ICAM-3.
Ten micrograms of total RNA isolated from parental J77 and J77.50.3 cells was subjected to blotting, hybridization, and washing as described (13). Labeled probes were generated by random priming of the entire ICAM-3 or glyceraldehyde-3-phosphate dehydrogenase cDNA (14).
ICAM-3 Deletion and Point Mutation ConstructsCoding sequences for HA epitope-tagged ICAM-3 proteins were generated as described (15). To engineer epitope-tagged full-length ICAM-3 construct, three separate PCR fragments that encoded 1) the signal sequence (preceded by a unique HindIII site and Kozak sequence), 2) Ig domains (IgD) I-II of ICAM-3, and 3) a triple (3×) influenza hemagglutinin (HA) epitope tag sequence were synthesized and gel-purified (16). The fragments were combined using PCR in the following order: ICAM-3 signal sequence, HA tag, and ICAM-3 IgD I-II. This product was ligated as a HindIII/ScaI fragment with a ScaI/EcoRI cDNA fragment containing the remainder of the ICAM-3 coding sequence into the HindIII/EcoRI sites of expression vector pMH-neo and all PCR products sequenced (17).
Cytoplasmic region deletions were generated as follows. The region of
coding sequence for the extracellular domains described above contained
on a HindIII/SacI fragment was combined with
SacI/EcoRI PCR fragments encoding cytoplasmic
domain truncations and ligated to the
HindIII/EcoRI sites of pMH-neo. The PCR fragments
were synthesized using the following primers: 1) 5 common anchoring primer CATAATGGTACTTATCAGTGC, and 2) 3
primers D505 (
1/3CT), ATATAGCGGCCGCGGATCCTCACTGCATAGACGTGAG; D493 (
2/3CT),
ATATAGCGGCCGCGGATCCTCACCTAACATGGTAACT; and D484 (
CT),
ATCACTATGCGGCCGCTCAGTGTCTCCTGAAGACGTACAT. Primer D484 contained a
change at codon 483 to increase the membrane anchor region of the
maximal cytoplasmic region truncation. Amino acid numbering uses
the mature amino terminus for the first residue.
To generate point mutations, the ICAM-3 cDNA was subcloned as a NotI/EcoRI fragment into M13 BM21 replicative form DNA (Boehringer Mannheim) and primers used for mutagenesis by the Kunkel method were designed to make alanine changes at the following codons: serine 487, serine 489, leucine 499, serine 496, serine 503, and serine 515 (18). Leucine 499 was chosen as a control for mutational effect, since it is not a potential phosphorylation site and a conservative change to alanine is expected to maintain similar overall charge. All mutants were sequenced, and each was subcloned as a SacI/EcoRI fragment along with the HindIII/SacI fragment described above into pMH-neo.
Expression and Selection in J77.50.3 CellsJ77.50.3 cells (107) in 0.2 ml of PBS were electroporated in the presence of 100 µg/ml plasmid DNA using a BTX 6000 device (126 V, 1710 microfarads, 72 milliamps, 0.2-cm electrode gap). Drug-resistant cells (G418, 1.25 mg/ml) were screened by indirect fluorescence flow cytometry with ICR1.1, OKT3, or isotype-matched control antibody. Two independent cell lines were chosen for functional studies from each transfection based on similar mean channel fluorescence measurements.
Cell PanningTo enrich for cell lines that expressed higher mean ICAM-3 levels, the drug-resistant lines were panned on mAb-coated plastic. Bacteriologic Petri plates were incubated with 8 ml of ICR2.1 (10 µg/ml) in PBS for 2 h at 37 °C. The plates were rinsed with PBS and cells seeded onto the mAb-coated plastic surface for 8 min at 25 °C. To remove non-adherent cells, the plates were rocked and aspirated. The plates were rinsed with PBS, checked visually to determine the absence of non-adherent cells, and adherent cells removed by trituration. Cells harvested in this manner were expanded and subjected to indirect fluorescence analysis using flow cytometry.
ICAM-3/TCR Costimulation AssayPlates (96-well, Corning 25860) were coated with OKT3 (0.5 µg/ml, 50 µl/well) in PBS for 16 h at 4 °C. The coating was removed and replaced with PBS alone or mAb at 10 µg/ml in PBS and incubated at 37 °C for 2 h. Wells were rinsed twice with PBS and 2 × 105 cells added in 0.25 ml of RPMI/well. Plates were incubated at 37 °C for 16 h. Conditioned medium from duplicate wells was pooled, diluted serially, and assayed for IL2 concentration by enzyme-linked immunosorbent assay (Biosource International, Camarillo, CA). A dose for OKT3 (25 ng/well) and ICR1.1 (500 ng/well) was chosen for co-stimulation experiments. Assays were repeated a minimum of three times, with similar results observed in each experiment.
Cell Spreading AssayDishes (T, 0.5-mm glass; Bioptechs
Inc., Butler, PA) were coated with 0.5 ml of mAb in PBS (10 µg/ml).
Dishes was incubated at 37 °C for 2 h and rinsed twice with
PBS. Cells (2 × 104) were seeded onto coated surfaces
for 15 min at 37 °C. Plates were held at 37 °C in the Bioptechs
stage insert while being photographed with Ilford Pan F film using a
Nikon Diaphot microscope and DIC optics. For PKC inhibitor studies,
chelerythrine chloride (in Me2SO) was added to cells at a
final concentration of 50 µM and incubated at 37 °C
for 10 min prior to seeding into coated dishes.
Cells were pelleted and resuspended
at 8 × 105/ml in complete medium and 0.25 ml
distributed to duplicate wells of a 96-well flat bottom plate. mAb
ICR1.1 or MOPC 21 control were added to a final concentration of 10 µg/ml. After 1 h of incubation at 37 °C, the percentage of
aggregated cells was determined as described previously (19).
Quantitative determinations were made by counting free cells in five
separate squares of a gridded ocular centered over the well at 125 × magnification. Percent aggregation = {1 (no. of free
cells experimentally treated/no. of free cells control-treated)} × 100.
Cells (5 × 107) were starved of methionine and cysteine for 1 h prior to labeling with 100 µCi/ml [35S]methionine and -cysteine (Tran35S-Label; ICN Biochemicals, Irvine, CA) in RPMI, 10% dialyzed fetal bovine serum.
Cells for phosphorylation studies were rinsed in phosphate-free RPMI and labeled with 0.5 mCi of inorganic 32P for 4 h at 37 °C.
Cell Solubilization, Immunoprecipitation, and SDS-PAGELabeled cell pellets were suspended in 1 ml of cold lysis buffer (PBS containing 1% Triton X-100, 1 mM PMSF, 1 mM Na3VO4, 1 mM Na2MoO4) and incubated on ice for 20 min with occasional rocking. The insoluble fraction was pelleted by centrifugation in a table top microcentrifuge. Soluble proteins were transferred to a fresh tube and 0.1 ml of Sepharose 4CL beads added (50% slurry equilibrated in lysis buffer without PMSF; Pharmacia Biotech Inc., Uppsula). The tube was rocked for 16 h, after which the beads were briefly spun down. The clarified supernatant was transferred to a fresh tube and antibody added to 10 µg/ml final concentration. Immune complexes were formed by incubation on ice for 1 h and harvested by incubation with protein A beads. The immune complexes were pelleted and washed two times with 1 ml of cold 1% Triton X-100, 1 M NaCl, 1 mM PMSF, 1 mM Na3VO4, 1 mM Na2MoO4, and once with 1 ml of cold lysis buffer. The remaining proteins were eluted by addition of reducing SDS-PAGE loading buffer, boiled for 5 min, and separated by gel electrophoresis.
Immune complexes of labeled proteins were separated by SDS-PAGE, and
transferred to polyvinylidene difluoride membrane (Millipore Corp.,
Bedford, MA). Autoradiography of gels containing
32P-labeled proteins was conducted using X-Omat film
(Eastman Kodak Corp.) with a single intensifying screen at 70 °C.
Gels of 35S-labeled proteins were impregnated with fluor
and exposed to film at
70 °C.
Autoradiographs were used to localize 32P-labeled ICAM-3. Bands were excised from the polyvinylidene difluoride sheet and the proteins partially acid hydrolyzed and separated as described in (20). Briefly, the samples were dried in vacuo and resuspended in 6 µl of pH 1.9 buffer containing unlabeled phosphoamino acid standards (Sigma). A portion of each sample, representing equal Cerenkov counts, was spotted on cellulose TLC plates and phosphoamino acids separated by high voltage thin layer electrophoresis (HTLE-7000; CBS Scientific, Del Mar, CA). After ninhydrin staining of the standards, the plates were exposed to film for autoradiography.
Phosphorylation sites were determined by tryptic peptide mapping using in vivo 32P-labeled proteins. Labeled protein bands were incubated with L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin (Worthington Biochemical Corp., Freehold, NJ) for 18 h. The resulting peptide mixture was spotted on a TLC plate and subjected to charge separation and ascending chromatography.
Protein Kinase AssayHuman protein kinase C was
amplified by PCR with a primer that contained a 6-histidine tag at the
carboxyl terminus of the protein coding sequence as described (21). The
cDNA was subcloned into a BACMID vector (Life Technologies Inc.),
and recombinant virus was generated as described by the manufacturer.
Infected Sf9 cells were lysed in hypotonic buffer (20 mM
Tris, pH 7.5, 250 mM sucrose, 5 mM EDTA, 5 mM EGTA, 100 µg/ml each aprotinin and leupeptin, 1 mM aminoethyl benzenesulfonyl floride) by Dounce homogenization. Active soluble protein kinase was separated from insoluble material by centrifugation (16,000 × g) and
0.04-ml assays performed using 50 µg/ml total protein and 100 µM substrate peptide as described (21). The following
substrates were synthesized: ICAM-3 CT, amino acids 482-518
(REHQRSGSYHVREESTYLPLTSMQPTEAMGEEPSRAE), SCR CT
(ARSTEQQGMYAESESEELRPGYPEHRSTHTMLPRSVE), SGS
(biotin-FREHQRSGSYHVREE), and SGS-P
(biotin-FREHQRSGS(PO4)YHVREE).
Jurkat cells (J77) examined by indirect fluorescence
cytometry with antibodies to ICAM-3 routinely display two populations of cells (Fig. 1A). The bulk
(97%) of these are ICAM-3-positive, while a small population (3%) is
ICAM-3-negative. Enrichment of the ICAM-3-negative population by
sequential rounds of cell sorting generated a population displaying
>97% ICAM-3-negative cells that we have termed J77.50.3. RNA analysis
showed that, in J77.50.3 cells, synthesis of the 2.2-kb ICAM-3 message
was below detectable levels (Fig. 1B).
Surface expression of numerous proteins was assessed in J77.50.3 and the parental J77 line. Both populations exhibited similar fluorescence profiles for all mAb studied including CD3e, CD11a, CD18, and CD45 (Fig. 1A), indicating that J77.50.3 cells were similar to J77 except for the ICAM-3 deficiency.
Expression of ICAM-3 in J77.50.3 Cells Restores FunctionOptimal T cell activation is thought to require two
signals: one from the antigen receptor and the other from one or more of a large number of accessory molecules including ICAM-3 (22). To
confirm that ICAM-3/TCR engagement was costimulatory, Jurkat T
cells were seeded onto ICR1.1 coimmobilized with increasing concentrations of OKT3. A dose-dependent increase in IL2
production was observed (Fig.
2A). Cells exposed to either
immobilized mAb alone showed no induction of IL2 secretion.
To determined if loss of ICAM-3 expression would impair costimulation, J77.50.3 cells were seeded into mAb-coated wells under conditions that stimulated the parental cells. ICAM-3-deficient J77.50.3 did not secrete IL2 (Fig. 2B). Both cell lines responded to a greater concentration of OKT3 by secreting IL2. Neither cell line responded to a combination of anti-ICAM-1 mAb (18E3D) and OKT3. These data reveal that TCR signaling and the synthetic machinery for IL2 production in J77.50.3 cells was intact. Further, co-engagement of ICAM-1 and CD3 was insufficient to produce IL2.
To evaluate whether ICAM-3 expression in J77.50.3 cells would
complement the phenotypic defect, cells were transfected with either a
control vector or HA-tagged ICAM-3 (ICAM-3FL). Cells were selected, and
several independent lines that maintained stable surface expression
were identified (Fig. 3A).
Inclusion of the HA tag allowed for validation that the expressed form
of ICAM-3 in the transfected cells was from the introduced DNA
construct rather than re-expression of the endogenous gene. Indeed,
surface staining for either ICAM-3 or HA tag epitopes showed similar
levels of fluorescence in the populations (Fig. 3A). While
the control-transfected lines lacked the ability to be stimulated by
the co-immobilized mAb, J77.50.3 cells expressing ICAM-3FL responded to
the costimuli by secreting IL2 into the medium as did the parental J77
cells (Fig. 3B). Therefore, expression of ICAM-3 by J77.50.3
cells restored their ability to respond to ICAM-3/TCR costimulation.
ICAM-3 Cytoplasmic Region Truncations Have Distinct Functional Consequences
J77.50.3 cells expressing the following HA
epitope-tagged cytoplasmic tail truncations were generated to grossly
map the cytoplasmic region of ICAM-3: 1/3CT (Gln505
terminus),
2/3CT (Arg493 terminus), and
CT
(His484 terminus) (Fig. 4).
Characterization of cells expressing each deletion included monitoring
surface expression (Fig. 5A
and Table I) and immunoprecipitation from
lysates of cells that had been metabolically labeled with
35S (Fig. 5B). SDS-PAGE analysis showed that the
relative migration of the truncated proteins was of the expected sizes
of ~120-140 kDa.
|
The truncations were tested for their ability to trigger J77.50.3 cells
to secrete IL2 when costimulated with anti-ICAM-3/TCR mAb. Conditioned
media from cells expressing either ICAM-3FL or 1/3CT forms showed
5.4- and 4.8-fold induction, respectively, when normalized for IL2
secretion in negative control-treated wells (Table I). Cells expressing
either
2/3CT or
CT forms secreted about 60% less IL2 (2.2-fold
each). All of the cell lines tested responded to a more concentrated
dose of OKT3 alone by secreting similar levels of IL2 (data not
shown).
Immunoregulation of leukocytes has been hypothesized to occur via
aggregate formation in which paracrine effects of cytokines (both
positive and negative) regulate progression of a cellular immune
response (23). J77.50.3 cells expressing ICAM-3FL treated with ICR1.1
mAb responded by forming aggregates (58% aggregated, Table I). This is
not due to direct cross-linking of cells by mAb, since Fab fragments of
ICR1.1 trigger aggregation as
well.2 Expression of 1/3CT
resulted in 29% aggregated cells. Truncation to residue
Arg493 (
2/3CT) resulted in no further reduction, while
cells that expressed
CT aggregated to the same level as vector
control cells.
The interface between a T cell and antigen-presenting cell (APC) is an
area of intercellular adhesion that is both dynamic and highly intimate
(24). T cells spread over a large region of the APC surface during the
early phase of contact that coincides with a transient calcium flux
(25, 26). Since ICAM-3 is present prior to activation, it is plausible
that it would be involved in spreading. J77.50.3 cells expressing
ICAM-3FL rapidly flattened and spread on ICR1.1, but remained rounded
on the isotype control coating (Fig. 6,
A and B, and Table I). Cells expressing either 1/3CT or
2/3CT also spread (Fig. 6, C and D),
while cells expressing
CT were incapable of spreading (Fig.
6E).
Phosphoamino Acid Composition of ICAM-3
Intracellular protein
phosphorylation regulates many signaling cascades that lead to
activation or morphological changes. Consequently, we examined the
phosphorylation of ICAM-3 under conditions that triggered Jurkat cell
activation. ICAM-3FL immunoprecipitated from unstimulated J77.50.3
cells had a basal level of 32P incorporation (Fig.
7A, lane 2). Cells
treated with either OKT3 cross-linking or PMA had an increase of
32P uptake onto ICAM-3 (Fig. 7A, lanes
3 and 4). Basal and inducible phosphorylation was also
observed with peripheral blood leukocytes.2 Phosphoamino
acid analysis revealed that PMA and OKT3 cross-linking elicited
phosphorylation of ICAM-3 only on serine residues (Fig. 7B).
Ser489 Is Phosphorylated in Vivo
Truncation
analysis indicated that residues 485-515 were important for
costimulation, while aggregation and cell spreading depended on the
presence of residues 506-515 and 485-493. To investigate whether
these ICAM-3-triggered functions correlated with phosphorylation of
specific serine residues in these functionally important regions, phosphorylation site analysis was undertaken by point mutation and
peptide mapping. 32P-Labeled ICAM-3FL from PMA-treated
cells was subjected to tryptic peptide mapping, which revealed several
major phosphopeptides (Fig. 8,
A and B). The pattern of separated
phosphopeptides was similar whether ICAM-3 was derived from cells
stimulated with PMA or CD3 cross-linking (data not shown). The tryptic
phosphopeptide map of 32P-labeled ICAM-3 Ser487
Ala resulted in the wild type pattern of labeled phosphopeptides (Fig. 8, compare B and C). The Ser489
Ala mutation resulted in complete loss of signal for one of the
major phosphopeptides (Fig. 8D), indicating that
Ser489 is a bona fide phosphorylation site
in vivo. Alanine substitution mutants of serine residues
496, 503, and 515 were also subjected to in vivo labeling
and endoproteolytic peptide mapping analyses; however, the results were
inconclusive (data not shown).
Mutation of Ser489 Blocks ICAM-3 Function
Phenotypic effects of the alanine point mutants were
tested in homotypic aggregation, cell spreading and TCR costimulation assays. Aggregation of Ser489 Ala was about 20% of
ICAM-3FL levels, similar to the vector control transfectants (12, 58, and 16%, respectively; Table I). Ser489
Ala also had a
deleterious effect on cell spreading. Conversely, Ser487
Ala and Leu499
Ala had no detectable inhibition of
aggregation or spreading. Induction of IL2 secretion by
Ser489
Ala in the TCR costimulation assay was also
reduced about 50%, compared with Leu499
Ala and
ICAM-3FL (2.9-, 6.0-, and 5.4-fold, respectively; Table I). The effect
of Ser489
Ala on costimulation was comparable to the
CT truncation. Cells expressing Ser487
Ala were
equally impaired when assayed for costimulation but not for aggregation
or spreading. Ser496
Ala reduced spreading to 33%,
while costimulation and aggregation remained intact. Ser503
Ala blocked costimulation completely and spreading partially, yet
left aggregation intact. Ser515
Ala abrogated
costimulation, aggregation, and spreading.
Numerous protein kinases have been characterized as to the
sequence specificity of their substrates. For the family of protein kinase C (PKC) isoforms, a consensus substrate sequence is
RXXS/T (27). Examination of the ICAM-3 Ser489
sequence context suggested that it might be a PKC substrate. In
addition, TCR cross-linking and PMA treatments, both of which activate
multiple PKC isoforms, resulted in the induction of serine phosphorylation of ICAM-3 (Fig. 7). PKC was chosen for in
vitro kinase assay since 1) it is found in abundance in cells of
the hematopoietic lineage, 2) it can activate the AP-1 element of the
IL2 promoter when overexpressed, and 3) it selectively translocates to
the T cell/APC contact region in an antigen-dependent
manner (28-31). A 37-residue peptide representing the entire
cytoplasmic region (amino acids 482-518) was phosphorylated in
vitro by recombinant human PKC
(Fig.
9A). In contrast, a scrambled
ICAM-3 peptide had little detectable incorporation. To address whether
Ser489 was a PKC
phosphorylation site in
vitro, the phosphorylation of shorter substrate peptides (amino
acids 481-495) containing Ser489 (SGS) or
phospho-Ser489 (SGS-P) was also evaluated. Incorporation of
32P was found only with the SGS peptide and not with the
synthetically phosphorylated version, SGS-P (Fig. 9). These data
indicate that PKC
phosphorylation of the cytoplasmic tail was
sequence-specific and that Ser489 was a phosphorylation
site in vitro.
Since PKC phosphorylation of Ser489 occurred in
vitro and mutation of this amino acid blocked function in
vivo, we sought to implicate PKC activity in an ICAM-3-triggered
event like cell spreading. Cells expressing ICAM-3FL pretreated with
the PKC inhibitor chelerythrine chloride were effectively blocked for
spreading, whereas vehicle-treated cells spread as usual (Fig.
9B). This result suggests that PKC activity is required for
ICAM-3-dependent cell spreading.
We investigated the functional requirements of the cytoplasmic region of ICAM-3 by expression of truncated or point mutated proteins in a variant of the human T leukemic cell line Jurkat. A Jurkat cell line deficient in ICAM-3 expression (J77.50.3) was developed and characterized with regard to surface protein expression, message synthesis, and costimulatory phenotype (Figs. 1 and 2). Expression of surface ICAM-3 restored accessory molecule function as measured by secretion of the T cell activation marker IL2 (Fig. 3). Therefore, the functional deficit in the cell line can be complemented by expression of a single protein, ICAM-3.
The natural occurrence of a subpopulation of ICAM-3-deficient cells in unselected Jurkat cultures is curious. Others have reported that the Jurkat cell line was mixed for the expression of ICAM-3, suggesting that this phenotype is inherent to the cell line itself (32). The ICAM-3 deficiency was stable in continuous culture, and cells with wild type levels of ICAM-3 expression do not arise.
This model cell line was quite useful for investigating the
structure/function relationship of the cytoplasmic region of ICAM-3. The 37 residues of the cytoplasmic region are grouped as alternating hydrophilic-hydrophobic-hydrophilic segments, which were used to divide
the region for mapping studies (33). The 1/3CT truncation, which
lacked five charged residues that gave the distal portion of the native
tail its hydrophilic nature, terminated with Gln505 and
contained the hydrophobic core residues 498-501 (Fig. 4). The
2/3CT
truncation removed the hydrophobic core residues and terminated with
Arg493. The
CT truncation terminated with
His484 and allowed enough positive charge to achieve
membrane stop transfer in the absence of the rest of the cytoplasmic
tail.
Functional evaluation of the cytoplasmic region consisted of measuring
IL2 secretion following costimulation and anti-ICAM-3-triggered aggregation and cell spreading of stable transfectants expressing truncations or point mutations. IL2 secretion was significantly impaired in cells that expressed truncations of residues 485-505 (2/3CT and
CT) and partially impaired by truncation of residues 506-518 (
1/3CT). Point mutation analysis showed that particular serine residues (serines 487, 489, 503, and 515), when mutated, would
also block triggering of IL2 secretion. These observations suggest that
the TCR accessory function of ICAM-3 required the majority of its
cytoplasmic region. Since some costimulatory activity remained with
CT truncation compared with the vector control transfectants (2.2 and
0.52, respectively), this suggests that the rest of the protein has a
role. Whether this activity is via associated transmembrane proteins or
through simple ICAM-3 mAb-mediated enhancement of cell binding to the
OKT3-coated surface is unclear. Aggregation and spreading were
partially (50-70%) inhibited with deletion of residues 506-518 and
an additional 50% with further deletion of amino acids 485-493. Loss
of the hydrophobic core residues 494-505 did not result in
incrementally greater deficits on aggregation or spreading. In
agreement, the point mutant analysis showed that Ser489
Ala and Ser515
Ala blocked, while Ser496
Ala, Leu499
Ala, and Ser503
Ala had
no deleterious effect on aggregation. Effects of the point mutants on
spreading were also in agreement with those of the truncations. In
particular, mutation of serine residues in the membrane distal or
proximal regions either partially or completely inhibited spreading.
These results suggest that two distinct and well separated regions
(amino acids 506-518 and 485-493) contribute to ICAM-3-triggered
aggregation and spreading. Interestingly, mutational effects on
costimulation, aggregation, and cell spreading were not absolutely
coincidental. This suggests that these three functions are separable
and not necessarily interdependent. In general, however, these ICAM-3
triggered functions were effected by mutations within the hydrophilic
regions proximal and distal to the membrane-spanning segment.
Since many protein functions are regulated by phosphorylation status, the phosphoamino acid composition of ICAM-3 was determined under basal and stimulated conditions. In vivo labeling of ICAM-3, via the intracellular ATP pool, resulted in phosphoserine content only (Fig. 7). In addition, ICAM-3 from activated blood-derived leukocytes (including neutrophils) and leukocytic cell lines, failed to yield immunoreactivity with anti-phosphotyrosine mAb.2 In agreement with our results, Lozano et al. (32) have reported inducible serine phosphorylation of ICAM-3. In contrast, Skubitz et al. (34) reported inducible phosphorylation predominantly on tyrosine. This discrepancy may be attributed to differences in cell isolation procedures or the manner in which the cells were labeled.
We determined that Ser489 is a major phosphorylation site
in vivo by tryptic peptide mapping of serine to alanine
point mutants and that recombinant PKC phosphorylated this site
in vitro (Figs. 8 and 9). The requirement for
Ser489 in ICAM-3 function was demonstrated by mutation of
this site, which abrogated all ICAM-3-triggered events tested (Table
I). Ser489 is closely apposed to the cytoplasmic face of
the plasma membrane, which could facilitate its phosphorylation by PKC
upon recruitment to the lipid microenviroment of the membrane by events
like TCR stimulation or phorbol ester treatment. In fact, only under
conditions of TCR stimulation or PMA treatment have increased levels of
ICAM-3 phosphorylation been found by us. Since phosphorylation is a
dynamic event, in which rapid dephosphorylation can occur, specialized conditions may be required for the isolation of hyper-phosphorylated ICAM-3 from aggregated or spread cells. Alternatively, ICAM-3-triggered aggregation and spreading require Ser489 for structural
integrity independent of phosphorylation state. In this case, an
alanine mutant blocks function by altering important structural
features. Whether other serine residues (496, 503, and 515) function,
at least in part, by their phosphorylation status remains to be
determined.
The studies described here delineate portions of the cytoplasmic region of ICAM-3 that are important for T cell biology. Since the regions implicated in TCR costimulation partially encompass regions linked to aggregation and cell spreading, our data suggest that these functions might share underlying mechanisms, perhaps reorganization of the cytoskeleton. Links between cytoskeleton dynamics and activation have been observed in T cells during the process of APC/T adhesion and contact (35-38). At initial contact, engagement of TCR by MHC-antigen complexes activates PKC which leads to phosphorylation of ICAM-3, similar to that observed using anti-CD3 mAb. T cell spreading over the APC surface, perhaps driven by actin polymerization, provides threshold levels of TCR triggering by antigen-MHC and allows other co-stimulatory molecular interactions to occur, such as CD28-B7. For sustained T cell signaling, actin-based cytoskeletal changes are required and, as the present work suggests, could be induced by engagement of ICAM-3 (37). Preliminary studies suggested that ICAM-3 engagement induced actin polymerization.2 Later, dephosphorylation of ICAM-3 could contribute to cell dissociation and rounding, as observed with Ser489 mutation, contributing to clonal expansion. Further biochemical studies are required to delineate the mechanisms by which T cell behavior is regulated by the functionally important amino acids of the ICAM-3 cytoplasmic region identified here.
We thank Dr. Steven J. Burakoff for the Jurkat 77 cell line, Drs. Meri Hoekstra and Don Staunton and Louise Band for critical reading of the manuscript, and Dr. Owen Lockerbie for advice during the project. We thank Christi Wood and Dina Leviten for expert DNA sequencing and synthesis, Tony DiMaggio and Adam Kashishian for advice with the phosphorylation studies, David Hynds for peptide synthesis, and Rick Jasman and Anne Jensen for expert monoclonal antibody production and characterization.