Identification of a Proline-Rich Sequence in the CD2 Cytoplasmic Domain Critical for Regulation of Integrin-Mediated Adhesion and Activation of Phosphoinositide 3-Kinase

Abstract

The CD2 molecule is one of several lymphocyte receptors that rapidly initiates signaling events regulating integrin-mediated cell adhesion. CD2 stimulation of resting human T cells results within minutes in an increase in β1-integrin-mediated adhesion to fibronectin. We have utilized the HL60 cell line to map critical residues within the CD2 cytoplasmic domain involved in CD2 regulation of integrin function. A panel of CD2 cytoplasmic domain mutants was constructed and analyzed for their ability to upregulate integrin-mediated adhesion to fibronectin. Mutations in the CD2 cytoplasmic domain implicated in CD2-mediated interleukin-2 production or CD2 avidity do not affect CD2 regulation of integrin activity. A proline-rich sequence, K-G-P-P-L-P (amino acids 299 to 305), is essential for CD2-mediated regulation of β1 integrin activity. CD2-induced increases in β1 integrin activity could be blocked by two phosphoinositide 3-kinase (PI 3-K) inhibitors or by overexpression of a dominant negative form of the p85 subunit of PI 3-K. In addition, CD2 cytoplasmic domain mutations that abrogate CD2-induced increases in integrin-mediated adhesion also ablate CD2-induced increases in PI 3-K enzymatic activity. Surprisingly, CD2 cytoplasmic domain mutations that inhibit CD2 regulation of adhesion do not affect the constitutive association of the p85 subunit of PI 3-K association with CD2. Mutation of the proline residues in the K-G-P-P-L-P motif to alanines prevented CD2-mediated activation of integrin function and PI 3-K activity but not mitogen-activated protein (MAP) kinase activity. Furthermore, the MEK inhibitor PD 098059 blocked CD2-mediated activation of MAP kinase but had no effect on CD2-induced adhesion. These studies identify a proline-rich sequence in CD2 critical for PI 3-K-dependent regulation of β1 integrin adhesion by CD2. In addition, these studies suggest that CD2-mediated activation of MAP kinase is not involved in CD2 regulation of integrin adhesion.

T lymphocytes continuously migrate throughout the body, mediating immune responses to foreign antigens. The capacity for normal T cells to function, develop, and migrate depends upon adhesive contacts with other cells as well as with extracellular matrix (ECM) components (76). These adhesive contacts are regulated by and initiate a complex series of elegantly controlled molecular signaling events. Members of the integrin superfamily of adhesion molecules are involved in the cell-cell and cell-ECM interactions that are essential for T-lymphocyte function and migration. The integrins are a family of αβ heterodimeric cell surface receptors with a vast tissue distribution (67). There are at least 20 different integrin heterodimers, with each heterodimer containing 1 of 8 β chains and 1 of 14 α chains. Members of the β1 integrin subfamily bind to components of the ECM, such as fibronectin (FN), as well as cell surface counter-receptors, such as VCAM-1 (34). The functional activity of integrins on circulating leukocytes is dynamically regulated by signaling events (21, 58). For example, resting human T cells express the α4β1 and α5β1 integrins but adhere only weakly to FN. Activation of resting T cells results in a rapid, transient increase in β1-integrin-mediated adhesion to FN that does not involve an alteration in the level of β1 integrin cell surface expression (74). Activation-dependent regulation of integrin activity is critical to effective T-cell recognition of antigen during T-cell activation, as well as efficient and specific movement of lymphocytes across vascular endothelium into extravascular tissue sites (40, 77).

A number of different activation conditions are capable of increasing integrin-mediated adhesion of T cells. These activation conditions include pharmacological activators of intracellular signaling pathways, such as the phorbol ester phorbol 12-myristate 13-acetate (PMA) (24, 74). Direct activation of integrins can also be achieved with the use of unique integrin-specific antibodies or alterations in the divalent cations present in the extracellular environment (3, 28, 51, 71). These modes of increasing integrin adhesiveness are presumed to bypass requirements for intracellular signaling that are normally needed to activate integrins. More significantly, activation of a number of different cell surface receptors on T cells can initiate signaling events that result in activation of β1 and β2 integrin-mediated adhesion (11, 24, 73, 74, 80, 81, 84).

One of these “integrin regulators” is the CD2 molecule, a 45- to 55-kDa immunoglobulin superfamily member that was first identified as an important signaling molecule on human T cells (7, 20, 56). Studies of human T cells demonstrated that unique pairs of CD2-specific monoclonal antibodies (MAbs) can induce T-cell proliferation in the absence of engagement of the CD3–T-cell receptor (CD3-TCR) complex (56). CD2 stimulation can also induce increased adhesion of human peripheral T cells and CD2⁺ human T cell lines to ICAM-1, FN, and laminin (57, 74, 84). CD2 itself is also an adhesion molecule and, as with integrins, the adhesive activity of CD2 can be regulated by T-cell activation (31, 33). Biochemical studies have demonstrated that CD2 stimulation can initiate a multitude of intracellular signaling events, including (i) tyrosine phosphorylation (64); (ii) Ca²⁺ flux (90); (iii) increased intracellular cyclic AMP (cAMP) (32); and (iv) the activation of phospholipase C-γ1 (61), protein kinase C (5), phosphoinositide 3-kinase (PI 3-K) (72), the tyrosine kinases p56^lck, p59^fyn, and Itk (19, 46, 55), mitogen-activated protein (MAP) kinase (59), and p21^ras (29).

The structural basis for CD2-mediated signaling lies within the 116-amino-acid CD2 cytoplasmic domain, which is characterized by the presence of at least four well conserved proline-rich regions and the striking absence of tyrosine residues. Several conserved regions of the cytoplasmic domain of human CD2 have been found to be important for CD2-mediated stimulation of interleukin-2 (IL-2) production in mouse T-cell hybridomas (7, 13, 14, 30, 33). Deletion of either of two repeated P-P-P-G-H-R motifs, beginning at amino acid 260 and amino acid 274 (Fig. 1), reduces CD2-mediated IL-2 production (7, 14), while deletion of both motifs results in a complete loss of CD2-mediated IL-2 production (7). The integrity of the P-P-P-G-H-R motifs is also essential for increases in intracellular cAMP levels (7). Moreover, the CD2 cytoplasmic domain contains two proline-rich regions that are similar to SH3 binding motifs with the general consensus sequence (h)-P-p-X-P, where “(h)” represents a hydrophobic residue and “p” represents a likely proline residue (17). These two proline-rich peptide segments have been shown to interact with the SH3 domain of the Src-like protein tyrosine kinase p56^lck in vitro (6). Mutation of the terminal asparagine residue in the CD2 cytoplasmic domain has also been shown to abrogate the ability of the antigen-specific CD3-TCR to increase CD2-mediated adhesion to its counter-receptor, LFA-3 (7, 33).

FIG. 1.

Schematic diagram of the CD2 cytoplasmic mutations used in this study. The position of each cytoplasmic truncation and internal deletion is denoted by a lack of sequence. The position of amino acid substitutions is indicated by bold letters. The TK amino acids at positions 211 and 212 were altered to KL (underlined) by the introduction of a HindIII restriction site at the junction between the transmembrane domain and the cytoplasmic sequence.

The signaling pathways that initiate integrin activation upon CD2 stimulation remain incompletely defined. The extracellular signal-regulated kinase (ERK) MAP kinases may be involved in CD2-mediated regulation of integrin function, since: (i) CD2 stimulation results in rapid activation of MAP kinases (59); (ii) expression of constitutively active H-ras has recently been shown to inhibit integrin activation (38); and (iii) modulation of PI 3-K activity has been shown to affect MAP kinase activity induced by a number of stimuli (25, 37, 44, 63, 66, 83, 88). Our prior studies have suggested a role for PI 3-K in the CD2-mediated regulation of integrin adhesion, since CD2 stimulation results in activation of PI 3-K, and CD2-mediated increases in β1 integrin activity in CD2⁺ HL60 transfectants and normal human T cells can be blocked by the PI 3-K inhibitor wortmannin (72).

The well-described form of PI 3-K consists of a heterodimer composed of two subunits, an 85-kDa regulatory subunit (p85) and a 110-kDa catalytic subunit (p110). At least three isoforms of p85 and four isoforms of p110 have been cloned (35, 36, 42, 60, 62, 79, 85). Functionally, PI 3-K phosphorylates the D-3 position of the inositol ring of PI, PI-4-phosphate, and PI-4,5-bisphosphate. The lipid products of PI 3-K have recently been shown to act as second messengers that bind to and regulate the activity of several intracellular signaling mediators, including the serine-threonine kinase Akt/PKB (22, 27, 49) and the intracellular protein GRP1 (48). GRP1 is structurally related to cytohesin-1, a protein that has been shown to specifically bind to and regulate the functional activity of the β2 integrin LFA-1 (50). A large number of functional responses have been demonstrated to be dependent on the activation of PI 3-K, including growth factor-dependent mitogenesis, membrane ruffling, cytoskeletal rearrangements, glucose uptake, prevention of apoptosis, cytokine production, and vesicular trafficking (for reviews, see references 10, 43, 52, and 89). Studies by our laboratory and others on agonist-induced activation of integrin activity (11, 12, 24, 53, 72–74, 81, 84, 91), coupled with the recent finding that PI 3-K lipid products can bind to cytohesin-1 (48), also suggest a critical role for PI 3-K signaling in the regulation of integrin-mediated adhesion (70).

This study focuses on the structural requirements for CD2-mediated regulation of β1-integrin-mediated adhesion. Site-directed mutagenesis and gene transfer was used to demonstrate a critical role for one of the proline-rich regions of the CD2 cytoplasmic tail in CD2-mediated increases in β1 integrin adhesiveness. This region of the CD2 cytoplasmic domain is distinct from the repeated P-P-P-G-H-R motifs and the cytoplasmic residue involved in CD3-TCR-mediated activation of CD2 adhesiveness. Cytoplasmic domain mutations that abrogated CD2-induced increases in integrin adhesiveness also blocked CD2-associated PI 3-K activity but did not affect association of CD2 with the p85 regulatory subunit. In addition, expression of a dominant negative isoform of the p85 regulatory subunit inhibited the ability of CD2 to upregulate adhesion to FN. In contrast, the region essential for CD2-mediated integrin upregulation is not required for CD2 stimulation of MAP kinase activity, implying that CD2-mediated activation of MAP kinase is not an essential component in the CD2 regulation of β1 integrin activity.

MATERIALS AND METHODS

Cell lines and culture reagents.

The cell line HL60 was maintained in RPMI 1640 medium supplemented with 10% fetal calf serum (FCS; Atlanta Biologicals, Norcross, Ga.), l-glutamine, and penicillin-streptomycin. Transfectants were maintained under 1-mg/ml G418 selection following electroporation. G418 was purchased from GIBCO-BRL (Gaithersburg, Md.). The HL60(CD2)-11.5 transfectant expressing wild-type human CD2 was generated as previously described (72).

Antibodies and reagents.

The following CD2-specific MAbs were used either as culture supernatant or as dilutions of ascites fluid: 95-5-49 (R. Gress, National Institutes of Health [NIH], Bethesda, Md.), 35.1 (American Type Culture Collection [ATCC], Rockville, Md.), TS2/18 (ATCC), and 9-1 (S. Y. Yang, Memorial Sloan Kettering Cancer Center, New York, N.Y.). The β1-integrin-specific MAb TS2/16 and the glycophorin-specific MAb 10F7 were obtained from the ATCC. The α4-integrin-specific MAb NIH49d-1 was provided by S. Shaw (NIH). The α5-integrin-specific MAb BIIG2 was provided by C. Damsky (University of California, San Francisco, Calif.). Rabbit polyclonal antibodies specific for the p85 subunit of PI 3-K were purchased from Upstate Biotechnology (Lake Placid, N.Y.). Rabbit polyclonal antibodies specific for the MAP kinase isoform ERK-2 (C-14) were purchased from Santa Cruz Biotechnology, Inc., (Santa Cruz, Calif.) Protein A-Sepharose was obtained from Zymed Laboratories, Inc. (San Francisco, Calif.). FN was purchased from the New York Blood Center (New York, N.Y.). PI-4-phosphate was obtained from Sigma (St. Louis, Mo.). PI was from Avanti Polar Lipids, Inc. (Alabaster, Ala.). Stock solutions of PMA (LC Laboratories, Woburn, Mass.), wortmannin (Sigma), LY294,002 (Alexis Corp., San Diego, Calif.), bisindolylmaleimide (Calbiochem, La Jolla, Calif.) and PD 098059 (1, 23) (kindly provided by S. J. Decker, Parke-Davis Pharmaceutical Research, Ann Arbor, Mich.) were dissolved in dimethyl sulfoxide (DMSO) and stored at −70°C.

DNA constructs.

The following constructs were generated in the lab of B. Bierer (7, 30): pFNeo (vector), pFNeo-16CD2, pFNeo-CD2-N327A, pFNeo-CD2-Δ260–265, pFNeo-CD2-Δ274–279, and pFNeo-Δ260 plus Δ274.

We have previously described a modified CD2 expression construct designated pMH-NeoH-CD2 that contains the extracellular and transmembrane regions of CD2 and permits replacement of the CD2 cytoplasmic domain with DNA fragments encoding alternative cytoplasmic domains via engineered HindIII and XhoI sites (91). Truncations within the CD2 cytoplasmic domain were introduced by PCR mutagenesis (91) or by site-directed mutagenesis (Clontech Laboratories, Inc., Palo Alto, Calif.). The full-length human CD2 cDNA in the plasmid expression vector pMH-NeoH (72) was mutagenized so that amber stop codons were introduced at codons 216, 266, 280, 292, and 306. A DNA fragment encoding each mutated cytoplasmic domain was generated by gene synthesis, as previously described (91), by using the following oligonucleotides: CD2 primer 1, 5′-TCAAAGCTTA GGAAAAAACA GAGGAGTCGG AGAAATGATGAG GAG-3′; CD2/215 primer 1, 5′-TCAAAGCTTA GGAAAAAATG AAGGAGTCGG AGAAATGATGAG GAG-3′; CD2 primer 2, 5′-AGGAGTCGGA GAAATGATGA GGAGCTGGAG ACAAGAGCCC ACAGAGTAGC TACTGAAGAA AGGGGCCGGA AGCCCCACCA AATTCCAGCT TCAACCC-3′; CD2 primer 3, 5′-CACCAAATTC CAGCTTCAAC CCCTCAGAAT CCAGCAACTT CCCAACATCC TCCTCCACCA CCTGGTCATC GTTCCCAGGC ACCTAGTCAT CGTCCCCC-3′; CD2/265 primer 3, 5′-CACCAAATTC CAGCTTCAAC CCCTCAGAAT CCAGCAACTT CCCAACATCC TCCTCCACCA CCTGGTCATC GTTGACAGGC ACCTAGTCAT CGTCCCCC-3′; CD2 primer 4, 5′-TTCTGCTGGT GAACTTGTGT GCCCGACGGA GCAGGAGGCC TCTTCTGAGG CTGGTGCTGA ACACGGTGTC CAGGAGGCGG GGGACGATGA CTAGGTGC-3′; CD2/279 primer 4, 5′-TTCTGCTGGT GAACTTGTGT GCCCGACGGA GCAGGAGGCC TCTTCTGAGG CTGGTGCTGT CAACGGTGTC CAGGAGGCGG GGGACGATGA CTAGGTGC-3′; CD2/291 primer 4, 5′-TTCTGCTGGT GAACTTGTGT GCCTCACGGA GCAGGAGGCC TCTTCTGAGG CTGGTGCTGA ACACGGTGTC CAGGAGGCGG GGGACGATGA CTAGGTGC-3′; CD2 primer 5, 5′-GAAGGGGACA ATGAGTTTTC TGCTGCCCCA TGGGGAGGTT TTGGCTGAAC TCGAGGTCTG GGGAGGGGCG GGCCTTTCTG CTGGTGAACT TGTGTGC-3′; CD2/305 primer 5, 5′-GAAGGGGACA ATGAGTTTTC TGCTGCCCCA TGGGGAGGTT TTGGCTGAAC TCGAGGTCAG GGGAGGGGCG GGCCTTTCTG CTGGTGAACT TGTGTGC-3′; and CD2 primer 6, 5′-CGACGTCGAC TCAATTAGAG GAAGGGGACA ATGAGTTTTCTG C-3′. Underlined nucleotides denote positions of stop codons. All PCR oligonucleotides were synthesized on an Applied Biosystems 342 oligonucleotide synthesizer (Foster City, Calif.). Primers 1 and 6 were used without purification. Primers 2, 3, 4, and 5 were purified with OPC cartridges (Applied Biosystems, Foster City, Calif.). PCRs were performed as described previously (91), except that CD2 primers 2, 3, 4, and 5 were included in the initial 5 cycles of amplification and primers 1 and 6 were used in the subsequent 25 cycles. Reaction products were purified by agarose gel fractionation and/or with Wizard PCR Preps (Promega, Madison, Wis.) and then digested with HindIII and SalI, purified by Wizard PCR Preps, and ligated into HindIII/XhoI-digested pMHNeoH-CD2 by standard techniques. Plasmids were sequenced to confirm the fidelity of the amplification reaction.

A truncation at position 299 within the CD2 cytoplasmic domain was generated by site-directed mutagenesis (Clontech) to create an in-frame stop codon directly following amino acid 299. Serial proline-to-alanine substitutions were also generated by site-directed mutagenesis (Clontech), creating a CD2 mutant that coded for alanine at codons 302, 303, and 305. The following mutagenic oligonucleotides were synthesized and purified by high-pressure liquid chromatography (Amitof, Boston, Mass.): CD2/299 mutagenic primer, 5′-GAGGGGCGGG CCTCACTGCT GGTGAAC-3′; CD2/P→A mutagenic primer, 5′-GAACTCGAGG TCTGGCGAGG GCCGCGCCTT TCTGCTG-3′; and S1 selection primer, 5′-CTTTTGCAAA CCGCGGCACG CTGCCG-3′. All constructs were sequenced to confirm the presence of the expected mutation.

The pEGFP-C2 vector was purchased from Clontech. The GFP-Δp85 construct was described previously (11). The GFP-WTp85 construct was made by subcloning the wild-type p85 cDNA (obtained from M. Kasuga and W. Ogawa, Kobe University, Kobe, Japan) into the KpnI/BamHI sites of the pEGFP-C2 expression vector. The resulting vector, pEGFP-WTp85, codes for a fusion protein consisting of GFP fused to the amino-terminal end of wild-type p85.

Stable transfection of HL60 cells.

All of the aforementioned CD2 constructs were transfected into the myelomonocytic HL60 cell line by electroporation as previously described (72). For each CD2 construct, at least three different bulk-positive transfectants were selected by using 1-mg/ml G418 resistance and subcloned by limiting dilution. G418-resistant subclones were analyzed for CD2 cell surface expression by flow cytometry with the CD2-specific MAb 95-5-49. In addition, each mutant CD2 protein was analyzed and quantified by Western blot analysis with the CD2-specific MAb TS2/18. Reverse transcription-PCR (RT-PCR) analysis, followed by DNA sequencing, determined that the CD2 transcripts contain the expected sequence. Finally, the level of α4β1 and α5β1 integrins expressed on the cell surface of CD2 transfectants was assessed by flow cytometry by using the β1-specific MAb TS2/16, the α4-specific MAb NIH49d-1, and the α5-specific MAb BIIG2 to guarantee that each transfectant had a comparable level of integrin expression.

Flow cytometry.

Single-color flow cytometry was performed as previously described (72, 91) with saturating amounts of primary unlabeled MAb and detection with fluorescein isothiocyanate-conjugated goat anti-mouse immunoglobulin G (IgG) or goat anti-rat IgG (Southern Biotechnology, Birmingham, Ala.). Samples were analyzed with a FACScan (Becton Dickinson, Mountain View, Calif.) and by using Cellquest software.

Adhesion assays.

The ability of each CD2 mutation to regulate β1 integrin activity was examined by measuring adhesion to FN as previously described (72). At least three independent subclones isolated from two separate bulk populations for each CD2 construct were analyzed. For PMA stimulation, cells were added to wells containing a final PMA concentration of 10 ng/ml. For CD2 stimulation, cells were added to wells containing a 1:10 dilution of the CD2-specific MAb 95-5-49 and a 1:6,000 dilution of the CD2-specific MAb 9-1. For direct stimulation of β1 integrins, cells were added to wells containing a 1:10 dilution of the activating β1-specific MAb TS2/16. After cells were allowed to settle for 60 min at 4°C, the plates were rapidly warmed to 37°C for 10 min, the nonadherent cells were washed off, and the percentage of bound cells was determined by lysing the well contents with detergent and counting gamma emissions. All data are expressed as the mean percentage of cells binding from three replicate wells.

CD2-associated PI 3-K activity assays.

For each HL60 transfectant, 20 × 10⁶ cells per condition were serum starved overnight in RPMI 1640–0.1% FCS and then washed once in phosphate-buffered saline (PBS). Cells were either left unstimulated or stimulated with an activating combination of CD2-specific MAbs for various times at 37°C. Cells were lysed in a modified radioimmunoprecipitation assay (RIPA) lysis buffer (1% sodium deoxycholate; 1% Triton X-100; 158 mM NaCl; 5 mM EDTA; 10 mM Tris-HCl, pH 7.2) supplemented with 10 μg of aprotinin per ml, 10 μg of leupeptin per ml, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 mM Na₃VO₄ [sodium orthovanadate]). The level of protein in each cell lysate was normalized as determined by BCA protein assay (Pierce, Rockford, Ill.). CD2 and any CD2-associated molecules were immunoprecipitated with the CD2-specific MAb 35.1. Samples were then tested for PI 3-K activity in vitro by assessing the phosphorylation of PI as previously described (72, 91). Radioactive lipid product was visualized by autoradiography and quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, Calif.). The lipid standard was visualized by exposing the thin-layer chromatography plate to iodine vapor.

Glutathione S-transferase (GST) fusion protein production.

The GST-p85α (SH3) fusion protein construct was a generous gift of M. Waterfield (Ludwig Institute for Cancer Research, London, England). The GST fusion construct was propagated in the BL21DE Escherichia coli cell line (Pharmacia Biotech, Inc., Piscataway, N.J.). The GST-p85α (SH3) fusion protein was generated, purified, and coupled to glutathione-Sepharose 4B (Pharmacia) as previously described (41).

GST binding reactions.

Cells (25 × 10⁶ cells per sample) were spun down and washed once in PBS–1% bovine calf serum (BCS; HyClone, Logan, Utah). Cells were either left unstimulated or were stimulated with an activating combination of CD2-specific MAbs 95-5-49 and 9-1 for 10 min; cells were then lysed in PI 3-K lysis buffer (50 mM HEPES, pH 7.5; 1% Triton X-100; 150 mM NaCl; 10% glycerol; 1.6 mM MgCl₂; 1 mM EGTA) supplemented with 10 μg of aprotinin per ml, 10 μg of leupeptin per ml, 1 mM PMSF, and 1 mM Na₃VO₄. Cell lysates were incubated on ice for 15 min and then centrifuged at 4°C for 20 min at 11,000 × g.

Ten micrograms of GST-p85α(SH3) fusion protein immobilized on glutathione-Sepharose 4B was incubated with each cell lysate for at least 2 h at 4°C. The beads were then washed once in ice-cold supplemented PI 3-K lysis buffer; bound proteins were then eluted by boiling in 4× sodium dodecyl sulfate (SDS) sample buffer under reducing conditions, separated on an SDS–10% polyacrylamide gel, and transferred to a polyvinylidene difluoride (PVDF) membrane. The blot was probed with the CD2-specific MAb TS2/18. The blot was then stripped and reprobed with a polyclonal antibody for GST (Upstate Biotechnology) by using GST fusion protein as an internal loading control. Detection was done by enhanced chemiluminescence (Pierce).

Immunoprecipitations.

Cells (1.25 × 10⁶ cells per sample) were washed once in PBS–1% BCS (HyClone) and then stimulated for 10 min at 37°C with an activating combination of CD2-specific MAbs. Samples were then lysed in PI 3-K lysis buffer as described above. One microgram of polyclonal p85 antibody (Upstate Biotechnology) per immunoprecipitation was immobilized on goat anti-rabbit IgG-coupled Sepharose (ICN-Cappel, West Chester, Pa.). The anti-p85 antibody-coupled beads were washed twice in ice-cold supplemented PI 3-K lysis buffer, and then the antibody-coupled beads were incubated with 1 μg of normal rabbit IgG (Santa Cruz) per immunoprecipitation for at least 2 h at 4°C to mask any available unbound sites on the goat anti-rabbit IgG-coupled Sepharose beads. The anti-p85 antibody-coupled beads were washed twice in ice-cold supplemented PI 3-K lysis buffer to remove excess normal rabbit IgG and were then incubated with each cell lysate for at least 2 h at 4°C. The beads were then washed twice in ice-cold supplemented PI 3-K lysis buffer; bound proteins were eluted by boiling in 4× SDS sample buffer under reducing conditions, separated on a 10% SDS polyacrylamide gel, and transferred to a PVDF membrane. The blot was probed with the CD2-specific MAb TS2/18, stripped, and then reprobed with a polyclonal antibody for p85 to ensure equal loading of each immunoprecipitate. Detection was by enhanced chemiluminescence.

Transient transfection of HL60(CD2/WT) cells.

Transfection of HL60(CD2/WT) cells was done as described previously (11) with the following modifications. Logarithmically growing HL60(CD2/WT) cells were washed once with Opti-MEM (GIBCO-BRL) and resuspended at 40 × 10⁶ cells/ml in 0.8 ml of Opti-MEM containing 120 μg of pEGFP-C2 (empty vector), pEGFP-Δp85, or pEGFP-WTp85 for 10 min at room temperature. Cells were electroporated with a BTX Gene Pulser set to low-voltage (LV) mode, at 230 V with one 15-ms pulse. After electroporation, cells were incubated for 30 min at room temperature before resuspension at 10⁶ cells/ml in RPMI 1640 medium supplemented with 10% FCS, l-glutamine, and penicillin-streptomycin. Cells were harvested after 16 to 18 h and used in the adhesion assay described below.

Adhesion assay of transiently transfected cells.

The adhesion of transiently transfected HL60(CD2/WT) cells to FN (1 μg/well) was analyzed as previously described (11). Adhesion was quantitated by the collection of adherent cells and analysis by flow cytometry. Cells from six replicate wells were pooled and resuspended in 200 μl of Hanks balanced salt solution supplemented with 1% BCS and 0.2% sodium azide. A 50-μl aliquot (10,082 beads) of PKH26 reference microbeads (Sigma) and 25 μl of PI were added to each tube, for a total sample volume of 275 μl. Each sample was analyzed on the flow cytometer, acquiring a minimum of 30,000 total events.

For each sample, analysis of adherent cells was performed by using the forward-scatter and side-scatter profiles to gate independently the reference microbeads and transfected cells. FL2 events within the microbead gate were used to calculate the total number of reference microbeads in each acquired sample. FL1/FL2 density plots were used to discriminate the live, PI-negative cells from the fluorescent microbeads. Gates were established by using FL1 fluorescence to calculate the number of cells in each of three populations: green fluorescent protein (GFP) negative (GFP⁻), GFP low positive (GFP⁺), and GFP positive (GFP⁺⁺). The total number of adherent cells was calculated by using the events in the microbead gate and the events in each of the GFP gates as previously described (11).

MAP kinase assays.

Assays for MAP kinase activity were performed as previously described (15) with modifications. 10⁷ HL60 cells per sample were preincubated with PMA (10 ng/ml) or with saturating concentrations of a mitogenic pair of CD2-specific MAbs, 95-5-49 and 9-1, for 30 min in 0.5 ml of PBS-BCS at 4°C. The cells were then stimulated for the indicated periods by incubation at 37°C. The stimulation was stopped by the addition of an equal volume of 2× modified RIPA lysis buffer (300 mM NaCl; 2% Nonidet P-40; 1% deoxycholate; 100 mM HEPES, pH 7.5; 2 mM Na₃VO₄; 100 mM NaF; 2 mM PMSF; 20 μg of aprotinin per ml; 10 mM benzamidine). The lysates were incubated on ice for 30 min with frequent vortexing and then centrifuged for 30 min in a microcentrifuge at 4°C. The lysates were transferred to new tubes containing 2 μg of ERK-2-specific MAb and 40 μl of protein A-Sepharose and incubated at 4°C for a minimum of 3 h. One-tenth of each immunoprecipitated sample was removed; bound proteins were then eluted by boiling in 4× SDS sample buffer under reducing conditions, separated on an SDS–10% polyacrylamide gel, and transferred to a PVDF membrane. The blot was probed with the ERK-2 specific antibody, and the resulting data were quantified with a Molecular Dynamics densitometer for use as a normalized ERK-2 loading control.

The remaining immunoprecipitates were washed two times with 0.25 M Tris (pH 7.6) and once with 0.1 M NaCl–50 mM HEPES (pH 8.0) and then incubated for 30 min at 30°C in 90 μl of reaction mixture (1 μCi of [γ-³²P]ATP; 50 μM ATP; 10 mM MgCl₂; 1 mM dithiothreitol; 1 mM benzamidine; 25 mM HEPES, pH 8.0; and 0.3 mg of myelin basic protein [MBP]). The reaction was stopped by centrifuging the samples for 1 min in a microcentrifuge and transferring the supernatant to a new tube containing 30 μl of 4× SDS-polyacrylamide gel electrophoresis (PAGE) sample buffer. The samples were boiled 3 min and subjected to 15% PAGE. The gel was dried and the labeled MBP was detected by autoradiography. Data were quantified with a Molecular Dynamics PhosphorImager or densitometer.

RESULTS

In order to conduct an analysis of CD2-mediated integrin activation, we have developed and characterized a gene transfer system that allows for an assessment of the adhesion regulatory signaling properties of mutated CD2 molecules (72). Human CD2 was expressed in the CD2-negative human myelomonocytic cell line HL60 (Fig. 2). Stimulation of CD2⁺ HL60 transfectants with an activating pair of CD2-specific MAbs resulted in an upregulation of β1 integrin-dependent adhesion of HL60 cells to the β1 integrin ligand FN that was comparable to the adhesion induced by treatment with the phorbol ester PMA (Fig. 3) or by direct activation of β1 integrins with an activating β1-specific MAb (data not shown).

FIG. 2.

CD2 expression on CD2⁺ HL60 transfectants. Single-color flow-cytometric analysis was performed as described in Materials and Methods. The binding of the CD2-specific MAb 95-5-49 (first column), the β1-specific MAb TS2/16 (second column), the α4-specific MAb NIH49d-1 (third column), and the α5-specific MAb BIIG2 (fourth column) is shown as filled histograms in each panel. Open histograms represent negative control staining of cells with MAb 10F7.

FIG. 3.

Mutations that affect CD2-regulated IL-2 production or CD2 avidity do not affect CD2-mediated adhesion to FN. Binding of ⁵¹Cr-labeled CD2⁺ HL60 CD2 wild-type or CD2 mutant transfectants to 0.3 μg of FN/well was assessed in the presence of no stimulus (open bars), 10 ng of PMA per ml (shaded bars), or an activating combination of CD2-specific MAbs (solid bars). Cells were activated for 10 min at 37°C before the nonadherent cells were washed away. Results shown are from four independent experiments. N.T., not tested.

Cytoplasmic regions of CD2 that regulate CD2-mediated interleukin-2 (IL-2) production in T cells or CD2 avidity do not modulate integrin activity.

We utilized a number of previously described CD2 cytoplasmic mutation constructs (7, 30) to determine whether regions of the CD2 cytoplasmic domain that have been shown to regulate other CD2-mediated functional responses are also involved in CD2-mediated upregulation of β1 integrin activity (Fig. 1). These mutant CD2 constructs included the following: a deletion of the proximal P-P-P-G-H-R motif (pFNeo-CD2/Δ260–265), a deletion of the distal P-P-P-G-H-R motif (pFNeo-CD2/Δ274–279), deletions of both proximal and distal P-P-P-G-H-R motifs (pFNeo-CD2/Δ260 + Δ274), and an asparagine-to-alanine amino acid mutation at the last position of the CD2 cytoplasmic domain (pFNeo-CD2/N327A).

Each construct, or wild-type CD2, was transfected into HL60 cells by electroporation. Transfected cell lines expressing CD2 at levels comparable to each other were identified by flow cytometry (Fig. 2). Levels of β1, α4, and α5 integrin were equivalent among the various CD2 transfectants analyzed (Fig. 2), indicating that β1 integrin expression was not altered by the transfection procedure. CD2-expressing HL60 transfectants were analyzed in adhesion assays for adhesion to FN under various stimulation conditions. As shown in Fig. 3, stimulation of wild-type CD2⁺ transfectants, but not a transfectant expressing the expression vector only, resulted in increases in FN that were comparable to that seen following PMA stimulation. Although the P-P-P-G-H-R motifs have been shown to be critical to CD2-mediated regulation of IL-2 production in T-cell lines (7, 14), mutation of either or both of the P-P-P-G-H-R motifs in the CD2 cytoplasmic domain did not consistently inhibit the ability of activating pairs of CD2-specific MAbs to upregulate HL60 cell adhesion to FN (Fig. 3). In addition, a mutation that has been shown to inhibit CD3-TCR-mediated regulation of CD2 avidity (CD2-N327A) (7, 33) did not affect CD2-induced increases in HL60 adhesion to FN (Fig. 3).

The proline-rich sequence between amino acids 299 and 305 is essential for CD2-mediated upregulation of β1 integrin activity.

In order to identify the region of the CD2 cytoplasmic domain that is involved in CD2-mediated increases in β1 integrin adhesiveness, additional CD2 cytoplasmic domain mutants were created and expressed in HL60 cells (Fig. 1). These mutations involved the introduction of stop codons at various places in the CD2 cytoplasmic domain to create truncations containing 3 (CD2/215), 53 (CD2/265), 67 (CD2/279), 79 (CD2/291), and 93 (CD2/305) amino acids of the CD2 cytoplasmic domain. These truncation mutants were expressed in HL60 cells at levels comparable to that of the wild-type CD2 (Fig. 4A), indicating that truncation of the CD2 cytoplasmic sequence does not alter its ability to be expressed at the cell surface.

FIG. 4.

The CD2 cytoplasmic domain is required for CD2-mediated activation of β1-integrin-mediated adhesion. (A) Single-color flow-cytometric analysis on the indicated transfectants was performed as described in Materials and Methods. The histogram setup is identical to that of the histograms described in the legend to Fig. 2. (B) Binding of ⁵¹Cr-labeled CD2⁺ HL60 CD2 wild-type or CD2 mutant transfectants to 0.3 μg of FN/well was assessed in the presence of no stimulus (open bars), 10 ng of PMA per ml (shaded bars), or an activating combination of CD2-specific MAbs (solid bars). Cells were activated for 10 min at 37°C before the nonadherent cells were washed away. Results shown are from one representative of at least three independent replicate experiments.

The largest CD2 truncation mutant, CD2/215, retains only three amino acids of the CD2 cytoplasmic domain (Fig. 1). Stimulation of HL60 transfectants expressing CD2/215 with activating pairs of CD2-specific MAbs did not result in an increase in adhesion to FN when compared to unstimulated cells and wild-type CD2⁺ transfectants stimulated with CD2 MAbs (Fig. 4B). This difference in CD2-induced regulation of β1 integrin activity was not due to defects in β1 integrin function in the CD2/215 transfectants, since PMA stimulation resulted in an increase in adhesion to FN comparable to control transfectant cells expressing the vector only (Fig. 4B). Furthermore, levels of β1 integrin on the CD2/215 transfectants were comparable to HL60 transfectants expressing wild-type CD2 or vector only (Fig. 4A), indicating that differences in the amount of β1 integrin could not explain the inability of CD2 stimulation to increase the adhesion of HL60(CD2/215) transfectants to FN.

Stimulation of transfectants expressing the CD2/265, CD2/279, and CD2/291 truncation mutants with an activating pair of CD2-specific MAbs also failed to result in increased adhesion to FN (Fig. 4B). As with the CD2/215 transfectant, PMA stimulation of these other transfectants did result in increased adhesion to FN, indicating that the β1 integrins expressed on these transfectants were responsive to integrin-activating signals. Conversely, CD2 stimulation of transfectants expressing the CD2/305 truncation mutant increased adhesion comparable to that observed following CD2 stimulation of wild-type CD2⁺ transfectants (Fig. 4B). This suggests that a region between amino acids 291 and 305 of the CD2 cytoplasmic domain is essential for CD2-mediated regulation of β1-integrin-dependent adhesion to FN.

Since it is also plausible that the truncation mutations cause an alteration in the kinetics of CD2-mediated stimulation of β1 integrin activity, we analyzed the adhesion of transfectants to FN at various time points following CD2 stimulation, ranging from 10 to 80 min. At all of the time points tested, only CD2/305 and CD2/WT possessed the ability to upregulate β1 integrin activity upon CD2 engagement (data not shown).

Two additional CD2 cytoplasmic mutations were generated to further define the region responsible for CD2-mediated regulation of β1 integrin activity. A stop codon was generated in frame following amino acid 299 to generate the CD2 mutant construct CD2/299 (Fig. 1). Transfection of the CD2/299 construct into HL60 cells resulted in expression of CD2 at the cell surface at a level comparable to wild-type CD2 and the other CD2 truncation mutants tested (Fig. 5A). However, stimulation of CD2/299-expressing HL60 transfectants with the activating pairs of CD2-specific MAbs did not result in increased adhesion to FN at any of the stimulation time points tested (Fig. 5B and data not shown). As with other transfectants expressing mutant CD2 molecules that were unable to upregulate β1 integrin adhesiveness, the responsiveness of β1 integrins on CD2/299 HL60 transfectants to activation was demonstrated by increased adhesion to FN after PMA stimulation (Fig. 5B). These results indicate that the K-G-P-P-L-P sequence in the CD2 cytoplasmic domain (amino acids 300 to 305) is critical for CD2-mediated increases in β1-integrin-dependent adhesion to FN.

FIG. 5.

A proline-rich region of CD2 between amino acids 299 and 305 is essential for CD2 regulation of adhesion to FN. (A) Single-color flow-cytometric analysis on the indicated transfectants was performed as described in Materials and Methods. The histogram setup is identical to that of the histograms described in the legend to Fig. 2. (B) Binding of ⁵¹Cr-labeled CD2⁺ HL60 CD2 wild-type or CD2 mutant transfectants to 0.3 μg of FN/well was assessed in the presence of no stimulus (open bars), 10 ng of PMA per ml (shaded bars), or an activating combination of CD2-specific MAbs (solid bars). Cells were activated for 10 min at 37°C before the nonadherent cells were washed away. Results shown are from one representative of at least three independent replicate experiments.

The K-G-P-P-L-P sequence is one of four proline-rich regions in the CD2 cytoplasmic domain. To demonstrate a role for the proline residues in this motif in CD2-mediated regulation of β1 integrin adhesiveness, each of the three proline residues in this motif—amino acids 302, 303, and 305—was mutated to alanine residues in the context of the wild-type CD2 cytoplasmic tail (Fig. 1). The resulting CD2 mutant construct, designated CD2/P→A, was transfected into HL60 cells, and expression was found to be comparable to that of wild-type CD2 and the other CD2 mutant constructs analyzed in this study (Fig. 5A). In adhesion assays, CD2 stimulation failed to increase the adhesion of HL60 transfectants expressing the CD2/P→A mutant construct (Fig. 5B). This result provides further evidence for the importance of the K-G-P-P-L-P motif in general, and the three proline residues in particular, in CD2-mediated regulation of β1 integrin adhesiveness.

Two structurally distinct PI 3-K inhibitors abolish CD2-mediated upregulation of β1 integrin activation.

We have previously demonstrated that the PI 3-K inhibitor wortmannin is capable of abolishing CD2-mediated activation of β1-integrin-dependent adhesion (72). However, the specificity of wortmannin as a selective inhibitor of PI 3-K has recently come under scrutiny, since wortmannin has been implicated as an inhibitor of phospholipase A₂ (18), phospholipase D, myosin light-chain kinase, mammalian target of rapamycin (9), and pleckstrin phosphorylation (2). Thus, we also tested the ability of LY294,002, another PI 3-K inhibitor whose mode of action on PI 3-K is distinct from wortmannin (87), to block CD2-induced increases in HL60 adhesion to FN. As shown in Fig. 6, either 25 μM LY294,002 or 100 nM of wortmannin blocked adhesion induced by stimulation of HL60 transfectants expressing either wild-type CD2 or the CD2/305 truncation mutant. At the depicted concentrations, the presence of these inhibitors did not affect adhesion induced by either PMA treatment (Fig. 6) or activating β1-specific MAbs (data not shown), suggesting that they were not acting via a toxic effect on the cells or a general impairment of cell adhesive function. Dose-response studies indicated that the 50% inhibitory concentration (IC₅₀) of LY294,002 is in the range of 5 to 15 μM (data not shown). This IC₅₀ value is consistent with the concentrations of LY294,002 found to specifically inhibit PI 3-K activity in other systems (9, 82, 86).

FIG. 6.

CD2-induced adhesion of HL60 transfectants to FN is blocked by two PI 3-K inhibitors, wortmannin and LY294,002. Binding of ⁵¹Cr-labeled CD2⁺ HL60 CD2/WT or CD2/305 transfectants to 0.3 μg of FN/well was assessed in the presence of no stimulus, 10 ng of PMA per ml, or an activating combination of CD2-specific MAbs as described in the legend to Fig. 5 in the continuous presence of DMSO (solid bars), 100 nM wortmannin (open bars), or 25 μM LY294,002 (shaded bars). The data depicted in this figure are from one representative of three independent replicate experiments.

Expression of dominant negative p85 causes inhibition of CD2-mediated upregulation of adhesion to FN.

We also employed a genetic approach to further define the role of PI 3-K in CD2-mediated regulation of β1 integrin adhesiveness. The effect of GFP fusion proteins expressing wild-type or a dominant negative form of p85 on CD2-induced increases in HL60(CD2/WT) cell adhesion to FN was assessed. The GFP tag, together with flow cytometric analysis, was used to determine the adhesion of transiently transfected cells expressing various levels of the GFP fusion protein (11). As shown in Fig. 7A, 5 to 10% of electroporated HL60(CD2/WT) cells expressed GFP or expressed GFP fusion proteins with wild-type p85 (GFP-WTp85) or dominant negative p85 (GFP-Δp85). Typically, GFP alone was expressed at higher levels in individual cells than either GFP-WTp85 or GFP-Δp85.

FIG. 7.

Expression of dominant negative p85 inhibits CD2-mediated activation of adhesion to FN. (A) Flow-cytometric analysis of untransfected HL60(CD2/WT) cells or transiently transfected HL60(CD2/WT) cells expressing GFP, GFP-WTp85, or GFP-Δp85. Histogram gates demarcating preadherent GFP⁻, GFP⁺, or GFP⁺⁺ expression are shown with the percentage of live cells present in each gate indicated in each panel. (B) Adhesion of transiently transfected HL60(CD2/WT) cells expressing either GFP (left panel), GFP-WTp85 (middle panel), or GFP-Δp85 (right panel) to FN (1 μg/well) was assessed as previously described (11). The percentage of cell adhesion to FN was quantitated by flow cytometry (11), and the results indicate the percentage of adhesion of GFP⁻ cells (open bars), GFP⁺ cells (shaded bars), and GFP⁺⁺ cells (solid bars) in each transfected population with no stimulation (U), PMA stimulation (PMA), or CD2 stimulation (CD2). Results shown are from one of four independent replicate experiments.

By using the three gates shown in Fig. 7A, the adhesion of transfected HL60(CD2/WT) cells expressing no GFP, low levels of GFP, or higher levels of GFP under various stimulation conditions was determined. Figure 7B shows that expression of GFP or GFP-WTp85 did not affect the adhesion of HL60(CD2/WT) cells that were either left unstimulated or stimulated with PMA or CD2. In contrast, expression of the GFP-Δp85 fusion protein inhibited CD2-mediated increases in adhesion to FN (Fig. 7B). The inhibitory effect of GFP-Δp85 was observed at high levels of GFP-Δp85 expression. However, cells expressing high levels of the GFP-Δp85 protein still exhibited adhesion comparable to that of GFP-negative cells after PMA stimulation or stimulation with the activating β1-integrin-specific MAb TS2/16 (Fig. 7B and data not shown). This suggests that β1 integrins were still functional and responsive to stimulation despite high levels of GFP-Δp85 expression. Thus, these results provide further support to the pivotal role that PI 3-K plays in CD2 regulation of β1 integrin activity.

Effects of CD2 cytoplasmic domain mutations on CD2-induced increases in PI 3-K activity.

CD2 stimulation of CD2⁺ HL60 transfectants and human T cells results within minutes in an increase in CD2-associated PI 3-K activity (72). If PI 3-K is involved in CD2-mediated increases in β1 integrin function, we reasoned that CD2 cytoplasmic domain mutations that abrogated CD2-induced increases in adhesion to FN should also show a loss of CD2-induced increases in associated PI 3-K activity. Thus, we assessed PI 3-K activity upon CD2 stimulation in a panel of HL60 transfectants expressing wild-type CD2 and various CD2 cytoplasmic domain mutants. CD2 stimulation of wild-type CD2⁺ and CD2/305⁺ transfectants resulted in increased PI 3-K activity in CD2 immunoprecipitates (Fig. 8). Both of these transfectants also demonstrated increased adhesion to FN after CD2 stimulation (Fig. 4B and 5B). Interestingly, transfectants expressing mutant CD2 proteins that did not induce increases in adhesion upon CD2 stimulation also did not exhibit changes in CD2-associated PI 3-K after stimulation with activating pairs of CD2-specific MAbs (compare Fig. 5B with Fig. 8). This included the CD2/299 truncation mutant, as well as the CD2/P→A substitution mutant (Fig. 8). Therefore, a direct relationship exists between CD2-mediated induction of CD2-associated PI 3-K activity and CD2-mediated β1 integrin regulation.

FIG. 8.

CD2 molecules that can mediate β1 integrin activity (CD2/WT or CD2/305) also induce CD2-associated PI 3-K activity. CD2⁺ HL60 transfectants were either unstimulated (open bars) or stimulated with an activating combination of CD2-specific MAbs for 2 min (shaded bars) or 5 min (solid bars) at 37°C, lysed, and then immunoprecipitated with the CD2-specific MAb 35.1. Samples were then tested for PI 3-K activity by assessing phosphorylation of PI followed by visualization of the phosphorylated in vitro product by autoradiography with a PhosphorImager after thin-layer chromatography. For each transfectant, the fold induction of PI 3-K activity was normalized to the PI 3-K activity in cells stimulated for 0 min, which was given an arbitrary value of 1. The fold induction of PI 3-K activity from two independent, identical experiments was averaged.

Association between CD2 and the p85 subunit of PI 3-K is independent of the K-G-P-P-L-P motif.

To further delineate the structural basis for the association of CD2 with PI 3-K, we examined the effect of CD2 cytoplasmic domain mutations on the in vitro association of CD2 with the p85 regulatory subunit. We reasoned that the proline-rich motif K-G-P-P-L-P between amino acids 299 and 305 might mediate the association of CD2 with the SH3 domain of the p85 subunit. Therefore, we analyzed whether a GST fusion protein expressing the SH3 domain of p85α could precipitate wild-type CD2 and various CD2 cytoplasmic domain mutant proteins. Cell lysates were prepared from unstimulated or CD2-stimulated transfectants and then incubated with beads coated with GST-p85α(SH3). Figure 9A shows that GST-p85α(SH3) was also able to precipitate wild-type CD2 under both unstimulated and stimulated conditions. The GST-p85α(SH3) was not able to precipitate the CD2/215 truncation mutant protein (Fig. 9A), which lacks all but three amino acids of the CD2 cytoplasmic domain. This result suggests that the association of CD2 with p85 is dependent upon the presence of the CD2 cytoplasmic domain. Stimulation of the cells with CD2 MAbs prior to p85 precipitation did not affect the amount of CD2 precipitated by the GST-p85α(SH3) fusion protein (Fig. 9A). The GST-p85α(SH3) fusion protein was also able to precipitate the CD2/299 and CD2/305 truncation mutant proteins, as well as the CD2/P→A substitution mutant protein (Fig. 9A).

FIG. 9.

The region responsible for CD2 association with the p85 subunit of PI 3-K does not overlap with amino acids 299 through 305. HL60(CD2) transfectants were either left unstimulated or stimulated with an activating combination of CD2-specific MAbs for 10 min at 37°C and lysed. (A) Lysates were precipitated with GST-p85α (SH3) fusion protein-coupled beads, separated by SDS–10% PAGE, and transferred to a PVDF membrane. The membrane was immunoblotted with the CD2-specific MAb TS2/18 (top panel) or stripped and reprobed with an antibody specific for GST (bottom panel). U, unstimulated; S, CD2 stimulated. (B) Lysates were immunoprecipitated with polyclonal anti-p85 antibody-coupled beads, separated on by SDS–10% PAGE, and transferred to a PVDF membrane. The membrane was immunoblotted with the CD2-specific MAb TS2/18 (top panel) or stripped and reprobed with an antibody specific for p85 (bottom panel).

Because of potential promiscuous binding capabilities of the GST-p85α(SH3) fusion protein, we also examined the association of wild-type CD2 and various CD2 cytoplasmic domain mutants with the p85 subunit by determining whether CD2 could be coprecipitated in an anti-p85 immunoprecipitate. The results shown in Fig. 9B were equivalent to those observed in the GST-p85α(SH3) pull-down experiments. Wild-type CD2 and all of the CD2 cytoplasmic domain mutants, with the exception of CD2/215, could be coprecipitated with p85. In addition, stimulation of the cells with CD2 MAbs prior to immunoprecipitation did not affect the amount of CD2 coprecipitated with p85 (Fig. 9B). The results presented in Fig. 9 suggest that the K-G-P-P-L-P motif in the CD2 cytoplasmic domain is not the primary structural element mediating the association of CD2 with p85. More significantly, the results depicted in Fig. 9 illustrate that the association of p85 with CD2 appears to be independent of the ability of CD2 to regulate β1 integrin adhesiveness.

CD2 stimulation results in wortmannin-sensitive induction of MAP kinase activity.

In order to investigate the role of MAP kinase in CD2-mediated activation of integrin-mediated adhesion, we also analyzed the ability of CD2-expressing HL60 cells to activate MAP kinase upon CD2 stimulation. Figure 10A shows that ligation of CD2 by an activating pair of CD2-specific MAbs resulted in the rapid induction of MAP kinase activity, peaking at 10 min of stimulation and declining to baseline levels by 30 min. In a way similar to that of CD2-mediated upregulation of integrin activity, CD2-mediated activation of MAP kinase requires the presence of both CD2-specific MAbs (72; data not shown). As a positive control, PMA stimulation of CD2/WT expressing HL60 cells resulted in increased MAP kinase activity (Fig. 10B). Both the kinetics of PMA-induced MAP kinase activation and the magnitude of increase in MAP kinase activity compared to unstimulated cells were similar to CD2-mediated activation of MAP kinase (data not shown). We next tested the effects of wortmannin on CD2-mediated activation of MAP kinase. While wortmannin had a minimal effect on the ability of PMA to induce MAP kinase activity, wortmannin completely inhibited CD2-mediated increases of MAP kinase activity (Fig. 10B). In contrast, treatment of CD2⁺ HL60 cells with the protein kinase C inhibitor bisindolylmaleimide blocked PMA-induced activation of MAP kinase, but such treatment had no effect on CD2-induced activation of MAP kinase (Fig. 10B).

FIG. 10.

Wortmannin-sensitive activation of MAP kinase after CD2 stimulation. (A) ERK-2 kinase activity was assessed for 60 min of stimulation with an activating pair of CD2-specific MAbs as described in Materials and Methods. Phosphorylation of the MAP kinase substrate MBP is shown after various times of CD2 stimulation. (B) CD2⁺ HL60 cells were stimulated for 10 min at 37°C with 10 ng of PMA per ml or an activating pair of CD2-specific MAbs in the presence of no inhibitor (open bars), 100 nM wortmannin (solid bars), or 3 μM bisindolylmaleimide (cross-hatched bars). The cells were lysed and immunoprecipitated with an ERK-2-specific polyclonal antibody, and the MBP-specific kinase activity was assessed as described in Materials and Methods. Results from one representative experiment of three total are shown.

Effects of inhibiting MEK on CD2- and PMA-dependent induction of integrin function and MAP kinase.

In order to determine the functional consequences of CD2-dependent induction of MAP kinase as they pertain to integrin-mediated adhesion, we employed an inhibitor of the MAP kinase activation cascade, PD 098059 (1, 23). This compound specifically inhibits MAPK/ERK kinase (MEK), the upstream regulatory kinase that phosphorylates and activates the MAP kinases ERK-1 and ERK-2. Figure 11 shows that PD 098059 inhibited both PMA- and CD2-induced MAP kinase activity in a dose-dependent manner, with an IC₅₀ of 10 to 20 μM for both stimuli. These IC₅₀s are similar to those reported for the inhibition of MAP kinase activity in vitro and growth factor-mediated activation of MAP kinase in vivo by PD 098059 in other studies (1, 23). In adhesion assays, PD 098059 inhibited PMA-induced adhesion, suggesting a role for PMA-induced activation of the MAP kinase pathway in phorbol ester-mediated activation of integrin function (Fig. 12). Although PD 098059 blocked CD2-mediated activation of MAP kinase, it failed to inhibit CD2-mediated activation of integrin adhesion (Fig. 12).

FIG. 11.

The MEK inhibitor PD 098059 inhibits both PMA- and CD2-mediated MAP kinase activation. CD2⁺ HL60 cells were stimulated for 10 min at 37°C with 10 ng of PMA per ml or an activating pair of CD2-specific MAbs in the presence of no inhibitor (open bars) or the indicated concentrations of PD 098059. The cells were lysed and immunoprecipitated with an ERK-2-specific polyclonal antibody, and the MBP-specific kinase activity was assessed as described in Materials and Methods. Results from one representative experiment of three total are shown.

FIG. 12.

MEK activity is required for the regulation of integrin function induced by PMA but not by stimulation through CD2. The adhesion of CD2/WT HL60 cells was assessed in the presence of no inhibitor (open bars) or the indicated concentrations of the MEK inhibitor PD 098059. The cells were left unstimulated or were stimulated for 10 min at 37°C with 10 ng of PMA per ml or with an activating pair of CD2-specific MAbs, and the adhesion to 1 μg of FN per well was determined.

Effects of various mutations within the CD2 cytoplasmic domain on CD2-mediated activation of MAP kinase.

In order to determine if CD2-mediated MAP kinase activation could be dissociated from CD2-mediated activation of β1-integrin-mediated adhesion, the CD2/WT, CD2/299, CD2/305, and CD2/P→A constructs (Fig. 1) were analyzed for their ability to activate MAP kinase in HL60 cells. In comparison to the wild-type CD2 transfectant, the CD2/299 truncation mutant failed to support any appreciable increase in MAP kinase activity (Fig. 13). In contrast, both the CD2/P→A and the CD2/305 mutations supported CD2-mediated MAP kinase activation, although at lower levels than wild-type CD2 did. These differences in CD2-mediated activation of MAP kinase do not appear to be due to differences in MAP kinase itself, since PMA stimulation resulted in a comparable activation of MAP kinase in all of the transfectants (Fig. 13). In addition, Western blotting analysis revealed similar levels of MAP kinase expressed in the transfectants (data not shown). Thus, these results provide additional evidence that CD2-mediated activation of MAP kinase is not involved in CD2-mediated activation of integrin-mediated adhesion.

FIG. 13.

Effect of various cytoplasmic mutations on CD2-mediated activation of MAP kinase. HL60 cells expressing CD2/299, CD2/305, CD2/P→A, or CD2/WT were activated for the indicated periods of time with 10 ng of PMA per ml (squares) or an activating pair of CD2-specific MAbs (circles). The cells were lysed and immunoprecipitated with an ERK-2-specific polyclonal antibody, and the MBP-specific kinase activity was assessed as described in Materials and Methods. For each stimulation condition, ERK-2 activity levels (as assessed by quantitation of the autoradiograph with a PhosphorImager) were normalized to the ERK-2 kinase activity in cells stimulated for 0 min, which was given an arbitrary value of 1. Results from one representative experiment of five total are shown.

DISCUSSION

In this study, we used mutational analysis to identify the structural basis for the regulation of β1 integrin functional activity by the CD2 molecule. This analysis demonstrated that the CD2 cytoplasmic domain is necessary for CD2-mediated activation of β1-integrin-mediated adhesion to FN. Furthermore, the proline-rich region between amino acids 299 and 305 (K-G-P-P-L-P) of the CD2 cytoplasmic domain was found to be essential for CD2-mediated regulation of β1 integrin activity, since the substitution of the three proline residues at amino acid positions 302, 303, and 305 with alanines abrogated the ability of CD2 to regulate β1 integrin activity. The region of the CD2 cytoplasmic tail critical for CD2-mediated activation of PI 3-K also mapped to the K-G-P-P-L-P motif. Expression of the dominant negative p85 fusion isoform GFP-Δp85, but not GFP-WTp85, inhibited CD2-mediated induction of FN adhesion, providing additional evidence for a role for PI 3-K in regulation of integrin adhesiveness by CD2. However, structure-function analysis revealed that the coupling of CD2 to PI 3-K does not appear to involve the well-described binding of the p85 regulatory subunit of PI 3-K to a sequence in the CD2 cytoplasmic tail. Nor does this domain appear to couple CD2-mediated activation of MAP kinase to CD2 regulation of β1 integrin activity.

Previous studies of CD2-mediated signaling have demonstrated a role for the two well-conserved P-P-P-G-H-R motifs in CD2-regulated IL-2 production and increases in intracellular cAMP (7). In contrast to those earlier studies, deletion of either or both of the P-P-P-G-H-R motifs in the CD2 cytoplasmic tail did not have any effect on the ability of CD2 stimulation to increase β1-integrin-mediated adhesion of HL60 cell transfectants to FN. Furthermore, mutation of the terminal asparagine residue to alanine, which blocks CD3-TCR-mediated modulation of CD2-mediated adhesion, also had no effect on CD2-mediated regulation of β1 integrin function in our experimental system. Thus, our data suggest that the regulation of β1 integrin adhesiveness by the CD2 molecule can be distinguished structurally from other CD2-dependent functional responses, such as IL-2 production and avidity regulation of CD2 itself. However, we were unable to directly compare the effect of CD2 cytoplasmic domain mutations on these various responses, since HL60 cells do not produce IL-2 or express CD3-TCR. Our studies are also consistent with other studies of CD2 signaling demonstrating that interaction of CD2 with its cellular counter-receptor, LFA-3, cannot by itself initiate the signals observed with the activating pairs of CD2-specific MAbs used in this study (39). The presence or absence of the CD3-TCR complex may influence CD2 signaling, since some, but not all, studies have suggested that CD2-mediated signaling is dependent on the expression of CD3-TCR (8). However, CD2 is expressed on natural killer and certain thymocyte subsets in the absence of CD3-TCR expression (26, 75). Furthermore, CD2-mediated signaling is observed in these cells, as well as in CD3-negative murine mast cells (4), indicating that CD2 can function as a signaling molecule in the absence of the CD3-TCR. Our studies also support a role for CD2 in transducing signals in the absence of CD3-TCR and suggest that fairly discrete regions of the CD2 cytoplasmic tail regulate specific intracellular signaling pathways.

Previous studies of CD2 regulation of integrin function demonstrated that treatment of CD2⁺ HL60 transfectants or T cells with wortmannin blocked CD2-mediated activation of β1-integrin-mediated adhesion (72). This study provides additional evidence that PI 3-K is in fact involved in CD2-mediated activation of β1 integrin function. Treatment of transfectants expressing either wild-type CD2 or the CD2/305 truncation mutant with the structurally distinct PI 3-K inhibitor, LY294,002, resulted in inhibition of increased adhesion induced by CD2 stimulation but not by activating β1-specific MAbs or by PMA stimulation. The IC₅₀s for both inhibitors was in the range previously shown to inhibit PI 3-K activity in other systems (9, 16, 68, 82, 86), providing additional evidence implicating PI 3-K as the common target of the inhibitory effect of these two drugs on CD2-mediated activation of β1 integrin activity. More significantly, the ability of dominant negative p85 to specifically inhibit CD2-induced adhesion of HL60 cells to FN provides additional genetic evidence that PI 3-K is critical to CD2 regulation of β1 integrin function.

If PI 3-K activation is in fact critical for CD2 induction of β1 integrin activity, then CD2 cytoplasmic domain mutations that block CD2 regulation of integrin function should also block the ability of CD2 to activate the PI 3-K signaling pathway. Our structure-function analysis of CD2-mediated regulation of β1 integrin adhesiveness confirms this prediction. CD2 mutant proteins that were unable to activate β1-integrin-mediated adhesion upon stimulation, such as the CD2/P→A mutant and the CD2/299 truncation mutant, also failed to show CD2-dependent increases in PI 3-K activity associated with CD2 (Table 1). Although engagement of either the CD2/WT molecule or the CD2/305 molecule expressed on HL60 cells induced only a twofold increase in CD2-associated PI 3-K activity (Fig. 8) or phosphotyrosine-associated PI 3-K activity (data not shown), other studies have recently demonstrated that similar 1.5- to 2-fold increases in PI 3-K activity were associated with PI 3-K-dependent increases in the invasiveness of breast carcinoma cells (69). Thus, modest changes in PI 3-K activity, as assessed by in vitro labeling of an exogenous substrate, can lead to clear changes in cell adhesion. However, it should also be noted that the in vitro PI 3-K assay employed in this study may not completely reflect the changes in PI 3-K activity that occur upon CD2 stimulation in vivo.

TABLE 1.

Summary of effects of CD2 cytoplasmic domain mutations on activation of MAP kinase and integrin function

CD2 construct	Activation ofa:
	β1 integrin function	PI 3-K activity	MAP kinase activity
CD2/WT	+	+	++
CD2/305	+	+	+
CD2P→A	−	−	+
CD2/299	−	−	−

^a+ or ++, moderate or significant increase, respectively, in the indicated response upon CD2 stimulation when compared to unstimulated cells; −, no increase in the indicated response upon CD2 stimulation when compared to unstimulated cells. 

The well-described form of PI 3-K consists of two subunits, a regulatory subunit (p85) and a catalytic subunit (p110). The p85 subunit of PI 3-K contains several protein binding motifs, including SH2 domains and an SH3 domain, as well as proline-rich sequences that fit the consensus sequences for binding sites recognized by SH3 domains. The association of CD2 with PI 3-K cannot occur via direct binding of the SH2 domain of p85 with a sequence in the CD2 cytoplasmic domain, since SH2 domains bind to a specific phosphotyrosine-containing motif (Y-X-X-M) and the CD2 cytoplasmic domain lacks tyrosine residues. However, the K-G-P-P-L-P motif that we have identified as being critical to CD2-mediated regulation of β1 integrin adhesiveness is similar to the consensus sequence for SH3 binding motifs, (h)-P-p-X-P. Furthermore, the SH3 domain of the tyrosine kinase p56^lck has been demonstrated to associate in vitro with a proline-rich region of the rat CD2 cytoplasmic domain that shares amino acid identity with the K-G-P-P-L-P sequence in human CD2 (6). Therefore, we reasoned that CD2 might associate directly with PI 3-K via binding of the SH3 domain of the p85 subunit with the K-G-P-P-L-P motif in the CD2 cytoplasmic domain. To our surprise, pull-down experiments with both GST-p85α(SH3) and immunoprecipitated p85 revealed that p85 was able to efficiently precipitate wild-type CD2 and all of the CD2 cytoplasmic domain mutants tested, with the exception of CD2/215, which essentially lacks all of the cytoplasmic domain. The efficiency of precipitation of CD2 by the SH3 domain of p85 or anti-p85 antibodies was not altered by prior stimulation of the cells with CD2 MAbs. Thus, the interaction between CD2 and the SH3 domain of p85 is independent of the K-G-P-P-L-P motif that is critical for CD2-mediated regulation of β1 integrin activity. This suggests that the direct association of p85 with the CD2 cytoplasmic domain is not sufficient for CD2-mediated activation of PI 3-K or integrin function. While a constitutive association between the CD2 cytoplasmic domain and p85 was previously documented in our laboratory (72), there are few other examples of a constitutive association between p85 and other cell surface receptors. However, the constitutive association of p85 with intracellular proteins, such as the docking protein p120^Cbl, has been reported (65).

The mechanistic basis for coupling of CD2 to the PI 3-K signaling pathway is currently unknown. One possibility is that the K-G-P-P-L-P motif in the CD2 cytoplasmic domain associates with a SH3 domain-containing protein that subsequently binds to and activates the p85-p110 PI 3-K complex. Alternatively, CD2 stimulation may result in the activation of a wortmannin-sensitive PI 3-K that is not dependent on the p85 subunit, such as the recently described p110γ (79). The p110γ form of PI 3-K requires Gβγ protein subunits to become catalytically active (54, 78), and recent studies have demonstrated a role for p110γ in the activation of the MAP kinase signaling pathway by G-protein-coupled receptors (54). However, pertussis toxin treatment of wild-type CD2⁺-expressing HL60 cells failed to block CD2-mediated activation of β1 integrin function (data not shown). Another possibility that deserves exploration is that CD2 stimulation may lead to phosphorylation or dephosphorylation of the catalytic p110 subunit that results in activation of constitutively associated PI 3-K.

Since stimulation with PMA and CD2/WT also results in activation of the MAP kinase signaling cascade (Fig. 10 and data not shown), we also explored the involvement of the MAP kinase signaling cascade in activation-dependent regulation of β1 integrin activity. Both pharmacological and genetic evidence suggests that CD2-mediated activation of MAP kinase is not an essential component in the CD2 regulation of β1 integrin activity. The MEK inhibitor PD 098059 (1, 23) is effective at inhibiting MAP kinase activation in HL60 cells stimulated with either CD2-specific MAbs or PMA. However, PD 098059 demonstrated differential effects on the activation-dependent upregulation of β1 integrin function in HL60 cells: it inhibited PMA-induced, but not CD2-induced, adhesion to FN. This result indicates a potential role for the MAP kinase signaling cascade in PMA-mediated activation of integrin-dependent adhesion, but it also suggests that CD2-mediated activation of MAP kinase is not essential in the signaling pathway by which CD2 regulates integrin function. PD 098059 does not appear to be nonspecifically toxic to HL60 cells, since increases in integrin function mediated by CD2 stimulation or direct activation of β1 integrins with an activating β1-specific MAb were not inhibited by the concentration of PD 098059 that effectively blocked PMA-induced adhesion (Fig. 12 and data not shown).

Structure-function analysis of the CD2 cytoplasmic domain provided additional evidence that CD2-mediated activation of MAP kinase can be dissociated from CD2-mediated activation of integrin-mediated adhesion (Table 1). Activation of MAP kinase by the CD2/305 truncation mutant was reduced when compared to wild-type CD2. However, this truncation of the CD2 cytoplasmic domain had no effect on CD2-induced activation of adhesion to FN. The CD2/P→A mutant also had a reduction in CD2-mediated activation of MAP kinase, yet this mutation completely blocked the ability of CD2 stimulation to upregulate integrin function. Finally, the CD2/299 truncation mutant was unable to activate MAP kinase or upregulate integrin function upon CD2 stimulation. This suggests that the carboxy-terminal 28 amino acids of the CD2 cytoplasmic domain are critical for both of these intracellular responses. However, the CD2/305 and CD2/P→A mutations suggest that genetic inhibition of CD2-mediated activation of MAP kinase does not impair the ability of CD2 to regulate cell adhesion. Combined with the inhibitory effects of PD 098059 on PMA-induced adhesion, our results suggest the existence of multiple signaling cascades that independently regulate integrin activity and that are utilized in a stimulus-specific fashion.

In contrast to the results with the MEK inhibitor, wortmannin had no effect on PMA-induced activation of MAP kinase but did inhibit CD2-mediated activation of MAP kinase. The inhibition of CD2-mediated activation of MAP kinase by wortmannin parallels the findings shown in Fig. 6, in that CD2 regulation of integrin function is wortmannin sensitive. The simplest interpretation of our results is that CD2-dependent activation of PI 3-K results in the activation of two independent signaling pathways, one that activates MAP kinase (presumably via MEK) and one that upregulates integrin function. A number of recent studies in a variety of cell systems are consistent with a role for PI 3-K in the activation of MAP kinase. First, expression of a constitutively active form of the p110 subunit of PI 3-K has been shown to result in the activation of certain ras-dependent responses (37). This result is consistent with an upstream requirement for PI 3-K in the ras signaling pathway, although other studies suggest that activation of ras itself may lead to activation of PI 3-K (45, 47, 63). Second, wortmannin has been shown to inhibit the activation of MAP kinase induced by a number of different stimuli, including platelet-activating factor, IL-2, IL-3, granulocyte-macrophage colony-stimulating factor, and antibody cross-linking of the CD3-TCR complex (25, 44, 66, 88). In addition, studies with wortmannin have implicated PI 3-K in the activation of MAP kinase by the polyomavirus middle T antigen (83).

In summary, we have identified a six-amino-acid sequence of the CD2 cytoplasmic domain that is essential for CD2-mediated upregulation of β1 integrin activity. This region, K-G-P-P-L-P, represents amino acids 300 through 305 of the CD2 cytoplasmic domain and is distinct from the cytoplasmic domains of CD2 responsible for CD2-regulated IL-2 production and CD2 avidity. While proline rich, this region does not mediate the interaction of CD2 with the p85 subunit of PI 3-K, yet it is critical for an induction of CD2-associated PI 3-K activity caused by CD2 engagement. CD2-mediated activation of MAP kinase can be perturbed by pharmacological or genetic approaches without affecting the ability of CD2 to activate β1 integrins. These studies provide the basis for identification of proteins that interact with the K-G-P-P-L-P sequence and for studying the role of such proteins in coupling CD2 engagement to the induction of PI 3-K activity and regulation of β1-integrin-mediated cell adhesion.