Loss of Original Antigenic Specificity in T Cell Hybridomas Transduced with a Chimeric Receptor Containing Single-Chain Fv of an Anti-Collagen Antibody and FcεRI-Signaling γ Subunit¹

Type II collagen is expressed predominantly in articular cartilage and highly conserved across species. Mice immunized with bovine type II collagen develop autoimmune arthritis (1), which bears similarity with rheumatoid arthritis. Collagen-induced arthritis in mice is partially mediated by type II collagen-reactive CD4⁺ lymphocytes producing Th1 cytokines (2). These cells can transfer arthritis to normal animals. If these collagen-specific T lymphocytes are genetically modified ex vivo to produce constitutively an antiinflammatory protein, such as soluble TNF receptor (3), or TGF-β (4), their pathogenic potential can be outbalanced by this antiinflammatory protein, because these cells reduce inflammatory reaction in the joints when injected into arthritic animals. Adoptive transfer of pathogenic self-reactive Th1 lymphocytes genetically modified to produce IL-10 (5), or transfer of self Ag-specific Th2 lymphocytes (6), has been shown to exert a therapeutic effect in other models of autoimmune diseases dominated by Th1 response.

Significant amounts of autologous self-reactive lymphocytes to use as delivery vehicles for therapeutic antiinflammatory genes are unlikely to be available in patients. However, it is perhaps possible to achieve tissue-specific homing of autologous lymphocytes by transducing them with a fusion receptor capable of coupling autoantigen recognition with TCR-mediated intracellular signaling. Such fusion receptors representing scFv of a tumor Ag-specific Ab fused with a signaling subunit CD3ζ or FcεRIγ have been used for targeting CTL to tumor cells (7, 8). These signaling subunits contain ITAM,³ the consensus motif whose tyrosine phosphorylation is one of the major membrane-proximal events in the activation of TCR-signaling pathway (9, 10).

To engraft T cells with MHC-nonrestricted specificity to collagen type II, which might be sufficient to make them home to the joints, we cloned a chimeric receptor representing collagen type II-binding ectodomain in the form of scFv of the anti-collagen mAb C2 (11) fused with the signaling γ subunit of FcεRI. To characterize this chimeric receptor, scC2Fv/γ, biochemically and functionally, it was transduced into T cell hybridomas, where it is expressed as a homodimer as well as a heterodimer with endogenous CD3ζ subunit. The transduced hybridomas, MD.45 (12) and HCQ6 (13), responded to collagen type II by production of IL-2. Dependent on predominant expression of scC2Fv/γ homodimers or scC2Fv/γ-CD3ζ heterodimers, transduced hybridoma cells demonstrated different cross-linking requirements for stimulation with collagen type II. Of interest, the chimeric receptor interferes with expression of endogenous CD3 subunits in transduced T cell hybridomas. As a result of this interference, their response to cognate Ags was considerably lower than in wild-type cells.

MD.45 is a H-2^d-restricted CTL hybridoma generated from a BALB/c mouse (12). HCQ6 is a type II collagen-specific H-2^q-restricted CD4⁺ hybridoma generated from a DBA/1 mouse (13), generously donated by R. Holmdahl (Lund, Sweden). Hybridoma-producing type II collagen-specific mAb C2 was a gift from L. Klareskog (Stockholm, Sweden). All hybridoma cells were cultured in DMEM supplemented with 10% FCS (Life Technologies, Renfrewshire, U.K.). The mAb C2 was purified from ascitic fluid of mice injected with the C2 hybridoma cells. Polyclonal anti-mAb C2 antiserum was raised in a rabbit by immunization with purified mAb C2. GP⁺E86 cells, a National Institutes of Health 3T3-based ecotropic murine-packaging cell line (14), were cultured in DMEM supplemented with 10% newborn calf serum (Life Technologies).

Total RNA was extracted from C2 hybridoma cells using guanidinium isothiocyanate (15). cDNA for the H and L chains were synthesized from this RNA preparation using SuperScript II RNase H-Reverse Transcriptase (Life Technologies), and primers complementary to mouse IgG C region (mCγ1, 5′-AGAGTTCCAKGTCAAGGTCACT; K = G or T), or κ-chain C region (mCκ1, 5′-GTAGAAGTTGTTCAAGAAGCACAC-3′). The cDNA was poly(G) tailed at the 3′ end using the enzyme terminal deoxynucleotidyl transferase (Pharmacia, Hertfordshire, U.K.), and amplified using anchored PCR. A primer complementary to the poly(G) tail (anch2pc, 5′-ACGAATTCTAGAGTCGACCCCCCCCCCCCCC) was used together with a primer complementary to mouse IgG C region upstream of the mCγ1 site (nested primer mCγ2; 5′-GAAATARCCCTTGACCAGGC; R = G or A), or a κ L chain region upstream of the constant mCκ1 site (mCκ2, 5′-AGATGTTAACTGCTCACTGGA). All primers were from Oswell DNA Service Laboratory (Southampton, U.K.). The anchor primer was allowed to extend for 15 min at 70°C, before the addition of the second primers mCγ2 or mCκ2. The PCR was conducted as follows: 94°C for 1 min, 62°C for 1 min, and 72°C for 1 min, with a final extension at 72°C for 10 min in a Hybaid Omnigene programmable thermocycler (Hybaid, Middlesex, U.K.). The PCR products were cloned using the Invitrogen TA cloning vector (pCR^TMII; Invitrogen BV, Leek, The Netherlands). Plasmid DNA was sequenced using Sequenase T7 DNA polymerase (United States Biochemical, Amersham International, Buckinghamshire, U.K.) with mCγ2 or mCκ2.

The L chain V region of the collagen type II-specific mAb C2 (C2VL) was amplified by PCR from the plasmid pCRII using oligonucleotide primers 5′-AAGATCTGGACATTGTGCTGACACAG and 5′-CCTGTTTCAACCTTTATTTTGGCCAGTGGA, which contain BglII and BstEII sites, respectively. The PCR product was cut with the same enzymes.

A DNA sequence encoding the transmembrane and cytoplasmic domain of the human FcεRIγ subunit with BstEII (upstream) and XhoI (downstream) sites was cut out of pRSVscFvRγ (7) by digestion with XhoI, blunting with Klenow fragment of DNA polymerase (Boehringer Mannheim, East Sussex, U.K.), and digestion with BstEII. The plasmid pIg16 (16) contains a DNA sequence encoding a polyglycine-serine linker (GGGGS)₃ bounded with XbaI site at the 5′ end and BglII site at the 3′ end. In this plasmid, (GGGGS)₃ is followed by a DNA sequence encoding the 3′ portion of the insert with Asp⁷¹⁸ restriction site at its 3′ end. To prepare pIg16 for subcloning of C2VL and γ subunit into it, its DNA was cut with Asp⁷¹⁸, filled in with Klenow fragment to form blunt ends, and digested with BglII. The C2VL and γ subunit were subcloned at one time into the Asp⁷¹⁸-blunted/BglII-cleaved pIg16 immediately downstream of (GGGGS)₃, C2VL in front of the γ subunit. The product contained a DNA fragment coding for (GGGGS)₃-C2VL-γ subunit. This fragment containing XbaI (upstream) and restored XhoI (downstream) sites was excised with the same enzymes.

A DNA fragment encoding the H chain V region of the mAb C2 (C2VH) was amplified by PCR from the plasmid pCRII using oligonucleotide primers 5′-TGGATCCTGAACACACATCCCTTACCATGG and 5′-GTCTAGATGAGGAGACTGTGAGAGT, the latter containing XbaI site. The PCR product was filled in with Klenow fragment to form blunt ends, and digested with XbaI. The C2VH was subcloned at one time with the (GGGGS)₃-C2VL-γ subunit into SnaB1/SalI-cleaved retroviral vector pBabeNeo (17), C2VH in front of the (GGGGS)₃-C2VL-γ subunit, to yield pBabeNeo+scC2Fv/γ containing the final insert C2VH-(GGGGS)₃-C2VL-γ subunit (EMBL database accession number X9676).

A nonsignaling version of scC2Fv/γ was produced by truncation of its intracellular domain. DNA from pBabeNeo+scC2Fv/γ was cut with SalI, filled in with Klenow fragment, and recircularized with T4 DNA-ligase to introduce a frame shift and stop codon. As a result of this modification, the entire intracellular domain apart from its first arginine was replaced with an irrelevant sequence of 14 amino acids yielding scC2Fv/γIC⁻.

The GP+E86 cells were transfected using calcium phosphate precipitation and glycerol shock, as described previously (4). The cells were transfected with 20 μg of plasmid DNA from pBabeNeo+scC2Fv/γ or pBabeNeo+scC2Fv/γIC⁻. Transfected cells were selected in 1 mg/ml G418. T cell hybridomas were infected with viruses containing genes of the chimeric receptors by coculture overnight with the transfected GP+E86 cells. Hybridoma cells transduced with scC2Fv/γ or scC2Fv/γIC⁻ were selected with 3 mg/ml of G418. This resulted in four transduced cell lines, including MD.45 cells expressing scC2Fv/γ or scC2Fv/γIC⁻, and HCQ6 cells expressing scC2Fv/γ or scC2Fv/γIC⁻, called, respectively, MD.45(C2γ), MD.45(C2γIC⁻), HCQ6(C2γ), and HCQ6(C2γIC⁻). Hybridoma cells transduced with an empty retroviral vector pBabeNeo only, MD.45(BabeNeo) and HCQ6(BabeNeo), were used as a control.

To label cell surface proteins with biotin, viable cells were washed five times in PBS with 1 mM CaCl₂ and 1 mM MgCl₂ and resuspended in the same buffer (5 × 10⁶ cells/ml). Sulfosuccinimidyl 6-(biotin-amido) hexanoate (Pierce, Rockford, IL) was added from stock solution in DMSO to a final concentration of 0.1 mg/ml. The cells were incubated for 1 h on ice and washed five times in PBS.

Surface-biotinylated or nonlabeled cells were solubilized by incubating for 30 min on ice in lysis buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40 (BDH, Poole, U.K.), 1 μg/ml leupeptin, 1 μg/ml chymostatin, 1 μg/ml pepstatin A, 1 mM PMSF, 1 mM aprotinin, and 10 mM iodoacetamide (all from Sigma, St. Louis, MO). IgG from rabbit anti-C2 antiserum was covalently bound to protein A-Sepharose beads (Pharmacia) by incubating the beads with the rabbit anti-C2 antiserum and 20 mM dimethylpimelimidate (Pierce) in 0.2 M borate buffer, pH 9, for 45 min at room temperature. The reaction was stopped with 0.2 M ethanolamine, pH 8, and the beads were washed five times in PBS. The protein A-Sepharose beads with covalently bound anti-C2 IgG were used for immunoprecipitation of nuclear-free supernatants of lysed surface-biotinylated cells. The immunoprecipitation was performed for 4 h at 4°C. For immunoprecipitation with the anti-CD3ζ mAb 6B10.2 (Santa Cruz Biotechnology, Santa Cruz, CA), nuclear-free supernatants of nonlabeled cells were incubated with 1 μg/ml of the mAb for 1 h on ice, and the immune complexes were precipitated with protein A-Sepharose beads for 1 h at 4°C. After immunoprecipitation, protein A-Sepharose beads were washed five times with lysis buffer.

Nuclear-free cell lysates, or immunoprecipitated proteins, were separated by SDS-PAGE and transferred onto nitrocellulose membrane, which was then blocked with nonfat milk. Biotinylated scC2Fv chimeras were detected by blotting with streptavidin-biotinylated horseradish peroxidase complex (Amersham). Nonlabeled scC2Fv chimeras were detected by immunoblotting with 1/1000 dilution of rabbit anti-C2 antiserum, followed by F(ab′)₂ of donkey anti-rabbit IgG conjugated with horseradish peroxidase (Amersham). For detection of CD3ζ, blots were incubated with 1 μg/ml of the anti-CD3ζ mAb 6B10.2, followed by horseradish peroxidase-conjugated sheep anti-mouse IgG (Amersham International). All blots were developed using a chemoluminescence kit (ECL, Amersham) and exposed to an x-ray film (Kodak-Pathe, Paris, France).

The x-ray films were semiquantitatively analyzed with Bio-Rad GS-670 Imaging Densitometer (Bio-Rad, Hertfordshire, U.K.). In transduced hybridoma cells, scC2Fv chimeras form homodimers as well as heterodimers with endogenous CD3ζ. The following formula was used to calculate the molar ratio of the homodimers to the heterodimers:

where M_hmd is the intensity of the band representing homodimer of scC2Fv/γ or scC2Fv/γIC⁻; and M_htrd is the intensity of the band representing heterodimer of the scC2Fv chimera and CD3ζ.

For detection of cell surface-expressed scC2Fv/γ and scC2Fv/γIC⁻, cells were stained with 1/200 dilution of the rabbit anti-C2 antiserum, followed by FITC-labeled F(ab′)₂ fragment of goat anti-rabbit IgG Ab (Sigma). The samples were analyzed with a FACS (Becton Dickinson, Mountain View, CA) using Consort 30 software. Cell surface-expressed CD3 was detected by incubating cells with 10 μg/ml of anti-mouse CD3 mAb 145-2C11, followed by incubation with FITC-conjugated affinity-purified goat anti-Armenian hamster IgG (Stratech Scientific, Luton Beds, U.K.). Samples were analyzed with a FACS (Becton Dickinson) using Cell Quest software.

Collagen type II was purified from bovine articular cartilage (18). T cell hybridomas were stimulated with collagen attached to the wells of a flat-bottom 96-well plate, or added to culture medium. Rabbit IgG enriched with Abs specific to Fv domain of the mAb C2 (further referred to as anti-mAb C2 IgG) was purified from the antiserum by a two-step affinity chromatography. First, anti-mouse IgG Fc Abs were depleted from the antiserum by passing through the column of mouse IgG coupled to Sepharose (Pharmacia). Anti-mAb C2 IgG was then isolated from this antiserum by affinity chromatography on mAb C2 bound to Sepharose (Pharmacia). Anti-mAb C2 IgG was also used for cell stimulation either attached to the microtiter wells or added to culture medium. In some experiments, the hybridomas were stimulated with the anti-CD3ε mAb 145-2C11 attached to the microtiter wells. For stimulation with their cognate Ags, MD.45 cells were cocultured with their target EL-4 cells, and HCQ6 cells were cultured with collagen added to culture medium in the presence of freshly isolated DBA/1 splenocytes. Concentration of murine IL-2 in culture supernatants was measured for quantitation of cell activation. It was determined by ELISA using mAb JES6-1A12 (PharMingen, San Diego, CA) as coating Ab and biotinylated mAb JES6-5H4 (PharMingen) for detection. The lower detection limit of this ELISA was 15 pg/ml.

T cell hybridomas MD.45 and HCQ6 transduced with scC2Fv/γ or scC2Fv/γIC⁻ expressed these chimeric receptors on the cell surface, as demonstrated by FACS analysis of cells stained with the polyclonal anti-mAb C2 antiserum (Fig. 1). Levels of scC2Fv/γ and scC2Fv/γIC⁻ expression were equivalent in transduced HCQ6 cells, and comparable in transduced MD.45 cells.

To determine the m.w. of scC2Fv/γ, pBabeNeo+scC2Fv/γ-transfected GP+E86 packaging fibroblasts were surface biotinylated, and their lysates were immunoprecipitated with rabbit anti-C2 antiserum. Fig. 2 demonstrates that transfected GP+E86 cells express scC2Fv/γ in the form of a disulfide-linked homodimer. As averaged from seven independent experiments (immunoprecipitation and immunoblotting), its molecular mass is 74 ± 5 kDa (mean ± SD). Under reducing conditions, scC2Fv/γ migrated as a single band of 36.7 kDa (Fig. 2), which corresponds to the predicted molecular mass of the monomeric protein.

We analyzed the subunit structure of scC2Fv/γ chimera in transduced lymphoid cells, bearing in mind that these molecules contain the transmembrane and intracellular domains of the FcεRI-signaling γ subunit that has been shown to associate with the CD3ζ subunit (19, 20). To this end, cell lysates were separated by SDS-PAGE under nonreducing conditions, transferred onto nitrocellulose membrane, and immunoblotted with rabbit anti-C2 antiserum. Fig. 3 demonstrates that in both T cell hybridomas, in addition to homodimerization, scC2Fv/γ forms a heterodimer of 54 kDa with an endogenous protein. Analogous homo- and heterodimers were present in cells transduced with scC2Fv/γIC⁻, suggesting that the cytoplasmic domain of scC2Fv/γ was not involved in the intermolecular association (Fig. 3). Based on the difference in the m.w. between monomeric scC2Fv/γ and the heterodimer, the latter could represent scC2Fv/γ associated with endogenous CD3ζ, which has been directly demonstrated by immunoprecipitation (see below). The four types of homodimeric and CD3ζ-associated scC2Fv chimeras expressed in the T cell hybridomas are presented schematically in Fig. 4.

When transduced into MD.45, scC2Fv chimeras were expressed predominantly as homodimers, whereas heterodimers of chimeric molecules and CD3ζ were more abundant in HCQ6 cells. As determined by densitometric analysis of immunoblots presented in Fig. 3, the molar ratio of homodimers to heterodimers in MD.45(C2γ) and MD.45(C2γIC⁻) cells was 2.57 and 2.7, respectively. In HCQ6 cells transduced with scC2Fv/γ or scC2Fv/γIC⁻, this ratio was 0.19 and 0.22, respectively. In other words, 1 of 3.5 chimeric receptors in transduced MD.45 cells is a heterodimer, and 1 of 6 receptors in transduced HCQ6 cells is a homodimer. The higher proportion of CD3ζ-associated scC2Fv chimeras in HCQ6 cells may be accounted for by a fivefold higher CD3ζ content in wild-type HCQ6 cells than in wild-type MD.45 cells (not shown data of immunoblotting).

Fig. 5 demonstrates that the heterodimers found in transduced T cell hybridomas represent the CD3ζ-associated scC2Fv chimeras, because they could be immunoprecipitated with an anti-CD3ζ mAb from Nonidet P-40-lysed HCQ6(C2γ) and HCQ6(C2γIC⁻) cells. No band of this m.w. was precipitated from lysates of wild-type HCQ6 cells by the anti-CD3ζ mAb, or from lysates of HCQ6(C2γ) cells by an isotype-matched control IgG1 (Fig. 5). Although proteins were separated under nonreducing conditions in this experiment, slight degradation of Igs used for immunoprecipitation did occur. As a result of this degradation, immunoblotting with rabbit anti-mouse mAb C2 antiserum visualized the heavy chains of the mouse anti-CD3ζ mAb 6B10.2, or control mouse IgG1. This signal is significantly weaker than the band corresponding to undegraded IgG on this film (not shown).

To assess the ability of scC2Fv chimeras to function as cell surface receptors, transduced and nontransduced MD.45 and HCQ6 cells were stimulated with type II collagen, and IL-2 levels in their culture supernatants were determined. Collagen was either added to culture medium or attached to plastic. It is reasonable to assume that, compared with plastic-immobilized collagen, collagen in solution has lower capacity to cross-link scC2Fv chimeric receptors. In the absence of collagen, none of the wild-type cell lines or transduced cells produced detectable levels of IL-2. Collagen in solution, or immobilized to plastic, did not stimulate IL-2 secretion in wild-type MD.45 or HCQ6 cells. At the same time, as expected, wild-type hybridoma cells produced high levels of IL-2 in response to stimulation with the anti-CD3 mAb 145-2C11 (Fig. 10).

In response to stimulation with plastic-immobilized collagen, MD.45(C2γ) and HCQ6(C2γ) cells secreted IL-2, demonstrating biphasic dose-response relationship (Fig. 6, A and D). Optimal concentration of collagen for coating microtiter wells was within the range of 1–10 μg/ml. IL-2 production progressively declined, as the concentration rose further. scC2Fv/γ binds to collagen through the antigenic epitope recognized by native mAb C2, because preliminary treatment of collagen-coated wells with the mAb C2 inhibited IL-2 production by MD.45(C2γ) cells (Fig. 6 C). This did not result from a nonspecific loss of responsiveness or viability in the hybridoma cells in the presense of the mAb C2, because IL-2 response of wild-type MD.45 cells to the anti-CD3 mAb 145-2C11 in control experiments was unaffected by the mAb C2. Similarly, if microtiter wells coated with 1 μg/ml collagen were pretreated with 50 μg/ml mAb C2 before adding cells, IL-2 production by HCQ6(C2γ) and HCQ6(C2γIC⁻) was inhibited by 93 and 97%, respectively (not shown).

As expected, MD.45(C2γIC⁻) cells did not produce significant levels of IL-2 in response to stimulation with plastic-immobilized collagen (Fig. 6,A). However, HCQ6(C2γIC⁻) cells stimulated with plastic-immobilized collagen secreted substantial levels of IL-2, which were about 70% of IL-2 levels secreted by HCQ6(C2γ) cells (Fig. 6,C and Table I). The high level of responsiveness to plastic-immobilized collagen in HCQ6(C2γIC⁻) cells finds explanation in a relatively high expression of scC2Fv/γIC⁻-CD3ζ heterodimers in these hybridoma cells (Fig. 3).

MD.45(C2γ) cells were responsive to soluble collagen (Fig. 6,B). As summarized in Table I, their IL-2 production induced by optimal concentration of soluble collagen was 81% of the IL-2 production induced by stimulation with optimal concentration of plastic-immobilized collagen. By contrast, HCQ6(C2γ) cells stimulated with soluble collagen produced only 2.4% of the amount of IL-2 that they produced in response to stimulation with plastic-immobilized collagen (Table I). When transduced HCQ6 cells were stimulated with polyclonal anti-mAb C2 IgG instead of collagen type II, they demonstrated similar cross-linking requirements for activation. Thus, anti-mAb C2 IgG attached to microtiter wells was stimulatory for these cells (Fig. 6,E), whereas IgG added to culture medium (the highest concentration tested was 20 μg/ml) did not induce detectable levels of IL-2 in them (not shown). In similarity to response to plastic-immobilized collagen, response to plastic-immobilized anti-mAb C2 IgG was totally inhibited by pretreatment of the microtiter wells with mAb C2 (Fig. 6,E). MD.45(C2γIC⁻) and HCQ6(C2γIC⁻) cells were unresponsive to collagen in solution (Fig. 6,B and Table I).

Subunit structure of the scC2Fv chimeras in transduced cells is perhaps an important factor determining cross-linking requirements for activation of these cells by collagen. The capacity to respond to collagen in solution appears to correlate with the presence of scC2Fv/γ homodimers in transduced cells, because MD.45(C2γ) cells, the only cells that responded to soluble collagen, contained the highest level of these homodimers. Meanwhile, cells that were unable to respond to soluble collagen either did not contain them, like MD.45 and HCQ6 cells transduced with scC2Fv/γIC⁻, or, like HCQ6(C2γ) cells, expressed lower level of the homodimers than MD.45(C2γ) cells (Fig. 3). This allows speculation that heterodimers of endogenous CD3ζ and scC2Fv chimeras do not initiate intracellular signaling events in response to stimulation with soluble collagen. To determine whether or not the heterodimers were capable of binding collagen from solution, we studied the effect of preincubation with soluble ligand on subsequent cell activation with plastic-immobilized collagen. Results of such experiments performed using HCQ6(C2γIC⁻) cells in which CD3ζ-scC2Fv/γIC⁻ heterodimer is the only ITAM-containing receptor for unprocessed collagen are shown in Fig. 7. Soluble collagen inhibited IL-2 production induced by plastic-immobilized collagen, which suggests that it blocks binding of the CD3ζ-associated scC2Fv/γIC⁻ to plastic-immobilized collagen by occupying the receptor or causing its internalization without initiation of downstream signaling events leading to IL-2 production. Another possible explanation is delivery of a negative regulatory signal by binding of soluble collagen to the heterodimer. IL-2 production by wild-type HCQ6 cells stimulated with plastic-immobilized anti-CD3 mAb 145-2C11 was unaffected by soluble collagen (not shown).

We asked whether or not response to cognate Ags was altered in T cell hybridomas transduced with scC2Fv chimeras. TCR-mediated activation of HCQ6 cells can be quantitatively estimated by IL-2 release in response to stimulation with type II collagen processed and presented by syngeneic APC. As scC2Fv chimeras expressed in HCQ6(C2γ) and HCQ6(C2γIC⁻) cells were unresponsive to soluble collagen, IL-2 production induced in these cells by soluble collagen in the presence of APC could be fully attributed to TCR-mediated stimulation. In the experiment presented in Fig. 8, wild-type or scC2Fv chimera-transduced HCQ6 cells stimulated with soluble collagen in the absence of APC did not secrete detectable levels of IL-2 (not shown). In response to stimulation with soluble collagen presented by DBA/1 splenocytes (Fig. 8,A), HCQ6(C2γ) and HCQ6(C2γIC⁻) cells produced much lower levels of IL-2 than nontransduced HCQ6 cells. The overall ability of HCQ6(C2γ) and HCQ6(C2γIC⁻) cells to produce IL-2, however, was not impaired, because under optimal conditions their response to unprocessed plastic-immobilized collagen (Fig. 8,B) was equivalent to the response of wild-type HCQ6 cells stimulated with APC-presented collagen at 100 μg/ml (Fig. 8,A). Preliminary experiments demonstrated that this concentration of collagen was optimal for the splenocyte-dependent activation of HCQ6 cells (not shown). To study the effect of scC2Fv chimeras on TCR-mediated functions in MD.45 cells, wild-type and transduced MD.45 cells were stimulated with their specific target EL-4 cells (12) (Fig. 8 C). Profound decrease in IL-2 response was observed in transduced cells. Response of hybridomas transduced with an empty retroviral vector was not different from that of wild-type cells (not shown).

Trying to elucidate the underlying mechanism of the reduced responsiveness to cognate Ags in T cell hybridomas expressing the chimeric receptors, we compared expression of endogenous CD3ζ and CD3ε in transduced and wild-type hybridoma cells. Immunoblot analysis presented in Fig. 9,A demonstrates that, compared with wild-type cells, scC2Fv chimera-transduced cells contained reduced amount of CD3ζ. Even in transduced HCQ6 cells, in which heterodimers of scC2Fv chimeras and CD3ζ were readily detectable by immunoblotting with the anti-mAb C2 antiserum (Fig. 3), the chimera-associated CD3ζ could not be revealed by immunoblotting with the anti-CD3ζ mAb 6B10.2, presumably due to an insufficient sensitivity of this staining. Therefore, overall decrease in CD3ζ content in the Nonidet P-40-soluble fraction of the cell lysates, rather than its redistribution in favor of the scC2Fv chimera-bound form, occurred in the transduced hybridomas. Results of FACS analysis (Fig. 9,B) demonstrated that CD3ε expression was also decreased on the surface of hybridoma cells transduced with scC2Fv chimeras, but not in control cells transduced with an empty retroviral vector. Data presented in Fig. 9,B are from one of five independent experiments. The decrease in CD3ε expression was highly reproducible, but its degree varied between these experiments from slight reduction compared with control to nearly undetectable levels. Such variation of CD3ε expression in scC2Fv chimera-transduced cells is compatible with functional down-modulation of endogenous CD3 subunits in the presence of the chimeric receptors, rather than with selection of clones expressing low CD3ε levels during generation of transduced cells. Fig. 10 demonstrates that reduced expression of the CD3ε subunit in hybridomas expressing scC2Fv chimeras was accompanied by a decreased response to stimulation with anti-CD3ε mAb 145-2C11. The reduction in responsiveness to anti-CD3ε was perhaps greater than one would expect, considering only partial decrease in the expression of CD3ε in scFv chimera-transduced cells. The response of cells, transduced with an empty retroviral vector, however, remained unaltered (Fig. 10).

The present study demonstrates that transduction with the gene encoding scC2Fv/γ, a chimeric receptor containing scFv domain of the anti-collagen mAb C2 and the FcεRIγ subunit, confers Ab-type specificity to collagen on T cell hybridomas. In addition, expression of endogenous CD3 subunits and response to the cognate Ag in T cell hybridomas expressing scC2Fv/γ are reduced. This receptor forms homodimers as well as heterodimers with endogenous CD3ζ subunit. The homodimerization of scC2Fv/γ and its association with CD3ζ do not require integrity of its cytoplasmic domain, as this ability is fully maintained in its mutein, scC2Fv/γIC⁻, lacking the FcεRIγ-signaling domain. Presumably, homodimerization of scC2Fv chimeras and their covalent association with CD3ζ are mediated by their FcεRIγ-derived transmembrane domains, because a cysteine residue in this domain is known to mediate homodimerization of native FcεRIγ, as well as its association with CD3ζ (19). Decreased expression of endogenous CD3ζ subunit in T cell hybridomas transduced with scC2Fv chimeras is a major reason for decreased TCR-mediated responsiveness because this subunit is a limiting factor for expression of TCR/CD3 complex (21). The exact mechanism of reduction in CD3ζ expression in T cells transduced with FcεRIγ-containing chimeras remains unclear. However, it does not appear to result from sequestration of CD3ζ with the engrafted chimera, as overall decrease in CD3ζ expression was observed in scC2Fv chimera-transduced hybridomas. One could hypothesize that for an unknown reason, production of CD3ζ is suppressed in the presence of scC2Fv chimeras, or CD3ζ-scC2Fv/γ and CD3ζ-scC2Fv/γIC⁻ heterodimers have higher turnover rate, compared with homodimeric CD3ζ, thus leading to a rapid depletion of a newly synthesized CD3ζ. Increased association of CD3ζ with the cytoskeleton and its disappearance from the Nonidet P-40-soluble fraction (22) in transduced hybridomas is also a possibility. With regard to decreased CD3ζ expression, the T cell hybridomas transduced with scC2Fv chimeras are reminiscent of unresponsive T cells in conditions such as rheumatoid arthritis, HIV infection, or cancer (23, 24, 25). The decrease in the TCR-mediated response seems to be specific for chimeras containing FcεRIγ subunit, since MD.45 cells expressing a chimera containing scFv and TCR α or β subunit apparently remained fully responsive to stimulation with EL-4 cells (26).

Taking into account that dependence on CD3ζ is a ubiquitous regulatory mechanism of TCR/CD3 expression (27), TCR-dependent response of primary lymphocytes transduced with FcεRIγ subunit-containing chimeras would perhaps also be reduced. Mitogenic stimulation of T lymphocytes in vitro, to facilitate their infection with scC2Fv/γ-containing retrovirus, might lead to expansion of autoimmune T cell clones. Loss of sensitivity to the original Ag in the transduced cells, however, would prevent such clones from revealing their harmful potential.

The FcεRIγ and CD3ζ belong to the same molecular family, and their transmembrane domains have 58% amino acid identity (28). These domains mediate noncovalent interaction of FcεRIγ and CD3ζ with other components of Ig FcR or the TCR/CD3 complex (27, 29, 30). In addition, the ectodomain of CD3ζ is also involved in association with other CD3 subunits (31). This suggests a possibility that apart from CD3ζ, other endogenous CD3 subunits are noncovalently associated with homo- and heterodimers of the chimeric receptors and contribute to intracellular signal transduction upon oligomerization of these receptors. As far as the homodimers of scC2Fv chimeras are concerned, the observation that MD.45(C2γIC⁻) cells that predominantly express homodimers of the ITAM-less chimera scC2Fv/γIC⁻ do not respond to collagen (Fig. 6, A and B) argues against this possibility. A number of published reports suggest that FcεRIγ and CD3ζ are interchangeable signaling subunits. Thus, FcεRIγ can substitute CD3ζ in supporting assembly and function of the TCR/CD3 complex in CD3ζ knockout mice (32). Another line of evidence suggesting functional similarity between the two signaling molecules is the observation that in basophils, the FcεRIγ subunit can activate Syk kinase (33), the tyrosine kinase playing an important role in TCR-mediated signaling (34, 35). Therefore, scC2Fv/γ homodimers and heterodimers of scC2Fv chimeras with CD3ζ may initiate qualitatively similar downstream signaling events in T cells.

Apparently, scC2Fv/γ homodimers and heterodimers of scC2Fv chimeras and CD3ζ have different cross-linking requirements for activation, because the capacity of transduced hybridoma cells to respond to collagen in solution correlated with predominant expression of the homodimers (Figs. 3 and 6). The reason that heterodimers of scC2Fv chimeras and CD3ζ need plastic-attached ligand for efficient oligomerization is probably the fact that they have only one ligand-binding site per receptor, hence, lower avidity to collagen. Soluble collagen, however, binds to the monovalent heterodimers, as it inhibits cell activation by plastic-immobilized collagen (Fig. 7). It remains unclear whether soluble collagen is unable to induce optimal oligomerization of the heterodimers, or its interaction with the receptor results in the generation of a negative regulatory signal. Such negative regulatory signal delivered through heterodimers could play a role in reduced response of chimera-transduced HCQ6 cells to collagen presented by splenocytes (Fig. 8,A). This, however, seems to be unlikely because in the absence of collagen, transduced hybridomas also exhibited decreased TCR/CD3-mediated responses (Figs. 8 C and 10).

Our recent observation on T cell hybridomas transduced with scC2Fv/CD8/ζ, a chimera representing scC2Fv ectodomain fused with CD8α hinge region and CD3ζ transmembrane and cytoplasmic domains (manuscript in preparation), lends support to the notion that expression of divalent receptors is a requirement for responsiveness to soluble collagen. Thus, scC2Fv/CD8/ζ is expressed only as a homodimer in both MD.45 and HCQ6 cells, without association with endogenous CD3ζ in either of them, and both hybridomas transduced with scC2Fv/CD8/ζ respond to soluble collagen by IL-2 production. Collagen in solution can activate cells expressing scC2Fv/γ homodimers because it is a multivalent Ag. Multivalent soluble Ags, however, may differ in the ability to induce oligomerization of scFv-containing chimeras optimal for generating intracellular signals in T cells. For instance, MD.45 cells transduced with the chimera consisting of the trinitrophenol-specific scFv and FcεRIγ were unresponsive to fowl γ-globulin-coupled trinitrophenol in solution (7). The ratio between scC2Fv/γ homodimers and scC2Fv/γ-CD3ζ heterodimers in transduced primary lymphocytes would determine their cross-linking requirements for activation with collagen. Therefore, had such genetically modified cells been used for gene therapy of arthritis, this ratio would be an important factor that would affect their behavior in vivo.

Does the difference between the homo- and heterodimers in the composition of their cytoplasmic domains contribute to their differential cross-linking requirements? With this regard, of interest is the observation of Sunder-Plassmann et al. (36), who have shown that T cells expressing a chimeric receptor containing all three ITAM of CD3ζ could be activated by cross-linking with a mAb specific to the receptor ectodomain, whereas the chimera with single ITAM required additional cross-linking with a secondary reagent for activation. However, no such correlation has been found in the present study, because the scC2Fv/γ homodimers, receptors that are more readily cross-linked with collagen, contain fewer ITAM per receptor than the heterodimers of CD3ζ and scC2Fv chimeras. Apparently, in this case, the structure of the ectodomain is more important in determining cross-linking requirements for receptor activation than the number of ITAM in the cytoplasmic domain.

In summary, transduction with the gene of the chimeric receptor scC2Fv/γ composed of a single-chain collagen-specific Ab fused with FcεRIγ confers Ab-type specificity to collagen on T cell hybridomas. This receptor is expressed as a homodimer as well as a heterodimer with endogenous CD3ζ. The two forms of the receptor differ in their cross-linking requirements for oligomerization. The original TCR-mediated reactivity of the transduced cells is reduced profoundly, as a result of which they become specific predominantly to collagen type II.

We thank Prof. M. Feldmann and Dr. B. Foxwell for critically reading the manuscript.