T Cell Activation by Antibody-Like Immunoreceptors: The Position of the Binding Epitope within the Target Molecule Determines the Efficiency of Activation of Redirected T Cells¹

T cells can be grafted with predefined specificity by expression of a recombinant TCR molecule (immunoreceptor) that binds Ag by an extracellular, Ab-derived binding domain and mediates T cell activation by an intracellular, CD3ζ-derived signaling domain. The strategy thereby combines the advantages of specific Ag recognition with receptor-triggered T cell effector functions, including cytokine secretion, T cell proliferation, and cytolysis of defined target cells (1, 2, 3). By using a single-chain fragment of variable regions (scFv)⁵ Ab as binding domain, the design of the immunoreceptor allows us to target Ags of any chemical composition, e.g., peptides or carbohydrates (4, 5), independently of simultaneous MHC recognition. Accordingly, T cells equipped with this type of Ab-derived immunoreceptors can be directed against a broad variety of cells that express the target Ag (3, 6). The efficiency of T cell activation is thought to be due to the efficiency of receptor cross-linking and intracellular signaling upon Ag binding. The efficiency in Ag binding is determined by the affinity of the scFv domain, although increase in affinity above threshold does not necessarily result in an increase in T cell activation (7). The efficiency in immunoreceptor-mediated T cell activation may also be determined by the targeted epitope itself, particularly its position within the Ag. This question, however, is so far not resolved, although a number of recombinant immunoreceptors has been generated during the last years and some of them entered clinical trials (6).

To address this issue, we generated two immunoreceptors with binding specificity for carcinoembryonic Ag (CEA), one immunoreceptor targeting the membrane distal N domain, the other the proximal A3 epitope. Upon binding to immobilized, isolated CEA activation of peripheral blood T cells is dependent on the binding affinity of the recombinant immunoreceptor. If CEA is expressed on the cell surface, interestingly, T cell activation is more efficient when targeting the membrane proximal A3 than the distal N epitope. This is not due to the epitope itself or the binding affinity, but is dependent on the position of the epitope within the CEA molecule, indicating that the position of the targeted epitope has obviously major impact on the efficacy of T cell activation independently of the binding efficiency of the immunoreceptor.

293T cells are human embryonic kidney cells that express the SV40 large T Ag (8). The CEA-expressing colon carcinoma cell lines LS174T (ATCC CCL 188) and SW948 (ATCC CCL 237) and the CEA-negative cell lines A375 (ATCC CRL-1619) and Colo320 (ATCC CCL 220.1) were obtained from American Type Culture Collection. The mouse tumor cell line C15A3 that expresses human CEA after transfection was described earlier (9). OKT3 (ATCC CRL 8001) is a hybridoma cell line that produces the anti-CD3 mAb OKT3. 293T cells were cultured in DMEM supplemented with 10% (v/v) FCS; all other cell lines were cultured in RPMI 1640 medium, 10% (v/v) FCS (all Invitrogen Life Technologies). OKT3 mAb was affinity purified from hybridoma supernatants using goat anti-mouse IgG2a Abs (Southern Biotechnology Associates) that were immobilized on N-hydroxysuccinimide ester-activated Sepharose as recommended by the manufacturer (Amersham Biosciences). The goat anti-human IgG Ab and its biotin-, FITC-, or PE-conjugated F(ab′)₂ derivatives were purchased from Southern Biotechnology Associates. The anti-human IFN-γ mAb NIB42 and the biotinylated anti-human IFN-γ mAb 4S.B3 were purchased from BD Biosciences and the anti-human TCRζ mAb 2H2D9 from Beckmann Coulter. Purified CEA was purchased from Calbiotech and bovine submaxillary mucin (BSM) from Sigma-Aldrich.

The anti-CEA scFv BW431/26 immunoreceptors were described earlier (10, 11). The BW431/26-scFv-Fc fusion protein was generated by deletion of the intracellular signaling domain. The anti-CEA scFv H10 was isolated from a human nonimmune library by phage display using a synthetic N-A3 CEA fragment as binding substrate (12). Briefly, a human scFv phage display library of a diversity of 8 × 10⁸ was screened for three rounds for binding on purified N-A3 and 96 individual clones were retested by ELISA. The cDNA of clone scFv H10 was subcloned as a SfiI/NotI fragment into the pUC119-based pSin1 vector (P. Schumacher and R. Finnern, unpublished observations). To generate the anti-CEA Ab H10-scFv-Fc, H10-scFv DNA was amplified from the vector pSyn1-H10-scFv using the following oligonucleotide primers: Lκ-5′, 5′-CCACGTACCATGGATTTTCAGGTGCAGATTTTC-3′ and M12-scFv-3′Bgl, 5′-TTCTAGATCTGCACCTAGGACGGTCAGCTTGGT-3′ introducing the SnabI, NcoI, and BglII restriction sites, respectively (underlined). The PCR product was ligated into the SnaBI and BamHI restriction sites of pRSV-BW431/26-scFv-Fc, thereby replacing the BW431/26-scFv and forming the expression plasmid pRSV-H10-scFv-Fc. For stable expression of the recombinant Abs BW431/26-scFv-Fc and H10-scFv-Fc, respectively, 293T cells were cotransfected with 12 μg of pRSV-BW431/26-scFv-Fc or pRSV-H10-scFv-Fc DNA and 6 μg of pGT60hB7–2v.03 DNA encoding the hygromycin resistance gene (InvivoGen). Stably transfected cells were selected in the presence of 200 μg/ml hygromycin B and cell clones were analyzed by ELISA for secretion of the respective Ab scFv-Fc fusion protein. Briefly, fusion proteins were bound to a solid-phase goat anti-human IgG Ab (1 μg/ml) and detected by a biotinylated goat anti-human IgG Fc Ab (0.25 μg/ml). The reaction product was visualized by reaction of a peroxidase-streptavidin-conjugate (1:10,000) with ABTS (both Roche Diagnostics).

BW431/26-scFv-Fc and H10-scFv-Fc fusion proteins were incubated on microtiter plates (Polysorb; Nunc) that have been coated overnight at 4°C with purified CEA or BSM for control (each 4 μg/ml). Bound fusion proteins were detected by ELISA. Cross-competition experiments were performed as follows. 293T cells (2 × 10⁶) were transfected transiently as described below with plasmid DNA coding for the BW431/26-scFv-Fc-ζ and H10-scFv-Fc-ζ receptor, respectively. Transfected cells were lysed by 1% (v/v) Nonidet P-40 and lysates of 293T cells (4 × 10⁶ cells/ml lysate) that express the BW431/26-scFv-Fc-ζ and H10-scFv-Fc-ζ receptor, respectively, were incubated in the presence of serial dilutions of supernatants containing the BW431/26-scFv-Fc and H10-scFv-Fc fusion protein, respectively. Bound receptor proteins were detected by the anti-TCR ζ mAb 2H2D9 (2 μg/ml) and a biotinylated goat anti-mouse IgG Ab.

The generation of the retroviral expression cassettes for the recombinant BW431/26-scFv-Fc-ζ and BW431/26-scFv-Fc-γ immunoreceptors was described in detail previously (13). To generate the retroviral expression cassettes for the recombinant H10-scFv-Fc-ζ and H10-scFv-Fc-γ immunoreceptor, respectively, the DNA coding for H10-scFv was amplified by PCR using the oligonucleotide primers Lκ-5′ and M12-scFv-3′Bgl and thereby flanked by NcoI (5′) and BamHI (3′) restriction sites, respectively. The BW431/26-scFv DNA within the BW431/26-scFv-Fc-ζ and BW431/26-scFv-Fc-γ DNA, respectively, (13, 14) was replaced by H10-scFv DNA. Retroviral transduction of T cells with recombinant receptors was described in detail elsewhere (8, 11, 13, 14, 15), and receptor expression was monitored by flow cytometric analysis. Recombinant receptors were also expressed in 293T cells after transfection of the vector DNA using Polyfect transfection reagent (Qiagen) (9 μg of DNA/2 × 10⁶ cells). Cells were harvested after 48 h and lysed (1 × 10⁷ cells/ml) in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, and 0.1% Nonidet P-40, pH 7.4) supplemented with protease inhibitors (Complete; Roche).

The generation of the N variant of CEA was described elsewhere (16). 293T cells were cotransfected with the retroviral expression vector rkat (17) containing wild-type CEA or CEA-N, respectively, and the retroviral helper plasmids pHIT and pCOLT (8). After three to five rounds of transduction, cells with stable expression of wild-type CEA and CEA-N, respectively, were isolated.

CEA and CEA-N expression was determined by flow cytometry using the CEA-specific fusion proteins BW431/26-scFv-Fc and H10-scFv-Fc. Bound fusion proteins were detected by a PE-conjugated F(ab′)₂ anti-human IgG Ab (0.1 μg/ml; Southern Biotechnology Associates). Recombinant receptor grafted T cells were identified by two-color immunofluorescence using a PE-conjugated F(ab′)₂ anti-human IgG1 Ab (1 μg/ml) and a FITC-conjugated anti-CD3 mAb (UCHT-1, 1/20). Immunofluorescence was analyzed using a FACScan cytofluorometer equipped with the CellQuest research software (BD Biosciences).

T cells were grafted with recombinant anti-CEA-scFv-Fc-ζ immunoreceptors and cultivated in microtiter plates (Polysorb; Nunc) (2.5 × 10⁴ receptor-grafted T cells/well) that were precoated with purified CEA or BSM for control (each 0.625–20 μg/ml). Cell proliferation was determined using the 5-Bromo-2′-deoxyuridine Labeling and Detection Kit III (Roche) according to the manufacturer’s recommendations. In a second set of experiments, immunoreceptor-grafted T cells were cocultivated for 48 h in 96-well round-bottom plates with CEA-positive (LS174T, SW948, 293T-wtCEA, 293T-N-CEA) and negative (A375, 293T) tumor cells (each 2.5 × 10⁴ cells/well). After 48 h, culture supernatants were analyzed by ELISA for IFN-γ. Briefly, IFN-γ was bound to the solid-phase anti-human IFN-γ mAb NIB42 (1 μg/ml) and detected by the biotinylated anti-human IFN-γ mAb 4S.B3 (0.5 μg/ml). The reaction product was visualized by a peroxidase-streptavidin-conjugate (1:10,000) and ABTS. Specific cytotoxicity of receptor-grafted T cells against target cells was monitored by a 2,3-bis(2-methoxy-4-nitro-5-sulfonyl)-5[(phenyl-amino)- carbonyl]-2H-tetrazolium hydroxide (XTT)-based colorimetric assay according to Jost et al. (18).

Statistical analysis was performed using the paired Student’s t test and the SPSS analysis software 11.0.1 (SPSS). Findings were regarded as significant with two-sided p values <0.05.

We asked whether the epitope specificity of recombinant immunoreceptors that target the same molecule has an impact on the efficacy of receptor-mediated T cell activation. We therefore generated CEA-specific recombinant immunoreceptors harboring the BW431/26-scFv- and H10-scFv-binding domain, respectively, that differ in their epitope specificity, but consist of the same extracellular spacer, transmembrane and intracellular ζ- or γ-signaling domains (Fig. 1,A). To allow monitoring of the binding specificity of the immunoreceptors, we also generated expression cassettes coding for the extracellular BW431/26-scFv-Fc and H10-scFv-Fc moiety, respectively (Fig. 1,B). CEA is composed of an Ig-like domain structure comprising N-A1-B1-A2-B2-A3-B3 (Fig. 1,C). The H10-scFv recognizes the membrane-distal N domain of CEA as demonstrated by binding to 293T cells that express exclusively the N domain. The BW431/26-scFv, in contrast, reacts with the membrane-proximal A3 domain (19). Accordingly, BW431/26-scFv-Fc did not react with 293T cells that express exclusively the CEA N domain (data not shown). We moreover performed cross-competition assays using the extracellular scFv-Fc-binding domain and solubilized full-length receptor proteins, respectively. As summarized in Fig. 2, the H10-scFv-binding domain did not block binding of the BW431/26-scFv-Fc-ζ immunoreceptor to isolated, immobilized CEA, but of the H10-scFv-Fc-ζ receptor as control. Vice versa, the BW431/26-scFv domain did not block binding of the BW431/26-scFv-Fc-ζ receptor. Obviously, the BW431/26-scFv and H10-scFv domains recognize distinct, nonoverlapping epitopes on the CEA molecule.

We monitored binding of the H10-scFv and BW431/26-scFv to CEA as well as Ag-specific T cell activation via immunoreceptors that harbor these binding domains. Hereby, we analyzed two different situations: reactivity 1) against solid-phase bound Ag with randomly arranged CEA domains and 2) against membrane-bound CEA with more spatially oriented domains (Fig. 3). The latter has the consequence that the positions of the individual epitopes differ in their distances to the cell membrane. Both BW431/26-scFv-Fc and H10-scFv-Fc proteins bind specifically to solid-phase CEA, but not to BSM for control (Fig. 4, A and B). We used the binding proteins in the same titers in these assays as indicated by binding to an Ab that is directed toward the Fc domain of both proteins (Fig. 4,C). Noteworthy, H10-scFv-Fc binds more efficiently to solid- phase CEA than BW431/26-scFv-Fc. Accordingly, H10-scFv-Fc binds also to human CEA⁺ LS174T colorectal carcinoma cells with higher efficiency (Fig. 4,D) and, moreover, to the mouse tumor cells C15A3 as well that express CEA after transfection (9) compared with the BW431/26-scFv-Fc domain (Fig. 4,E). As controls, both BW431/26-scFv-Fc and H10-scFv-Fc domains do not bind to CEA⁻ Colo320 cells (Fig. 4 F). The differences of the scFv moieties in their efficiency to bind CEA are obviously not dependent on the endogenous (LS174T) or heterologous (C15A3) fashion of CEA expression, making it unlikely that the observed differences result from Ag variability or cross-reactivity to other surface Ags.

We asked whether a higher binding efficiency to target Ag results in more efficient activation of T cells. We therefore recorded the induction of proliferation and IFN-γ secretion of receptor-grafted T cells upon binding to isolated, solid-phase bound CEA. The immunoreceptors were expressed with similar efficiencies on the surface of peripheral blood T cells upon retroviral gene transfer as exemplarily shown in Fig. 5. Both BW431/26-scFv-Fc-ζ and H10-scFv-Fc-ζ receptor-grafted T cells are specifically induced to proliferate and to secrete IFN-γ upon binding to solid-phase CEA. T cells with H10-scFv-Fc-ζ receptor, however, proliferate more vigorously and secrete more IFN-γ than BW431/26-scFv-Fc-ζ receptor-grafted T cells (Fig. 6). As controls, anti-CEA receptor-grafted T cells are not activated upon binding to BSM. We conclude that the efficacy in receptor-triggered T cell activation upon binding to immobilized CEA correlates with the binding efficiency of the receptor.

Whereas activation of receptor-grafted T cells by binding to randomly arranged, solid-phase CEA correlated with the binding efficiency, it remains an open question, whether the targeted epitope in higher order of spatial orientation has an impact on receptor-mediated T cell activation. To resolve this issue, we monitored receptor-triggered T cell activation upon binding to CEA expressed on the cell surface of tumor cells. Coincubation with CEA⁺ SW948 colon carcinoma cells induced anti-CEA receptor-grafted T cells to specific cytolysis (Fig. 7,A). BW431/26-scFv-Fc-γ receptor-grafted T cells, however, lysed SW948 cells with higher efficiency than H10-scFv-Fc-γ receptor-grafted T cells. This is unexpected because the BW431/26-scFv domain of the receptor has substantially lower binding capacity to both solid-phase and membrane-bound CEA than the H10-scFv domain (cf Fig. 4). We obtained essentially the same results when T cells have been grafted with the corresponding CD3ζ chain-signaling receptors H10-scFv-Fc-ζ and BW431/26-scFv-Fc-ζ, respectively (Fig. 7,E). Cytolysis is Ag specific because CEA⁻ A375 cells were not lysed by receptor-grafted T cells (Fig. 7, B and F) and T cells without receptor or with receptors of an irrelevant specificity, i.e., anti-CD30 HRS3scFv-Fc-ζ receptor, lysed none of the target cells (data not shown). The differences in the efficacy of T cell effector functions is not restricted to the cytolytic activity but is also evident in the IFN-γ secretion (Fig. 7, C and G). Again, the same pattern in the efficiency of cytokine secretion was recorded for both ζ- and γ-chain signaling immunoreceptors. We obtained the same results using T cells from different donors and using CEA⁺ LS174T cells as target cells (data not shown). Taken together, the data imply that upon binding to CEA⁺ cells, immunoreceptors with the BW431/26-scFv-binding domain have a higher T cell activation capacity compared with those with the H10-scFv domain, although H10-scFv has a higher binding efficiency to CEA than BW431/26-scFv. This is independent of the intracellular signaling domain for T cell activation. We conclude that the observed differences in target cell binding and in receptor-triggered T cell activation are due to the interaction of the binding domain with the targeted epitope in the membrane-bound fashion and not to other domains of the recombinant immunoreceptor nor to other intrinsic properties of the target cells.

We asked whether the position of the targeted epitope within the CEA molecule has an impact on the efficacy of receptor-mediated T cell activation. H10-scFv recognizes the N domain of CEA, which is in a far distal membrane position (cf Fig. 3). We therefore generated a truncated variant of CEA that consists of the N domain directly linked via a transmembrane domain to the cell surface, thereby translocating the distal epitope of the H10-scFv into a more membrane-proximal position (Fig. 8,A). Wild-type CEA and the N variant of CEA, respectively, were expressed on the surface of 293T cells (Fig. 8,B). The H10-scFv-Fc-ζ receptor triggered more efficiently specific cytolysis and IFN-γ secretion upon binding to the proximal N epitope than to wild-type CEA with the distal located N epitope (Fig. 8,C), although wild-type CEA was expressed in higher numbers (59.7% vs 53.8%) and density (mean fluorescence intensity (MFI), 3117 vs 2136) than the CEA-N variant (Fig. 8 B). This demonstrates that the epitope position itself has a direct impact on the efficiency of receptor-mediated T cell activation.

This study explores whether the epitope and/or the position of the epitope within the targeted molecule has an impact on the efficiency in immunoreceptor-triggered T cell activation. Using CEA as a model Ag, T cells were redirected by recombinant immunoreceptors that bind different epitopes of CEA, i.e., the BW431/26-scFv-Fc-ζ receptor binds to the membrane-proximal A3 domain (19), whereas the H10-scFv-Fc-ζ receptor binds to the membrane-distal N domain. The scFv domains of the respective receptors, moreover, bind with different efficiencies to CEA, irrespectively whether purified CEA is coated onto plastic surfaces and whether CEA is expressed on the surface of tumor cells (cf Fig. 4). The H10-scFv, recognizing the distal N domain, binds to CEA with higher efficiency than the BW431/26-scFv, recognizing the proximal A3 domain. This situation is also reflected by the capacity of immunoreceptors with the respective scFv domains in activating grafted T cells upon binding to solid-phase bound CEA. In this situation, T cells expressing the H10-scFv-containing immunoreceptor are more efficiently activated to proliferate and to secrete IFN-γ than T cells with the BW431/26-scFv-containing receptor. Upon binding to purified immobilized CEA, the efficiency in T cell activation obviously correlates with the binding efficiency of the scFv domain. Upon binding to CEA expressed on the surface of target cells, in contrast, the efficiency in T cell activation does not reflect the binding efficiency of the respective scFv domain. Particularly, the BW431/26-scFv-Fc-ζ receptor that binds less efficiently to CEA activates T cells more efficiently than the corresponding H10-scFv-Fc-ζ receptor with the higher binding capacity (cf Fig. 7). This effect is not due to the signaling domain of the recombinant immunoreceptor because immunoreceptors with the CD3ζ and FcεRIγ-derived signaling domain act in the same fashion. Upon targeting CEA⁺ cells, the binding efficiency does not necessarily correlate with the efficiency in receptor-mediated T cell activation which is in contrast to binding to purified CEA. We therefore conclude that the position of the targeted epitope in the membrane-bound configuration of Ag determines the efficiency of T cell activation more than the binding efficiency of the scFv domain. This conclusion is supported by our observation that the N domain of CEA when expressed in a membrane-proximal position triggers H10-scFv-Fc-ζ receptor-mediated T cell activation more efficiently than in the membrane-distal position (cf Fig. 8).

TCR-triggered T cell activation is thought to require extensive cross-linking upon binding to Ag to form a synapse efficiently, thereby incorporating downstream signaling molecules (20). We assume that T cell activation by recombinant immunoreceptors occurs in a similar fashion. In the case of binding to isolated CEA immobilized on plastic surfaces, immunoreceptors that bind their epitope with higher efficiency drive more efficiently receptor cross-linking and T cell activation than less efficient binders. In this situation, the efficiency in T cell activation correlates with the efficiency in receptor binding. In the case of membrane-bound Ag, however, the position of the targeted epitope has a major impact on the efficiency of T cell activation. This result reflects the situation that the epitope in the membrane-proximal position is likely to have a higher capacity in receptor cross-linking and thus a higher T cell activation capacity than in a more distal position of the membrane-anchored molecule. Noteworthy, the position effect of the targeted epitope has, at least in the example of CEA targeting, a more prominent impact on T cell activation than the binding efficiency of the targeting domain. This is in accordance with a recent investigation demonstrating the impact of the position of the target epitope on the structural requirements of the recombinant immunoreceptor for optimal signaling upon Ag encounter (16).

In addition to these parameters, the accessibility of the epitope and the binding affinity of scFv domain may have direct impact on the efficiency in receptor-mediated T cell activation. In the situation analyzed here, the accessibility of the epitope seems not to be limiting because the distal H10-scFv epitope, which is thought to be more accessible than the more proximal BW431/26 epitope, is superior in binding but less capable in mediating T cell activation. In contrast, the binding affinity of the scFv domain itself may have an impact on receptor-mediated T cell activation. We recently recorded T cell activation upon binding of immunoreceptors with different affinities to the same epitope on a target Ag (7). This setting revealed that the efficiency of T cell activation positively correlates with the affinity of the scFv receptor domain when the target Ag is present in an immobilized fashion coated onto plastic surfaces. In contrast, when expressed on the cell surface of target cells, the efficiency of receptor-mediated T cell activation does not increase with the binding affinity above threshold. The conditions that define the activation threshold, however, are so far not understood but these results furthermore demonstrate the impact how the target Ag is presented to the T cell.

Compared with the physiological TCR complex, recombinant immunoreceptors exhibit substantial differences in structure. Whereas immunoreceptors with an Ab-derived binding domain are designed to bind Ags of any structure and chemical composition for which an Ab exists, the natural TCR complex recognizes its specific peptide Ag exclusively in the context of MHC. Due to only marginal differences in the overall structure of peptide-MHC-TCR complexes, the TCR exhibits more restricted Ag recognition to initiate the signaling process. Immunoreceptors with scFv-binding domains that recognize a variety of epitopes in different spatial positions on the membrane-anchored target Ag allow, in contrast, a high diversity in target recognition which in turn needs a variety of prerequisites to trigger efficient T cell activation. Compared with MHC-independent immunoreceptors analyzed here, receptors with specificity for MHC-bound peptides are assumed to be less dependent in their activation potency on structural prerequisites of the targeted epitope.

For clinical applications in adoptive immunotherapy, a recombinant immunoreceptor has to fulfill multiple requirements, including highly specific target recognition and efficient T cell activation. The Ag recognition domain of the immunoreceptor obviously defines the binding specificity, the binding affinity, and, as shown here, the activation capacity by defining the targeted epitope. Ab-derived binding domains intended to be used for immunoreceptor targeting are commonly selected for binding to solid-phase immobilized Ag and for high binding affinity. Our data, however, demonstrate that the position of the epitope in the membrane-anchored topology of the targeted molecule has substantial impact on the efficiency of receptor-triggered T cell activation, independently of the scFv binding efficiency.

We thank Dr. Michael Neumaier (Institute for Clinical Chemistry, Medical Faculty Mannheim of the University of Heidelberg, Mannheim, Germany) for C15A3 cells. We thank Heike Brand, Petra Hofmann, Birgit Hops, and Frank Steiger for excellent technical assistance.

The authors have no financial conflict of interest.